University of Southern California Dissertations and Theses

The grammaticalization of present and past in Basque

(USC Thesis Other)

The grammaticalization of present and past in Basque

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content THE GRAMMATICALIZATION OF PRESENT AND PAST
IN BASQUE
Copyright 2002
t> y
Gontzal Aldai
A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(LINGUISTICS)
December 2002
Gontzal Aldai
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UMI Number: 3093727
Copyright 2002 by
Aldai, Gontzal
All rights reserved.
®
UMI
UMI Microform 3093727
Copyright 2003 by ProQuest Information and Learning Company.
All rights reserved. This microform edition is protected against
unauthorized copying under Title 17, United States Code.
ProQuest Information and Learning Company
300 North Zeeb Road
P.O. Box 1346
Ann Arbor, Ml 48106-1346
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UNIVERSITY OF SOUTHERN CALIFORNIA
The Graduate School
University Park
LOS ANGELES, CALIFORNIA 90089A695
This dissertation , w ritten b y
& O M T 2 A , L A L b A I _______________
U nder th e direction o f A.'.L. D issertation
C om m ittee, an d approved b y a ll its m em bers,
has been p resen ted to and accep ted b y The
G raduate School, in p a rtia l fu lfillm en t o f
requirem ents fo r th e degree o f
DOCTOR OF PHILOSOPHY
: ...
/ f - Dean o f G raduate S tu d ies
D ate December 18, 2002
D1SSERTA IIO N COMMITTEE
A - Akttfc
s' D
C hairperson
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
DEDICATION
To my family
And
To Gloria Katcher
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ACKNOWLEDGEMENTS
I would like to thank my professors both in the Basque Country and in California, for
helping me pursue and enjoy my academic goals. I am most especially indebted with
two of my professors, Bernard Comrie and Jack Hawkins, who have guided me in
the long and winding road towards the discovery of one’s own linguistic world. I
have greatly benefited from their knowledge, encouragement and dedication. I would
like to include among my professors one I never got to meet, Luis Michelena, to
whom I owe my first joy with linguistics and part of my passion for it.
This dissertation has greatly profited from the work of the Association for
Linguistic Typology (ALT) circle of linguists. I offer to them my most sincere
appreciation for having maintained their linguistic believes during decades of
tempting sirens’ chants. I need to particularly acknowledge my debt with Joan
Bybee, whose work has been very inspiring to me.
To the USC linguistic community, my thanks for their contribution in the
process of completing my dissertation. My gratitude goes to Roger Woodard, Mario
Saltarelli, Elaine Andersen, Mariellen McDonald, Marsha Polinsky, Bill Rutherford,
Carmen Silva-Corvalan, Mark Seidenberg and to all the other professors in the
department.
I am very pleased to show my recognition to my partners in crime, Jon Aske,
David Zarazua, Andres Enrique, Aaron Sonnenschein, Susan Robinson, Despoina
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
iv
Theodorou, and many more, with whom I have had the most irreverent linguistic
discussions. Special thanks to Santos Aviles for his help in the revision of earlier
versions of this dissertation.
Last but not least, I need to acknowledge the Deusto Seminar circle of
Basque linguists for their enthusiasm and stimulus which was partially responsible in
my becoming a linguist.
This dissertation has been partly possible thanks to a four-year grant from the
Basque Government’s Department of Education-Universities-Research and to a one-
year dissertation fellowship from the Del Amo Foundation.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
V
TABLE OF CONTENTS
Dedication ....................................................................................................................... ii
Acknowledgments ............................................................................................. iii
List of Tables..................................................................................................................viii
Abbreviations................................................................................................................. xi
Abstract........................................................................................................................ xiii
CHAPTER 1: GENERAL BACKGROUND................................................................. 1
1. Introduction 1
2. The theory of grammaticalization 7
3. Hypotheses on grammaticalization 14
4. Methodology: testing the hypotheses 21
5. The lexical semantic dimension: classes of predicates 29
6. Tools: searching for sources of grammaticalization and establishing a
diachronic depth 32
7. Data used in this work 41
8. Summary of hypotheses 43
CHAPTER 2: THE T.A.M SYSTEM OF MODERN BASQUE...............................45
1. Introduction 45
2. Overview: the TAM system of Modem Basque 46
3. Non-finite forms 54
4. Synthetic forms 62
5. Main analytic forms (of the indicative) 68
6. Peripheral analytic forms of the indicative 74
7. Subjunctive forms 84
8. Summary of the forms studied in this work 85
CHAPTER 3: THE GRAMMATICALIZATION OF PRESENT ACROSS
LANGUAGES AND IN BASQUE................................................................... 87
1. Introduction 87
2. From present progressive to present habitual and generic 88
3. From present habitual and generic to future time-reference subordinate
contexts 95
4. Diachronic ordering of the cline starting with present progressive and
possible competition with other types of sources 99
5. Present telic sources 103
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6. Sources of grammaticalization of the Basque constructions in the
domain of non-past 107
CHAPTER 4: THE DOMAIN OF PRESENT IN HISTORICAL BASQUE 125
1. Introduction 125
2. Chronological ordering 127
3. General predictions 138
4. Competition for the youngest semantic subtypes in the cline starting
with present progressive 143
5. Competition for the domain of present habitual / gnomic / (stative
present) 146
6. Competition for the domain of future time-reference less-dynamic
subordinate contexts 150
7. Competition for the domain of future time-reference dynamic
subordinate contexts 156
8. Present Progressive vs. Periphrastic Present: testing the hypotheses 161
9. Conservation of the Synthetic Present: factors 162
10. Competition for stative present meaning: type-frequency 169
CHAPTER 5: THE GRAMMATICALIZATION OF PAST PERFECTIVE
ACROSS LANGUAGES AND IN BASQUE.............................................. 177
1. Introduction 177
2. The clines starting with present resultative 179
3. The cline starting with past resultative 187
4. Dynamic-telic sources 193
5. Narrative-discourse sources 197
6. Double-compound perfects 202
7. Sources of grammaticalization of the Basque constructions in the
domain of past non-imperfective 213
CHAPTER 6: THE DOMAIN OF PAST PERFECTIVE IN HISTORICAL
BASQUE........................................................ 222
1. Introduction 222
2. Chronological ordering 224
3. General predictions 236
4. Competition for the youngest semantic subtypes in the cline starting
with present resultative 239
5. Competition for the youngest semantic subtypes in the cline starting
with past resultative 245
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
vii
6. Competition for degrees of remoteness 247
7. (Historical) competition for perfective meaning 249
8. Evidential 253
9. Narrative anteriority 258
10. Conclusions 262
CHAPTER 7: CONCLUSIONS.................................................................................. 268
1. Introduction 268
2. General characteristics of TAM systems 269
3. Hypotheses on grammaticalization 292
4. Towards general motivations or principles 301
GLOSSARY..................................................................................................................321
BIBLIOGRAPHY.............................................. 356
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
LIST OF TABLES
Table 2.1 Main analytic forms of the indicative: examples 47
Table 2.2 Main analytic forms of the indicative: structure and meaning 48
Table 2.3 Peripheral analytic forms of the indicative: examples 50
Table 2.4 Peripheral analytic forms of the indicative: structure and meaning 51
Table 2.5 Subjunctive forms of Modem Basque: examples 52
Table 2.6 Subjunctive forms of Modem Basque: form and meaning 52
Table 2.7 Synthetic forms of Modem Basque: examples 53
Table 2.8 Morphology of the Verbal Noun 55
Table 2.9 Morphology of the Imperfective Participle 56
Table 2.10 Morphology of the Perfect Participle 57
Table 2.11 Morphology of the Future Participle 59
Table 2.12 Morphology of the Prospective Participle 60
Table 2.13 Morphology of the Verbal Radical 61
Table 2.14 Formal expression of the constructions studied in this work 86
Table 3.1 Semantic sub-types in the cline starting with present progressive 100
Table 4.1 Present habitual (Main Corpus) 147
Table 4.2 Stative gnomic (Main Corpus) 148
Table 4.3 Dynamic gnomic (Main Corpus) 148
Table 4.4 Non-stative open-conditional and concessive protases
(Main Corpus) 152
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
IX
Table 4.4.bis Non-stative open-conditional and concessive protases
(Special Corpus) 153
Table 4.5 Stative open-conditional and concessive protases (Main Corpus) 154
Table 4.6 Non-stative w/ien-future (Main Corpus) 154
Table 4.7 Purpose clauses and complement clauses to ‘want’ and ‘order’
(Main Corpus) 158
Table 4.8 Stative purpose clauses and complement clauses to ‘want’
and ‘order’ (Main Corpus) 160
Table 4.9 Token-frequency of most frequent lexical verbs (Main Corpus) 163
Table 4.10 Stative present of core-stative predicates (Main Corpus) 171
Table 4.11 Core-stative predicates in the cline of present (Main Corpus).
(Incomplete) 173
Table 4.12 Dynamic (non-stative) predicates in the cline of present
(Main Corpus). (Incomplete) 175
Table 5.1 Diachronic semantic subtypes in the main cline starting with
present resultative 182
Table 5.2 Diachronic semantic subtypes in the cline starting with past
resultative 188
Table 5.3 Dynamic and directional sources of grammaticalization
(including narrative-discourse sources) 202
Table 6.1 Youngest subtypes in the cline starting with present resultative
(Main Corpus) 241
Table 6.2 Youngest subtypes in the cline starting with past resultative
(Main Corpus) 247
Table 6.3 Pre-hodiemal perfective non-fictional non-passive (Main
Corpus) 250
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 6.4 Evidential: unwitnessed, quotation and fictional narrative
contexts (Main Corpus) 255
Table 6.5 Finite forms in the function of narrative anteriority (Main
Corpus) 260
Table 7.1 Main constructions studied in this work 296
Table G.l Prototypical features of gnomic or generic meaning 333
Table G.2 Prototypical features correlating with perfective meaning 343
Table G.3 Prototypical features of present habitual meaning 345
Table G.4 Prototypical features of present progressive meaning 347
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ABBREVIATIONS
A: Aorist, Old Basque Periphrastic Aorist
ABS: absolutive case [-0 ending]
ALL: allative case [-ra ending]
AUX: auxiliary
Cf.: compare (from Latin confer)
COM: comitative case [-rekin ending]
(often called ‘sociative’ in Basque grammars)
COMP: complementizer, subordinator [~(e)n, -(e)la endings]
DAT: dative case [-(r)i ending]
DcPr: Double-compound Present Perfect
DcPs: Double-compound Past Perfect
DET: determiner, article [-a, ak endings]
E.g.: for example (from Latin exempli gratia)
ERG: ergative case [-k ending]
GEN: genitive case [-(r)en, -ko endings]
(see Chapter 2 section 3.4)
HARAN: Haraneder 18th century database
(see Chapter 1 section 7, Chapter 4 note 3)
I.e.: that is to say (from Latin id est)
INST: instrumental case [-z ending]
ITUN: Itunberria 20th century database
(see Chapter 1 section 7, Chapter 4 note 3)
-ke: potentiality marker, old futurity marker
LEIZAR: Leizarraga 16th century database
(see Chapter 1 section 7, Chapter 4 note 3)
Lit.: literally (used mostly to refer to sources of grammaticalization)
LOC: locative case [-an ending]
(often called ‘inessive’ in Basque grammars)
Mt: Gospel according to Saint Matthew
PART: partitive case [-{r)ik ending]
PL: plural
PP: Prepositional Phrase
PPr: Periphrastic Present
PrP: Present Perfect-(Perfective)
PrR: Present Resultative
PsP. Past Perfect-Perfective
PsR: Past Resultative
RAD: Radical [-0 ending]
RS: Refranes y Sentencias (1596), 16th century Basque text
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
SG: singular
-TA: Participial ending
(see Chapter 2 section
T.A.M.: T ense-Aspect-Mood
-TZE: Verbal Noun ending
(see Chapter 2 section
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
xiii
ABSTRACT
The present dissertation examines the Tense-Aspect-Mood (TAM) system of
historical Basque, focusing on the spheres of present and past perfective. This work
is based on the framework of the theory of grammaticalization, which is part of the
more general assumption that grammars are conventionalizations of language use.
The main goals of this dissertation are the following:
1) To contribute to the understanding of the design (i.e. the relationship
between form and meaning) of the Basque TAM constructions under consideration,
from a grammaticalization and typological perspective.
2) To contribute to the understanding of cross-linguistically common sources
and clines of grammaticalization.
3) And to test against the Basque data several hypotheses arising from the
clines of grammaticalization.
Regarding the first goal, this work shows how the theory of
grammaticalization can account for the Basque data, based on the comparison with
the evolutions of cross-linguistic parallels. Some of the issues in the Basque TAM
system which are nicely explained from this perspective are the present habitual
meaning of the Periphrastic Present (a construction grammaticalized as a locative
present progressive), the perfective meaning of the Past Perfect-Perfective (a
construction grammaticalized as pluperfect), the existence in Old Basque of the
Periphrastic Aorist (a construction grammaticalized with dynamic auxiliaries which
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
xiv
are only used in non-indicative contexts in Modem Basque), and the evidential use in
Old Basque of the Present Perfect-Perfective (a construction grammaticalized as
present perfect).
Regarding the latter goals, some of the contributions of the Basque data come
from the conservation of the high token-frequency forms of the Synthetic Present,
and from the clear existence in Basque of pairs of layers with the same source of
grammaticalization. An interesting areal phenomenon is provided by the Basque
Double-compound Perfects, a type of construction reported in Occitan, French and
German.
The last chapter of this dissertation has a more ambitious but more tentative
goal. It is devoted to discussing general issues on the design of TAM systems, such
as (non)discreteness, competing motivations, frequency, conventionalization, etc.
This part is intended to offer insights for further research.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1
CHAPTER 1
GENERAL BACKGROUND
1. INTRODUCTION.
1.1. General introduction.
The present dissertation aims to account for part of the Tense-Aspect-Mood (TAM)
system of Basque focusing on the relationship between form and meaning. I will
refer to this relationship as the ‘structure’ or ‘design’ of grammatical material. The
general framework adopted in this work departs from, and is a challenge to, current
mainstream linguistics in several respects. The most important of these is probably
the assumption that grammars (competence) are created in language use
(performance), by means of a complex process of conventionalization and
grammaticalization. (It is impossible at our present knowledge to tell whether
everything in grammars comes from a conventionalization of performance. We can
at least assume that much of grammars does.)
In the case of TAM systems, the assumption that grammars are created in
language use amounts to saying that TAM systems are characteristically subject to a
specific and cross-linguistically comparable process of grammar-creation, which
involves, broadly speaking, formal reduction and semantic generalization of
erstwhile lexical material. I will refer to this specific process of grammar-creation as
‘grammaticalization of lexical items’ (alternatively, ‘lexical grammaticalization’). (If
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2
not otherwise indicated, the term ‘grammaticalization’ will be used in this work as
grammaticalization of lexical items. See Hawkins 1994:19ff, Comrie 1998 for
broader uses of the term.) The process of lexical grammaticalization affects typically,
besides TAM systems, parts of grammars such as case marking, verbal agreement,
function words, etc. These are usually related with inflectional morphology and
morphologization. (It is not necessarily the case that all languages recur to
grammaticalization to the same degree. Yet, we can confidently state that the process
of grammaticalization is very common across languages.)
I also assume that the design of TAM systems, as that of other parts of
grammars that typically undergo lexical grammaticalization, can constitute an object
of study in itself. This study is related to, but also logically independent from, the
study of TAM systems in a synchronic state in the speaker’s mind. There seem to be
several reasons which recommend that TAM design and TAM mental representation
be studied apart. I cannot get into details on this complex issue here. It may suffice to
say now that the synchronic processing of TAM systems (as that of other
morphological categories and markers) seems to be quite indirectly related to the
historical process of grammaticalization that designed them. This is not the case,
generally speaking, of syntactic design and syntactic processing, which are often
closely mirrored in one another (Hawkins 1994, 1999, 2001; Phillips 1996).
Moreover, many cross-linguistic generalizations regarding the design of TAM
systems, etc. are best captured from a grammaticalization-historical perspective, as
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3
we will see throughout this work. These are thus reasons to maintain that a complete
account of these parts of grammar ought to come from a perspective that embraces
both the synchronic and the diachronic dimension (cf. Bybee 1988b).
The general framework of the theory of grammaticalization I adopt is based
on works such as Heine, Claudi & Hiinnemeyer (1991); Hopper & Traugott (1993);
Bybee, Perkins & Pagliuca (1994); and Lehmann (1995). This framework appeals to
cross-linguistic comparison to try to uncover the formal sources for the creation of
grammar and the subsequent diachronic changes that lead to the design of TAM
systems and other morphological categories in the way we find them in languages.
1.2. This work.
There are several specific goals in the present work, all of them concerning the TAM
domains of present and past in Basque.
1) To provide a synchronic description of most of the TAM system of Modem
Basque. (See Chapter 2.)
2) To establish the history of Present and Past constructions in the Basque TAM
system during the last five centuries.
3) To propose the source of grammaticalization and a chronological ordering for
the Basque constructions at issue.
4) To use the Basque case to illustrate and test some of the hypotheses on
grammaticalization that have been proposed in the literature. I will test these
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4
hypotheses against Basque textual data of different ages, from the 16th to the
20th century.
5) To contribute to the understanding of sources and clines of
grammaticalization of TAM markers.
More ambitious but also more tentative objectives will be:
6) To extract from the data general conclusions concerning synchronic analyses
of TAM systems, thus partly contributing to the understanding of TAM
categories and to the philosophy of language.
7) To propose non-autonomous general principles and motivations (cf. Clark &
Malt 1984:211), either synchronic or diachronic, which would be responsible
for the changes that will be examined in this work. These general motivations
would ultimately constitute the last explanation behind the structural
generalizations of TAM systems. (See Chapter 7.)
In this work, therefore, I deal with synchronic and diachronic issues. Also, both form
and meaning are examined, since the two are interrelated within the theory of
grammaticalization. In the synchronic study, I focus on the semantic side rather than
on the formal side.
The semantic study is prominently based on the concept of ‘TAM meanings’
or ‘semantic subtypes’, i.e. specific meanings (rather than super-categories or formal
contrasts), such as present habitual, (past) perfective, present perfect, pluperfect,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5
present progressive, etc, or even finer divisions such as perfect of result, perfect of
recent past, experiential perfect, etc. (See the Glossary for clarification.) Another
relevant concept for the semantic study is that of ‘competition’, i.e. the situation of
partly overlapping forms competing for similar semantic subtypes. Thus, the most
important kind of data collected in this work will record the different constructions
of historical Basque that appear in competition for expressing one specific ‘TAM
meaning.’ (Hence, the collection and analysis of the data will be correctly carried out
only if the pertinent factors and the appropriate semantic divisions have been
previously taken into account.) The semantic divisions and terminology labels used
here will be taken from the typological-functional literature on TAM: most
importantly from Comrie (1976,1985); Dahl (1985); Bybee & Dahl (1989); and
Bybee, Perkins & Pagliuca (1994).
The formal study, on the other hand, will not be examined at length. Formal
synchronic analyses of the compositionality of specific TAM constructions of
Basque (e.g. whether they are monoclausal or biclausal) will not be pursued in depth.
As will be discussed in Chapter 7, dissecting a detailed compositional analysis of the
constructions at issue does not appear mandatory for a semantic account of TAM
systems to be correctly achieved.
The diachronic study, in turn, will be central in this work. I will focus on
attempting to account for the historical processes, both formal and semantic changes,
which are responsible for the design of TAM and constitute the main object of study
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6
of the theory of grammaticalization. This diachronic part of the work is also
primarily concerned with the semantic side. Concretely, 1 will examine semantic
clines or paths of grammaticalization that the TAM constructions follow in a
strikingly similar way across languages. These paths are based on semantic or
pragmatic chains o f ‘associations’, i.e. metaphors, contextual implicatures, etc.
Explaining these paths, as I will try to do in this work, amounts to accounting for
how the relations between form and meaning in TAM systems come into being.
Some of the clines of grammaticalization put forward in the literature will be tested
against the Basque data on semantic change. On the other hand, formal data about
the relative ‘degree of rigidity’ and ‘boundness’ of the Basque TAM constructions,
i.e. about their formal age, will also be collected and tested, to the extent the
attestation makes it possible. (See the Glossary for terminological clarification.)
1.3. The present chapter.
The presentation of this preliminary chapter is as follows. In section 2 ,1 offer a
general introduction on the main lines of the theory of grammaticalization. In section
3 ,1 put forward in more detail the hypotheses on grammaticalization of TAM
systems that will be illustrated and tested in this work. Section 4 provides a still more
detailed account o f the methodology followed to test the hypotheses on
grammaticalization, and to recognize different scenarios in the competition of TAM
forms. Section 5 is devoted to studying the issue of the interaction and restrictions of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
7
TAM markers with verbal classes, i.e. what I will be calling the ‘lexical-semantic
dimension.’ In section 6 ,1 focus on the tools we have in order to put into practice a
diachronic project like the present one. In section 7 ,1 describe the data used in this
work. Section 8, finally, groups together all the hypotheses on grammaticalization
put forward in the preceding sections.
2. THE THEORY OF GRAMMATICALIZATION.
The present section is devoted to providing a general introduction on the theory of
grammaticalization, with a focus on the grammaticalization of TAM. This section is
based on the literature on the topic, although I will also present at this point the
particular issues on grammaticalization of TAM systems that will be given most
importance in the present work. In the following sections, I will develop in more
detail the specific hypotheses on grammaticalization that will be illustrated and
tested throughout this work.
2.1. Grammaticalization of lexical items.
The theory of grammaticalization “begins with the observation that grammatical
morphemes develop gradually out o f lexical morphemes or combinations of lexical
morphemes with lexical or grammatical morphemes” (Bybee et al. 1994:4). (For
similar definitions, see Meillet 1912:132, and Kurylowicz 1965:52, cited in
Campbell 1998:238.) Furthermore, this process of evolution is strikingly similar
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
8
across unrelated languages, depending importantly on the lexical source of
grammaticalization. We may distinguish two principal sides to this development:
formal and semantic. In all, I will propose the following four variables in the process
of grammaticalization of lexical items: 1) formal reduction, 2) changes in meaning,
3) generality of use, and 4) the process of ‘renewal.’ (See the Glossary for the
terminology used in the following sections.)
2.2. Formal reduction: hypothesis of Formal Unidirectionality.
In the process of grammaticalization, the major formal development is from
independent constituents towards mutual dependency, rigidification and boundness
(fusion), and then towards phonetic reduction. An illustrative example could be the
evolution of the Romance synthetic Futures from Latin. The formal development of
these constructions suffered an attested reduction in the line just pointed out. Thus,
apparently the Classical Latin loose periphrasis [[cantare] habeo] underwent a
process of reanalysis and rigidification that yielded the Late Latin rigid periphrasis
[cantare habeo]. Then, further processes of “fusion across morpheme boundaries and
phonological attrition” took place which rendered, for instance, the French Future
[chanterai] (Hopper & Traugott 1993:44).
Formal changes are overwhelmingly unidirectional across languages: i.e. the
change towards formal reduction is considerably the most common. This observation
can be called the hypothesis of Formal Unidirectionality. Therefore, the age of a
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
9
construction often correlates with its degree of boundness. (See however Bybee et al.
1994:118 for possible “typological constraints on grammaticalization”, particularly
in isolating languages.)
2.3. Changes in meaning: hypothesis of Semantic Unidirectionality.
The major semantic developments evolve from lexical and specific meaning towards
abstraction and generalization. This is especially true of the first stages of the process
of grammaticalization. Generalization of meaning is also called “semantic
reduction”, “bleaching”, “abstraction”, “weakening of semantic content.” (See Heine
et al. 1991:21; Bybee et al. 1994:6.)' Two illustrations of the process of semantic
abstraction may be the evolution from progressive to habitual, and that from
resultative to perfective, both of which will be studied in this work (in Chapters 3
and 5, respectively). Thus, progressive and resultative can be considered more
lexical and specific meanings than habitual and past perfective, which appear as
more grammatical and abstract.
The issue of abstraction and generalization in semantic changes is,
nevertheless, a difficult one (see note 1). In the present work, the concern will not be
so much with classifying semantic changes as with trying to explain them. From this
1 There is some disagreement in the literature on whether all these terms are in fact interchangeable.
According to Heine et al. (1991:41), “bleaching is inadequate as a descriptive or explanatory
parameter of grammaticalization.” Furthermore, there are also different uses for a term such as
“abstraction” (1991:43). (See Heine et al. 1991:108-113, Hopper & Traugott 1993:96-100 for
discussion on these issues.)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
10
perspective, it may not be totally indispensable for this work to label specific
changes as “bleaching”, “abstraction” or “generalization.”
On the other hand, it will be important to try to explicate what
psycholinguistic associations are responsible for linking one given meaning with
another. This is so because understanding what kind of association occurs behind
each of the semantic changes entails accounting for the direction of the change. As a
matter of fact, empirical evidence shows that semantic changes are mainly
unidirectional (hypothesis of Semantic Unidirectionality). Therefore, semantic
unidirectionality is a byproduct of the general cross-linguistic tendency to create very
similar associations (i.e. metaphors, implicatures, etc.) departing from similar
sources of grammaticalization. These psycholinguistic processes give rise to ‘dines’
or ‘paths’ of semantic evolution that recur across languages.
As we have seen above, many, probably most, changes in meaning (either
semantic or pragmatic) progress towards abstraction and generalization, including
time-reference generalization (i.e. more inclusion). Yet, there seem to be other
changes whose outcome is not a generalization in time-reference (i.e. in the level of
inclusion). Tentatively, this might be the case of the change from habitual and
gnomic subordinate clauses to (definite) future time-reference subordinate clauses
(e.g. when the game ends, we always go to the bar -A when the game ends tomorrow,
we will go to the bar: see Chapter 3). In any event, semantic unidirectionality still
holds in these latter cases too; i.e. the direction of the change is habitual future,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
11
and seldom the opposite. Hence, the main point to capture regarding semantic
changes seems to be the (uni)directionality of the paths of grammaticalization, rather
than whether they evolve towards abstraction, generalization, inclusion, etc, or they
do not.
2.4. Generality of use.
By the term ‘generality of use’ I will refer to a variable in the process of
grammaticalization which indicates the degree of lexical applicability that a given
construction has. That is, in the case of TAM systems, ‘generality of use’ amounts to
how many lexical verbs the construction is applicable to. Thus, we could also dub
this concept as “lexical generality” (Bybee 1985:16). Generality of use, therefore,
applies in the ‘lexical dimension’ (see the Glossary for clarification).
2.4.1. Generality o f use in the process o f abstraction and generalization.
In this work, I will mainly use the term ‘generality of use’ while referring to the
process of abstraction and generalization, which is fundamental in the process of
grammaticalization, especially in its early and central stages. Thus, within the early
stages of the process of grammaticalization of TAM constructions, the competition
for ‘generality of use’ tends to evolve towards the generalization of one construction
for all lexical items. This generalization, although not easy to observe in practice,
occurs usually as a semantic extension to new lexical-semantic classes of predicates.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
12
Hence, I will often talk about the ‘lexical-semantic dimension’ to refer to this
variable in the process of grammaticalization. (See the Glossary for clarification.)
As Bybee (1985:17) puts it, “in order for a morphological process to be
general, it must have only minimal semantic content. If a semantic element has high
content, i.e. is very specific, it simply will not be applicable to a large number of
stems.” Therefore, the more abstract a TAM construction is, the wider its domain of
applicability (Bybee et al. 1994:19). In other words, the first steps in the process of
grammaticalization, which unequivocally progress towards semantic generalization
and abstraction (i.e. towards less lexical and more grammatical meaning), lead also
to an increase in generality of use. This is the case of the above-mentioned changes
from progressive to habitual and from resultative to perfective. While progressive
and resultative have high lexical-semantic restrictions, habitual and perfective are
practically applicable to all classes of predicates. (For more discussion on the lexical-
semantic dimension, see section 5 below.)
Summarizing the last sections, in the first and central stages of the
grammaticalization process (which can be considered ‘grammaticalization properly’:
see the Glossary for clarification), we find: 1) a formal evolution towards reduction;
2) a semantic evolution towards generalization or abstraction; and 3) an evolution
regarding generality of use that progresses towards high lexical applicability.
Concerning the latter, in the best case, a morphological marker or construction
undergoing grammaticalization will achieve total applicability or total productivity:
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
13
i.e. a ‘winner-take-all’ case. These three evolutions can be considered as constituting
a ‘process of generalization’ which is most characteristic of grammaticalization (see
Chapter 7 section 4.2, and the Glossary for clarification).
2.4.2. Generality o f use in the process o f decline and conservation.
Regarding the very late stages of the evolution of grammatical material, unlike in the
‘process of generalization’, if decline of forms occurs, formal reduction and
boundness are not paralleled by generality of use. In other words, declining forms,
although being usually highly hound, evolve towards loss of productivity; i.e.
towards decreasing in generality of use. It could be discussed, nonetheless, whether
the process of decline of forms should be included within the concept of
grammaticalization. It may seem it should not. In any event, in the present work I
will deal with conservation of declining forms separately from issues pertaining to
‘grammaticalization properly.’ In this case, the variable of ‘generality of use’ does
not necessarily correlate with lexical-semantic classes of predicates. One could thus
talk about the ‘lexical dimension’ in this case, but not about the ‘lexical-semantic
dimension.’ (See the Glossary for clarification. For discussion, see note 3, and
sections 3.5 and 4.3 below.)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
14
2.5. The process of ‘renewal.’
The fourth variable I will take into account in order to describe the whole picture of
the evolution of grammar is the competition among forms and the process of
“renewal” (cf. Hopper & Traugott 1993:121-123). Thus, new forms undergoing the
first stages of the process of grammaticalization will compete with older forms
within their semantic sphere.
The renewal of forms gives rise to “layers” (cf. Bybee et al. 1994:21-22;
Hopper & Traugott 1993:123-126) of constructions with different ages or degrees of
grammaticalization. The process of renewal occurs frequently along analogous paths
of evolution, so that the old and young layers share and compete for overlapping
semantic territories. However, there can be layers with similar and with different
sources of grammaticalization. (The process of renewal and layering is studied in
depth in sections 3.3, 4.1 and 4.2 below.)
3. HYPOTHESES ON GRAMMATICALIZATION.
This section is devoted to presenting in more detail the hypotheses that this work will
test. These are the hypothesis of Source Determination (section 3.2), the hypothesis
of Layering-Displacement (section 3.3), the hypothesis of Form-Meaning
Covariation (section 3.4), and the hypothesis of Conservation of high token-
frequency forms (section 3.5). In section 4 ,1 will further propose the methodology
for testing these hypotheses against the Basque data at hand.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
15
3.1. Predictions vs. explanations a posteriori.
A question that arises when dealing with the hypotheses on grammaticalization is
whether the grammaticalization framework based on cross-linguistic findings is
and/or should be able to predict the behavior of the TAM system of a specific
language such as Basque. The approach adopted in this work maintains that strict
prediction in advance of the behavior of TAM forms is impossible. There are
different reasons behind this position, some of which I introduce below (see section
4 of Chapter 7 for more discussion on this issue).
We have seen that the potential changes that a given TAM form may undergo
are mostly due to semantic associations triggered out of the particular content of the
form by itself (i.e. its “semantic substance”, cf. Bybee 1988a). These semantic
associations, in turn, are quite unidirectional, and may seem partly predictable.
However, the actual change depends also on factors such as the competition with
other constructions in the language. Areal features and borrowing are also variables
to take into account, which can be subject to random variation. Thus, there is an
important component of chance in the potential changes of TAM forms. Moreover,
as we will see throughout this work, there are clines that can bifurcate, so that there
is sometimes more than one possible evolution for a given source of
grammaticalization.
On the other hand, not everything in TAM evolution is random. Once the
grammaticalization clines of the types we will see in this work become established
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
16
and understood, we have a very solid criterion to rank putative evolutions from very
possible to extremely implausible. This criterion is mostly based on the Source
Determination hypothesis, as we will see below. Furthermore, what the
grammaticalization framework can at least confidently provide is a compelling
explanation ‘a posteriori’ of the relations between form and meaning; i.e. explaining
why one form conveys a specific meaning based on how that meaning was acquired
by the form, once the actual changes have occurred.
The case where we can most confidently propose some ‘predictions’ (always
in a loose sense) is that where we can establish layers of constructions with the same
source of grammaticalization. (See sections 3.3 and 4.1 below.) Even in this case, the
predictions proposed will not be absolute and in advance, but will rather relate the
evolution of a given construction with the possible evolution of a competing one.
3.2. Hypothesis of Source Determination.
A first premise I will assume within the theory of grammaticalization is the Source
Determination hypothesis. I present the hypothesis of Source Determination in (1)
below.
(1) Hypothesis of Source Determination:
The source construction that enters into grammaticalization conditions very
importantly the path of grammaticalization that the form will travel in its
semantic development.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
17
To be more precise, “it is the entire construction, and not simply the lexical meaning
of the stem, which is the precursor, and hence the source, of the grammatical
meaning.” Thus, “we must attend to the syntax and morphology of the source
construction” (Bybee et al. 1994:11), and not only to, for instance, the kinds of
auxiliaries (e.g. be, have, go, come, etc) that appear in it.
As already stated, the hypothesis in (1) does not entail we can strictly predict
in advance the change that any given construction will undergo. What we can
assume rather confidently is that if the change occurs at all, it will most probably
follow the path or paths that the source of grammaticalization determines. This
statement does not imply either that the construction at hand has to evolve
necessarily till the last step in its cline of grammaticalization. In any event, one of the
strengths of the hypothesis of Source Determination is its ability to exclude
evolutions from the universe of possible changes that a given form may undergo.
Unfortunately, at the present moment, we do not have the same confidence
about all sources of grammaticalization of TAM meanings. In other words, while
some sources are widely attested and very well understood, others still remain to be
fully comprehended. For instance, the combination of stative auxiliaries and
nonfinite forms in the locative case, or in general all constructions that parallel
locative expressions, are very well known as sources of grammaticalization of
progressive meaning. Also, the combination of stative auxiliaries and a past passive
participle is well understood as being the source of resultative meaning. These two
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
18
types of sources are among those where we confidently know the path that the
constructions at issue may follow.
Another problem with the hypothesis of Source Determination is a practical
one. Namely, it is often difficult to determine, i.e. to reconstruct, the source of
grammaticalization of specific constructions. Obviously, when this occurs, no
predictions at all can be offered. This situation is especially common of old, highly
grammaticalized, formally reduced constructions. This in turn leads to a further
problem. While the youngest constructions in a given language are among those
whose source of grammaticalization is more likely to be confidently determined, it is
frequently old constructions that have experienced the best-attested semantic
changes. In other words, a common picture is unfortunately that where the
constructions with the best attested semantic changes have unknown sources,
whereas the constructions with best-known sources have a very short history of
documented semantic changes.
3.3. Hypothesis of Layering-Displacement.
In this section, I present the hypothesis of Layering-Displacement for the ideal case
of competition among forms with very similar (i.e. theoretically, ‘the same’) source
of grammaticalization but at different degrees of grammaticalization. I will use the
term ‘same-source layering’ to refer to this specific situation. By ‘the same source of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
19
grammaticalization’, I mean to denote the case where we can confidently assure that
both sources are initiators of the same cline.
When two constructions make up a case of same-source layering, the
hypothesis of Source Determination in (1) predicts that the two forms will evolve
through the same cline of grammaticalization; i.e. they will travel along the same
path, but will be at different points at any given stage. “New layers are continually
emerging; in the process, the older layers are not necessarily discarded but may
remain to coexist with and interact with new layers” (Hopper & Traugott 1993:124).
Thus, the most common synchronic situation of two same-source layers is that where
they partly overlap. From a diachronic perspective, however, apart from a situation
of inertia, a common picture of change is a displacement of older forms by younger
forms. I will therefore propose the following hypothesis of Layering-Displacement,
stated for the case of same-source layers.
(2) Hypothesis of Layering-Displacement (same-source layers):
When the renewal of TAM systems occurs in the way of layers of
constructions with the same source but different degrees of
grammaticalization, if the younger layer ‘catches on’ and starts to attain some
generalization (i.e. in a situation of no-inertia), then the tendency is for the
younger layer to compete with the older layer and to gradually displace it
from specific semantic subtypes, while both evolve along the same cline.
The hypothesis in (2) allows for the description of the dual character of the
competition of layers. From a synchronic perspective, the young and the old layers
partly overlap while they compete for specific TAM semantic subtypes. From a
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
20
diachronic perspective, however, the scenario is one of displacement of the old layer
by the young layer. (I detail in section 4.1.1 below the methodology for testing the
hypothesis of Layering-Displacement against the Basque data.)
3.4. Hypothesis of Form-Meaning Covariation.
In this section, I put forward the hypothesis of Form-Meaning Covariation, which
encompasses the hypotheses of Formal Unidirectionality and Semantic
Unidirectionality. (As already touched upon above, the hypothesis of Form-Meaning
Covariation does not take into account whether semantic changes progress always
towards generalization and abstraction or not. This hypothesis is only based on the
fact that semantic changes are unidirectional, to a very important extent.) The
hypothesis of Form-Meaning Covariation (Bybee et al. 1994:279) is stated in (3)
below.
(3) Hypothesis of Form-Meaning Covariation:
There is a highly significant correlation between semantic age and formal
age.
The hypothesis of Form-Meaning Covariation states that there is an important
correlation between how far a given form has progressed along its corresponding
semantic cline (i.e. its semantic age) and the form’s degree of boundness or formal
grammaticalization (i.e. its formal age). (In section 4.1.2,1 will present the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
21
methodology for testing the hypothesis of Form-Meaning Covariation against the
Basque data, for the case of same-source layers.)
3.5. Hypothesis of Conservation of high token-frequency forms.
The last prediction I propose takes frequency measures into account. This is the
hypothesis of Conservation of high token-frequency forms, which is stated in (4)
below.
(4) Hypothesis of Conservation of high token-frequency forms:
The conservation of specific items pertaining to an old unproductive
declining construction correlates with their token-frequency. In other words,
high token-frequency items tend to persist after the loss of their lower-
frequency counterparts.
According to the hypothesis in (4), thus, the conservation of old declining
constructions depends, not on the type-frequency or applicability of the construction,
but on the absolute frequency of specific lexical verbs of that construction. That is,
conservativeness correlates with the total of times that a specific verbal form is used.
4. METHODOLOGY: TESTING THE HYPOTHESES.
In this section, I present the methodology that will be followed in this work in order
to test the hypotheses on grammaticalization put forward in the previous section.
This methodology will lead to finding out that the TAM constructions of Basque
constitute different types of scenarios which have to be approached in a partly
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
22
differentiated manner. I will now introduce the first and general methodological
steps in this work. Then, 1 deal below with the different scenarios of competition that
the Basque data may represent.
1) In a first methodological step, I will put forward (based on the literature)
sources and clines of grammaticalization in the domains object o f study which have
been proved to be cross-linguistically common. The semantic paths found in the
literature will be shown to progress in a unidirectional fashion. I will also try to
account for the semantic unidirectionality in each of the clines, based on the nature
of the associations that trigger them.
2) Second, I will try to independently ascertain the sources of
grammaticalization of each of the Basque TAM constructions at issue, and compare
them with the cross-linguistically well-established sources of grammaticalization.
(See section 6.1 below for more details.)
3) Then, I will appeal to the hypothesis of Source Determination (see (1)
above), which will determine the cline of grammaticalization that the construction at
issue will follow.
4.1. Competition among same-source layers.
If we are able to find in the Basque case more than one construction with ‘the same’
source of grammaticalization but with different degrees of grammaticalization, we
can conclude that there exist same-source layers of forms in competition, which will
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
23
most often travel along the same path, but at different stages of development. As a
matter of fact, the Basque constructions that are object of study in this work include
various cases where the existence of pairs of same-source layers is readily
recognizable and safely established.
We need to provide at this point an independent relative chronology for the
groups (pairs) of constructions that constitute same-source layers. I will base this
chronology mostly on formal grounds. For that purpose, I will assume the hypothesis
of Formal Unidirectionality and therefore that formal rigidity and boundness is an
indicator of the relative age of a layer. (See section 6.2 below for other kinds of
evidence and more details.) Actually, given that the source of each of these pairs of
forms is very similar and there is evidence indicating their different degree of
grammaticalization, it is easy to come up with a relative chronological order of these
layers.
As we saw above, this scenario is the ideal one for testing the hypotheses of
Layering-Displacement and Form-Meaning Covariation.
4.1.1. Testing the hypothesis o f Layering-Displacement.
I formulate with more detail in (5) below the hypothesis of Layering-Displacement
formulated in (2) above, indicating the predictions and the conditions for testing the
diachronic tendency towards displacement in the competition between same-source
layers.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
24
(5) Testing the hypothesis of Layering-Displacement:
Given two TAM constructions, A and B, both appearing in two (sufficiently
distant) synchronic stages of a language, X and Y (X prior to Y), so that A
and B share and compete for similar TAM meanings in both stage X and
stage Y. If we have solid evidence that A and B have had the same source of
grammaticalization, and if we have solid evidence to order chronologically A
as immediately prior to B in their path of grammaticalization, then if there
occurs an expansion in the meaning of B from stage X to stage Y (to the right
in the corresponding cline of semantic grammaticalization), it tends to be
paralleled by an analogous narrowing of A from the same contexts in the
passage from stage X to stage Y.
As stated before, these conditions are securely established in the case of Basque.
Concretely the Present Progressive and the Periphrastic Present, on the one hand (see
Chapter 4), and the Resultatives and the Perfect-Perfectives, on the other (see
Chapter 6), make up clear cases of layering where we can test the hypothesis of
Layering-Displacement. The hypothesis in (5) has to be tested against strictly
diachronic data (see the Glossary for clarification). Unfortunately, the main problem
with the Basque case is that the young layers have very short attestation of semantic
changes that have taken place. In any event, as we will see in the corresponding
chapters, the prediction in (5) is not falsified by the Basque data.
4.1.2. Testing the hypothesis o f Form-Meaning Covariation.
I formulate below in more detail the hypothesis of Form-Meaning Covariation that
was given in (3) above, indicating the ideal conditions for its testing; i.e. the
existence of a competition between same-source layers.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
25
(6) Testing the hypothesis of Form-Meaning Covariation:
Given two TAM constructions, A and B, in a synchronic stage of a language
X, so that A and B are in competition for similar TAM meanings in stage X.
If we have solid evidence that A and B have had the same source of
grammaticalization, then: If A is formally more bound than B (i.e. if A is
formally older than B), then its meaning in stage X should have progressed to
the right in the corresponding cline of semantic grammaticalization no less
than the meaning of B.
The statement in (6), thus, assumes that when the existence of same-source layers in
a language is established with high confidence, we can consider that by comparing
two layering constructions we are actually witnessing the evolution along one given
cline. Therefore, the hypothesis in (6), unlike that in (5), can be tested against
‘synchronic data of change’ (see the Glossary for clarification).
4.2. Competition among different-source constructions.
Unfortunately, not all the Basque constructions we need to study in the present work
can be proved to constitute layers with the same source of grammaticalization. When
this occurs, the evolution of the competition among the forms involved is often
difficult to establish, and even more difficult to predict. Moreover, in the competition
among different-origin forms, it is often more difficult to come up with a relative
chronological ordering of the forms at issue.
For these reasons, I will not propose in this work any predictions regarding
the competition of constructions with different sources (or with no well-established
sources). In these cases, I will just present the data on the corresponding evolution.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
26
Nevertheless, I will try to propose the source of grammaticalization and the relative
age of all of the constructions examined, even if no predictions will be offered. It
might be the case that in a situation of different-source layering, when we can
provide a chronological order for the layers in competition, the scenario we get is
approximately similar to the one we postulated for the ideal case of same-source
layering. If this were the case, then we could also talk about ‘displacement’ and
‘form-meaning covariation’ for the scenario of different-source layering. However,
this will not be overtly defended in the present work.
Instances of different-source layering in Basque are provided by the cases of
competition between the Periphrastic Present and the (Periphrastic) Present
Subjunctive (Chapter 4), on the one hand, and the Past Perfect-Perfective and the
Periphrastic Aorist, on the other (Chapter 6).
4.3. Conservation of old declining forms.
We have already mentioned that the conservativeness of declining forms can be
understood as a different scenario from that of the constructions undergoing
processes of generalization (or ‘grammaticalization properly’: see section 2.4 above).
This is so because the conservation of old declining forms seems to be mainly
conditioned by the frequency of appearance of specific items (i.e. their token-
frequency: see section 3.5 above). The Basque synthetic forms appear to constitute a
clear case of conservation of otherwise declining and unproductive forms. (In this
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
27
case, however, it seems that the lexical-semantic dimension may play some role.) In
any case, in section 9.1 of Chapter 4 ,1 will test the hypothesis of Conservation of
high token-frequency forms for the case of the synthetic verbs of Basque.
However, before we try to explain the low degree of generality of use of the
Basque synthetic forms in terms of conservativeness, we need to ensure that they
indeed constitute a case of declining old constructions. I put forward in section 6.2
below the kind of evidence we have to maintain that the synthetic forms of Basque
present the characteristics to be considered very old constructions undergoing an
advanced process of decline. We will see there that there is evidence to think that in
earlier stages of Basque the synthetic forms constituted a productive rule, which
became unproductive for reasons that perhaps we may only speculate about. (Since
they do not constitute a productive rule any longer, we may hypothesize that they are
partly stored based on memory processing. Thus, the factors that condition the
appearance of these forms are very different from those conditioning the use of
productive constructions; hence, their being studied apart.)
4.4. Competition for generality of use: the expression of stative predicates.
A last differentiated scenario I will propose is that pertaining to the competition for
generality of use. Since in this work I use the term ‘generality of use’ while referring
to the process of generalization (i.e. ‘generalization in use’), the conservation of
declining forms will be excluded from it. In the process of generalization, generality
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
28
of use seems to progress along lexical-semantic classes of predicates. I devote the
next section to introducing the lexical-semantic dimension. (See also the Glossary
for clarification on the relationship between these terms.)
The competition for generality of use is more difficult to observe in the
database than other scenarios of competition. This is so because the competition for
generality of use often occurs among very young and similar constructions (e.g.
progressives, resultatives), which are not well attested in written texts. (See section 4
of Chapter 4, and sections 4 and 5 of Chapter 6 for the competition among the
youngest constructions examined in this work.)
An interesting issue we find within the Basque data is that referring to the
competition for stative meaning. Many of the conservative cases of Synthetic Present
forms presented in the previous section correspond to stative verbs (see section 5.3
below). In addition, there are special ‘verbal locutions’ for many other stative
predicates (see section 5.1 below). For the case of present tense, I will talk about
Stative Present Verbal Locutions. These lexical rather than grammatical formations
are studied in section 6.5 of Chapter 3. The Stative Present Verbal Locutions are
formations with low degree of generality of use, and compete with more generalized
constructions for the expression of stative meaning. This competition (including the
role of type-frequency) is examined in section 10 of Chapter 4.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
29
5. THE LEXICAL-SEMANTIC DIMENSION: CLASSES OF PREDICATES.
One of the most widely accepted classifications of verbal predicates is that of
Vendler (1967), who considers four main classes: states, activities, accomplishments
and achievements. Among these four classes of predicates, it is probably the class of
states that is the most clearly defined and differentiated from the others. We could
say that states do not fit very well with the prototype of verb, which is formed around
dynamic actions. Thus, it is common to divide verbal predicates into stative and non-
stative or dynamic (Comrie 1976:48-49). In this work, I will use both classifications
just mentioned. I will sometimes refer to Vendler’s four classes, while other times I
will consider a division between stative and dynamic predicates, whenever the class
of states alone seems to behave differently from the others.
There are three main issues where the lexical-semantic dimension may be
relevant for this work: 1) lexical locutions, 2) incompatibilities and restrictions with
main constructions, and perhaps 3) conservativeness of old forms.
5.1. Lexical locutions.
A first topic where lexical-semantic classes seem significant for the present work
regards the so-called “verbal locutions” of Basque (cf. Lafitte 1962:344-350).
Basque has an important number of constructions consisting of a nominal item plus a
light verb (e.g. nahi *edun ‘to want’ lit. ‘have desire’, bizi izan ‘to live’ lit. ‘be
alive’, lan egin ‘to work’ lit. ‘do work’), appearing in cases where the neighboring
2 For the use of the asterisk notation in *edun, see Chapter 2: note 1.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
30
languages use a normal simple verb. Thus, in principle, these locutions do not
contrast with the major TAM oppositions in the system, but are rather a lexical
phenomenon. Now, the distribution of some of these verbal locutions seems to
respond to lexical-semantic motivations, although (perhaps expectedly) there is
nothing like a one-to-one correspondence between the Basque groups of locutions
and, for instance, Vendler’s classes of verbs. In some cases, the Basque locutions
appear more related to transitivity issues than to TAM distinctions, but it might be
that both phenomena are involved in these instances.
The only Basque locutions I will examine in this work are those that appear
in stative predicates. These will be introduced in Chapter 2 and studied in more depth
in Chapters 3 (section 6.5) and 4 (section 10). As mentioned above, it should not be
surprising to find that stative predicates behave somewhat differently. Unfortunately,
not all the Basque stative verbs are verbal locutions. Still, it appears sound to
consider that the existence of an important group of stative locutions in Basque
ultimately stems from the inherent differentiation of stative predicates. (See Dahl
1985:28 for a cross-linguistic report).
5.2. Incompatibilities and restrictions.
The most important and cross-linguistically significant of the phenomena related
with lexical-semantic classes of predicates are those concerning incompatibilities,
restrictions and special readings of the major TAM constructions in specific classes
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
31
of verbs. Here again we find that the class of stative predicates is, in general terms,
the one presenting most incompatibilities and restrictions with the major TAM
constructions in a language. As exposed in section 4.4, the main lexical-semantic
incompatibilities do not appear with mature constructions but rather with young
constructions that are not highly generalized in use: e.g. progressive forms. (These
are actually the kinds of constructions used in the literature as evidence for
distinguishing among classes of predicates: cf. Vendler 1967, Dowty 1979.)
In this work, I will examine these lexical-semantic restrictions when dealing
with the corresponding clines of grammaticalization. Most specially, I will present
the restrictions and special readings appearing in the path of grammaticalization
starting with present progressive (Chapter 3). These affect mostly stative predicates
but also achievements (Vendler 1967:102-103), the latter conveying only iterative or
prospective-future senses when appearing in a Progressive construction.
5.3. Conservativeness and lexical-semantics.
A final phenomenon where the lexical-semantic dimension might be a pertinent
factor regards the conservativeness of declining old constructions. Recall that
declining forms are confined to specific verbs and have therefore lower generality of
use than mature constructions. However, unlike in the low degree of generality of
young constructions, it is not obvious that lexical-semantics plays any role in the
case at hand. (This is why I talk about the ‘lexical dimension’ in this case.) What
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
32
seems more secure is that token-frequency is a very relevant variable in the
conservativeness of forms in decline (see section 4.3 above); perhaps the most
significant factor.
Nevertheless, we will see in Chapter 4, section 8.2, that the Basque Synthetic
Present forms are being conserved in high-frequency items, but also specifically in
stative predicates. This might point, thus, toward some contribution of lexical-
semantic factors in the conservativeness of declining forms, along with high token-
frequency. In any event, “membership in a structural class is not determined solely
by its meaning”, but also by the conservation strength of particular items within the
class and also “by chronological coincidence” (Bybee & Dahl 1989:61).3
6. TOOLS: SEARCHING FOR SOURCES OF GRAMMATICALIZATION
AND ESTABLISHING A DIACHRONIC DEPTH.
As explained in the preceding sections, many of the hypotheses I put forward and
test in this work are assumed to hold for the case of same-source layers. Thus,
reconstructing sources of grammaticalization and establishing a relative chronology
between those constructions that have the same source are two fundamental tasks
towards testing the hypotheses on grammaticalization. In addition, I believe that
3 Note that, if the above is correct, it appears very important to be able to distinguish whether a case of
low degree of generality of use is occurring in a young-generalizing construction or in an old-
declining construction. If the former, then the morphological classes appearing from the non-total
generality of use of the major construction are due to lexical-semantic incompatibilities with the
source of grammaticalization. If the latter, then the structural classes are mostly due to frequency
differences. In this case, it does not necessarily make sense to look for lexical-semantic distinctions.
(This latter may be perhaps the case of the so-called French “unaccusative” verbs having ‘to be’ as
auxiliary in the Perfect forms.)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
33
establishing a chronology of the constructions in a given TAM system provides a
diachronic depth that has valuable power in illuminating the synchronic stage of the
language. This strengthens the idea that the diachronic dimension is needed as part of
a complete explanation.
However, to achieve this diachronic dimension we need reliable tools of
reconstruction. Unfortunately the success of reconstructing depends to an important
extent on the data we have in each case. On the other hand, for the purpose of testing
the hypotheses on grammaticalization presented in this chapter, we do not need deep
detailed reconstructions. Actually, we only look for the origin of the constructions,
and then for a relative chronology between those having the same origin. In any case,
the tools we have for reconstructing the sources of grammaticalization of the Basque
TAM constructions are presented below.
6.1. Searching for sources of grammaticalization.
The general tools for reconstructing the sources of grammaticalization of the Basque
TAM constructions are: internal reconstruction (mostly based on the comparison of
verbal and nominal morphemes), attested historical data, and cross-linguistic
comparison of both form and meaning.
The great majority of the Basque TAM constructions are analytic. The
analytic forms of the indicative consist of two words (in a few cases three): a
nominal nonfinite form and a stative auxiliary verb. The most common auxiliaries
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
34
are the verbs izan ‘to be’ and *edun ‘to have’, which exist as independent verbs. The
nonfinite forms are also readily identifiable for the most part. Some of them have
morphemes that are identical to nominal case-markers, such as the locative marker -n
and the genitive markers -ko / -en. Therefore, to discover the source of
grammaticalization of these constructions, we only need a shallow reconstruction,
which in most of the cases is very simple. We will see each of these reconstructions
in detail throughout this work. For now, it just suffices to say that, besides the
simplicity of the reconstructions, many of these forms have exact parallels in other
languages, especially European. Thus, the Basque Perfect has parallels in Romance,
Germanic and other Indo-European families; the Basque Future with the genitive
marker has a parallel in the Spanish haber de construction; and the Basque
constructions based on a locative expression have parallels in a fair number of
language families.
Some periphrastic constructions of the Basque TAM system are not this
transparent, though. Among the most opaque are the Present Subjunctive and the
Periphrastic Aorist (we might call them provisionally “subjunctive forms”: see
discussion in Chapter 5). I discuss the origin of these constructions in Chapters 3 and
5. In any case, and unfortunately, since these constructions do not have a source of
grammaticalization similar to other constructions in the system of Modem Basque,
we will not be able to offer a reliable relative chronology for them. I will assume,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
35
without much evidence, that the Present Subjunctive and the Periphrastic Aorist are
the oldest analytic constructions I will examine in this work.
Finally, the most opaque of all Basque constructions are the Synthetic forms.
I will also propose sources of grammaticalization for these forms in Chapters 3 and
5, although in these cases the reconstructions are rather insecure. Nevertheless, the
Synthetic Present and the Synthetic Past of Basque will not be properly analyzed as
belonging to a grammaticalization (i.e. generalization) cline. Rather, as already
discussed in section 4.3, they will be understood as very old constructions, in
competition for specific areas of TAM meaning, but experiencing an advanced
process of decline. Thus, more importantly than determining their sources of
grammaticalization, what we need regarding the Synthetic forms of Basque is to
prove that they are very old in the language.
I examine now the tools we have for establishing this relative chronology, or,
what is the same, for providing a partial diachronic depth to the TAM system of
Modem Basque.
6.2. Providing a diachronic depth to the TAM system of Basque.
It ought to be clear by now that I do not aim to give a general chronology relating all
forms in the TAM system of Basque. I will only establish a relative chronology
between the constructions that follow the same path of grammaticalization, after
having ensured that their source of grammaticalization is the same or very similar.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
36
Concretely, the paths of grammaticalization where we can safely hypothesize the
existence of layers in Basque are the path of grammaticalization starting with present
progressive (Chapter 4) and the paths starting with resultatives (Chapter 6). The task
of proposing a chronology for the clines in the TAM system of Basque will be
presented, for each path, in the beginning of its corresponding chapter. In what
follows we will see some generalizations valid for all of the paths.
In the areas of competition we will examine in this work, there may be
distinguished four groups of forms (see Chapter 2 for more details on the description
of the constructions at issue): 1) the main analytic forms of the indicative, 2) the
peripheral analytic forms of the indicative, 3) the “subjunctive” analytic forms, and
4) the synthetic forms. The two first groups of forms are strikingly parallel, but have
also interesting differences with each other. The most important difference between
the forms in 1) and the forms in 2) is their degree of rigidity: i.e. the difference in
relative mobility between nonfinite form and auxiliary. Thus, we may also call them
‘rigid periphrases’ and ‘loose periphrases’, respectively.
Throughout this work, I will try to prove that the ‘main analytic forms’ of the
indicative (rigid periphrases) are older in the system than the ‘peripheral analytic
forms’ of the indicative (loose periphrases). Moreover, I will try to prove that the
latter constitute a younger layer of grammaticalization which is evolving along the
same path the main indicative forms previously evolved. I also want to prove, on the
other hand, that the synthetic forms are older in the language than any periphrastic
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
37
form. As for the so-called “subjunctive” forms, I already said that their relative
chronology is very difficult to determine. They appear, in any case, as older than the
main analytic forms of the indicative.
The following general evidence will be used as tools for establishing the
relative chronology between the just-introduced four layers of constructions: i.e. the
synthetic forms, the subjunctive forms, the main analytic forms of the indicative, and
the peripheral analytic forms of the indicative.
6.2.1. Degree offormal rigidity and fusion.
The degree of formal rigidity / boundness is definitely a very important factor
(though not the only factor) I will use as evidence to order chronologically the
Basque TAM constructions in each grammaticalization cline.
- Synthetic vs. analytic forms: According to the above assumption, between
the Basque synthetic and analytic constructions in the same path, this factor will
predict that the synthetic forms are older.
- “Subjunctive” forms vs. indicative forms: There is some formal evidence
pointing to a higher degree of rigidity and lower degree of mobility of the
Periphrastic Aorist over the rigid periphrases of the indicative (see section 1.2 of
Chapter 6). As for the Present Subjunctive, the data are not that clear in this respect.
In any case, the observation on the Periphrastic Aorist is a piece of evidence for
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
38
considering that the “subjunctive” forms are older than all the analytic forms of the
indicative.
- Rigid periphrases vs. loose periphrases of the indicative: Similarly, the
degree of formal rigidity will indicate that, between the rigid and loose periphrases
of Modem Basque, the former are older than the latter.
6.2.2. Other evidence to chronologically order synthetic vs. analytic forms.
1) All periphrastic forms of Basque are construed upon synthetic forms. In other
words, the auxiliaries of the analytic forms are in turn independent forms themselves,
which have either exactly the same or a very similar morpheme ordering to other
synthetic constructions. This is an indication that the periphrastic constructions were
made up by the combination of previously independent synthetic constructions with
nonfinite forms.
2) The nonfinite forms (concretely the participles) of most inherited verbs of
Basque, as opposed to younger borrowings, always show the same shape: they all
begin with a prefix *e- and end in one of two suffixes -i or -n (with few exceptions
ending in -o). The younger borrowings do not begin with *e~, and always have the
ending -tu, of Latin-Romance origin. Now, it is only among the first group that we
find synthetic forms, both currently and in the oldest texts. It seems then that most
probably all these inherited verbs of the shape *e-root-i/-n, and only they, have ever
had any synthetic form. In other words, these data seem a proof that the “synthetic
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
39
conjugation” was actually a productive rule in earlier stages of Basque, but it became
unproductive by the time the massive verbal borrowing from Latin-Romance started.
6.2.3. Other evidence to chronologically order the three layers o f analytic forms.
From the comparison between the rigid and loose periphrases of the indicative, we
can find Basque-specific formal features distinguishing between each pair of these
(very similar but partly different) periphrases. These formal features independently
point to a different seniority in the language.
Probably the clearest case is that appearing from the comparison between
Resultative and Perfect. These constructions are very similar in Basque, although the
Resultative can show a great amount of variation. In the most similar case, the two
forms are almost identical; the only difference between them is that the Resultative
has the article -a in the participle: e.g. ikusi dut ‘I have seen it (perfect)’ vs. ikusia
dut ‘I have seen it (resultative).’ Now, independent evidence can be adduced to show
that the article is a relatively new innovation in adjectives and predicative
complements of Basque. In other words, independent evidence shows that the
construction having a participle with no article is older than that having a participle
with article.
As for the “subjunctive” forms, we could mention the fact that these Basque
constructions are built upon dynamic auxiliaries which are much more unusual in the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
40
geographical area than the analytic constructions of the indicative. This may
constitute a small piece of evidence for considering the former older than the latter.
6.3. Making one general age correspond with one Basque construction type.
In many of the TAM areas of competition we will be studying in this work, there
appear four main types of constructions, which we have chronologically ordered
already. It would seem illustrative to try and make these constructions of Basque
roughly correspond with four idealized general ‘ages’ of TAM constructions (see the
Glossary for clarification).
1) Thus, the loose periphrases of the Basque indicative seem to be young
restricted constructions still currently undergoing a process of generalization of use.
These constructions are unfortunately not very common in texts (specially in old
texts), and therefore cannot be studied in total depth in this work. Furthermore, they
present important dialectal variation, perhaps also related to their not being very
generalized. I believe then that a comprehensive study of these forms needs recorded
and elicited data of current speech, which is beyond the scope of this work.
2) Rigid periphrases, i.e. the main analytic forms of the indicative in Modem
Basque, in turn, are the most frequent and most generalized constructions in the
system. (In some descriptions of the TAM system of Basque these are the only
constructions taken into account.) They seem to be mature constructions regarding a
four-way chronological ordering. These are the forms we will encounter the most in
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
41
the database and are thus the ones I will consider in most detail. I will especially
examine the semantic changes they have undergone in the historical period.
3) I will assume (unfortunately without much proof) that the Present
Subjunctive and the Periphrastic Aorist (so-called “subjunctive forms”) are older in
the system than the rigid periphrases of the indicative. In fact, the “subjunctive
forms” would have been ousted from the domain of the indicative by the
generalization of the main analytic forms of the indicative. They thus appear as
rather old constructions experiencing some recession in the system.
4) Synthetic forms of Basque, finally, have all the characteristics of very old
constructions, actually in a state of advanced decline, towards their eventual loss.
Hence, they appear to have been productive in earlier stages of Basque, but are not
productive any longer. Their study, therefore, has to be viewed from this perspective,
and needs to take into account token-frequency as a main factor of their conservation
in the system.
7. DATA USED IN THIS WORK.
The kinds of data utilized in this work are as follows. Chapter 2, dealing with the
description of the TAM system of Modem Basque, uses basically data taken from
the literature on the topic, and sometimes from my own introspection. (Likewise, in
chapters 3 and 5, which examine cross-linguistic phenomena, the main source of data
is the literature on the issues under consideration.) On the other hand, Chapters 4 and
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
42
6 make use of Basque data collected specially for this work. The data examined in
these two chapters is of written origin only. The most important corpus of data
(‘Main Corpus’) will come from the comparison of three texts. These are three
versions of the Gospel according to Saint Matthew: the first from the 16th century,
the second from the 18th, and the third from the 20th century. Other texts will also be
taken into account when needed. More specifically, the database that will be used for
frequency measures and other detailed comparisons will come from the three texts
just mentioned. The other texts will be used in a more impressionistic fashion.
I will refer to the three texts in the Main Corpus (i.e. the versions of the
Gospel according to St Matthew) as LEIZAR (16th c.), HARAN (18th c.), and ITUN
(20th c.). These three texts are part of the following New Testament books.
1) LEIZAR: From the 1571 version of the New Testament by Joannes
Leizarraga from Beraskoiz (Jean de Licarrague de Briscous, Labourd): Jesus Christ
Gure Jaunaren Testamentu Berria, Pierre Hautin publisher, La Rochelle, 1571.
Reprinted in facsimile by Euskaltzaindia [Royal Academy of the Basque Language],
Bilbao, 1990; with a 1900 introduction by Hugo Schuchardt.
2) HARAN: From the late-18th century manuscript version of the New
Testament by Joannes Haraneder from Donibane Lohitsu (Jean de Haraneder, de
Saint-Jean-de-Luz, Labourd). Edited by Patxi Altuna: Euskaltzaindia, Bilbao, 1990.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
43
3) ITUN: From the 1997 version of the New Testament in Modem Standard
Basque: Elizen arteko Biblia: Itunberria, Euskal Herriko Elizbarrutiak [Dioceses of
the Basque Country], San Sebastian, 1997.4
8. SUMMARY OF HYPOTHESES.
I repeat below for convenience the hypotheses on grammaticalization I have put
forward in this chapter.
(1) Hypothesis of Source Determination:
The source construction that enters into grammaticalization conditions very
importantly the path of grammaticalization that the form will travel in its
semantic development.
(2) Hypothesis of Layering-Displacement (same-source layers):
When the renewal of TAM systems occurs in the way of layers of
constructions with the same source but different degrees of
grammaticalization, if the younger layer ‘catches on’ and starts to attain some
generalization (i.e. in a situation of no-inertia), then the tendency is for the
younger layer to compete with the older layer and to gradually displace it
from specific semantic subtypes, while both evolve along the same cline.
(3) Hypothesis of Form-Meaning Covariation:
There is a highly significant correlation between semantic age and formal
age.
(4) Hypothesis of Conservation of high token-frequency forms:
The conservation of specific items pertaining to an old unproductive
declining construction correlates with their token-frequency. In other words,
4 Note that the two oldest texts in the Main Corpus are written in the Labourdin dialect, whereas the
20th century text is written in Modem Standard Basque. Modem Standard Basque, although close to
Labourdin, has more influence from the western dialects, especially from Guipuzcoan (and even from
Biscayan). Thus, the database can sometimes be partly biased towards one dialectal area or another
regarding specific phenomena.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
44
high token-frequency items tend to persist after the loss of their lower-
frequency counterparts.
(5) Testing the hypothesis of Layering-Displacement:
Given two TAM constructions, A and B, both appearing in two (sufficiently
distant) synchronic stages of a language, X and Y (X prior to Y), so that A
and B share and compete for similar TAM meanings in both stage X and
stage Y. If we have solid evidence that A and B have had the same source of
grammaticalization, and if we have solid evidence to order chronologically A
as immediately prior to B in their path of grammaticalization, then if there
occurs an expansion in the meaning of B from stage X to stage Y (to the right
in the corresponding cline of semantic grammaticalization), it tends to be
paralleled by an analogous narrowing of A from the same contexts in the
passage from stage X to stage Y.
(6) Testing the hypothesis of Form-Meaning Covariation:
Given two TAM constructions, A and B, in a synchronic stage of a language
X, so that A and B are in competition for similar TAM meanings in stage X.
If we have solid evidence that A and B have had the same source of
grammaticalization, then: If A is formally more bound than B (i.e. if A is
formally older than B), then its meaning in stage X should have progressed to
the right in the corresponding cline of semantic grammaticalization no less
than the meaning of B.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
45
CHAPTER 2
THE T.A.M. SYSTEM OF MODERN BASQUE
1. INTRODUCTION.
This chapter is devoted to the synchronic description of the TAM system of Modem
Basque. As stated in Chapter 1, it is not the whole of the TAM system of Basque that
I aim to study in this work but only the domains of present (Chapters 3 and 4) and
past perfective (Chapters 5 and 6). Therefore, the present chapter will mostly focus
on introducing the Basque constructions competing for these domains of TAM
meaning. Even though this chapter deals with synchronic issues only, it has in part
the purpose of preparing some of the grammaticalization-diachronic questions that
will be examined in depth throughout this work. (More comprehensive descriptions
of the TAM system of Modem Basque are the following: Lafitte 1962:185-412 and
N ’Diaye 1970:153-213, in French; Royal Academy of Basque = Euskaltzaindia
1987, in Basque. In English, we could mention Saltarelli 1988:218-247, Haase 1994,
Trask 1997:102-109, 211-239.)
This chapter has a purpose and orientation different from those of Chapters 4
and 6, although sometimes their contents may overlap. In a way, it can be said that
the present chapter is complementary to Chapters 4 and 6. The differences between
this chapter and chapters 4 and 6 are the following. First, this chapter is mainly a
revision of the literature, with some personal remarks; Chapters 4 and 6, in turn, are
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
46
based on the textual data I have collected. Second, while the present chapter has a
synchronic orientation, chapters 4 and 6 have a grammaticalization perspective: i.e.
mostly diachronic, though synchronic issues are also examined there. Third, this
chapter is intended to be just a presentation of the TAM system of Basque (paying
particular attention to the formal description), whereas Chapters 4 and 6 aim to study
several issues (mostly semantic) in more detail and with the aid of actual data. As a
matter of fact, there are some features of the TAM system of (Old) Basque that the
data show, which have never been reported in the literature (e.g. the existence in Old
Basque of a non-evidential perfective). On the other hand, due to the fact that the
data used in Chapters 4 and 6 are collected from written texts only, it is probably the
case that this database fails to capture in their real magnitude some of the synchronic
traits of current spoken Basque (e.g. Progressive constructions).
2. OVERVIEW: THE T.A.M. SYSTEM OF MODERN BASQUE.
2.1. The main analytic forms of the indicative.
The core of the verbal system of Modem Basque consists of three analytic types of
constructions. Each of them can appear in both present and past tenses, thus giving
six different verbal forms. I will usually refer to them as the main or core analytic
forms of the indicative. I also called them ‘rigid periphrases’ in Chapter 1. These six
verbal forms are presented in Table 2.1 below.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
47
TABLE 2.1: Main analytic forms of the indicative: examples.
NAME EXAMPLES APPROXIMATE ENGLISH
EQUIVALENT
Periphrastic Present sartzen naiz I enter
idazten dut I write
Present Perfect-
perfective)
sartu naiz I have entered
idatzi dut I have written
(Periphrastic) Future sartuko naiz I will enter
idatziko dut I will write
Past Imperfective sartzen nintzen I used to enter
idazten nuen I used to write
Past (Perfect)-
Perfective
sartu nintzen I entered
idatzi nuen I wrote
(Periphrastic)
Future-in-the-Past
sartuko nintzen I would enter
idatziko nuen I would write
In the present work, I will only examine in detail three of the above forms:
the Periphrastic Present (Chapters 3 and 4), the Present Perfect-Perfective and the
Past Perfect-Perfective (Chapters 5 and 6). In any case, all of the six main analytic
constructions of the indicative are built by means of the combination of a nonfinite
form plus a conjugated form of the stative auxiliaries izan ‘to be’ (e.g. naiz ‘I am’,
nintzen ‘I was’) or *edun1 ‘to have’ (e.g. dut ‘I have’, nuen ‘I had’). The selection of
izan or *edun as auxiliary depends on whether the subject is, respectively, in the
absolutive case (as with the verb sartu ‘to enter’) or in the ergative case (as with the
verb idatzi ‘to write’).2
1 The indicative transitive auxiliary *e(d)un ‘to have’, like the subjunctive auxiliaries *edin and *ezan,
does not have any clearly-attested nonfinite form. Thus, their citation form (the Perfect Participle) has
been reconstructed following the pattern of the other verbs subject to synthetic inflection.
21 cannot pursue in this work the complex issue whether Modem Basque should be considered
(predominantly) ergative-absolutive or active-inactive. The terminology I will be using is the
traditional one which assumes that Basque is morphologically ergative. However, this question is
open to discussion and clearly needs more empirical work. In addition, the dialectal division seems
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
48
The three nonfinite forms that partake in the composition of the core
indicative constructions could be called Imperfective Participle, e.g. sartzen
‘entering’, idazten ‘writing’, joaten ‘going’ (-tze-n, -te-n endings); Perfect Participle,
e.g. sartu ‘entered’, idatzi ‘written\jo a n ‘gone’ (-tii, -i, -n endings); and Future
Participle, e.g. sartuko, idatziko, joango (-tu-ko, -i-ko, -n-go endings). (See section 3
for more details on the nonfinite forms.) As for the main analytic forms themselves,
there are different names in the literature to designate them. I will be using in this
work the names given in Tables 2.1 and 2.2. Table 2.2 below shows the kind of
periphrasis that makes up each of the six main forms of the indicative, and the
current approximate meaning of these constructions. (See section 5 for more details.)
TABLE 2.2: Main analytic forms of the indicative: structure and meaning.
NAME FORMAL EXPRESSION
(TYPE OF PERIPHRASIS)
APPROXIMATE CURRENT
MEANING
Periphrastic Present Imperfective Participle +
present stative auxiliary
present non-progressive
(mostly habitual)
Present Perfect-
(Perfective)
Perfect Participle +
present stative auxiliary
perfect and hodiernal perfective
(Periphrastic) Future Future Participle +
present stative auxiliary
future
(and present probability)
Past Imperfective Imperfective Participle +
past stative auxiliary
imperfective past non-progressive
(mostly habitual)
Past (Perfect)-
Perfective
Perfect Participle +
past stative auxiliary
pre-hodiemal perfective
(sometimes pluperfect)
(Periphrastic)
Future-in-the-Past
Future Participle +
past stative auxiliary
future-in-the-past and unreal
conditional
fundamental in this case, to the extent that we may perhaps need to talk about eastern dialects being
ergative and western dialects being active-inactive or agentive-nonagentive.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
49
According to some synchronic formal descriptions of these constructions
(which I do not necessarily endorse), we could represent the six forms above in two
“super-categories”: “tense” and “aspect.” The former category would have two
members (“Present” and “Past”) and the latter three (“Perfective”, “Imperfective”
and “Future”), thus giving the six possible combinations above (cf. Euskaltzaindia
1987:404). Hence, the auxiliary would bear the “tense” feature of the whole verbal
complex (i.e. “Present” or “Past”), and can also have agreement in person and
number (also gender for informal 2n d person ergative) with absolutive, ergative and
dative cases3. In turn, according to the same formal descriptions, the main verb (its
nonfinite form) would bear the “aspect” feature (i.e. “Imperfective”, “Perfective” or
“Future”)4.
2.2. The peripheral analytic forms of the indicative.
Besides the main forms of the indicative, there are other analytic forms appearing in
indicative contexts. These constructions have a lower degree of boundness between
the two components of the periphrasis than the main analytic constructions, and
3 There can appear also a fourth kind of person agreement in the synthetic forms (either auxiliaries or
independent verbs): namely, allocutive agreement, whereby the addressee of an utterance is also
cross-referenced in the verb with a second-person marker. Allocutive forms occur only in main
clauses and in colloquial (and ‘individually addressed’) speech.
4 The present work is not especially sympathetic to descriptions-classifications such as the one just
presented (based on the widely-adopted classification by Euskaltzaindia 1987:404), which assumes an
identity between formal organization and semantic categorization. Concretely, if the form-meaning
isomorphism in categorization is a theoretical primitive, the Basque Future has to be included in the
same super-category (termed “Aspect”) with the “Perfective” and the “Imperfective.” That is,
Euskaltzaindia’s classification assumes that, just because the Future, the “Perfective” and the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
50
convey non-generalized specific meanings. They are also less frequent, at least in
written texts. I will refer to them as the peripheral analytic forms of the indicative. I
also called them ‘loose periphrases’ in Chapter 1. (As presented in Chapter 1 ,1 will
try to prove in this work that these peripheral constructions are rather young in the
language and seem to represent the renewal of the indicative system.)
In this section, I present three types of peripheral analytic constructions. Each
of them can be construed, among other possible combinations, with present and past
stative ‘quasi-auxiliaries’ (see the Glossary and section 6 below for the use of this
term), thus yielding six fonns we are interested in. Table 2.3 below introduces these
six constructions.
TABLE 2.3: Peripheral analytic forms of the indicative: examples.
NAME EXAMPLES APPROXIMATE ENGLISH
EQUIVALENT
Present Progressive sartzen ari naiz I am entering
idazten ari naiz I am writing
Present Resultative sartua naiz (I am entered)
idatzia dut (I have it written)
Present Prospective sartzeko naiz I am to enter
idazteko dut I am to write
Past Progressive sartzen ari nintzen I was entering
idazten ari nintzen I was writing
Past Resultative sartua nintzen I had entered
idatzia nuen I had written
Past Prospective sartzeko nintzen I was to enter
idazteko nuen I was to write
“Imperfective” are formally marked in a parallel way, they have to belong in the same semantic super
category. (See Chapter 7 for discussion on these issues.)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
51
From the description in Table 2.3, the reader can already note at this point the
high similarity in formal expression that each of these six peripheral constructions
shows with one of the main constructions in Table 2.1. In truth, the formal
expression of all these six constructions presents some cross-dialectal variation; what
appears in Tables 2.3 and 2.4 are the forms in Modem Standard Basque. Throughout
this work, I will only examine in depth the Present Progressive (Chapters 3 and 4)
and the Present and Past Resultatives (Chapters 5 and 6). In any case, Table 2.4
below gives the structure of the six peripheral main forms of the indicative, and their
current approximate meaning. (See section 6 for more details.)
TABLE 2.4: Peripheral analytic forms of the indicative: structure and meaning.
NAME FORMAL EXPRESSION
(TYPE OF PERIPHRASIS)
APPROXIMATE CURRENT
MEANING
Present Progressive Imperfective Participle + ari +
present stative quasi-auxiliary
present progressive
Present Resultative Perfect Participle + article +
present stative quasi-auxiliary
present resultative
Present Prospective Prospective Participle +
present stative quasi-auxiliary
prospective present
Past Progressive Imperfective Participle + ari +
past stative quasi-auxiliary
past progressive
Past Resultative Perfect Participle + article +
past stative quasi-auxiliary
past resultative and pluperfect
Past Prospective Prospective Participle +
past stative quasi-auxiliary
prospective past
2.3. The analytic forms of the non-indicative.
There is a third group of analytic forms that in Modem Basque only show up in
contexts other than indicative. Among these periphrastic constructions we find the
Imperative, the Subjunctive, the Irrealis and the Potential. In this work, we will be
concerned with the (Periphrastic) Present Subjunctive, which is presented in section
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
52
7 below, and examined in depth in Chapters 3 and 4. We will also study in Chapters
5 and 6 a construction identical to the Modem Basque Past Subjunctive, which had a
different meaning in earlier stages: the Old Basque Periphrastic Aorist. The rest of
non-indicative forms will not be studied in the present work. Table 2.5 gives
examples of the Present Subjunctive and the Past Subjunctive of Modem Basque.
TABLE 2.5: Subjunctive forms of Modem Basque: examples.
NAME EXAMPLES APPROXIMATE ENGLISH
EQUIVALENT
(Periphrastic)
Present Subjunctive
sar nadin so that I enter
idatz dezadan so that I write
(Periphrastic) Past
Subjunctive
sar nendin so that I entered
idatz nezan so that I wrote
The (Periphrastic) Present Subjunctive and the (Periphrastic) Past
Subjunctive (as well as the analytic Imperative and Potential, not examined here) are
constructed by the combination of a nonfinite form called the Radical (-0 ending:
e.g. sar ‘to enter’, idatz ‘to write’) and the achievement-dynamic auxiliaries *edin
(e.g. nadin, nendin) and *ezan (e.g. dezadan, nezan). Table 2.6 outlines the form and
meaning of the Present Subjunctive and the Past Subjunctive constructions of
Modem Basque. (See section 7 for more details.)
TABLE 2.6: Subjunctive forms of Modem Basque: form and meaning.
NAME FORM (PERIPHRASIS) APPROXIMATE CURRENT USE
(Periphrastic)
Present Subjunctive
Radical + present
achievement auxiliary
purpose clauses and
complement clauses to verbs of ordering
(Periphrastic)
Past Subjunctive
Radical + past
achievement auxiliary
past purpose clauses and
complement clauses to verbs of ordering
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
53
2.4. The synthetic forms.
Finally, for no more than twenty high-frequency verbs (including some used as
auxiliaries, such as ‘to be’ or ‘to have’) there are four synthetic or inflectional forms
(Present, Past, Irrealis and Imperative), which are also in competition with the
analytic forms for specific domains. Table 2.7 gives examples of the synthetic
constructions.
TABLE 2.7: Synthetic forms of Modem Basque: examples.
NAME EXAMPLES APPROXIMATE ENGLISH
EQUIVALENT
Synthetic Present nator I am coming, I come
dakart I am bringing, I bring
Synthetic Past nentorren I was coming
nekarren I was bringing
Synthetic Irrealis ba-nentor if I came
ba-nekar if I brought
Synthetic Imperative hator come! (2n d p. colloquial)
ekark bring it! (2n d p. colloq. masc.)
Only the Synthetic Present and the Synthetic Past will be considered in this
work. Currently, unlike in Old Basque, both Synthetic Present and Synthetic Past are
confined to imperfective aspect. (See section 4 for more details.)
2.5. Division of the chapter.
The presentation of this chapter is as follows. In section 3 ,1 present the nonfinite
forms that are part of the analytic constructions we aim to examine. I will study the
nonfinite forms insofar they appear as part of finite analytic constructions. In section
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
54
4 ,1 present the synthetic finite forms of Basque, which are relevant for this work in
two respects: as auxiliaries of the analytic forms and as independent synthetic
constructions for a handful of verbs. I will limit to give a quick formal account of the
synthetic forms. Once we have studied the two types of components of the analytic
forms (nonfinite forms under section 3 and auxiliaries under section 4), section 5 will
be devoted to describing the use of the main analytic forms of the indicative. Section
6 will present the peripheral periphrases of the indicative. Section 7 will be devoted
to outlining the constructions of the subjunctive, and its dynamic auxiliaries. In
section 8, finally, I give a summary of the shape of the constructions examined in
this work.5
3. NON-FINITE FORMS.
Given that the main object of study of this work is some of the finite forms of
Basque, the nonfinite forms will not be treated in detail. However, as most of the
verbal system of Basque is analytic in nature, it is mandatory to give at least a
superficial account of the nonfinite forms that are a part of those periphrastic
constructions. This is especially true of the peripheral constructions of the indicative
51 cannot conclude this overview without evoking the question of dialectal variation. The labels
“Western” and “Eastern” I will be using hereafter will make reference to the prototypical use in each
area. I do not mean to imply with these terms that there are two clear-cut areas with strict
complementary distribution. In fact, the differences between dialects are rather diffuse and the
borderline between East and West is really difficult (if not impossible) to draw. It is rather a
continuum of different features. Still, if we compare Biscayan on the one extreme and French-Basque
dialects (most especially Souletin) on the other, there are remarkable differences in the matters under
consideration in this work.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
55
(section 6), which make use of a more mobile nonfinite form. In other words, these
peripheral constructions utilize nonfinite forms that are used in the language as
independent items. The main analytic constructions of Basque, on the other hand,
have more grammaticalized nonfinite forms, some of which do not have any
independent existence in the current language.
I will present in this section only the five nonfinite forms appearing as part of
the periphrases in Tables 2.1-2.6 above, plus the Verbal Noun. The latter can be
considered, in formal terms, as the base of other nonfinite forms. (See Trask 1995 for
a detailed account of the Basque nonfinite forms and the morphological classes of
verbs.)
3.1. Verbal Noun.
TABLE 2.8: Morphology of the Verbal Noun.
ENDING EXAMPLES (and approximate English equivalent)
OPEN CLASS OF
VERBS
-tze(a) apurtze(a) ‘(the) breaking’, sartze(a) ‘(the) entering’,
kopiatze(a) ‘(the) copying’, deitze(a) ‘(the) calling’
OTHER MORPH.
CLASSES
-te(a) izate(a) ‘(the) being’, joate(a) ‘(the) going’, esate(a) ‘(the)
saying’, ikuste(a) ‘(the) seeing’, eroste(a) ‘(the) buying’
DIALECTAL
VARIANTS
various See Trask (1995:219-223) for a detailed account.
The Verbal Noun of Basque, like in other languages, shares features of nominal and
verbal nature. It behaves like a noun in form, and thus it appears with the article or
determiner -a most of the time. It can also take any nominal case-marker, as any
other NP does. An example of the Verbal Noun in the ergative case is given in (1)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
56
below. (For the use of the genitive case in the direct object within the nonfinite
embedded clause, see Heath 1972 and examples (12) in section 6.2.1 below.)
(1) [idaz-le zaharr-en irakur-tze]-a-k monta handi-a du
[writ-er old-GEN read-TZE\-DET-ERG importance great-DET it-has
“Reading the old authors is very important” (Eastern dialects)
Lit. “The reading of the old authors has great importance”
(Euskaltzaindia 1987:99)6
The Verbal Noun conveys commonly temporal meanings when bearing local
case-markers, especially with the locative (-an ending): e.g. liburua irakurtzean
‘while reading the book / as soon as I finish reading the book.’ (Note that the Modem
Basque locative case-marker ends in -an, and is therefore different from the ancient
locative in -n, which appears fossilized in the Imperfective Participle, studied in
section 3.2 below).
3.2. Imperfective Participle, Present Participle or Gerund.
TABLE 2.9: Morphology of the Imperfective Participle.
ENDING EXAMPLES (and approximate English equivalent)
OPEN CLASS OF
VERBS
-tzen apurtzen ‘breaking’, sartzen ‘entering’, kopiatzen ‘copying’,
deitzen ‘calling’
OTHER MORPH.
CLASSES
-ten izaten ‘being’, joaten ‘going’, esaten ‘saying’, ikusten ‘seeing’
DIALECTAL
VARIANTS
various See Trask (1995:219-225) and Lakarra (1996:264-265) for a
detailed account of the dialectal variants.
6 Most of the examples I give throughout this chapter will be taken from the three following
grammars: Lafitte (1962), Euskaltzaindia (1987) and Saltarelli (1988). I indicate the source next to
each example. (Non-attributed examples, my own.)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
57
The Present Participle seldom appears independently, except in western dialects
where it can be a modal gerund. Normally it is always subordinated to a main verb
that specifically subcategorizes the Imperfective Participle as its complement. In
most of its subordinated appearances, the Imperfective Participle has a progressive
meaning. The main verbs that subcategorize for it are of the kind of: ari izan ‘to be
engaged in’ (Eastern dialects), izan ‘to be (at, in Eastern dialects)’, egon ‘to be at, to
be located, to stand’, ibili ‘to wander, to walk, to be doing something’, ihardun ‘to be
engaged in, to pursue’, ohitu ‘to be used to’, usatu ‘to be used to’, etc. (For
examples, see Euskaltzaindia 1987:103). Some of these complex constructions will
be studied when dealing with Progressive forms (section 6.2 below).
The most important use of the Imperfective Participle, anyhow, is as part of
main analytic forms: e.g. the Periphrastic Present and the Imperfective Past (see
section 5.1 below).
3.3. Perfect Participle or Past Participle.
TABLE 2.10: Morphology of the Perfect Participle.
ENDING EXAMPLES (and approximate English equivalent)
OPEN CLASS OF
VERBS
-tu apurtu ‘broken’, sartu ‘entered’, kopiatu ‘copied’, deitu
‘called’
OTHER MORPH.
CLASSES
-i, -n
(-ki, -0)
izan ‘been’, joan ‘gone’, esan ‘said’, ikusi ‘seen’, etorri
‘come.’ See Trask (1995:208-211) for details.
DIALECTAL
VARIANTS
-(a)du, -(a)u,
-(i)du, -tun
Western dialects
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
58
The Perfect Participle, unlike in its neighboring languages, is the citation form of
Basque verbs7 (and is probably understood as the most unmarked of all nonfinite
forms). There is an important difference between eastern and western dialects in
what concerns the use of the Perfect Participle independently. The two main
independent uses of the Perfect Participle are in temporal (anterior) / causal
subordinate clauses, and as a verbal adjective-adverb.
In temporal-causal subordinate clauses, the Perfect Participle conveys the
meaning of ‘anteriority.’ However, in this function, the Perfect Participle always
bears some kind of suffix. All dialects use the partitive-ablative case-marker -(r)ik
for this function, as in example (2a) below. Along with the former but probably more
often than it, western dialects make use of a -ta ending suffixed to the Perfect
Participle, as in (2b). The ending -ta seems to have its origin in the copulative
conjunction (e)ta ‘and’ (see Chapter 5 for more details on these anteriority markers).
(2a) ele horiek entzun-ik, etxe-ra itzuli nintzen
word those heard-PART house-ADL returned I-was
“Having heard those words, I came back home” (Common)
(Euskaltzaindia 1987:91)
(2b) berba horrek entzun-da, etxe-ra bueltatu nintzan
word those heard- TA house-ADL returned I-was
“Having heard those words, I came back home” (Biscayan-Western)
7 This creates a problem of ambiguity when one needs to translate the Basque Past Participle into, say,
English. On the one hand, it can be translated as the citation form, that is, as an English infinitive. (As
a matter of fact, I have done so in many places throughout this work.) On the other hand, it is
sometimes properly a past participle and it should be translated as so.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
59
The use of the Perfect Participle as a verbal adjective-adverb (e.g. in
resultative constructions) draws even more remarkable differences among dialects.
Eastern dialects can use the Perfect Participle itself as an adjective, while Modem
Western dialects can hardly utilize the Perfect Participle as a normal adjective.
Focusing on the Resultative constructions, which is the case we are mainly interested
in (see section 6.1 below), in most Eastern varieties the Perfect Participle behaves
formally like a normal adjective in a predicative use, and thus it appears with the
article or determiner -a (sg) / -ak (pi). On the other hand, in Western varieties and
also in Souletin, the Resultative constructions do not use the Participle plus article as
the predicative complement; instead they utilize the Participle plus one of the two
endings we saw in examples (2) above. The ending -rik is common to all dialects; the
ending -ta is only known in western dialects, especially in Biscayan. These endings
are invariable: i.e. there is no number agreement in the Participle in these cases. (See
section 6.1 for more details on the Basque Resultatives.)
3.4. Future Participle.
TABLE 2.11: Morphology of the Future Participle.
ENDING EXAMPLES
OPEN CLASS OF
VERBS
-tuko apurtuko, sartuko, kopiatuko, deituko
i.e. Perfect Participle + -ko: Common
OTHER MORPH.
CLASSES
-iko, -ngo izango, joango, esango, ikusiko, etorriko
i.e. Perfect Participle + -ko
DIALECTAL
VARIANTS
-iren, -nen izanen, joanen, esanen, ikusiren, etorriren
i.e. Perfect Participle + ~(r)en: Eastern dialects
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
60
The Future Participle is not an independent nonfmite form in Modem Basque.
Within the indicative mood, it only appears in the Future analytic constructions, with
both present and past auxiliaries. Etymologically, the Future Participle is clearly the
‘genitive’ of the Past Participle. Thus, to render the Future Participle, western
dialects add the ending -ko to the Perfect Participle, whereas eastern dialects can add
not only -ko but also -(r)en. Now, the ending -ko is formally identical to the locative
genitive (which could be called more properly a general linker). Similarly, the
ending -(r)en is formally identical to the possessive genitive. Interestingly, both of
these case-markers seem to have had some other values in older stages of Basque:
e.g. benefactive and/or purposive.
■ »
Since the Future Participle is not an independent nonfmite form in Modem
Basque, it will only be studied as part of the Periphrastic Future (section 5.4). In any
event, the most remarkable characteristic of the Future Participle is that it is
construed upon the Past Participle, instead of the Verbal Noun.
3.5. Prospective Participle.
TABLE 2.12: Morphology of the Prospective Participle.
ENDING EXAMPLES
OPEN CLASS OF
VERBS
-tzeko apurtzeko, sartzeko, kopiatzeko, deitzeko
i.e. Verbal Noun + -ko
OTHER MORPH.
CLASSES
-teko izateko, joateko, esateko, ikusteko, etorteko
i.e. Verbal Noun + -ko
The Prospective Participle is formally the locative genitive in -ko of the
Verbal Noun. Its origin seems to be somewhat parallel to the Future Participle
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6 1
ending in -tuko, the difference being that this latter adds the -ko suffix to the Past
Participle in -tu, whereas the Prospective Participle in -tzeko does it to the Verbal
Noun, which ends in -tze.
The Prospective Participle appears most importantly in Prospective or Weak-
obligation periphrases, by the combination with a stative quasi-auxiliary (cf. Tables
2.3-2.4 above; for more discussion on these constructions see Euskaltzaindia 1987:
468-470). Besides this use in peripheral analytic constructions, the Prospective
Participle is used independently as a complement of a noun, a kind of nonfmite
relative clause. Its meaning in this noun-complement function has very often a future
time-reference (e.g. egiteko lana ‘work to do’, lit. ‘work of doing’). Another
independent use of the Prospective Participle is as a verbal complement with
destinative, purposive or benefactive values (e.g. zuk egiteko ‘so that you do it’).
3.6. Verbal Radical.
TABLE 2.13: Morphology of the Verbal Radical.
ENDING EXAMPLES (and approximate English equivalent)
OPEN CLASS OF -0 apur, sar, kopia, dei
VERBS
OTHER MORPH.
-0,
izan, joan, esan, ikus, etor
CLASSES
(-n)
(See Trask 1990 for discussion)
The last of the nonfmite forms of Modem Basque we will examine here is the so-
called verbal Radical. The Radical is part of the analytic constructions of the
subjunctive in Standard Basque and in Eastern dialects. The western dialects, in turn,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
62
do not use the Radical but the Past Participle in these Periphrastic Subjunctive forms,
(See section 7 for the description of the Modem Basque Subjunctive constructions.)
The Radical can also appear in imperatives, in old refrains or other fossilized
expressions, in coordinate clauses, and in different types of subordinate clauses. For
an account of the independent uses of the Radical, see Lafitte (1962: 207-211),
Euskaltzaindia (1987: 67-77), Trask (1995: 211-217). In general, it can be said that
the appearance of the Radical in the spoken language is scanty. Moreover, in the
western dialects it is almost non-existent. (For the diachronic orientation of this
work, the Radical appears as more opaque than the other nonfmite forms presented
so far. Its putative origin will be studied in depth in Chapter 3.)
4. SYNTHETIC FORMS.
There are in Modem Basque a handful of verbs that, besides the six main analytic
constructions that all verbs have in the indicative, show also synthetic forms. All
these verbs subject to synthetic inflection have a Past Participle beginning in *e- (e-,
0 -) and ending in -i or -n (cf. Table 2.10 above). In current spoken Basque
there are no more than twenty verbs that have synthetic forms8 . Moreover, some of
them only have a small number of rather fossilized forms. In Old Basque texts,
however, the number of verbs having synthetic forms is higher. (Lafon 1943 lists 58
8 For an exhaustive relation of the independent verbs subject to synthetic inflection, see
Euskaltzaindia (1987:409-410).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
different verbs subject to synthetic inflection in the 16th century texts.) All of the
ever-attested verbs with synthetic forms have a Past Participle of the shape described
above, and none of them seems to be a recognizable loan. On the other hand, none of
the verbs belonging to the currently productive class in -tu has any synthetic form
(unless it is also attested as having formerly belonged to the *e-ROOT-z'/-n classes9 ).
(See section 6.4.1 of Chapter 3 for more discussion on related issues.)
The stative auxiliaries of the main analytic forms of the indicative, izan ‘to
be’ and *edun ‘to have’, are among the handful of verbs that have independent
synthetic forms, though the latter only does so in the Eastern dialects. The rest of
auxiliaries of the indicative and subjunctive do not have any independent synthetic
forms in Modem Basque, but most of them do have independent synthetic forms
attested in the Old Basque texts: e.g. *edin, egin.
Besides izan and *edun, the most common (by far) among the verbs now
subject to synthetic inflection are the following eight high-frequency verbs: egon ‘to
stay, to be located, to be at’, e(d)uki ‘to have\ja kin ‘to know\jo a n ‘to go’, etorri ‘to
come’, ibili ‘to walk, to wander, to be doing something’, eraman (Eastern) / eroan
(Western) ‘to take to, to carry; formal causative of jo a n \ and ekarri ‘to bring.’
All the synthetic forms, including those of the auxiliaries, show an easy to
segment morpheme alignment, which is furthermore extremely consistent throughout
the analogous forms of all different verbs. It has to be admitted, though, that there
9 As in the case of ezagutu ‘to know somebody’, formerly ezagun. (Cf. the current noun ezagun-a
‘acquaintance.’)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
64
are some dialectal variants for some non-principal morphemes. Particularly, number
marking (which is always a distinct morpheme separate from the corresponding case)
and dative “flags” (see Trask 1997: 227-229) are among the morphemes that have
more dialectal and cross-verbal variation. The transitive auxiliaries *edun and *ezan,
for instance, have a morpheme for absolutive-plural agreement which is different in
both form and linear-position from the one that the parallel forms of other non
auxiliary transitive verbs have.
The verbal inflection of Basque is extremely complex, showing agreement
with ergative, absolutive and dative cases. It is not the purpose of this work to study
this formal complexity in any detail. Hence, in what follows I will restrict myself to a
brief description of the shape of the most frequent forms of the Synthetic Present
(section 4.1) and the Synthetic Past (section 4.2). (For a comprehensive description
of the morphemes in the verbal inflection of Basque, see Lafitte 1962:237-333,
Euskaltzaindia 1987:168-347. See also Aldai 2000 for discussion on these issues,
especially on the morphological ergativity-split appearing within the Synthetic Past
forms.)
Regarding the meaning of the Synthetic Present and the Synthetic Past, we
should say that both of these constructions convey only imperfective aspect in
Modem Basque. (This was not the case in Old Basque, as we will see throughout this
work.) More specifically, the Synthetic Present and the Synthetic Past are practically
confined to two kinds of contexts and two groups of verbs in Modem Basque: a)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
65
stative continuous meaning, and b) progressive-prospective meaning of some motion
verbs. Therefore, the Synthetic Present and Past of Modem Basque seldom convey
habitual meaning.
4.1. Synthetic Present.
As stated before, the Synthetic Present of Modem Basque conveys imperfective
present meaning, and is confined to a small group of high-frequency verbs, mostly
stative and motion verbs. Thus, the Synthetic Present is currently in competition with
the other forms appearing in the sphere of imperfective present: namely, with the
Periphrastic Present and with the Present Progressive. This competition will be one
of the objects of study of Chapter 4.
The Synthetic Present is formally distinguished from the Synthetic Past by
the marked presence in the latter of an ~(e)n past-tense suffix (sometimes -an,
especially in Biscayan), while the former is formally unmarked. There is another
formal difference between the two tenses, although this is not as systematic as the -0
/ ~(e)n opposition. This difference is the appearance of an -a- prefix in the Synthetic
Present versus -e- in the Synthetic Past. The third persons show also an added formal
distinction between Synthetic Present and Past. In the Synthetic Present, all third
person absolutive forms (both with ergative and without ergative agreement) bear a
d- prefix.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
66
According to the above presented, we may propose the following description
of the shape of intransitive and transitive Synthetic Present forms, limited to the
simplest and most frequent forms in each group. (Thus, I will consider here neither
dative agreement, nor direct object plural agreement, nor non-third person direct
object forms.)
(3a) Synthetic Present intransitive-absolutive (non-ergative) forms:
1. n-a-ROOT 4. g-a-ROOT-PluralAbs
2. h-a-ROOT
2’. z-a-ROOT-PluralAbs 5. z-a-ROOT-PluralAbs-(t)e
3. d-a-ROOT 6. d-a-ROOT-PluralAbs
(3b) Synthetic Present transitive (ergative) forms:
1. d-a-ROOT-t 4. d-a-ROOT-gu
2. d-a-ROOT-k/na
2’. d-a-ROOT-zu 5. d-a-ROOT-zu-(t)e
3. d-a-ROOT 6. d-a-ROOT-(t)e
4.2. Synthetic Past.
The Synthetic Past forms (other than those of the auxiliaries) are considerably less
frequent than the Synthetic Present forms. As just mentioned, the past marker of
Modem Basque, which formally distinguishes Synthetic Past from Synthetic Present
forms, is an ~(e)n suffix. (It might be interesting to point out that in colloquial
Basque it is not a rare phenomenon to re-make some of the Synthetic Past forms just
by adding the ~(e)n suffix to the corresponding Synthetic Present form. Among the
Synthetic Past forms, it is the least frequent in the system, such as transitive forms
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
67
with 1st and 2n d person absolutive agreement that are more readily subject to this
colloquial reshaping.)
Synthetic Past forms show the above-alluded ergativity split, which will not
be studied here (see Aldai 2000 for discussion and details). Also, intransitive forms
with 3rd person absolutive (subject) agreement, and transitive forms that show both
3rd person ergative (subject) and 3ld person absolutive agreement bear a z- prefix
rd
(normally 0- in western Biscayan) that contrasts with the 3 person absolutive d-
prefix of Present tense. To complete the forma] description of the Synthetic Past
forms, it has to be added that all forms with non-3)d person absolutive agreement
have an -n- infix that appears after the -e- prefix mentioned in section 3.1. A partly
similar infix appears in 1st and 2n d plural (including formal singular, former plural)
ergative agreement forms with 3rd person absolutive agreement.
(4a) Synthetic Past intransitive-absolutive (non-ergative) form s:
1. n-e-n-ROOT-en 4. g-e-n-ROOT-PluralAbs-en
2. h-e-n-ROOT-en
2’. z-e-n-ROOT-PluralAbs-en 5. z-e-n-ROOT-PluralAbs-te-n
3. z-e-ROOT-en 6. z-e-ROOT-PluralAbs-en
(4b) Synthetic Past transitive (ergative) forms:
1. n-e-ROOT-en 4. g-e-n-e-ROOT-en
2. h-e-ROOT-en
2’. z-e-n-e-ROOT-en 5. z-e-n-e-ROOT-te-n
3. z-e-ROOT-en 6. z-e-ROOT-te-n
The Synthetic Past has only imperfective past meaning in Modem Basque. It
appears most often in the case of high-frequency stative verbs. In Old Basque,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
68
however, it was unmarked for aspect, much as the English Simple Past is. Thus, in
Old Basque texts, it is common to find the Synthetic Past with perfective meaning.
Since I will only examine in this work the domain of perfective past (Chapters 5 and
6) and not the sphere of imperfective past, the Synthetic Past of Old Basque will be
object of study in Chapter 6 inasmuch it conveys perfective past meaning.
5. MAIN ANALYTIC FORMS (OF THE INDICATIVE).
The constructions I will be dealing with in this section are the main analytic forms of
the indicative we saw in Tables 2.1 and 2.2, which are construed by way of the
combination of the three major nonfinite forms with present and past forms of the
stative auxiliaries. In this section, I will just introduce the form and uses of the main
analytic forms. (Other formal issues in relation to the periphrases, such as the relative
order between nonfinite form and auxiliary, will be examined in sections 2 of
Chapters 4 and 6.)
As a matter of fact, I will not examine in this section all six main forms of the
indicative. I will limit the presentation to the constructions that will be studied in
Chapters 4 and 6. That is, those appearing in the domains of present and past
perfective, respectively. In other words, I will only describe next the meanings of the
Periphrastic Present (section 5.1) and of the Present Perfect-Perfective and the Past
Perfect-Perfective (section 5.2). I will also introduce two more analytic constructions
formally related to the latter two: the Double-compound Present Perfect and the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6 9
Double-compound Past Perfect (section 5.3). 1 will finally present the subordinate
uses of the Periphrastic Future (section 5.4), because in these contexts the Future is
in competition with other forms that have primarily present-tense meanings and it
will be thus examined in Chapter 4.
5.1. Periphrastic Present.
The Periphrastic Present is constructed by means of the combination of the
Imperfective Participle (-tzen, -ten) and a Synthetic Present form of a stative
auxiliary izan ‘to be’ or *edun ‘to have.’ Its most usual meaning is that of present
habitual, but this is not the only meaning this construction conveys.1 0
For the majority of verbs (those not having synthetic forms), the Periphrastic
Present can appear with gnomic value (txoriek began egiten dute ‘birds fly’), in
performative sentences (nik ematen dizut nire fedea ‘I give you my faith’), and in
simultaneous reports (Beckham-ek ematen dio baloia Owen-i ‘Beckham passes the
ball to Owen). It can also appear in subordinate clauses with some non-indicative
uses, as in, for instance, protases of open conditionals (hori lortzen baduzu, saria
emango dizute ‘if you achieve that, they will give you a prize’).
The Periphrastic Present, however, does not often express stative present
meaning (i.e. present continuous non-progressive: see the Glossary for clarification
1 0 There is actually a particle ohi that can be introduced between the main verb and the auxiliary to
convey specifically habitual aspect: Plazetan ja rri/ jartzen ohi dira noizetik noizera ‘they are usually
posted in the squares, from time to time’ (Buruxkak; Euskaltzaindia 1987:462). The Periphrastic
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
70
on the terminology). This is so because most of the stative predicates of Basque are
either verbal ‘locutions’, consisting of a noun plus a stative main verb, or are
otherwise expressed by synthetic forms. For those few stative predicates that are
neither verbal locutions nor subject to synthetic inflection (recent loans for the most
part), the Periphrastic Present does convey present continuous non-progressive
meaning (e.g. txokolatea gustatzen zaio ‘s/he likes chocolate’).
Thus, it seems that the Periphrastic Present could be regarded as a general
present, or better stated a non-progressive present, because it does not convey
present progressive meaning. Present progressive is expressed by the peripheral
analytic constructions we will see in section 6.2. Therefore, the Basque Periphrastic
Present is roughly similar in meaning (unlike in form) to the English or Spanish
Simple Presents. However, for the handful of verbs subject to synthetic conjugation
(mainly stative predicates and motion verbs), the Periphrastic Present is rather
marked for habitual aspect. In this case, it seems it is the Synthetic Present the
‘unmarked’ form that appears in most of the contexts mentioned above.
5.2. Perfect-Perfective constructions.
The Present Perfect-(Perfective) and the Past (Perfect)-Perfective are rendered by the
combination of the Past Participle (ending in -tu, -i, -n) and, respectively, a Synthetic
Present or Synthetic Past form of the stative auxiliaries izan or *edun: e.g. dantzatu
Present is not, thus, the marked form for habitual aspect (at least for the majority of verbs), but the ohi
form is.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
71
du ‘s/he has danced’, dantzatu zuen ‘s/he danced’ (literally ‘s/he had danced’). In
order to show some consistency in the terminology between formally related
constructions, I have decided to name in the same fashion the two forms that use the
Past Participle as the nonfinite form; namely, ‘PerfechPerfective’ constructions. That
is to say, the one that uses present forms of the stative auxiliaries will be called the
Present Perfect-Perfective {dantzatu du), and the one with past-tense auxiliaries, Past
Perfect-Perfective {dantzatu zuen).
Both forms are of clear perfect origin (probably grammaticalized in a first
step as resultatives: see Chapter 5), formally parallel to constructions that we find in
many languages of Europe: i.e. the Present Perfect and the Pluperfect of languages
such as English or Spanish. However, the past-tense form of Modem Basque is
mainly a pre-hodiemal perfective (‘s/he danced’: see the Glossary for clarification on
the terminology), rather than a pluperfect (although this last meaning is not totally
excluded, especially in subordinate clauses). As the corresponding present tense
form can also convey hodiernal perfective meaning {gaur 7retan puntuan altxatu
naiz', like we find in, for instance, Spanish hoy me he levantado a las 7 en punto,
literally ‘I have woken up at 7 o’clock today’), then it does not seem too forced to
call both forms Perfect-Perfective. It should be bom in mind, however, that there is a
slight mismatch between present and past. The Present Perfect-Perfective of Modem
Basque is mainly a perfect (and sometimes I will refer to it as the Present Perfect:
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
72
egin dut ‘I have done’). Whereas the Past Perfect-Perfective is for the most part a
perfective (egin nuen ‘I did’).
The Perfect-Perfective constructions compete with the Resultatives, with the
Double-compound constructions (see 5.3 next) and also between themselves for
several areas or TAM domains that we will examine in depth in Chapter 6.
5.3. Double-compound Perfect constructions.
There is another group of analytic constructions, formally related to the Perfect-
Perfective forms, which add an extra auxiliary-participle between the Past Participle
of the lexical verb and the conjugated form of the auxiliaries. I will term these forms
as Double-compound Perfects. Thus, I will call Double-compound Present Perfect
(e.g. sartu izan naiz, idatzi izan dut) the extra-auxiliary counterpart of the Present
Perfect-Perfective {sartu naiz ‘I have entered’, idatzi dut ‘I have written’); and
Double-compound Past Perfect (e.g. sartu izan nintzen, idatzi izan nuen) the forms
corresponding to the Past Perfect-Perfective {sartu nintzen ‘I entered’, idatzi nuen ‘I
wrote’). In Modem Basque, Double-compound constructions add always an extra
participle of the verb izan ‘to be’, regardless o f whether the construction is
intransitive or transitive (note that the transitive auxiliary *edun ‘to have’ does not
have any nonfinite forms).
The Double-compound Perfects of Basque have the same origin as the
corresponding constructions in French and Occitan. Also similar to the case of these
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
73
languages, their specific uses are not very well known. In any case, the meaning of
the Double-compound Present Perfect is somewhat similar to that of the Present
Perfect; and the meaning and functions of the Double-compound Past Perfect are
similar to those of the Past Perfect-Perfective. Double-compound Perfects will be
examined in detail in Chapters 5 and 6.
5.4. (Periphrastic) Future.
The Future is construed by combining the Future Participle (-tuko, -iko, -iren, -ngo, -
nen) with an inflected Synthetic Present form of the auxiliaries izan or *edun: e.g.
pasatuko naiz ‘I will pass’, egingo duzu ‘you will do.’ As discussed in section 3.4,
the Future Participle is in origin clearly the genitive {-ko, -(r)en) of the Past
Participle {-tu, -i, -n). Synchronically, however, the Future Participle does not exist
as an independent nonfinite form, its only use being as a part of the analytic form we
are studying now.
The only use of the Basque Future we will examine in this work is that in
future time-reference subordinate clauses (excluding protases of real conditionals,
where it seldom appears or appears conveying a special sense: see Comrie 1982,
Bybee et al. 1994:208). In temporal subordinate clauses and in relative clauses with
future-gnomic time-reference, the Basque Future can appear more freely; but still
there is a tendency towards using the Periphrastic Present rather than the Future in
these contexts. (We will examine this competition in Chapter 4.)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
74
6. PERIPHERAL ANALYTIC FORMS OF THE INDICATIVE.
The peripheral analytic constructions of the indicative constitute an intricate topic,
whose complete understanding is beyond the scope of this work. Among other
complications that these constructions present, it turns out that there are actually very
few data in the database regarding them. In any case, these peripheral constructions
constitute a complex yet appealing object of study in an interface between syntax and
semantics, which also includes voice devices (i.e. active vs. passive and antipassive).
On the one hand, the semantic analysis of these constructions seems to be best
approached by taking them as one unit, just as any other kind of TAM marker. On
the other hand, although their syntactic analysis is not straightforward, the literature
has mostly considered them as biclausal constructions, consisting of a main verb and
a subordinate nonfinite clause, (while the main analytic constructions of the
indicative are always regarded as monoclausal, having only one verb which is made
up by the union of nonfmite form plus auxiliary).
These synchronic syntactic analyses are in agreement with a point I aim to
defend in this work; namely, that the peripheral constructions are less
grammaticalized than their corresponding main forms. However, given the primarily
diachronic orientation of this work, I will not necessarily endorse these or any other
synchronic analysis. On the other hand, the terminology used to describe the
peripheral constructions will point to one or another formal analysis. Thus, I will be
using the term ‘quasi-auxiliary’ to refer to the verbal item (e.g. ‘to be’, ‘to have’, etc)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
75
which would be analyzed as main verb in a biclausal analysis and as auxiliary in a
monoclausal analysis. In spite of this, the term ‘quasi-auxiliary’ should be
understood just as a compromise between the two possible analyses in synchrony.
What I do need to prove in this work is that the main constructions are ‘more
monoclausal’ than their corresponding peripheral constructions. In other words, that
the main constructions have higher rigidity and less relative mobility between their
components than the peripheral constructions. This task has already been outlined in
Chapter 1. As mentioned in section 5 above, I will examine these formal issues in
more depth in sections 2 of Chapters 4 and 6.
Although I have introduced six peripheral constructions of the indicative in
Tables 2.3 and 2.4 above, I will only study three of them in this section and
throughout this work. These are the Present Progressive (section 6.2) and the Present
and Past Resultatives (section 6.1). As a matter of fact, due to the cross-dialectal
variation and the different possibilities regarding voice these constructions show, we
will see more than one formal expression for each case. In any event, all the
peripheral constructions we will examine here have a stative quasi-auxiliary, which
in general terms can be said to correspond to either ‘to be at, to be temporarily, to be
engaged in doing something’ or ‘to have.’ Since there are already dialectal
differences in the uses of these stative quasi-auxiliaries, we will have a very
important deal of variation in the expression o f the peripheral constructions, as we
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
76
will see below. (There are further other types of peripheral constructions that I will
not mention here: see Euskaltzaindia 1987:466-473, 480-482; Lafitte 1962:350-352.)
6.1. Resultatives.
The different resultative constructions we find in the dialects of Basque are built by
means of the combination of the Past Participle (either with the article -a (sg) / -ak
(pi), or with the invariable endings -ta or -rik: see section 3.3 above) and a quasi
auxiliary of the temporary-sense ‘to be’ or ‘to have’: e.g .joana da ‘s/he is gone’,
egina du ‘s/he has it done.’ Thus, this type of construction is very similar in formal
expression to the Perfect-Perfective forms (e.g.joan da ‘s/he has gone’, egin du ‘s/he
has done it’). However, both the Past Participle in Resultatives and the quasi
auxiliaries present significant variation across dialects.
As we saw in section 3.3, in the Biscayan dialect, the Past Participle always
appears with the morphemes -ta or -rik in the Resultative constructions (and
elsewhere, except in the Perfect forms). The ending -ta is the most typical of
Biscayan. The Souletin dialect, in turn, always uses the Past Participle ending in -rik
for the Resultative constructions. The more standard Resultative, however, utilizes
the article -a / -ak in the Perfect Participle (Lafitte 1962:384, Euskaltzaindia
1987:476). As for the Resultative quasi-auxiliaries, these are egon ‘to be at’ and euki
‘to have’ in Biscayan; i.e. the only ones that can convey temporary stative meaning
in that area. In the central and eastern dialects, besides the former two, izan ‘to be’
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
77
and *edun ‘to have’ can also convey a temporary sense. Thus, in these latter
varieties, there are four possible quasi-auxiliaries for Resultatives, although izan and
*edun are the most prototypically used.
I will now present a brief formal description of the prototypical Resultative
constructions in Standard Basque, including the variants expressing different voice
functions. (Clearly, a deep study of these functions and of their formal expression is
beyond the scope of this work.)
6.1.1. Standard Basque Resultatives.
The prototypical Resultative constructions in Standard Basque utilize the Perfect
Participle with the article -a (sg) / -ak (pi) and the quasi-auxiliaries *edun ‘to have’
for ergative predicates and izan ‘to be’ for intransitive-absolutive predicates.
a) Transitive Resultatives:
(5a) hori entzun-a dut (ni-k)
that heard-DET I-have 1-ERG
Lit. “I have that heard”
(5b) zu-k esan aurretik, entzun-a nuen (nik)
you-ERG said before heard-DET I-had-(it) I-ERG
“I had (already) heard it, before you mentioned it”
b) Intransitive-absolutive Resultatives:
(6a) Paris-en egon-a naiz
Paris-LOC been-DET I-am
“I have been to Paris”
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
78
(6b) Paris-era joan-a nintzen ordu-ko
Paris-ALL gone-DET I-was by-then
“I had moved to Paris by that time”
c) Intransitive-ergative Resultatives:
As for the Basque intransitive verbs that usually have ergative subject (i.e.
intransitive-ergatives; “unergatives” in some formalist terminologies), it would seem
it is easier to find izan ‘to be’ than *edun ‘to have’ as the quasi-auxiliary of their
Resultative constructions. However, this observation is provisional at this point.
d) Antipassive Resultatives:
While the most normal behavior for transitive Resultatives is to use *edun ‘to have’
as the quasi-auxiliary, as we have seen in examples (5), it is also possible for
transitive Resultatives to present the quasi-auxiliary izan ‘to be’, apparently to
emphasize the result of the action in the subject, instead of in the object. This
construction, thus, is a sort of antipassive (see Rebuschi 1979:219-221 for the
Biscayan Antipassive Resultative, with the quasi-auxiliary egon ‘to be at’).
(7) Komeri edeix-ak ikusi-ak gera gazte denbora-n
comedy nice-DET,PL seen-DET,PL we-are young time-LOC
Literally: “We are seen beautiful comedies in our youth”
(Euskaltzaindia 1987:475)
A further formal possibility of Basque Resultative constructions is a sort of passive,
which I will not consider here: cf. Ortiz de Urbina & Uribe-Etxebarria (1991). See
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
79
Rebuschi (1979) for discussion, examples and details on the syntax of the three
Biscayan Resultatives for transitive verbs: i.e. transitive, passive and antipassive.
There is in any event much to research on the distribution of all these possible
Resultative constructions, regarding both cross-dialectal and lexical-semantic
variation. The following example by de Rijk, for instance, shows an antipassive
Present Resultative with the Perfect Participle ending in -(r)ik and a Present form of
the intransitive quasi-auxiliary egon.
(8) Honen-beste-z esan-ik nago
this-much-INST said-PART I-am-now
“I have already said what I wanted”
Literally: “With this much, I am said”
(de Rijk 1981:101)
The Standard Basque Resultative construction has number agreement in both the
Perfect Participle (unlike the construction with -ta and -rik endings) and the quasi
auxiliary, with the subject of intransitive sentences and with the object of transitive
ones. (Example (9) below shows also dative agreement in the quasi-auxiliary.)
(9) sagarr-ak amari eman-ak dizkiot
apple-DET,PL mother-DAT given-DET,PL I-have-them-to-her
“I have (already) given the apples to my mother”
(Euskaltzaindia 1987: 83)
(10) mutil-ak etorri-ak dira
boy-DET,PL come-DET,PL they-are
Lit. “The boys are come”
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
80
Regarding the meaning of the Resultative constructions, we find again many
uncertainties. It seems that the Resultative Present can convey the following
meanings: predicative-stative present, present resultative, perfect of result, and
experiential perfect. This construction, thus, will be in competition with its formally
similar Present Perfect-Perfective. The Past Resultative, in turn, seems to express the
following meanings: predicative-stative past, past resultative, and pluperfect (at least
perfect-in-the-past). This construction will be in competition with the Past Perfect-
Perfective. I will examine in depth these cases of competition in Chapter 6.
6.2. Present Progressive(s).
In a similar way to what we have seen for Resultatives, we also find cross-dialectal
differences in the formal expression of the Progressive constructions of Modem
Basque. The main distinction could be, as usual, roughly divided into western and
eastern patterns. More exactly, the Western Progressive construction can also be
found in eastern dialects, but the prototypical Eastern Progressive construction (the
one adopted in Modem Standard Basque) does not appear in Biscayan. Both,
prototypical-Eastem and prototypical-Westem Progressives, make use of the same
nonfinite form: the Imperfective Participle which ends in -tzen, -ten. The quasi
auxiliary, however, varies. Eastern dialects typically use the modal verb ari izan ‘to
be engaged in’ as quasi-auxiliary for progressive meaning. The Biscayan dialect, on
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
81
the other hand, does not have this modal verb, and uses the common quasi-auxiliary
egon ‘to be at, to be temporarily.’
Beyond these differences, there is one formal similarity that should be
pointed out. Unlike all other analytic forms of Basque we have seen so far (both
belonging to the core of the Basque verbal system or to the peripheral group of
constructions), the most frequent Progressive constructions do not present alternation
in the quasi-auxiliary. The quasi-auxiliary is always an intransitive-absolutive verb
(but see section 6.2.3 below), and there is no transitive counterpart.1 1 Thus,
Progressives are generally the only constructions of Basque with neither auxiliary
selection nor subject distinction (see however note 11). This is the reason why the
transitive Progressive has been sometimes considered a kind of antipassive
construction of Basque (cf. Heath 1981:440-441).
6.2.1. ‘ari ’ Progressives.
As just mentioned, Eastern dialects (including Guipuzcoan this time) and Standard
Basque, use the intransitive modal verb ari izan ‘to be engaged in’ as the most
common quasi-auxiliary for progressive meaning, for both intransitive and transitive
verbs. Likewise, the subject is always in the absolutive case.
1 1 The verb ari can appear sometimes with the ergative auxiliary *edun, hut this use is very little
widespread cross-dialectally. It seems almost restricted to weather expressions, such as: euria egiten
ari da / euria egiten ari du ‘it is raining.’ Among the XIX century Labourdin authors, however,
apparently unlike in earlier and later stages, the use of ari with selection of auxiliaries (izan for non
ergative agreement and *edun for ergative agreement) became rather common (cf. Pikabea 1993:74-
76). This is an issue that needs further research.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
82
(11a) Peru etxe-ra itzultzen ari da
Peter house-ALL returning he-is-engaged-in
“Peter is returning home”
(lib ) Peru liburu-a irakurtzen ari da
Peter book-DET reading he-is-engaged-in
“Peter is reading the book”
In the eastern dialects (though seemingly this used to be a phenomenon common to
all dialects in Old Basque: cf. Lakarra 1996:191-192), what can be analyzed as the
direct object of the progressive transitive sentence, viewed as a whole unit, appears
in the genitive case (cf. Heath 1972). It usually precedes the nonfinite form in -tzen
(which is etymologically a verbal noun in the locative case: see 3.1 above and
example (1) presented there).
(12a) liburu-a-0 irakurtzen ari da
book-DET-ABS reading s/he-is-engaged-in
“S/he is reading the book” (Guipuzcoan, Navarrese)
(12b) liburu-a-ren irakurtzen ari da
book-DET-GEN reading s/he-is-engaged-in
“S/he is reading the book” (French-Basque dialects)
6.2.2. ‘egon ’ Progressives.
The Biscayan dialect does not have the construction with ari izan. In its place, the
Progressive is usually made by means of the combination of the Imperfective
Participle (ending in -tzen) with a form of the verb egon ‘to be temporarily’; i.e. the
same quasi-auxiliary that we saw for the Biscayan Resultative constructions.
Actually, the verb egon is used in all dialects in combination with the -tzen nonfinite
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
83
form to convey the sense of progressive. The frequency of use of this construction
(and perhaps the degree of grammaticalization), however, is higher in Biscayan.
The egon Progressive does not show auxiliary selection and its subject
appears always in the absolutive case, just like in the case of the prototypical
Progressive of Standard Basque (i.e. as in the ari Progressive).
(13) Peru liburu-a irakurtzen dago
Peter book-DET reading he-is-at
“Peter is reading the book”
6.2.3. Other Progressive constructions.
Along with the egon quasi-auxiliary, there are some other verbs that also combine
with the Imperfective Participle to yield progressive meaning. The Biscayan dialect
is the one that has more of these quasi-auxiliaries, such as, eragon ‘pursue; formal
causative of egon', eutsi ‘grasp, hold on, pursue’, ibili ‘walk, wander, be doing
something’, erabili ‘use, handle; formal causative of ibili.' The first two, confined to
the Biscayan dialect and tending to be lost, are ergative-agreement verbs. The verb
ibili is common in Basque and intransitive. Its formal causative, thus transitive and
with ergative subject, is mostly used in Biscayan with this progressive reading. (Cf.
Euskaltzaindia 1987:440-443.)
The most frequently cited ergative construction with Progressive meaning is
built upon the ergative-intransitive verb ihardun ‘to be engaged in.’ This verb is very
similar in meaning to ari izan, but formally it is subject to synthetic inflection and
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
8 4
has always an ergative subject. Thus, (nik) egiten dihardut ‘1 am doing’ is parallel in
meaning to (ni) egiten ari naiz, but has always a subject in the ergative case. (Cf.
Saltarelli 1988:229.) This construction, however, is not common across dialects, and
does not appear in the database in this work. Therefore, I will not pursue its study
any further.
7. SUBJUNCTIVE FORMS.
The Modem Basque Present Subjunctive and Past Subjunctive are construed by the
combination of the Radical (ending in -0) and, respectively, Synthetic Present and
Synthetic Past forms of the dynamic auxiliaries *edin or *ezan, depending on
whether the subject is in absolutive or ergative case. Modem Biscayan (unlike Old
Biscayan) does not have the ergative auxiliary *ezan, and uses in its place egin ‘to
do.’ The auxiliary egin, instead of *ezan, is also used in the Guipuzcoan dialect,
especially in ditransitive constructions (see Irizar 1991, Lakarra 1996:181-185, for
the distribution of egin and *ezan). Also, Modem Western dialects do not use the
Radical but the Past Participle as the nonfinite form in the Subjunctive constructions.
(The Radical and the dynamic-achievement auxiliaries just presented will be studied
in more detail, from a grammaticalization perspective, in Chapters 3 and 5.)
The Present Subjunctive and the Past Subjunctive only appear in Modem
Basque in subordinate contexts. In fact, they are practically confined to purpose
clauses and complement clauses of verbs expressing wanting and ordering. In these
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
85
contexts, they always bear the subordinator or complementizer ~(e)n, formally
identical to that appearing in relative clauses, temporal clauses, and other subordinate
clauses of Basque.
(14) etor zaitez-en nahi dut
come(RAD) you-*edmAUX-COMP desire I-have
“I want you to come”
Lit. “I want that y o u co m e (SUBJUNCTIVE)”
(Saltarelli 1988:237)
(15) Patxi-k ikus deza-n ekarri dut liburu-a
Patxi-ERG see(RAD) he-*ezcwAUX-COMP brought I-have book-DET
“I have brought the book so that Patxi sees it”
(Euskaltzaindia 1987:422)
8. SUMMARY OF THE FORMS STUDIED IN THIS WORK.
I present below a chart showing the most important constructions of Modem Basque
that will be object of study in this work. I use six different lexical verbs to illustrate
some of the formal and syntactic differences that may appear in these constructions.
The verbs izan ‘to be’ and *edun ‘to have’ are, besides independent verbs, the most
important auxiliaries of Basque. The verb Joan ‘to go’ illustrates the -n class of verbs
subject to synthetic inflection; it also illustrates the intransitive-absolutive pattern.
The verb ekarri ‘to bring’ illustrates the -i class of verbs subject to synthetic
inflection, and the transitive (ergative) pattern. The verbs sartu ‘to enter’ and hartu
‘to take’ belong to the -tu class of verbs, and therefore do not have any synthetic
forms; the former illustrates the intransitive-absolutive paradigm and the latter the
transitive (ergative). Intransitive-ergative verbs are avoided in the chart due to the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
86
existence of cross-dialectal variation in the type of auxiliary (‘to be’ vs. ‘to have’,
etc) they use (see note 2). All forms in the chart correspond to the first person
singular subject (i.e. 1s t sg absolutive for intransitive verbs and 1st sg ergative for
transitive verbs). They are all forms of Modem Standard Basque.
TABLE 2.14: Formal expression of the constructions studied in this work.
IZAN
‘to be’
*EDUN
‘to have’
JOAN
‘to go’
EKARRI
‘to bring’
SARTU
‘to enter’
HARTU
‘to take’
Synthetic
Present
naiz dut noa dakart
Periphrastic
Present
izaten
naiz
izaten dut joaten
naiz
ekartzen dut sartzen
naiz
hartzen
dut
Present
Progressive
izaten ari
naiz
izaten ari
naiz
joaten ari
naiz
ekartzen ari
naiz
sartzen ari
naiz
hartzen
ari naiz
(Periphrastic)
Future
izango
naiz
izango dut joango
naiz
ekarriko dut sartuko
naiz
hartuko
dut
Synthetic
Past
nintzen nuen nindoan nekarren
Present
Perfect
izan naiz izan dut joan naiz ekarri dut sartu naiz hartu dut
Double-comp.
Pres. Perfect
joan izan
naiz
ekarri izan
dut
sartu izan
naiz
hartu izan
dut
Past
Perfect(ive)
izan
nintzen
izan nuen joan
nintzen
ekarri nuen sartu
nintzen
hartu
nuen
Double-comp.
Past Perfect
joan izan
nintzen
ekarri izan
nuen
sartu izan
nintzen
hartu izan
nuen
Present
Resultative
izana
naiz
joana naiz ekarria dut sartua naiz hartua dut
Past
Resultative
izana
nintzen
joana
nintzen
ekarria nuen sartua
nintzen
hartua
nuen
Present
Subjunctive
izan
nadi-n
izan
dezada-n
joan
nadi-n
ekar
dezada-n
sar nadi-n har
dezada-n
Past
Subjunctive
izan
nendin
izan nezan joan
nendin
ekar nezan sar nendin har nezan
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
8 7
CHAPTER 3
THE GRAMMATICALIZATION OF PRESENT
ACROSS LANGUAGES AND IN BASQUE
1. INTRODUCTION.
This chapter and the next are devoted to the TAM sphere of present. (As we will see
below, we should perhaps more appropriately talk about ‘non-past’ instead of
‘present.’ See also the Glossary for clarification on the TAM terminology used in
these chapters). This chapter examines the sources and clines of grammaticalization
of non-past across languages, and the grammaticalization sources of the Basque
constructions in this domain of TAM. Chapter 4, in turn, studies, based on actual
data, the domain of non-past in historical Basque. In the present chapter, I will
present and justify unidirectional clines of grammaticalization in the TAM sphere at
issue (hypothesis of Semantic Unidirectionality). These clines of grammaticalization
are determined by the source of grammaticalization which initiates them (hypothesis
of Source Determination). Since I will ascertain here the grammaticalization source
of most of the Basque constructions under consideration, we will be able to verify in
the next chapter the hypotheses of Source Determination and Semantic
Unidirectionality for the Basque case. More importantly, due to the fact that some of
the Basque constructions have the same source of grammaticalization, we will be
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
88
able to establish and test, also in Chapter 4, the hypotheses of Layering-
Displacement and Form-Meaning Covariation against the Basque data.
The presentation of this chapter is as follows. Sections 2 and 3 are devoted to
presenting and justifying clines of grammaticalization in the sphere of non-past,
especially those most relevant for explaining the Basque data. Section 2 presents the
cline from present progressive to present habitual; section 3, the path from present
habitual and gnomic to future time-reference subordinate contexts. The two clines in
these sections can be actually viewed as partly progressing in sequence; i.e. they can
be considered intervals or sub-clines of one and the same general cline: the cline
starting with present progressive. In section 4 ,1 order diachronically the semantic
subtypes in the path starting with present progressive. I also put forward the
possibility of competition with other types of sources. Section 5 is devoted to
examining one of these non-progressive competing sources: namely, present telic
sources. Section 6, finally, examines in detail the formal expression of the Basque
constructions in the domain of non-past, in order to ascertain (when possible) their
source of grammaticalization.
2. FROM PRESENT PROGRESSIVE TO PRESENT HABITUAL AND
GENERIC.
The diachronic cline of evolution I will present next starts with present progressive
and proceeds to present habitual first, and generic later. As present progressive is
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
8 9
restricted to Vendler’s classes of accomplishment and activity predicates, the
beginning of the cline is incompatible (to different extents) with states and
achievements. However, as the evolution progresses, the TAM meanings reached
become more abstract and thus the constructions at hand acquire wider generality of
use, combining with practically all classes of verbs. We will see below that this cline
of evolution progresses via a chain of associations that lead from multiplicity to unity
(see Kemmer 1996 for a nominal case-marking parallel). It can also be said that the
evolution is an abstraction from the stage-level to the individual-level (and then
from the level of objects to the level of kinds).
I put forward in the schema in (1) below the path of grammaticalization we
want to examine. (Explaining the reasons for the directionality of the cline in (1) is
the goal of the following discussion.)
(1) present progressive -> present iterative present frequentative present
habitual gnomic / generic
The cline in (1) is well known in the literature on grammaticalization. “Cases of such
development have been proposed in the literature on the basis of historical and
comparative evidence” (Bybee et al. 1994:141). We have, for instance, the examples
of Yoruba, Scots Gaelic, Turkish, Punjabi and Hindi-Urdu, the Chamus dialect of
Maa (Eastern Nilotic), and even partly the case of English (cf. Bybee & Dahl
1989:82-83).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
90
The cline at issue starts with present progressive. As a matter of fact, we are
mostly interested in present progressive expressed by means of a locative
construction. Locative-progressive is not only the source of grammaticalization of
most of the Basque constructions we will be examining in section 6 but also the most
widespread source of grammaticalization of progressives across languages. Location
seems to be cognitively a very accessible metaphor for expressing dynamic ongoing
meaning. This is one instance of the common associations transposing from space to
time (cf. Comrie 1976:103; Bybee et al. 1994:137). Furthermore, present locative-
progressive is, cross-linguistically, the most important initiator of the cline of present
tense. As Bybee et al. (1994:127) put it, “we view progressive meaning as the
specific meaning that feeds into the chain of developments leading eventually to the
highly generalized imperfective or present meanings.”
The most restricted present progressive meaning would prototypically have
the following characteristics: ongoing, dynamic, internally complex, incomplete,
durative, temporary (see the entry ‘present progressive’ in the Glossary for
clarification). A prototypical example would be: John is working now.
The first step in the cline in (1) is the extension of restricted present
progressive to iterative meaning. Since progressive requires a constant input of
energy, and since many of the dynamic situations that are commonly used in
progressive contexts are not continuous but consist of separate instances, it seems
that the extension of meaning from restricted progressive to iterative is a fairly
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
91
natural one. The iterative reading is even more natural for progressives of inherently
punctual situations (cf. Comrie 1976:42): e.g. he is coughing. (The present
progressive of punctual achievement situations can also have the sense of a
preparatory phase leading to an end-point: e.g. he is reaching the summit, cf. Comrie
1976:47-48)
An inference from iterative meaning, and ultimately from the dynamic nature
of progressives, is what might be called insistence or effort. Depending on the
inherent nature of the situation, iteratives are sometimes understood as entailing
some sort of resistance that has to be overcome by repeated attempts, as in Span, el
coche no esta arrancando. From this reading, it appears to be a small step to
continuative meaning, which “specifies that the agent of the action is deliberately
keeping the action going” (Bybee et al. 1994:127).
A related reading might be termed gradual or cumulative, which conveys not
only that the situation is in iterative progress but also that its result is being
accumulated over time. It might as well express a “development by degrees, as in it’ s
getting dark” (Bybee et al. 1994:140). Verbs that have restrictions with the common
progressive reading, such as some stative verbs (understood here as achievements),
might naturally express cumulative meaning when used in a Progressive
construction: e.g. Span. Me esta. gustando mas cada vez ‘I am liking it more every
time’; I ’ m understanding more about quantum mechanics as each day goes by
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
92
(Comrie 1976:36). In any case, iterative, continuative and cumulative are meanings
that will mostly appear restricted to concrete lexical items.
The second step in the schema in (1) progresses to (recent-contingent-
temporary) present frequentative. Iteratives indicate that “the action is repeated on
the same occasion” (Bybee et al. 1994:159), which is ongoing at the reference time.
Now, if the concept of ‘repeated on the ongoing occasion’ is extended to cover
‘repeated on the ongoing (i.e. the most recent) span of time’ we arrive at a
frequentative. The directionality of the cline starts to become clearer at this point: the
cline evolves first towards acquiring wider temporal reference. That is, the restriction
of ongoingness is fading away, i.e. the link with the most immediate present is
getting lost. Playing perhaps the role of beachheads in this evolution, we find here
again that lexical verbs having incompatibilities with strict progressive meaning,
such as some changes and achievements, might naturally express a frequentative
reading when used in a Progressive construction: e.g. we ’ re going to the opera a lot
these days (Comrie 1976:37), Span, estoyyendo mucho al cine (ultimamente).
Once a one-time present progressive has reached the meaning of present
frequentative, there is just a small step to generalizing to present habitual. From the
evolution we have outlined so far, we see that the concept of repetition (including
insistence, effort, continuation, accumulation, gradualism, frequency, etc.) can be
taken as the most important conceptual beachhead that leads from prototypical
progressives to prototypical habituals.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
93
We arrive to present habitual (third step in the schema in (1) above) when
mere repetition of stages is abstracted and understood as characteristic behavior of
the individual to whom the stages belong. At this point, the original nature of a one
time progressive construction is importantly blurred. Concretely, the progressive
features of being ongoing, temporary and, to some extent, dynamic no longer hold in
the habitual meaning. Habitual is characteristic of an extended period, and can be
applied for wholly characterizing or describing the entity referred to by the subject,
as in Peter smokes.
Regarding the degree of abstraction, we can perceive a change in the cline in
(1) which finally evolves from multiplicity to unity. That is, first, internal phases
lead to gradualism and repetition in the same occasion or stage; this is further
extended in temporal reference to repetition in multiple occasions or stages. Second,
the repetition of stages becomes understood as one unit, and applied in another level
of abstraction to habitual or characteristic of an individual; habitual is thus taken as
descriptive.
At this point of the evolution, present habitual is not necessarily dynamic any
longer. Moreover, in its characteristic reading, present habitual can already be
applied not only to habits but to all classes of verbs, including perhaps permanent
states. Thus, present habitual conveys a sense of repetition only in the case of
accomplishments (e.g. she writes one book every year) and achievements (e.g. they
win the tournament every season). For the cases of activities (e.g. she plays
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
94
basketball very well, he runs very fast, driver carries only $5.00 change') and
permanent states (e.g. she lives in Miami), there is not necessarily such internal
division. More temporary states, however, can also have a frequentative reading (e.g.
he always sits in the same chair). Nevertheless, the case of permanent stative
predicates, such as to live or better to know, seems problematic. (For discussion and
more details, see the entries ‘present habitual’, ‘characteristic behavior’ and ‘stative
present’ in the Glossary). Notice, in any event, that, the evolution from restricted
present progressive to present habitual entails an increase in generality of use: from
being only applicable to accomplishments and activities, towards general
appropriateness.
Finally, according to the last step in (1) above, present habitual can be
generalized to cover gnomic or generic. Apparently, there is a small difference
between present habitual and gnomic. However, the evolution from habitual to
gnomic involves a relevant increase in the degree of abstraction: from the level of
objects to the level of kinds. This amounts to a descent in the degree of assertion,
because referring to the level of kinds implies an empirical prediction. The change
from habitual to gnomic also implies an important divergence in time reference:
generics are ‘timeless.’ Regarding temporal reference, then, the whole of the cline in
(1) can be understood as an evolution towards an extension or ‘relaxation’ of the
most immediate present: i.e. from ongoing and temporary to more permanent time
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
95
reference, and from the latter to timeless reference. (See section 3.1 below for more
discussion on the change from habitual to generic in subordinate clauses.)
3. FROM PRESENT HABITUAL AND GENERIC TO FUTURE TIME-
REFERENCE SUBORDINATE CONTEXTS.
In this section, I study the interval of cline that progresses from present habitual and
gnomic to future time-reference subordinate contexts. I will divide this evolution into
two different parts. These will involve the changes towards open conditionals,
temporal clauses, generic relative clauses, etc, on the one hand, and towards purpose
clauses and complement clauses of ordering verbs, on the other. Although Bybee et
al. (1994:231) claim that “the subordinate clause uses -purpose, protasis,
complement to ‘be necessary’, and adverbial temporal clause- all present situations
viewed, not as in progress even at some future time, but more as bounded entities”,
there seem to be differences among these contexts regarding their degree of
boundedness. Thus, the first group of contexts I have considered above seems less
bounded and less dynamic than the latter. I will dub the former group as ‘less-
dynamic subordinate future time-reference contexts’ (section 3.1), and will refer to
the latter as ‘dynamic subordinate future time-reference contexts’ (section 3.2). (See
the Glossary for clarification.)
The first evolution I will consider, i.e. that involving less-dynamic contexts
(section 3.1), will be examined in some depth. On the other hand, the second
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
96
evolution, i.e. that toward purpose clauses, etc (section 3.2), is hard to understand,
and I will just present some remarks on these contexts.
3.1. From present habitual to less-dynamic subordinate future contexts.
The interval of cline I want to examine in this section evolves from habitual (in
subordinate clauses) to non-assertive future time-reference subordinate contexts such
as if-clauses, when-clauses, and generic relative clauses. I cannot examine in depth
here all the complexity of if-clauses and other similar modality contexts. Moreover,
conditional clauses often behave, regarding TAM distinctions, in a different way to
future temporal clauses and relative clauses. What follows thus is a general account
that may be valid for most of the contexts at hand.
I propose in (2) below that the first step in this sub-cline is a change from
habitual to generic subordinate contexts. Then a subsequent change progresses from
generic time-reference to future time-reference (i.e. from unbounded and non
specific to bounded and specific at one point in the future, even though hypothetical).
(2) subordinate habitual -> subordinate generic subordinate future
The first step in the cline of (2) is completely parallel to the change from present
habitual to generic or gnomic, which we examined in section 2. The only difference
is that in this case we are referring to subordinate contexts. The change from habitual
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
97
to generic implies a generalization from the level of objects to the level of kinds.
However, since kinds, unlike objects, are in principle timeless, this generalization
entails also a change in time reference. That is, from a present which is characteristic
of a real-world habit to a present + future, applying to any possible world. An
example illustrating this change could be the following: when winter comes, my
friend Peter (alwaysj hibernates when winter comes, bears hibernate.
The second step in (2), i.e. from subordinate generic to subordinate future,
seems to involve a change in a direction opposite to generalization, i.e. towards more
specific reference; although on the other hand it progresses towards a lower degree
of assertion, due to the predictive nature of future time reference. This change thus
appears to entail a semantic narrowing of the type: Vx x, i.e. ‘if the generic, then
the specific.’ As an illustration of this change, we could give the following example:
when winter comes, bears hibernate -> when next winter comes, the bear in the zoo
will hibernate.
The two steps just proposed could also be illustrated by means of if-clauses,
as in the following examples: i f / whenever Robert plays with us, we lose (habitual)
i f Robert plays with us, we lose / will lose (generic, non-specific) i f Robert
plays with us tomorrow, we will lose (future, definite). (For more discussion on the
relationship between generics and habituals, on the one hand, and conditionals and
temporal clauses, on the other, see for instance G. Carlson 1979 and Langacker
1997.)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
98
In sum, unlike the changes we saw in the cline in (1) above (section 2) which
finally evolved from multiplicity to unity and always towards more inclusive (wider)
time reference, the path in (2) seems to ultimately progress towards less inclusive
(narrower) time reference, and from generality and unity to specificity, singularity
and punctuality (though with future time reference).
3.2. Dynamic subordinate future time-reference contexts.
As mentioned above, it is not totally clear how a present habitual or generic can
extend its meaning to cover dynamic subordinate future time-reference contexts such
as complement clauses to verbs of ordering and wanting, or purpose clauses. It
would seem, though, that this evolution occurs only after the passage from habitual
and generic to the less-dynamic contexts studied in 3.1 has taken place. In other
words, it appears that once a Present Habitual has generalized and bleached its
semantic content to cover if-clauses, when-clauses, until-clauses, etc, it can further
generalize to convey purpose clauses and complement clauses to ordering and
wanting verbs. It might be the case that until-clauses can function as a beachhead
between these two sub-groups of subordinate contexts. The complete evolution I
propose is schematized in (2’) below.
(2’) subordinate present habitual -> subordinate generic subordinate less-
dynamic future (if-clauses, when-clauses) -> (until-clauses) -> subordinate
dynamic future (“subjunctive”)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
99
The contexts of purpose clauses and complement clauses of ordering and wanting are
often called ‘subjunctive mood.’ These contexts are non-assertive but specific and
bounded. They are also dynamic and telic, i.e. they convey the idea of “attainment of
a limit” (Dahl 1985:76) applied to the future. (For the difference between the
concepts of attainment of a limit in non-past (present -> future) vs. past, see note 2
below.) Purpose clauses are often constructed by means of purposive or allative
expressions. This type of formal expression points again to the idea of dynamicity
and telicity.
4. DIACHRONIC ORDERING OF THE CLINE STARTING WITH
PRESENT PROGRESSIVE AND POSSIBLE COMPETITION WITH OTHER
TYPES OF SOURCES.
To recapitulate, so far we have seen two paths of grammaticalization (sections 2 and
3) that I repeat below for convenience.
(1) present progressive -> present iterative -> present frequentative -> present
habitual gnomic / generic
(2’) subordinate present habitual -> subordinate generic subordinate less-
dynamic future (if-clauses, when-clauses) -> (until-clauses) -> subordinate
dynamic future (“subjunctive”)
The cline in (1) evolves from present progressive to present habitual and gnomic.
The cline in (2’) regards subordinate contexts, and progresses from subordinate
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1 0 0
present habitual to different subordinate future time-reference contexts. It follows
from the above schemas that the cline in (2’) is actually a continuation of the cline in
(1). Therefore, we may consider that the clines above are just intervals of a complete
path of evolution starting with present progressive.
We may propose then a diachronic ordering of the whole cline starting with
present progressive, as in the schema in Table 3.1.
TABLE 3.1: Semantic sub-types in the cline starting with present progressive.
Young
sub-types
Mature
sub-types
Old subordinate
sub-types (future time-
reference)
progressive
iterative
gradual
frequentative
habitual
gnomic
(characteristic)
if-clauses
(i.e. open conditionals)
concessive clauses
when-clauses
relative clauses
until-clauses
permanent stative
present temporary stative
present
complement clauses to
ordering & wanting verbs
purpose clauses
Table 3.1 gives a chronologically ordered relation of some of the most important
contexts or sub-types that a construction grammaticalized as present progressive may
eventually get to cover as it evolves with time. It is important to notice that a form
once grammaticalized as a present progressive might also reach other meanings not
appearing in this table. These are either punctual-like or future-like main-clause
contexts such as performative sentences, simultaneous reports, historical present,
immediate future, and scheduled future, or even prediction-future and imperative.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
101
Table 3.1, therefore, shows the subtypes corresponding to what might be cross-
linguistically the most prototypical developments of a form with present progressive
origin. The developments in Table 3.1 are, in any case, those I will study for the case
of Basque. Conversely, not all the meanings appearing in this schema need to be
always expressed by an originally present progressive construction.
Thus, we can expect to find that constructions grammaticalized as present
progressive experience the competition of other forms with non-progressive sources
of grammaticalization. The major competition that we could expect to find comes
probably from constructions grammaticalized as young futures (cf. Bybee et al.
1994:279). A construction grammaticalized as a young future can compete with a
form of progressive-source for some of the mature subtypes (e.g. gnomic) and some
of the old subordinate subtypes (e.g. when-future clauses, generic relative clauses).
Other possible competitors of constructions with a progressive source are forms
grammaticalized directly with habitual meaning (see Bybee et al. 1994:154-158). Yet
another possible type of competitor comes from constructions grammaticalized as
Present Perfective or built upon present dynamic telic auxiliaries. This latter type of
construction is perhaps not very common cross-linguistically, but it appears in the
Basque case. Thus, I will study this kind of source of grammaticalization in section 5
below. We should also call attention to the case o f stative verbs, which may present a
type of predicate phrase different from that of dynamic verbs (see section 6.5 below).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
102
Now, even in the hypothetical case of a language that would express all the
meanings in Table 3.1 by means of constructions grammaticalized as present
progressive, it need not be one and the same form that conveys all of the meanings.
As we saw in Chapter 1, the renewal of the system may produce layers of
constructions with very similar sources of grammaticalization (e.g. locative
expressions), which evolve through a very similar path: concretely, that outlined in
Table 3.1. As Hopper & Traugott (1993:106) put it, “it is both difficult and
unnecessary to illustrate the whole of any one cline with a single form.”
There might be languages, though, which express all of the meanings in
Table 3.1 by way of one and the same construction, although expressive and
differentiating motivations would probably make this case rather uncommon. It
seems more expected to find across languages two or three constructions that would
cover the semantic sub-types presented in Table 3.1. Probably a common case would
be that of a language with three constructions to cover all these meanings. It would
not be uncommon to find (if possible to reconstruct) that all these constructions were
grammaticalized as present progressives. This is partly the Basque case we aim to
examine, (although a fourth construction with non-progressive origin competes in
historical Basque for the old subordinate subtypes). (For the source of
grammaticalization of the Basque constructions at issue see section 6 below.)
In the case of layers of constructions with the same progressive source, we
expect (‘predict’) that the constructions gradually evolve from left to right in the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
103
schema of Table 3.1. Thus, the oldest construction would appear in the right-most
sub-types and the youngest in the left-most.
The cross-linguistically most common ways of marking the meanings in
Table 3.1 can be found in Bybee et al. (1994). They report the following possibilities
of marking: 1) “It is possible to find [constructions] of ‘present tense’ which express
all [the following] meanings: progressive, habitual, [stative present], and gnomic”
(1994:141). 2) Another possibility is that of the “many cases of explicit markers of
progressive in the present [...] distinct from habitual/generic/stative meaning”
(1994:152). 3) A third possibility is illustrated by Kui, where we find a younger
Present Progressive opposing “the older simple Present [which expresses] habitual
and future” (1994:158). 4) The last possibility I will mention is illustrated by
Armenian, where a periphrastic form conveys present progressive and present
habitual, while an older construction is used with future and non-indicative
subordinate functions, such as purpose clauses, protases of reality conditions, until-
clauses, etc (1994:231).
5. PRESENT TELIC SOURCES.
As mentioned in the previous section, at least one of the constructions of Basque that
compete for the semantic subtypes in Table 3.1 was not grammaticalized as present
progressive. I am concretely referring to the (Periphrastic) Present Subjunctive,
which is constructed with dynamic achievement auxiliaries in present tense (for more
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
104
details on the source of grammaticalization of this form, see section 6.4 below). In
this section, I present a revision of the literature with the purpose of finding out what
meanings and evolutions are reported across languages for constructions with present
telic sources.
Unfortunately, this kind of construction is not very commonly documented.
Dynamic verb sources of grammaticalization are cross-linguistically common.
However, these auxiliaries are much more frequent in past tense than in present (cf.
Bybee et al. 1994:69-74). This is actually something we could expect from a source
of grammaticalization of this kind. Movement sources are also encountered for
present progressive meaning (Bybee et al. 1994:132-134). But these are cases which
contain a locative (and thus ongoing) element. What we are looking for here is telic
punctual-like constructions grammaticalized in present tense.
Heine (1993:35) considers a type of grammaticalization source that perfectly
fits with what we are in search of. He calls it “the Change-of-State Schema.”
Although he claims that this source of grammaticalization is “rarely made use of to
develop tense or aspect markers”, he provides “an instance of it”: “the German
werden [‘become’] construction, Hans wird Arzt ‘Hans is going to be a doctor’,
which has been grammaticalized to a future tense category: Hans wird kommen
‘Hans is going to come.’”
Bybee et al. (1994:190-191) report also a “group of auxiliary verbs” that is
formally very close to the Basque case, and which “comes to signal ability.” These
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
105
are “dynamic, telic verbs of various sorts”, such as ‘arrive at’ (in Guaymi and Lahu),
‘get, obtain’ (in Lahu), ‘take’ (in Uigur), ‘become’ (in Lahu: 1994:188), or
constructions such as the particle ‘finished’ (in Worora), the Present Punctiliar
Subjunctive (in Alawa), etc. Similarly, Bybee et al. (1994:206) also report a
“construction in Tigre involving ‘to become’ and an adposition meaning ‘if , giving
probability.” Also, the Baluchi Continuative construction, which is built upon “an
inflected form of the verb bm g ‘be, become’”, has a future use “characterized as
expected future progressive” (1994:250): i.e. a scheduled future.
To these cases of constructions seemingly grammaticalized with present telic
auxiliaries, we could add the Perfective Present derivational forms, extremely
common in languages such as those of the Slavic family. These Present Perfective
forms from “bounders” (Bybee & Dahl 1989:86; Bybee et al. 1994:87) are after all
quite similar to telic presents, because both types of constructions combine present
tense with the “attainment of a limit” (Dahl 1985:76). The most important meanings
conveyed by Perfective Presents are: future, subordinate clauses of time, narrative
present, definite frequentative-habitual, imperative, etc. (Comrie 1976:66-70;
Monnesland 1984:53-66).
In sum, constructions grammaticalized as dynamic-achievements are not very
common in the domain of present tense. This appears easy to explain, because,
unlike future and especially past, there is hardly room in present for a punctual-like
meaning indicating the attainment of a limit. In any case, the constructions
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
106
grammaticalized in this way, evolve towards future or gnomic time-reference
meanings, such as the ones we have just seen documented in the literature: future,
ability, probability, temporal subordinate clauses, imperative, etc.1 We could
distinguish two kinds of contexts among those which constructions with present telic
sources are suitable to cover: a) the uses in main clauses (scheduled future, future,
ability, probability, imperative), and b) the future-gnomic time-reference subordinate
contexts.
(3) present telic sources -> a. future time-reference in main clauses
b. future time-reference in subordinate clauses
The subordinate contexts reached in the evolution in (3b) roughly coincide with
those studied in section 3. We saw there that these contexts can also be achieved by a
construction grammaticalized as present progressive, (perhaps more readily when it
combines with achievement predicates). Thus, we find parallel evolutions between,
on the one hand, a construction grammaticalized as present progressive when
combined with achievement predicates or in bounded contexts, and on the other
hand, a construction built upon telic (achievement) auxiliaries when grammaticalized
in present tense. (A third parallel evolution is found in constructions
11 avoid the label “present perfective” as a conceptual meaning. I find it to be a bit vague: while “past
perfective” is a well-delimited meaning, “present perfective” is not. Furthermore, the use of the term
“perfective” for both non-past and past, based on the Slavic source of grammaticalization, has led to
some confusion. Thus, many times one does not know for sure if a construction labeled “Perfective”
in the literature bears also a Present marker, or a Past marker, or none (cf. Dahl 1984a:7-8, 20-21;
Dahl 1985:82; Bybee & Dahl 1989:83; Bybee et al. 1994:83). I will therefore keep the term
“perfective” for perfective past. 1 do employ, however, the name “Present Perfective” (with initial
capital) for the specific construction in Slavic and other languages (see the Glossary for clarification).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
107
grammaticalized in present tense with completion adverbs: i.e. the Perfective
Presents of Slavic languages; see note 2.)
All of these sources of grammaticalization seem to share a combination
between present tense and punctuality (i.e. specificity, boundedness, definiteness, or
moreover the “attainment of a limit”2 or end-point). This combination thus appears
as the reason for their evolution towards future time-reference senses.
6. SOURCES OF GRAMMATICALIZATION OF THE BASQUE
CONSTRUCTIONS IN THE DOMAIN OF NON-PAST.
6.1. The Synthetic Present.
In Chapter 4 ,1 will try to prove that the Synthetic Present of Basque (e.g. dator
‘he/she/it comes’; dakar ‘he/she/it brings him/her/it) is a very old form in an
advanced process of recession, and that it has remained in the modem language
mostly in a handful of high-frequency verbs (many of them stative verbs). The
Synthetic Present has a highly bound morphology, which is often opaque regarding
the morphemes involved. For this reason, it is a difficult task to put forward a
hypothesis of the source of grammaticalization of this construction.
We can mention, nonetheless, the two following hypotheses in the literature
for the origin of the Synthetic Present. A first hypothesis argues that the Synthetic
2 As for the idea of “attainment of a limit”, as well as the formal ways of expressing this concept (e.g.
Perfective markers), it is not clear that they represent the same TAM conceptual meaning in past and
in non-past. While in past these markers appear to signal a sense of ‘completedness’, in non-past they
would seem to convey a ‘towards’ reading. (See also section 4.3 of Chapter 5.)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
108
Present (e.g. dator) may have had its origin as a periphrasis *da+tor consisting of
some sort of nonfinite form (e.g. *-tor-) and the form *da, which could correspond to
Modem Basque da ‘he/she/it is’ (Trask 1977:204-205). A second hypothesis claims
that the *da morpheme could have meant “nunc”, i.e. “present moment”, in early
stages of Pre-Basque (de Rijk 1995:308-309).3
In principle, I do not find any major counter-evidence for either one of the
etymologies presented above. It could be interesting, in any event, to see if we can
relate the origin of the Synthetic Present forms like dator with a locative-progressive
construction. Trask’s hypothesis, of which I feel perhaps more supportive, does not
explicitly propose any locative source for dator, but only an analytic origin coming
from the combination of nonfinite form and to be. However, it might be the case that
the copula of Modem Basque (e.g. da ‘he/she/it is) comes originally from a
locational or postural verb.
Thus, Bybee et al. (1994:130-131) report, besides “the twenty-one cases of
progressives derived using a locative element, ... six cases of progressives from be-
auxiliaries plus a non-finite verb form.” And they add further: “even for these six
cases, it is possible that a locative notion is involved in the periphrasis.” This
hypothesis is especially plausible because “copulas often derive from locational or
postural verbs” (1994:131). As a matter of fact, some present-day varieties of Basque
31 do not feel totally sympathetic with some of the arguments in de Rijk’s article, especially when he
deals with nominal case-marking. However, his conclusion about the *da morpheme in the Synthetic
Present verbal forms seems plausible.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
109
accept for da the temporary stage-level meaning of ‘s/he is at’, besides its more
widespread permanent individual-level meaning ‘s/he is.’
De Rijk’s hypothesis, in turn, does not explicitly mention any locative origin
for dator either. But in this case the sense of ‘now’ he proposes for the original
meaning of the morpheme *da can be directly related to present progressive. In fact,
Bybee et al. (1994:129) report three progressives in their database related to the word
‘now.’ Moreover, ‘now’ and ‘here’ might be also diachronically related to one
another. (The morphemes ‘here’, ‘this’ and even ‘be here’ are other sources of
grammaticalization for present progressive meaning in Bybee et al.’s database;
1994:128.)
Therefore, I will only regard as possible the hypothesis that the Synthetic
Present of Modem Basque had its source of grammaticalization in an analytic
construction, expressing some kind of temporary location, and, more generally,
conveying present progressive meaning.
6.2. The Periphrastic Present.
If the origin of the Synthetic Present as a locative construction is speculative, there
can hardly be any doubt that the Periphrastic Present had its source of
grammaticalization in an analytic expression conveying location. This hypothesis has
been assumed by all scholars working on the field, starting with the famous work by
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
110
the Abbe Darrigol, which inspired Anderson 1973 (Darrigol 1829: see Anderson
1973: 5, 9).
The Periphrastic Present, e.g. itzultzen naiz ‘I return’, egiten dut ‘I do’,
consists of the combination of the nonfinite form itzultzen, egiten, which I called the
Imperfective Participle or Gerund in Chapter 2, and a finite form of the auxiliaries
izan ‘to be’ (e.g. naiz ‘I am’) or *edun ‘to have’ (e.g. dut ‘I have’). Now, regardless
of what the current synchronic analysis of the Imperfective Participle may be,
etymologically its relationship with the Verbal Noun, which ends in -tze(a) / -te(a), is
evident: the Imperfective Participle is formally the Verbal Noun in the archaic
locative case (-n). The archaic locative case-marker did not bear the determiner or
article -a, unlike the modem locative case which bears it (-an). (Cf. Mitxelena
1979/1987:452; Lafitte 1962:336.)
Hence, if the Imperfective Participle is an archaic locative, the whole
construction of the Periphrastic Present, -tzen + present of be/have, must have been
grammaticalized as a (biclausal constmction) parallel to the expression of present
temporary location (cf. French ‘etre en train de ’ ). Furthermore, this is probably the
most common source for present progressives across languages, as we saw in section
2 above.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
I ll
6.3. The Present Progressive.
As stated in Chapter 2, the Present Progressive is among the less generalized forms I
am going to study in this work. I also mentioned there that this form(s) is subject to
important variation in Modem Basque. In this section, I will examine the source of
grammaticalization of the two main Present Progressive constructions of Basque.
(For other types of progressive constructions in Modem Basque, and more details see
section 6.2 of Chapter 2). The two constructions I will study next are the Modem
Standard Basque ari construction, i.e. egiten ari naiz ‘I am doing’, and the egon
construction, i.e. egiten nago ‘I am doing.’ (The latter construction is historically
attested in all dialects of Basque, but probably it is not so frequent currently as is the
ari construction, with the exception of the Biscayan dialect).
Regarding the origin of the egiten ari naiz Present Progressive, the first
element of the construction is obviously the Imperfective Participle, whose source of
grammaticalization we have already studied in the previous subsection. We also saw
in Chapter 2 that usually the whole construction is synchronically analyzed in the
literature as biclausal, consisting of a ‘main verb’ ari naiz and a nonfinite
subordinate clause, which has the form egiten of the ‘lexical verb’ and its arguments
(the direct object appearing in the genitive case in the eastern dialects). What we are
in search of, then, is the origin of the ‘quasi-auxiliary’ ari naiz. The combination of
ari and a conjugated form of the verb izan ‘to be’ is actually an instance of what I
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
112
called ‘verbal locutions’ in Chapter 1. Its meaning is something like ‘to be engaged
in’, ‘to be busy doing something.’
The preceding information is enough to describe the source of
grammaticalization of the ari Present Progressive. I have not come across any
hypothesis for the origin of the word ari, though. This might not be totally necessary
because we know what the combination of ari and ‘to be’ means. However, I would
like to hypothesize that ari (attested also as hari) comes originally from the Perfect
Participle of the verb hartu ‘to take, to seize’ (historically documented also as haritu:
cf. Altuna 1979:22, 94). The Radical of this verb is har. The addition of the old
participle suffix -i would have rendered *har-i. If the -i participial ending is replaced
by the modem -tu, of Latin-Romance origin, we have har-tu. If instead of being
replaced, the modem ending is added to the old participle, we arrive at har-i-tu. (For
discussion on the substitution of the old participial suffix -i by the modem -tu, see
Mitxelena 1964/1988:32; Trask 1995:211, 217-218, 232.)
The Participle *hari, haritu, hartu would refer here to the subject of the
progressive construction, and thus would have meant something like Latin
occupatum, from oh- and the past participle of the verb capere ‘to take’; that is,
‘taken, busy, occupied.’ The literal original meaning of the construction etxea-ren
(GEN) egite-n ar-i naiz would have been therefore ‘I am taken/occupied in the
making of the house.’4
4 The verb lotu ‘to tie’ seems to be used in some varieties of Biscayan in a similar way. In this case,
however, the combination of Past Participle and ‘to be’ appears to have more of a ‘perfect of result’
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
113
I turn now to the other Present Progressive construction, e.g. egiten nago.
The source of grammaticalization in this case is straightforward. The construction is
a combination of the Imperfective Participle and the light-verb egon ‘to be at, to stay,
to stand.’ Unlike most of the instances of izan ‘to be’, which is one of the auxiliaries
in the Periphrastic Present, the light-verb egon has always a temporary stage-level
meaning. This is very common of the auxiliaries of progressives. As Bybee et al.
(1994:129) point out: “The verbal auxiliary [of a progressive construction] may
derive from a specific postural verb, such as ‘sit’, ‘stand’, or ‘lie’, or it may express
the notion of being in a location without reference to a specific posture but meaning
only ‘be at’, ‘stay.’”
The main use of egon, especially in the Biscayan dialect, is that of ‘be at.’
However, it also conveys the meaning o f ‘stay.’ Moreover, it is also documented as
meaning ‘stand’, as in the following example: dagoana belaunean onatu, dabilena
sabelean obatu (“el que esta cansa en la rodilla, el que anda en el vientre mejora”)
lit. ‘he who stands gets tired in his knee, he who walks improves in his stomach’ (RS
1596: 365).
meaning, as in etxean lotu da ‘he has remained home.’ Nevertheless, if we were to understand it as a
present resultative, then the meaning of e.g. umea jagoten lotu da (etxean) ‘he has remained (home) to
take care of the child’ can be understood not only as ‘he has bound him self but also as ‘he is (home)
engaged in taking care of the child.’ In other words, this expression would not be too far from having
a present progressive reading.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
114
6.4. The Present Subjunctive.
I already mentioned in sections 4 and 5 the existence of at least one Basque
construction in competition for the domain of non-past which was not
grammaticalized as present progressive. I was concretely referring to the
(Periphrastic) Present Subjunctive. The Present Subjunctive of Basque had a present
telic source of grammaticalization, instead of a present locative source. The formal
description of the Present Subjunctive was already introduced in Chapter 2 .1 also
presented in section 5 above sources of grammaticalization and subsequent
evolutions of cross-linguistic constructions that may be similar in origin to the
Basque Present Subjunctive. In this section, I examine in more detail the components
of the Present Subjunctive of Basque, which are not always easy to understand.
The (Periphrastic) Present Subjunctive (e.g. itzul -dadi-, ekar -deza-) is
formed by the combination of the Radical (e.g. itzul, ekar) and a finite form o f the
auxiliaries *edin (e.g. *dadi) or *ezan (e.g. *deza). The Present Subjunctive appears
always in subordinate contexts in Modem Basque. Therefore, the auxiliaries have
always a subordinator either prefixed or suffixed, which currently is mainly the
suffixed complementizer ~(e)n: e.g. itzul dadi-n, ekar deza-n. The Biscayan dialect
has the verb egin ‘to do’ as the transitive auxiliary, instead of *ezan (e.g. ekarr(i)
dagiari). But this appears to be an innovation, because *ezan is documented in the
early Biscayan texts and has disappeared from this dialect in recent times. Another
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
115
innovation is the use of the Perfect Participle instead of the Radical (e.g. ekarri
instead of ekar) in Biscayan and other western dialects.
6.4.1. The Radical.
The Radical is a rather obscure fonn for the current speaker of Basque. In fact, it has
practically disappeared from the western dialects, replaced by the Perfect Participle,
as just mentioned. However, Trask (1995:213-216) has shown that the *e-prefixed
Radical “was formerly some kind of verbal noun.” We might formally distinguish
two main groups of Radicals: Radicals beginning in *e- and Radicals not beginning
in *e-. The former group contains inherited ancient verbs, whereas within the latter
group we can differentiate between ancient verbs and younger borrowings from
Latin and Romance.
All of the verbs that historically are documented as having synthetic forms
have an %-prefixed Radical, i.e. what Trask identified with an ancient verbal noun.
It is my hypothesis that all of the verbs having this type of Radical, and only them,
have been subject to synthetic inflection. The rest of Modem Basque verbs have
never had synthetic forms. Actually, I believe the rest of verbs did not properly have
a Radical functioning as a verbal noun either.
Among the verbs not subject to synthetic inflection we have the important
number of borrowings from Latin-Romance. These verbs were assimilated to the *e-
prefixed Radicals, mostly (or only) to form the Present Subjunctive and related
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
116
constructions by the combination with the auxiliaries *edin and *ezan. More
specifically: there was in early times a productive opposition Radical (= verbal noun)
/ Perfect Participle that only affected ancient verbs starting in *e- (e.g. ikus / ikusi
‘see / seen’ respectively). This distinction was marked by the participial suffix -z.
Later in time, probably when the Radical had already lost most of its uses as an
independent verbal noun and was mainly used in periphrases, the Radical / Participle
opposition was rebuilt for Romance loans by means of the absence / presence of the
Latin participial suffix -tu. Thus, from the borrowed participlepensatu ‘thought’, the
Radical pensa (approximately ‘think’) was created by analogy. Consequently, pensa
was assimilated to the ancient Radicals and behaved like them when combining with
the auxiliaries *edin and *ezan.
A related evolution in the opposite direction was undergone by verbs deriving
from nouns and adjectives, especially achievement verbs of change of state: e.g. batu
‘pick up, gather, unite’ from bat ‘one’; loratu ‘blossom’ from lore ‘flower’; zorroztu
‘sharpen’ from zorrotz ‘sharp’, zahartu ‘get old’ from zahar ‘old.’ The creation of
these post-nominal verbs (i.e. participles) seemed to occur mainly after the suffix -tu
was borrowed to function as the productive marker of Past Participles. (In these
cases, therefore, the opposition Radical / Participle can be roughly equated originally
with nominal / verb. Tentatively, another group of verbs might have been created
similarly but from ancient adverbs: sar / sartu, bil / bildu, etc.)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
117
6.4.2. The auxiliaries *edin, *ezan and egin.
In a similar fashion to the Radical, the auxiliaries in the (Periphrastic) Present
Subjunctive are not very well understood from a diachronic perspective (nor from a
synchronic point of view either). Fortunately, the intransitive auxiliary (*edin), only
appearing in periphrases today, is documented as an independent verb in some Old
Basque texts (cf. Lafon 1943:37-38). Its meaning is that of ‘become’: e.g. as to dina,
zordun (“el que se hace fiador, deudor”) ‘he who becomes guarantor, debtor’ (RS
1596:249); lotsaga nindin, ogiz ase nindin (“sinvergiienza me hice y harteme de
pan”) ‘I became shameless, and stuffed myself with bread’ (RS 1596:263); arotz
dinean zirola, oski gitxi urratu doa (“cuando el zapatero se hace carpintero, pocos
zapatos se desgarran”) ‘when the shoemaker turns to carpenter, few shoes get
broken’ (RS 1596:275); autsa zala, euriaz loitza zidi (“siendo polvo, con la lluvia se
hizo lodo”) ‘while it was dust, with the rain it became mud’ (RS 1596:358).
The transitive counterpart (*ezan), however, is not independently attested
anywhere. Lafon (1943:38-39) proposed a meaning of ‘to do, to produce’ for *ezan,
based on two main reasons. First, the fact that egin ‘to do’ took on the role of *ezan
as an auxiliary in western dialects. Second, the semantic parallelism of *ezan with
*edin, its intransitive counterpart. It seems, then, that we could more precisely
suppose an original achievement meaning for *ezan (and for egin as an auxiliary: see
below), parallel to *edin ‘to become.’ I would propose something like ‘get, achieve,
attain, complete, (do)’ as the original meaning of *ezan.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
118
The third subjunctive auxiliary, namely egin ‘to do’, is more familiar than the
other two. As mentioned above, the use of egin as the transitive auxiliary in the
Present Subjunctive instead of *ezan seems an innovation that only occurred in
western varieties. On the other hand, egin is attested as a main verb in all dialects
and varieties of Old and Modem Basque. The main verb egin is mostly used in
activity predicates (e.g. zer ari zara egiten ‘what are you doing?’). Egin is also the
base for intransitive-ergative verbal locutions of the verbal class of activities, such as
korrika egin ‘to run’, lan egin ‘to work’, dantza egin ‘to dance.’ However, egin has
also more of an achievement reading in specific contexts, when the senses of
punctuality or of attainment of a limit are present. The following examples are taken
from Leizarraga (1571): manamenduak egin ‘observe the commands’ (Mt V,19),
borondatea egin ‘obey his will’ (Mt VII,21), bilo batxuri ezpa beltz egin ‘turn,
transform, convert a hair into white or black’ (Mt V,36),fruktuak egin ‘produce,
render fruits’ (Mt VII, 17).
Summarizing, it appears sound to assume that the Present Subjunctive
auxiliaries *edin, *ezan and egin were originally main verbs with dynamic and, more
specifically, achievement punctual-like meaning.
6.4.3. The combination o f the Radical and the achievement auxiliaries.
From the previous discussion, we conclude that the Present Subjunctive was
grammaticalized parallel to dynamic telic expressions (changes, etc) in present tense.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
119
Concretely, the Present Subjunctive was formed by the combination of a set of
present dynamic-achievement auxiliaries with a nonfinite item that conveys the
lexical meaning of the construction: i.e. the Radical. The Radical seems sometimes
“some kind of [ancient] verbal noun” (Trask 1995:213), and other times a noun or
adjective. In Latin loans the Radical appears as a secondary formation, created by
analogy, based on the Radical / Participle opposition of inherited verbs. In all of
these cases, nonetheless, the Radical expresses only the bare meaning of the lexical
verb, while the dynamic auxiliaries convey the grammatical meanings.
In sum, the Present Subjunctive of Basque was grammaticalized as a sort of
present telic expression, comparable to those we examined in section 5 above,
including the Slavic Present Perfectives. From the discussion in section 5, then, the
hypothesis of Source Determination would predict that the Present Subjunctive of
Basque has had an evolution parallel to those we saw for constructions with similar
sources across languages.5 Namely, the evolution in (3) that I repeat below for
convenience.
(3) present telic sources -> a. future time-reference in main clauses
b. future time-reference in subordinate clauses
More specifically, we expect that the Present Subjunctive would have evolved
towards covering (at least some of) the following main-clause meanings: definite
5 Similarly, we can expect that a construction parallel to the Present Subjunctive but with past-tense
auxiliaries will render a perfective past meaning, in an analogous fashion to the Slavic Past Perfective.
This construction is the so-called Old Basque Periphrastic Aorist, studied in Chapters 5 and 6.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
120
habitual, gnomic, future, ability, probability, and imperative; and the following
future-gnomic time-reference subordinate-clause meanings: if-clauses (open
conditionals), when-clauses, relative clauses, until-clauses, complement clauses of
verbs of wanting and ordering, and purpose clauses.
The predictions are corroborated by the data. In fact, there is attested
evidence (albeit scanty in some cases) which points out that the Present Subjunctive
of Old Basque conveyed all of the meanings just mentioned, at least in some dialects
(e.g. Biscayan). The Present Subjunctive, however, was ousted from many of those
contexts by other types of constructions. Also, for some of the above contexts (e.g.
ability, probability), it is the Present Subjunctive plus the ancient future marker -ke
that appears conveying the meanings at hand.
In any case, among the meanings where we are likely to find the Present
Subjunctive, I will only take into account subordinate contexts. This is so because
the main-clause assertive meanings above are mostly expressed in historical Basque
(and perhaps generally across languages too) by other types of forms, such as
Futures, Potentials, Imperatives, etc. For this reason, I do not study here (as in Table
3.1) main-clause future time-reference contexts. On the other hand, I do study
subordinate contexts with future-(gnomic) time-reference, because these are more
often covered in historical Basque by constructions grammaticalized as progressives,
(despite the existence of other competitors, such as the Present Subjunctive).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
121
6.5. The Stative Present Verbal Locutions.
A last Basque formation whose source of grammaticalization I will examine here is
what I will call the type of ‘Stative Present Verbal Locutions’ (for the term “verbal
locution”, cf. Lafitte 1962:344-350). The Stative Verbal Locutions (as other lexical
locutions of Basque: see section 5.1 of Chapter 1) are usually not considered TAM
constructions properly, since they do not oppose other types of TAM markers. They
rather seem to constitute a phenomenon pertaining to the lexicon (see section 10 of
Chapter 4 for discussion).
However, from the perspective of sources of grammaticalization, all verbal
formations in a language can be considered originally the same: i.e. formal
expressions consisting of lexical material in combination with grammatical material.
It is only factors such as high generality of use or obligatoriness that eventually make
specific formal expressions be analyzed as grammatical markers. On the other hand,
formal expressions that do not achieve high generality of use will not be analyzed as
obligatory, and therefore they are not often considered grammatical markers. This
appears to be the case o f the Stative Verbal Locutions of Basque.
What I will be calling the Stative Verbal Locutions of Basque are
combinations of a noun or adjective and a stative light-verb izan ‘to be’ or *edun ‘to
have.’ In the case of present that we are examining, the Stative Present Verbal
Locutions are expressions such as bizi naiz ‘I live’, maite du ‘s/he loves’, nahi dugu
‘we want’, behar duzu ‘you need’,falta du ‘s/he lacks’, plazer dute ‘they like’, etc.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
122
The etymology of these expressions is straightforward in most cases. Thus, the
examples just given have a literal meaning (that in some cases is still possible) such
as the following: ‘I am alive’, ‘s/he has as dear’, ‘we have desire’, ‘you have (the)
need’, ‘s/he has lack (of)’, ‘they have pleasure’, etc. Note that most of the Stative
Verbal Locutions are constructed upon the transitive (ergative) light-verb *edun ‘to
have.’
More specifically, these formations were grammaticalized (to the extent they
are grammatical) as predicative complements of the verb ‘to have’, with no article on
the nominal predicate. (For the absence / presence of the article -a (sg) / -ak (pi) in
predicative constructions, see section 2.3.2 of Chapter 6.) In other words, what I am
proposing to be the source of grammaticalization of the Stative Present Verbal
Locutions of Basque is a formation parallel to sentences such as the following:
Abraham dugu aita (Leizarraga Mt 111,9), Abraham guk aita dugu (Haraneder Mt
111,9) ‘we have Abraham to our father, we have Abraham as (our) father.’ In these
sentences, the verb ‘to have’ behaves as a copula between two nominals in a
predicative relation (e.g. Abraham and aita ‘father’; both of them appearing in the
absolutive case, the latter with no article). However, the verb ‘to have’ agrees also
with a third nominal, which syntactically appears as the subject of the sentence (e.g.
guk ‘we’, in the ergative case).
Therefore, the Stative Present Verbal Locutions cited above would have been
grammaticalized in sentences such as: GilenekMari maite du ‘William loves Mary’
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
123
lit. ‘William has Mary as (his) dear’; guk dirua nahi dugu ‘we want money’ lit. ‘we
have money as (our) desire’; zuk kotxe bat behar duzu ‘you need a car’ lit. ‘you have
a car as (your) need’; liburuak lehenengo orria falta du ‘the book lacks the first
page’ lit. ‘the book has lack (of) the first page.’
Dahl (1985:28) offers a cross-linguistic report of a situation that maybe
similar to the Basque case at hand:
... most languages divide up their predicate phrases in at least two types of
constructions, which from the semantic point of view often correspond fairly
well to a ‘dynamic-stative’ classification of predicates. The ‘dynamic’
construction type typically has a full verb as its head, whereas the ‘stative’
construction tends to involve nouns or adjectives in predicative function, with
or without a copula functioning as the dummy head of the predicate phrase.
Note, finally, that the source of grammaticalization proposed for the Stative Present
Verbal Locutions of Basque is somewhat comparable to that of Resultative
constructions (see Chapter 5, section 7.6; and Chapter 6, section 2.3). In both cases
the source of grammaticalization is a predicative formation with similar syntactic
characteristics: Peruk dirua nahi du ‘Peter wants the money’ lit. ‘Peter has the
money (as his) desire’, vs. Peruk dirua hartu du ‘Peter has taken the money’ lit.
‘Peter has the money taken.’ (Moreover, the presence / absence of the article is in
both cases the reflect of the same nominal phenomenon.)
However, the syntactic parallelism between the Stative Present Verbal
Locutions and the Resultative Present shows up as a total divergence regarding their
semantic content. While the Stative Present Verbal Locutions have (stative) present
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
124
meaning, the Resultative Present evolves towards past meaning (see Chapters 5 and
6). This phenomenon is superficially similar to the case of Preterite-Present verbs of
the Germanic languages, but seems to have a very different origin (see Bybee et al.
1994:77-78). The Basque phenomenon at hand might be closer to the case of a
number of West African languages, including Yoruba and Igbo, which appears also
in some Creoles: i.e. nonstative verbs with no marker convey perfective meaning
(usually past), while stative verbs with no marker convey imperfective aspect
(usually present): see Comrie (1976:82).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
125
CHAPTER 4
THE DOMAIN OF PRESENT
IN HISTORICAL BASQUE
1. INTRODUCTION.1
This chapter examines the TAM sphere of present / non-past in historical Basque.
The purpose of the chapter is twofold. On the one hand, the chapter accounts for the
history of the Basque constructions at issue during the last five centuries. On the
other hand, the matter in this chapter will constitute an illustration and testing of the
hypotheses on grammaticalization put forward in Chapter 1. In order to test these
hypotheses, we need two prerequisites: namely, a) to establish the source of
grammaticalization of the Basque constructions under consideration (this task was
already undertaken in section 6 of Chapter 3), and b) to chronologically order the
Basque constructions. Once these prerequisites are fulfilled, I will distinguish
different scenarios of competition among constructions. (See Chapter 1 for more
details on the methodology in this work.)
The Basque constructions appearing historically in the domain of present are
the Synthetic Present, the (Periphrastic) Present Subjunctive, the Periphrastic Present
and the Present Progressive(s). To these we may add the Stative Present Verbal
1 Part of this chapter (as part of Chapter 3) was included in the talk Layering, lexical-semantics and
frequency: a diachronic account o f the constructions o f present tense in Basque, delivered to the
audience of the XIV International Conference on Historical Linguistics, Vancouver, August 1999.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
126
Locution, which merits a different examination. The scenarios of competition I will
distinguish among these Basque forms are: 1) a case of same-source layers: namely,
the competition between the Present Progressive(s) and the Periphrastic Present; 2) a
case of different-source layers: namely, that between the Periphrastic Present and the
Present Subjunctive; 3) the conservation of the Synthetic Present; and 4) the lexical-
semantic competition for stative present meaning.
The presentation of the chapter is as follows. Section 2 is devoted to
chronologically ordering the Basque constructions in the domain of non-past. In
section 3 ,1 outline the different scenarios we will find and the general ‘predictions’
we can propose for this TAM domain in Basque. Sections 4, 5, 6 and 7 examine the
Basque data on the historical competition for specific areas of TAM within the
sphere of present. These sections will be dependent on the framework of the
grammaticalization clines of non-past put forward in Chapter 3. Each of sections 8, 9
and 10 examines a one scenario of competition out of those outlined in section 3.
Section 8 presents the case of same-source layering between the Periphrastic Present
and the Present Progressive. This is the occasion for testing the hypotheses of
Layering-Displacement and Form-Meaning Covariation, presented in Chapter 1. In
section 9 ,1 deal with the conservation o f the Synthetic Present. First, I will test the
hypothesis of Conservation of high token-frequency forms formulated in Chapter 1.
Then, I will discuss the issue of the possible relation between the conservation of the
Basque Synthetic Present forms and their lexical-semantics. In section 10, finally, I
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
127
examine the competition for stative present meaning. I will also touch upon the role
of type-frequency in the evolution of formations such as the Stative Present Verbal
Locution, already introduced in section 6.5 of Chapter 3. (See the Glossary for the
terms ‘token-frequency’ and ‘type-frequency’, and for other terminological
clarification.) ,
2. CHRONOLOGICAL ORDERING.
In this section, I provide theory-independent evidence for supporting a relative
chronological order of the Basque TAM constructions that will be studied in the
sphere of present / non-past. I already put forward in Chapter 1 general evidence to
order chronologically the four main layers o f constmctions I am studying in this
work. For the Basque constructions involved in the domain of non-past, the
diachronic order would be the following (where the symbol “>” stands for ‘older
than’):
(1) Synthetic Present > (Present Subjunctive) > Periphrastic Present > Present
Progressive
In fact, the only constructions we can chronologically order with confidence are
those which I have already proved to have the same source of grammaticalization.
As we saw in Chapter 3, among the four constructions at issue, we can only ensure
that the Periphrastic Present and the Present Progressive were grammaticalized with
the same source; namely, both had their origin as present progressives. (I will
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
128
examine the evidence for proposing a relative chronological order between
Periphrastic Present and Present Progressive in section 2.3 below.) The Synthetic
Present, in turn, could have been grammaticalized as a present progressive too. But
this is just a tentative hypothesis. The Present Subjunctive, finally, was not
grammaticalized as a present progressive, but with a present telic source.
2.1. The Synthetic Present.
I proposed in section 6.2 of Chapter 1 that the Synthetic Present, as well as all
synthetic forms of Modem Basque (e.g. nator ‘I come’, dakart ‘I bring’), is a very
old unproductive construction in an advanced process of decline. I repeat below for
convenience the evidence supporting this view of the Synthetic Present (and the
other synthetic forms of Basque).
1) The Synthetic Present is highly fused; i.e. many of the morphemes in it are
difficult to segment and/or present different variants (cf. Comrie 1989:46-51). The
Synthetic Present is also a rather opaque fonnation. It is often subject to phonetic
attrition. (In some cases, especially in the most frequent fonns, it seems the Synthetic
Present has experienced a loss of suprasegmental autonomy; cf. Bybee et al.
1994:110-111.)
2) The analytic forms of Basque are built upon synthetic forms, which serve
as auxiliaries of the former. To be more precise, the great majority of the auxiliaries
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
129
of the periphrastic fonns are or have been documented as independent synthetic
forms. (See section 4 of Chapter 2 and section 6 of Chapter 3 for more details.)
3) The great formal regularity of the synthetic forms support the claim that
the synthetic conjugation was a productive rule in earlier stages o f Basque. This rule
became unproductive for unknown reasons (e.g. competition with the periphrastic
constructions, opacity, etc.). The synthetic conjugation apparently affected only
those verbs of the form *e-ROOT-i/-n}
4) As we will see in the present chapter, the historical documentation shows
that the Synthetic Present has evolved in the last five centuries towards losing
presence in the language, specially concerning the number of verbs subject to
synthetic inflection. That is, mostly regarding type-frequency, (but not that much
regarding absolute token-frequency).
Now, if the Synthetic Present is a very old unproductive construction in an
advanced process of decline and confined to a handful of lexical verbs, then its study
should be carried out with a perspective different from that of the other three
constructions in the domain of non-past. The study of a form with these features may
be considered as not pertaining to the general process of grammaticalization.
A very small number of verbs not having the shape *e-ROOT-i/-n are also subject to synthetic
inflection. This is the case of ihardun ‘to act, to be engaged doing something, to chat.’ This verb
appears to be originally a compound formation of the verb *edun ‘to have’ (cf. Mitxelena 1977/1987:
333). Also, the 17th century Souletin author Oihenart has Synthetic Present forms such as donhetsa
(Oihenart 1657:390) from the compound verb on-etsi ‘love’, lit. ‘good-consider.’ Thus, it would seem
that the mle of synthetic inflection was also productive for some verbs constructed as compounds of
the*e-ROOT-i/-n basic verbs. (A different possibility is that authors such as Oihenart made up some
of the forms at hand; i.e. by remaking as partly ‘productive’ a rule that was already unproductive.)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
130
Therefore, from this point of view, it is not as important to determine the origin of
the Synthetic Present as to prove that it is a declining form. I will assume, thus, that
the Synthetic Present of Basque has to be understood differently from the other
constructions in competition for the domain of present (see, however, Trask
1977:212; Bybee et al. 1994:149). The study of the Synthetic Present is presented in
section 9 below.
2.2. The (Periphrastic) Present Subjunctive.
The (Periphrastic) Present Subjunctive (e.g. etor nadin ‘so that I come’, ekar
dezadan ‘so that I bring’) is difficult to order chronologically relative to the other
two analytic constructions at issue: the Periphrastic Present and the Present
Progressive. This is mainly so because their sources of grammaticalization are
different; i.e. the Present Subjunctive was not grammaticalized as a present
progressive.
In any event, we can be rather confident that the Present Subjunctive is
younger than the Synthetic Present, for two o f the auxiliaries in the former (*edin ‘to
become’ and egin ‘to do’) were after all independent instances of the latter. In other
words, the Present Subjunctive was built upon the Synthetic Present (see section
6.4.2 of Chapter 3). It seems also sound to consider that the Present Subjunctive is
older than the Present Progressive.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
131
However, it is hard to tell whether the Present Subjunctive is older, younger
or nearly coetaneous in relation to the Periphrastic Present (e.g. egiten dut ‘I do’). As
I have been assuming throughout this work, the most important kind of evidence for
ordering chronologically two constructions should be that regarding their degree of
boundness and mobility/rigidity. Yet, there seems to be a problem with this criterion
in the case of periphrastic constructions in subjunctive contexts. Thus, the clearest
evidence for proposing a low degree of boundness for a given analytic construction
of Basque is the frequent attestation of instances of the construction with the non-
basic inverted order [auxiliary + nonfinite form]. The recurrent existence of the non-
basic order implies in principle a high degree of mobility between auxiliary and
nonfinite form. However, it seems that the case of subjunctive clauses may be more
complicated. It would appear that the basic relative order of nonfinite form and
auxiliary is not necessarily [nonfinite form + auxiliary] in the subjunctive clauses of
Basque.
Thus, among the few exceptions to the order [nonfinite form + auxiliary] that
Lafitte (1962:337) gives, he cites the two following examples: dezagun kanta
gogotik (“chantons de bon coeur”) ‘let’s sing from the heart’, dugun errepika
(“repetons”) ‘let’s repeat it.’ The two examples are instances of the subjunctive-
imperative mood: the former is a Present Subjunctive (with the dynamic auxiliary
*ezan), while the latter is a somewhat strange formation that has the stative auxiliary
*edun ‘to have.’ Lafitte adds further: “dugun, employe comme auxiliaire imperatif
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
132
au lieu de dezagun, est toujours mis avant Pinfinitif radical” [‘when dugun (‘let’s’) is
used as imperative auxiliary instead of dezagun (‘let’s’), it is always placed before
the Radical’].
In sum, it seems that the order [auxiliary + nonfinite form] is not as
exceptional in the subjunctive and imperative clauses of Basque as it is elsewhere.
Therefore, it does not appear appropriate to compare the frequency of this order in
the Present Subjunctive with its frequency in main clauses of the Periphrastic
Present. In section 2.2 of Chapter 6 ,1 make a comparison of the relative order of
nonfinite form and auxiliary between the Periphrastic Aorist (the non-subjunctive
past counterpart of the Present Subjunctive), on the one hand, and the analytic forms
built with stative auxiliaries, on the other. It turns out that the Periphrastic Aorist
presents more consistently the order [nonfinite form + auxiliary] than the stative-
auxiliary constructions.
Thus, I tentatively conclude that there is some evidence to assume that the
constructions built by the combination of the Radical and the dynamic auxiliaries are
older than all the constructions with stative auxiliaries (i.e. that the so-called
“subjunctive” constructions are older than the “indicative” constructions).
2.3. Periphrastic Present vs. Present Progressive(s).
In this section, I present evidence in support of the relative age between the
Periphrastic Present and the Present Progressive. I want to prove that the former (e.g.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
133
mugitzen da ‘it moves’ intransitive) is older than the latter (e.g. mugitzen ari da /
mugitzen dago ‘it is moving’). The former is an instance of what I termed ‘rigid
periphrases of the indicative’ in Chapters 1 and 2. The latter in turn are ‘loose
periphrases of the indicative.’ The most important piece of evidence to substantiate
the chronological order proposed comes from the degree of rigidity / mobility of
nonfinite form and auxiliary / quasi-auxiliary. Yet I present below some other types
of evidence as well.
2.3.1. Degree offormal rigidity between nonfinite form and auxiliary.
I have already mentioned that the Present Progressive is analyzed in the literature as
a biclausal construction. Saltarelli (1988:229), for instance, states that the
progressive is expressed by the verb ari izan ‘to be engaged in’ “with nominalized
complements in the locative case.” That is, the lexical verb is in practice
subordinated to the quasi-auxiliary ari + ‘to be.’ The Periphrastic Present, in turn, is
always analyzed as a monoclausal construction.
More importantly, while the Periphrastic Present always appears in the order
[lexical verb + auxiliary], e.g. mugitzen da, the Present Progressive has more
freedom in the relative order of elements: mugitzen ari da / ari da mugitzen.
Saltarelli (1988:229) gives the following example: gauza asko ari gar a ikasten ‘we
are learning many things’, where the lexical verb ikasten ‘learning’ appears after the
quasi-auxiliary ari gara ‘we are (engaged).’ The same could be stated for the Present
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
134
Progressive with egon ‘to be at’: e.g. gauza asko ikasten gagoz / ganza asko gagoz
ikasten both orders are possible (although the relative frequency of each order should
be further examined in texts and in the spoken language).
The Periphrastic Present, therefore, has a higher degree of rigidity than the
Present Progressive. This is probably the most solid evidence for ordering the former
as older than the latter.
2.3.2. Combinability with other types o f constructions.
Another piece of evidence for considering the Periphrastic Present as older than the
Present Progressive is the fact that the latter type of construction combines with more
types of constructions than the former. Although in this work I will only examine the
Present Progressive, the Progressive type of construction (an instance of loose
periphrasis) can in principle combine with any of the rigid periphrases of Modem
Basque. However, the rigid periphrases, such as the Periphrastic Present, can hardly
combine with other rigid periphrases. This is an illustration of what Bybee et al.
(1994:7) call “mutual exclusivity.” Thus, while *?ikusten izango dut, *?ikusten izan
dut are hardly possible, the progressive formations ikusten ariko naiz, ikusten ari
izango naiz, ikusten egongo naiz ‘I will be watching’ or ikusten ari izan naiz, ikusten
egon naiz ‘I have been watching’ are normal in Modem Basque.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
135
2.3.3. Selection o f auxiliaries and ergative pattern.
I present next other more specific points throwing evidence for ordering the Present
Progressive as younger than the Periphrastic Present, although probably they are not
as important as the degree of rigidity. It should be noted, at any rate, that, even if the
criteria below are not totally safe, it is remarkable that all of them coincide in the
same chronological direction: it is hard to think of just one argument in favor of the
opposite ordering.
A first small piece of evidence might come from the comparison of
auxiliaries. The Periphrastic Present, as the Synthetic Present and all other old and
mature constructions (i.e. synthetic forms and rigid periphrases) of Modem Basque,
has an ergative pattern (i.e. ergative-case subjects and transitive auxiliaries for
transitive verbs, and absolutive-case subjects and intransitive auxiliaries for
intransitive verbs; see however the disclaimer in note 2 o f Chapter 2 and the entry
‘intransitive-ergative’ in the Glossary). The Present Progressive, on the other hand,
has a nominative pattern (i.e. always with absolutive subjects). It is not clear this
different behavior could be related to historical trends. In case it could be, it seems
the nominative pattern appears as an innovation, perhaps under the influence of the
surrounding nominative languages.
This different case-marking behavior of the Periphrastic Present (ergative)
and the Present Progressive (nominative) could perhaps be taken as further evidence
for analyzing the former as monoclausal and the latter as biclausal.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
136
2.3.4. Individual-level auxiliaries vs. stage-level quasi-auxiliaries.
Another small argument in favor of the chronological ordering proposed comes from
comparing the intransitive auxiliary in the Periphrastic Present (i.e. izan ‘to be’) with
the only quasi-auxiliary in the western Present Progressive (i.e. egon ‘to be at’). The
argument these verbs provide is rather theory-internal, but still can be taken as
offering some evidence for the case at issue.
If we compare these two verbs (izan ‘to be’ vs. egon ‘to be at, to stand, to
stay’), we find that, whereas the latter is only momentary, the former is attested as
both permanent and momentary, although, especially in Biscayan, it seems to have
lost its temporary reading. This suggests an evolution for the verb izan ‘to be’ from
momentary (i.e. ‘to be at’) towards only permanent, while being replaced by egon for
the momentary reading. This is, incidentally, an evolution which is widely attested
cross-linguistically. If this is the case, it explains why the verb egon is in Biscayan
more suitable than izan to appear in a young progressive construction. Recall that
progressives are grammaticalized with temporary meaning.
2.3.5. Amount o f variation.
A final piece of evidence I will present in favor of the chronological order put
forward above comes from the amount of variation across and within dialects. I think
this criterion is a bit slippery, but in this case it seems safer than in others.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
137
In any event, the Present Progressive shows a much greater amount of cross-
dialectal and other kinds of variation than the Periphrastic Present. It is significant
that we actually cannot properly talk about one Present Progressive construction. All
grammars of Basque give more than one construction for present progressive
meaning. Saltarelli (1988:229), for instance, gives three different constructions. Even
when talking about the ari construction, the most widespread of all, it is not always
the construction I have described here (i.e. ari izan quasi-auxiliary + Imperfective
Participle: e.g. ikasten ari naiz ‘I am studying’) the only one given in the grammars.
Thus, Lafitte (1962:347) describes three different types of progressive constructions
or “locutions” with ari izan, aside from the one constructed with the Imperfective
Participle (see also section 4 below). (Moreover, in western Biscayan there is no ari
construction attested in the historical period. That is, the verb ari izan is not
documented in western Biscayan.)
As mentioned above, interpreting cases of dialectal variation is not always
straightforward. In this case, though, it may represent a recent development that has
not reached generality enough as to be spread across dialects. The existence of
variation within dialects seems easier to understand, nonetheless. This factor points
to the existence of very young similar-age similar-source competing constructions
neither of which has yet reached high generality of use. This appears the case of the
different (but very similar) types of progressive constructions with ari izan.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
138
3. GENERAL PREDICTIONS.
I introduce in this section the most general ‘predictions’ (to the extent we can
predict) we can put forward for the sphere of non-past in historical Basque. (For the
approach to ‘prediction’ adopted in this work, see section 3.1 of Chapter 1, and
section 4.2.4 of Chapter 7.) As previously stated, the examination of the data
uncovers that the TAM constructions of Basque constitute different types of
scenarios which have to he approached in different ways; I present below four
different scenarios and the types of predictions and/or empirical observations that
each scenario requires to be properly understood.
3.1. Competition between Present Progressive and Periphrastic Present.
I showed in Chapter 3 (sections 6.2 and 6.3) that both the Periphrastic Present and
the Present Progressive were grammaticalized as present progressives. We have also
seen in section 2.3 above that the Periphrastic Present is older than the Present
Progressive. It follows from these premises that the Present Progressive and the
Periphrastic Present of Basque constitute an instance of ‘same-source layering’ (see
sections 3.3 and 4.1 of Chapter 1). This is thus the ideal case of competition between
forms, where we have the safest scenario for proposing hypotheses on clines of
grammaticalization and for testing them.
More specifically, the hypothesis of Source Determination predicts that the
Present Progressive and the Periphrastic Present travel along the same path of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
139
grammaticalization at different stages of development. We will therefore be able to
test (in section 8 below) the hypotheses of Layering-Displacement and Form-
Meaning Covariation against the data provided by this ideal case o f the same-source
layering competition between Present Progressive and Periphrastic Present (see
section 4.1 of Chapter 1). Unfortunately, the semantic changes that the Present
Progressive underwent in the historical period are minimum. Thus, the hypothesis of
Layering-Displacement will not find data to be tested against. On the other hand, the
hypothesis of Form-Meaning Covariation can be tested in the case at hand.
3.2. Competition between Periphrastic Present and Present Subjunctive.
We saw in section 6.4 of Chapter 3 that the Present Subjunctive was not
grammaticalized as a present progressive. Its source of grammaticalization was
rather a type of ‘present telic’ expression (see section 5 of Chapter 3). Therefore, the
Periphrastic Present and the Present Subjunctive do not constitute an instance of
same-source layering. I have also tried to prove in section 2.2 above that the Present
Subjunctive is older than the Periphrastic Present. Hence, we can conclude that the
Periphrastic Present and the Present Subjunctive make up a case of different-source
layering (see section 4.2 of Chapter 1).
In this case, it is not straightforward to understand the competition between
the layers. This case does not constitute either a good scenario to test the hypotheses
on clines of grammaticalization. This is so because we cannot ensure that the two
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
140
layers of constructions follow the same path. However, I speculated in Chapter 1 that
in the case of the competition between different-source layers we could understand
to have approximately a similar picture to that provided by the ideal case.
Therefore, although I will not test any hypotheses or propose any predictions
for the competition between the Periphrastic Present and the Present Subjunctive, it
seems we can understand their competition in the way that the former (i.e. the
younger layer) is displacing the latter (i.e. the older) from some of the contexts they
compete for. Given the present telic source of grammaticalization of the Present
Subjunctive, we can predict that the areas of competition in this case will be future
time-reference contexts (see section 6.4.3 of Chapter 3). The empirical observation
shows actually that the competition between the Periphrastic Present and the Present
Subjunctive occurs in future time-reference ‘less-dynamic’ subordinate contexts; i.e.
if-clauses, when-clauses, relative clauses, etc (see section 6 below).
3.3. Conservation of the Synthetic Present.
Along with the above-examined three constructions, the fourth main form competing
in Basque for the TAM domain of present is the Synthetic Present. I have tried to
prove in section 2.1 above that the Synthetic Present, as the other synthetic forms of
Basque, is a very old unproductive construction in decline. Hence, the Synthetic
Present will not be considered in the same way as the other three constructions at
issue. Concretely, the instances of the Synthetic Present appearing in historical
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
141
Basque will be considered cases of conservation rather than grammaticalization
properly. In other words, the competition of the Synthetic Present forms with the
other constructions in the domain of present does not occur in terms of normal layers
of grammaticalization, but is a defective competition: it only affects a handful of
verbal items.
In section 9 below, 1 will discuss the factors that may condition the
conservation of the Synthetic Present. I have already proposed in Chapter 1 that high
token-frequency is with all probability among the most important of these factors.
Thus, I will test in section 9.1 below the hypothesis of Conservation of high token-
frequency forms (section 3.5 of Chapter 1) against the Basque data on token-
frequency of the Synthetic Present.
In section 9.2, in turn, I will examine the lexical-semantics of the Synthetic
Present forms. Concretely, the majority of the conservative Synthetic Present forms
of Basque belong to either one of two lexical groups: stative verbs and motion verbs.
It is not totally clear, though, that lexical-semantics is a capital factor in the
conservation of the Synthetic Present.
3.4. Competition for stative present.
The last type of scenario I will propose within the TAM domain of present in Basque
is the competition for generality of use, which takes place along the lexical-semantic
dimension. Within the possible scenarios of competition for generality of use, I will
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
142
only examine the competition for expressing stative present meaning. (In section 4
below, I touch upon the competition for present progressive.) The competition for
stative present is a complex one because it involves not only full-fledged
grammatical constructions but also the Stative Present Verbal Locution, which can
be considered a lexical phenomenon.
We already saw in section 2 of Chapter 3 (and also in Table 3.1) that a
construction once grammaticalized as present progressive can eventually cover the
meaning of stative present, thus becoming applicable to all lexical-semantic classes
of verbs. It is common to find across languages that the construction expressing
present habitual conveys also stative present. This situation can be actually found in
some instances in the Basque data: i.e. the Periphrastic Present can express stative
present. Another Basque form that can convey stative present meaning is the
Synthetic Present. A third kind of formation that can also express stative present is
what I have been calling ‘Stative Present Verbal Locutions’, such as bizi naiz ‘I live’
or nahi dut ‘I want’ (see section 6.5 of Chapter 3). Recall that these are formed by
the combination of a noun / adjective and Synthetic Present forms of the verbs izan
‘to be’ or *edun ‘to have.’
As just mentioned, the case of the Stative Verbal Locutions is a difficult one.
In principle, they can be synchronically analyzed as a phenomenon pertaining to the
lexicon (see section 10.1). For this reason, in the beginning of the ensuing discussion
I will not differentiate between Stative Present Verbal Locutions and the Synthetic
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
143
Present forms (which host them), but will include the former among the latter. On
the other hand, in section 10.4 below, I will present some remarks towards
understanding the Stative Verbal Locutions from the perspective of the theory of
grammaticalization. Concretely, I will propose that these locutions can be understood
as a case of constructions grammaticalized to be applicable to a low type-frequency
class: i.e. the class of stative predicates. In section 10, therefore, I will explicitly
distinguish Stative Present Verbal Locutions from proper Synthetic Present forms.
4. COMPETITION FOR THE YOUNGEST SEMANTIC SUBTYPES IN THE
CLINE STARTING WITH PRESENT PROGRESSIVE
I present in this section the data in the Main Corpus on the competition for the
youngest subtypes in the cline of Table 3.1 (Chapter 3). The Present Progressive is
the construction that mostly covers the subtypes at issue. Another construction that
can express some of the meanings in the young subtypes in Table 3.1 is the Synthetic
Present. However, it is not totally clear that the so-called present progressive of
dynamic verbs of motion (e.g. Patxi autobusean dator ‘Patxi is coming in the bus’
Euskaltzaindia 1987: 407) is actually a pure present progressive meaning or has
rather a sense of a preparatory phase leading to an end-point. In any event, the
Synthetic Present will be examined in section 9 below.
As for the Present Progressive, the data regarding this form in the Main
Corpus database are so scanty that one can hardly extract any conclusions from
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1 4 4
them. This is not totally unexpected, though, the database is drawn from written
language, and in the case of the Main Corpus the texts are rather narrative, with few
passages of dialogue. However, present progressive is a meaning that one would
expect to find mostly in the spoken language, and not so often in written texts.
Therefore, to study in detail the use of the Present Progressive in current Basque (and
in general for all the early subtypes in this cline) we would need elicited and spoken
data. This could constitute a very interesting topic for another work. The same
disclaimer would probably apply to all the young constructions of Basque, i.e. those
with a low degree of generality of use (see Chapters 1 and 2).
In any case, I examine below the (scanty) data we have on present
progressive in this work. The most noticeable aspect of the Main Corpus regarding
the present progressive is that, among the few instances where we find a meaning at
least similar to present progressive, only a part of them are expressed by the two
constructions I have presented as the most representative of the Basque Present
Progressive: i.e. egiten ari naiz and egiten nago ‘I am doing.’
Thus, in LEIZAR3 (16th c.) we do not find any examples containing an ari
construction; we find two in HARAN (18th c.), and three in ITUN (20th c.).
However, not even all of the instances where the quasi-auxiliary ari izan appears are
31 will call the three databases in the Main Corpus LEIZAR (16th c.), HARAN (18th c.) and ITUN
(20th c.), from Leizarraga, Haraneder and Itunberria, respectively (see Chapter 1, section 7). I use
these abbreviations for the databases I have employed in the work, to distinguish them from the whole
book where they belong. In other words, when I say that I do not find any ari construction in
LEIZAR, I do not necessarily imply that the author Leizarraga never used this construction in his
whole book. Recall that the three databases are extracted from the Gospel according to Saint Matthew.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
145
cases of [Imperfective Participle + ari izan]: e.g. egiten ari naiz. Some of them are
combinations of ari izan plus adverbs, such as oihuka ‘shouting’ (oihu ‘shout (noun)’
+ adverbial suffix -ka), negarrez ‘crying’ (negar ‘cry (noun)’ + modal case-marker
~(e)z), lanean ‘at work, working’ (lan ‘work (noun)’ + locative case-marker -(e)an).
Thus, we have oihuka ari da ‘he is shouting’, ari da (bere haurrengatik) negarrez
‘she is crying (for her children)’, ez dire hari lanean ‘they are not working.’ (For a
description of these alternative constructions based on ari izan, see Lafitte
1962:347.)
As for the other major progressive form of Basque, namely the egon
construction, we find practically the same situation: we find egon used with present
progressive meaning in the two oldest texts, but hardly accompanied by the
Imperfective Participle. LEIZAR has one example: oihuz dago ‘he is shouting’;
HARAN, four examples: oihuz dago (twice), nigarrez dago ‘he is crying’, dagoela
auhenez ‘that he is grieving.’ Finally, ITUN has a present progressive formed upon
the light-verb ibili ‘to wander’: nahian dabil ‘he is desiring’ (nahian = nahi ‘desire
(noun)’ + locative case marker). Note that in this last case the progressive is used
with a stative verb, to convey an idea of repetition or gradualism (see Chapter 3,
section 2).
The great amount of variation we find in the Present Progressives
corroborates the picture of these ‘loose periphrases.’ The Present Progressives of
Basque seem to be with all probability lowly grammaticalized constructions with low
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1 4 6
generality of use and experiencing competition among very similar types of
expressions. (See also section 2.3.5 above.)
5. COMPETITION FOR THE DOMAIN OF PRESENT HABITUAL /
GNOMIC / (STATIVE PRESENT).
The mature subtypes in Table 3.1 (Chapter 3) are frequently covered across
languages by one construction that can primarily be considered a Present Habitual or
a non-Progressive Present (cf. Comrie 1976:77). The most important TAM meanings
conveyed by this construction are probably present habitual, generic and stative
present (cf. Bybee et al. 1994:151-152).
In the case of Basque, however, the expression of stative present is quite
consistently differentiated from present habitual. (A tendency towards expressing
habitual and stative by means of the same form, though, may be occurring in recent
times, especially in the eastern dialects, perhaps. This form is the Periphrastic
Present.) In general, most stative predicates of Basque are expressed by what might
be almost considered special forms for states: namely, the Synthetic Present and the
Stative Verbal Locutions. I will examine the conservation of the Synthetic Present in
section 9 below, and the expression of stative present meaning, including the Stative
Verbal Locutions and the Synthetic Present in section 10.
If we consider only the mature non-stative subtypes in Table 3.1, we find that
they are mostly expressed in Basque by the Periphrastic Present. Thus, this
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
147
construction is often considered in the literature as a Present Habitual, even though
there is a specific construction for habitual meaning (see below). (Recall that the
Periphrastic Present was grammaticalized as a present progressive.) The Synthetic
Present, in turn, is the major competitor for these semantic subtypes. However, as we
will see in section 9, the Synthetic Present appears in the case of high token-
frequency verbs. Hence, the picture we find in this area of TAM (i.e. mainly present
habitual and gnomic, excluding stative predicates) shows a major construction with
high type-frequency (i.e. the Periphrastic Present), and a competitor with low type-
frequency but often high token-frequency (i.e. the Synthetic Present).
5.1. Present habitual.
I present below the data on present habitual in the three texts of the Main Corpus.
TABLE 4.1: Present habitual (Main Corpus).
LEIZAR HARAN ITUN
TOKENS ITEMS TOKENS ITEMS TOKENS ITEMS
TOTAL 40 29 32 22 26 20
SYNTHETIC
PRESENT
2 2 1 1 1 1
PERIPHRASTIC
PRESENT
38
(95.0%)
27
(93.1%)
29
(90.6%)
19
(86.4%)
21
(80.8%)
17
(85%)
OTHER
CONSTRUCTIONS
0 0 2 2 4 2
The most important remark to make to these data is the high degree of consistency to
express habitual meaning in all three texts, i.e. the low degree of competition for
habitual meaning. The Periphrastic Present is by far the major construction for
expressing present habitual in the three databases. We should mention, however, that
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1 4 8
other minor constructions are also used conveying present habitual. The most
important of them is a special form for expressing habitual meaning, built upon the
verbal locution ohi izan / *edun ‘to have the habit o f, which appears four times in
ITUN (20th c.).
5.2. Generic or gnomic.
When studying gnomic meaning in Basque, we need to take into account whether it
refers to a stative or to a dynamic predicate. This is so because, as I have introduced
above, the use of ‘special’ forms for stative predicates is very common. Thus, I offer
below the data on stative gnomic and dynamic gnomic meanings separately.
TABLE 4.2: Stative gnomic (Main Corpus).
LEIZAR HARAN ITUN
TOKENS ITEMS TOKENS ITEMS TOKENS ITEMS
TOTAL 13 4 16 8 19 5
SYNTHETIC
PRESENT
13 4 14 6 19 5
PERIPHRASTIC
PRESENT
0 0 2 2 0 0
TABLE 4.3: Dynamic gnomic (Main Corpus)-
LEIZAR HARAN ITUN
TOKENS ITEMS TOKENS ITEMS TOKENS ITEMS
TOTAL 41 33 31 24 27 24
SYNTHETIC
PRESENT
2 2 4 3 2 2
PERIPHRASTIC
PRESENT (PPr)
39 31 24 18 25 22
OTHER
CONSTRUCTIONS
0 0 3
PPriCe = 2
3 0 0
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
149
The quite differentiated marking of stative and dynamic predicates in Basque
shows up clearly in the data above. In a first approach we may say that states
correlate with the Synthetic Present and are rather incompatible with the Periphrastic
Present. (Note however that at this point I am including the Stative Verbal Locutions
within the Synthetic Present: see section 3.4 above. In section 10 below, I will
examine these two types of formations individually.)
As a matter of fact, stative predicates tend to appear in the Synthetic Present
(together with Stative Verbal Locutions) in almost any context, including future
time-reference subordinate clauses. (The main exception to this observation is the
case of purpose clauses in Modem Basque, but not in Old Basque: see section 7
below.) In other words, the distinction stative / dynamic has to be carried on to
practically all of the semantic subtypes in the domain of non-past. It might also be
said that TAM distinctions appear rather “neutralized in stative contexts” (Dahl
1985:28).
On the other hand, Table 4.3 above shows that in the case of dynamic gnomic
meaning (parallel to what we saw for present habitual in section 5.1) the degree of
competition among forms is very low: the Periphrastic Present is clearly the major
construction for conveying dynamic gnomic. (In fact, even an uncommon alternative
construction that HARAN shows 2 times for expressing dynamic gnomic is related
to the Periphrastic Present. This alternative construction is the Periphrastic Present
plus the potentiality marker -ke, which I will not study in this work.)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
150
In sum, the distinction stative / dynamic has proved to be highly relevant for
the case of gnomic meaning in Basque. Thus, the evolution we saw in section 2 of
Chapter 3, i.e. present habitual gnomic, seems to be better stated at least in the
case of Basque as present habitual dynamic gnomic. Another evolution would
then progress in the direction: permanent stative present -> stative gnomic, as in this
ant is black ants are black. If this is the case, it might be interesting to draw a
general distinction between dynamic gnomic (e.g. copper conducts / will conduct
electricity) and stative gnomic {elephants have long trunks), where the former
appears more inductive than the latter (cf. Lightfoot 1975:117). In English, for
instance, “gnomic or general truth statements that involve permanent states do not
use the future” (Bybee 1988b:373).
6. COMPETITION FOR THE DOMAIN OF FUTURE TIME-REFERENCE
LESS-DYNAMIC SUBORDINATE CONTEXTS.
In Chapter 3 ,1 divided the sphere of future time-reference subordinate contexts into
two subgroups that I termed for convenience ‘dynamic contexts’ and ‘less-dynamic
contexts.’ In the present section, I examine the competition in historical Basque for
the future time-reference subordinate ‘less-dynamic’ contexts. Recall that these
contexts include open conditionals, future time-reference temporal clauses, generic
relative clauses, concessive clauses and until-clauses. The Basque constructions in
competition for this area of TAM meaning are the following.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
151
1) (Periphrastic) Present Subjunctive: I examined in Chapter 3, section 6.4,
the source of grammaticalization of the Present Subjunctive and concluded that it
was grammaticalized as a present telic construction. For this reason, I proposed in
section 6.4.3 of Chapter 3 that, based on the hypothesis of Source Determination, the
Present Subjunctive is a suitable form for conveying future time-reference meaning
in both main and subordinate clauses. I will focus here on the subordinate
appearances of the Present Subjunctive.
2) Periphrastic Present: The Periphrastic Present was grammaticalized as a
present progressive. Yet, we already saw in Chapter 3, section 3, that a construction
with present progressive source may reach future time-reference subordinate
contexts, especially the less-dynamic ones.
The Present Subjunctive and the Periphrastic Present, thus, constitute a case
of competition between different-source layers (see section 3.2 above). Now,
although this case is difficult to understand, I proposed above that this evolution
might be viewed as one where the younger layer (the Periphrastic Present) is
displacing the older layer (the Present Subjunctive). In fact, the data seem to
corroborate this view, especially for if-clauses (see section 6.1). However, the
situation of the future time-reference less-dynamic contexts is more complex than the
just exposed, as there are other forms in competition for this area of TAM.
3) Synthetic Present'. A third form in competition for the future time-
reference subordinate less-dynamic contexts is the Synthetic Present. However, as I
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
152
have already mentioned, this form will only appear in the case of a handful of high
token-frequency verbs, and particularly for stative predicates (including Stative
Verbal Locutions).
4) (Periphrastic) Future: The grammaticalization of the Periphrastic Future
could constitute the topic of a whole work. For this reason, I have not examined in
detail the source of grammaticalization of the Periphrastic Future (see nonetheless
section 5.4 of Chapter 2 for a synchronic description). In any event, we only find the
Periphrastic Future in two specific contexts in this work: future time-reference
temporal clauses (section 6.2) and, sometimes, generic relative clauses.
6.1. Open conditionals (and concessive protases).
I present below the data on if-clauses and concessive protases, which are formally
very similar in Basque. Since stative predicates are very often expressed by the
Synthetic Present, I make a distinction between stative and non-stative predicates.
TABLE 4.4: Non-stative open-conditional and concessive protases (Main Corpus).
LEIZAR HARAN ITUN
TOKENS ITEMS TOKENS ITEMS TOKENS ITEMS
TOTAL 14 12 14 11 14 13
SYNTHETIC
PRESENT
0 0 0 0 1 1
PERIPHRASTIC
PRESENT
5 5 13 10 13 12
PRESENT
SUBJUNCTIVE
9 7 1 1 0 0
Table 4.4 shows a clear evolution in historical Basque for the expression of if-
clauses. Concretely, the last centuries have witnessed a competition for this context
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
153
between the Present Subjunctive and the Periphrastic Present, from which the
Periphrastic Present has resulted winner, displacing the Present Subjunctive from
open conditionals. In LEIZAR the Present Subjunctive is predominant. In HARAN it
still appears, but marginally. In ITUN it has totally disappeared ousted by the
Periphrastic Present.
To complement Table 4 .4 ,1 offer also the distribution of open conditionals in
the other main 16th century texts. I present below the data on open conditional (and
concessive) if-clauses in Etxepare (1545), and Refranes y Sentencias (1596).
TABLE 4.4.bis: Non-stative open-conditional and concessive protases (Special Corpus).
ETXEPARE (1545) REFRANES (1596)
TOKENS ITEMS TOKENS ITEMS
TOTAL 21 14 11 6
SYNTHETIC
PRESENT
7 3 4 1
PERIPHRASTIC
PRESENT
(1) (1)
2 1
PRESENT
SUBJUNCTIVE
12 10 4 4
OTHER
CONSTRUCTIONS
1 1 1 1
As just mentioned, the Present Subjunctive was the predominant construction for
expressing if-clauses in Old Basque. In Modem Basque, however, it has been ousted
by the Periphrastic Present. Note also in Table 4.4.bis that the Synthetic Present is, in
terms of tokens, quite frequent in the two 16th-century texts presented now.
I add finally the data on open conditionals of stative predicates. Unlike the
above cases of non-stative predicates, stative predicates express if-clauses (as most
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
154
other TAM meanings) mainly by means of the Synthetic Present (including still
Stative Verbal Locutions)
TABLE 4.5: Stative open-conditional and concessive protases (Main Corpus).
LEIZAR HARAN ITUN
TOKENS ITEMS TOKENS ITEMS TOKENS ITEMS
TOTAL 11 4 12 6 10 6
SYNTHETIC
PRESENT
11 4 10 4 10 6
PERIPHRASTIC
PRESENT
0 0 2 2 0 0
6.2. Future time-reference temporal clauses.
I turn now to non-stative when-future clauses. As a matter of fact, it would seem that
when-future clauses might consist of a number of similar but not identical sub
contexts. In the case of non-stative predicates, these sub-contexts would oppose
chiefly extended-phases (i.e. when = while) vs. punctual-no-phases (i.e. when =
once). Therefore, there can be an important deal of variation in this specific context.
The data regarding when-fiiture clauses is given in Table 4.6 below.
TABLE 4.6: Non-stative when-future (Main Corpus).
LEIZAR HARAN ITUN
TOKENS ITEMS TOKENS ITEMS TOKENS ITEMS
TOTAL 16 12 13 10 5 5
SYNTHETIC
PRESENT
1 1 0 0 1 1
PERIPHRASTIC
PRESENT
6 3 6 5 1 1
PRESENT
SUBJUNCTIVE
2 2 0 0 0 0
(PERIPHRASTIC)
FUTURE
2 2 5 3 3 3
OTHER
CONSTRUCTIONS
5 4 2 2 0 0
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
155
Non-stative when-future sentences show indeed an important amount of variation
within each text. To the three main constructions, we have to add in this case the
(Periphrastic) Future, which appears a number of times in all three texts. Moreover,
there is a fifth construction that appears in LEIZAR (5 times) and HARAN (twice).
This is an Old Future Perfect that is currently lost. It is built as a Present Perfect plus
the old futurity marker -ke in the auxiliary ‘to be’ or ‘to have’, as in eriden
dukezuenean (Mt II, 8) ‘when you-pl have found him’, ‘once you have found him’,
‘once you find him. ’
In LEIZAR, the great competition we find for this context (five different
constructions) seems to correlate to some extent with the types of predicate. Thus,
activity predicates, built upon the light-verb egin ‘to do’, appear either in the
Synthetic Present (once) or in the Periphrastic Present (six times), as in othoitz egiten
duanean ‘while you are praying.’ More punctual predicates, on the other hand, seem
to correlate with the Present Subjunctive or the Old Future Perfect; conveying,
especially the latter, a sort of posterior-inchoative meaning, as in eriden dukezuenean
above or ethor dadinean ‘when s/he comes’ = ‘once s/he arrives.’ This correlation is
only partial, in any case. The Future in turn appears to also convey mostly a punctual
sense.
In HARAN and ITUN, nevertheless, the putative correlation just presented is
practically erased. We still might witness a tendency in HARAN for extended
predicates to appear in the Periphrastic Present, while punctual predicates tend to
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
156
appear in the Future. But this correlation is rather vague. We also find in HARAN
one example of the Old Perfect Future (aurkhitu dukezuenean ‘once you find him’),
although this construction appears already in recession at this stage. As for ITUN,
the number of examples is unfortunately too short. The reason for this is a tendency
in this text (which probably reflects the current situation) to use nonfinite
subordinate sentences in when-future contexts, as in aurkitu ondoren (Mt II, 8) ‘after
having found him.’ Future and Periphrastic Present seem, nonetheless, the most
common constructions in this context in current Basque. (The only example here of
the Synthetic Present belongs to the verb joan ‘to go.’)
What is perhaps most important from Table 4.6 is to partly find again the
tendency we uncovered in Table 4.4 (and 4.4.bis). Namely, the disappearance of the
Present Subjunctive from what I called the future time-reference subordinate ‘less-
dynamic’ subtypes, displaced by the Periphrastic Present (with the competition with
the Periphrastic Future in this specific case).
7. COMPETITION FOR THE DOMAIN OF FUTURE TIME-REFERENCE
DYNAMIC SUBORDINATE CONTEXTS.
This section is devoted to examining the domain of what I have termed future time-
reference subordinate ‘dynamic’ contexts. Recall that these include complement
clauses to verbs o f wanting and ordering, and purpose clauses (see section 3.2 of
Chapter 3). This domain of TAM is often called ‘subjunctive mood’, although the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
157
‘less-dynamic’ contexts I studied in the previous section are sometimes included in
this label too.
The competition in historical Basque for this domain is much more simple
than the scenario we found in the previous section. There are only two constructions
in historical Basque for conveying the subordinate dynamic contexts at issue: the
(Periphrastic) Present Subjunctive and the Synthetic Present. The Periphrastic
Present is not attested in these contexts. As a matter of fact, it appears that, for a
construction grammaticalized as present progressive (as is the case of the
Periphrastic Present) to reach these subordinate dynamic contexts, it has to be very
old and eroded in meaning. In other words, the contexts at issue seem to be among
the latest to be reached (if they are reached at all) in the cline of grammaticalization
starting with present progressive (Table 3.1 of Chapter 3). This may be a reason for
the conservativeness of TAM markers in future time-reference subordinate dynamic
contexts (see Bybee et al.1994:230-231).
On the other hand, the Present Subjunctive, which was grammaticalized as a
present telic expression appears as a very appropriate construction to convey
subjunctive meaning, i.e. subordinate dynamic contexts such as purpose clauses. The
Synthetic Present, in turn, could have had a present progressive source (see section
6.1 of Chapter 3). However, since this construction is apparently so old, it does not
seem strange for the Synthetic Present to express the subordinate dynamic contexts
at hand. Moreover, both the Present Subjunctive and the Synthetic Present express
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
158
purpose clauses and complement clauses to ‘want’ and ‘order’ by means of adding a
subordinator morpheme -(e)n. In fact, a Present Subjunctive form such as etor dadi-
is in practice cited as etor dadin ‘so that s/he comes’, with the subordinator ending.
This subordinator morpheme may have had its origin in a purposive marker (see
Haspelmath 1989 for cross-linguistic parallels).
I present in Table 4.7 below the data on the historical competition between
the Synthetic Present and the (Periphrastic) Present Subjunctive for purpose clauses
and complement clauses to verbs expressing ordering and wanting.
TABLE 4.7: Purpose clauses and complement clauses of ‘want’ and ‘order’ (Main Corpus).
LEIZAR HARAN ITUN
TOKENS ITEMS TOKENS ITEMS TOKENS ITEMS
TOTAL 38 26 30 27 10 10
SYNTHETIC
PRESENT
13 6 3 3 0 0
PERIPHRASTIC
PRESENT
0 0 0 0 0 0
PRESENT
SUBJUNCTIVE
25 20 27 24 10 10
In table 4.7, we witness a competition throughout the last centuries between
the Synthetic Present and the Present Subjunctive. In this competition, the Present
Subjunctive has totally displaced the Synthetic Present from purpose clauses. (The
low figures in ITUN are due to the current tendency to use nonfmite subordinate
clauses as purpose clauses: see section 3.5 of Chapter 2.)
It is worth pointing out the great number of Synthetic Present lexical verbs
(i.e. items) appearing in LEIZAR for the case at hand, in comparison with other
contexts. Indeed, it is specially in these subordinate dynamic contexts, and in general
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
159
in all subordinate contexts, where we find the highest type-frequency of Synthetic
Present forms in all texts of Old Basque. Most of the verbs that were subject to
synthetic inflection in Old Basque and have lost this possibility in Modem Basque
appeared attested only in these subordinate contexts (and in the imperative). Thus,
the 13 instances o f the Synthetic Present appearing in LEIZAR in Table 4.7 belong
to 6 different lexical verbs. Now, some of these verbs can hardly be found in other
contexts: e.g. eman ‘to give’, *i(n) ‘to give’, erran ‘to say’, egin ‘to do.’ The three
examples in HARAN, on the other hand, belong to verbs that can still currently
appear in main-clause contexts: jakin ‘to know’, ethorri ‘to come', joan ‘to go.’
In other words, the subordinate contexts at issue (along with the imperative
mood, which will not be studied in this work) witnessed the loss of the last synthetic
forms of a good number of lexical verbs that were still subject to synthetic inflection
in Old Basque. Therefore, we might say that purpose clauses (and in general all
future time-reference non-assertive subordinate clauses) constitute the most
conservative context in the sphere of non-past I am studying. However, it should not
be forgotten that there are other factors for the conservativeness of old declining
forms. The most important among them seems to be high token-frequency, as I will
test in the next section. Thus, the conservative subordinate contexts I am studying
were the last contexts of appearance of synthetic forms for a number of lexical verbs.
Yet, a handful of verbs still maintained synthetic forms in main clauses. (The factors
in this conservation will be the object o f study of the next section.)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
160
It ought to be mentioned too that, in the ease of the subjunctive contexts at
issue, unlike elsewhere, the loss of the Synthetic Present is paralleled by stative
predicates. In other words, the Synthetic Present of all verbs, even of those stative
verbs with the highest token-frequency, disappeared from purpose clauses and
complement clauses to ‘want’ and ‘order’, ousted by the Present Subjunctive. (The
reason for this replacement might be related to a reordering in the TAM system,
whereby the achievement-dynamic auxiliaries were reanalyzed as the auxiliaries of
non-indicative moods: see also section 7.2 of Chapter 5 and section 2.10 of Chapter
7). Therefore, the only construction that expresses these contexts in Modem Basque
is the Present Subjunctive, equally for non-stative predicates as for stative verbs.
Table 4.8 below shows the data on purpose clauses for stative predicates.
TABLE 4.8: Stative purpose clauses and complement clauses of ‘want’ and ‘order’ (Main Corpus).
LEIZAR HARAN ITUN
TOKENS ITEMS TOKENS ITEMS TOKENS ITEMS
TOTAL 6 4 3 3 2 2
SYNTHETIC
PRESENT
6 4 0 0 0 0
PERIPHRASTIC
PRESENT
0 0 0 0 0 0
PRESENT
SUBJUNCTIVE
0 0 3 3 2 2
As a summary of the last two sections, we may say that the displacement of
the Present Subjunctive by way of the Periphrastic Present in the less-dynamic
future-gnomic time-reference subordinate subtypes (section 6), and the displacement
of the Synthetic Present in the dynamic future time-reference subordinate subtypes
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
161
by means of the Present Subjunctive (section 7) are probably the two most notorious
changes in the last centuries of the Basque TAM system in the sphere of non-past.
8. PRESENT PROGRESSIVE VS. PERIPHRASTIC PRESENT: TESTING
THE HYPOTHESES.
We have seen above that the Present Progressive and the Periphrastic Present
constitute a case of same-source layering. Therefore, this is the ideal situation for
testing the hypotheses of Layering-Displacement and Form-Meaning Covariation,
which were formulated in Chapter 1. However, unfortunately the attestation in
historical Basque hardly shows any expansion of the meaning of the Present
Progressive. Therefore, the hypothesis of Layering-Displacement cannot be tested in
this case. On the other hand, the hypothesis of Form-Meaning Covariation can
suitably be tested.
8.1. Testing the hypothesis of Form-Meaning Covariation.
I repeat, from (6) in Chapter 1, the statement of the methodology for testing the
hypothesis of Form-Meaning Covariation.
(2) Testing the hypothesis of Form-Meaning Covariation:
Given two TAM constructions, A and B, in a synchronic stage o f a language
X, so that A and B are in competition for similar TAM meanings in stage X.
If we have solid evidence that A and B have had the same source of
grammaticalization, then: If A is formally more bound than B (i.e. if A is
formally older than B), then its meaning in stage X should have progressed to
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
162
the right in the corresponding cline of semantic grammaticalization no less
than the meaning of B.
All the premises in (2) are met for A = Periphrastic Present and B = Present
Progressive, A being more bound (formally older) than B. Therefore, the meaning of
the Periphrastic Present in any stage of Basque should have progressed to the right in
Table 3.1 (Chapter 3) no less than the meaning of the Present Progressive.
This prediction is fulfilled by the data in the corpus. As we have seen in the
preceding sections, the Present Progressive (B) is only documented in historical
Basque conveying meanings included among the young subtypes in the cline-schema
of Table 3.1. The Periphrastic Present (A), on the other hand, is always documented
expressing meanings more to the right in the cline. Concretely, the attestation shows
a Periphrastic Present conveying meanings included in the mature subtypes and also
in some o f the old subordinate subtypes.
9. CONSERVATION OF THE SYNTHETIC PRESENT: FACTORS.
This section is devoted to studying the factors that play a role in the conservation of
the Synthetic Present. One of the most important factors seems to be token-
frequency, as was proposed in sections 3.5 and 4.3 of Chapter 1. Thus, the first task
in this section will be to test the hypothesis of Conservation of high token-ffequency
forms, which correlates degree of conservativeness with token-ffequency (section
9.1). Then, I will discuss other possible factors that may play a role in the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
163
conservation of the Synthetic Present, especially lexical-semantic of verbs (section
9.2).
9.1. Testing the hypothesis of Conservation of high token-frequency forms.
In order to test the hypothesis of Conservation of high toke-frequency forms, i.e. (4)
in Chapter 1 ,1 present in Table 4.9 below the figures on token-frequency of the most
frequent lexical verbs in the database.
TABLE 4.9: Token-frequency of most frequent lexical verbs (Main Corpus).
LEIZAR HARAN ITUN TOTAL
IZAN ‘to be’ 103 111 62 276
ERRAN/ESAN ‘to say’ 67 88 39 194
ETORRI ‘to come’ 36 33 24 93
*IO ‘to say’ 51 11 22 84
EGIN ‘to do’ 37 35 10 82
*EDUN ‘to have’ 21 15 26 72
JOAN ‘to go’ 15 30 22 67
EMAN ‘to give’ 17 28 18 63
IKUSI ‘to see’ 16 21 17 54
HARTU ‘to take’ 11 16 13 40
ERAMAN ‘to carry, to take to’ 11 13 9 33
JARRAIKI ‘to follow’ 13 11 9 33
SARTU ‘to enter’ 11 13 9 33
UTZI ‘to leave, to quit’ 10 14 5 29
NAHI *EDUN ‘to want’ 10 11 7 28
JAKIN ‘to know’ 7 6 14 27
EGON ‘to stay, to be at’ 4 10 13 27
Obviously, the figures in Table 4.9 can only be taken as approximate. The nature of
the texts used makes one think that the data in Table 4.9 can deviate from the real
frequency of use in the spoken language. The two verbs meaning ‘to say’ in Table
4.9, for instance, may possibly display a somewhat distorted ranking, due to the high
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
164
frequency in the texts of introductory expressions such as ‘and then he said.’ ’ In any
event, the data in Table 4.9 does not appear too incorrect either.
It is remarkable that out of the 17 lexical verbs I give in Table 4.9 (i.e. the
most frequent in the database) only three are not documented as being or having
been subject to synthetic inflection. These are hartu ‘to take’, sartu ‘to enter’ and
nahi *edun ‘to want.’ The latter is actually a stative verbal locution of the type that
will be studied in section 10. Thus, if we analyze nahi *edun as being a lexical
phenomenon that grammatically depends on the light-verb hosting it (i.e. *edun ‘to
have’), then nahi *edun could also he considered as being subject to synthetic
inflection. Another verb that has very few attested cases of synthetic forms (if any,
other than imperatives) is utzi ‘to leave, to quit.’ In this case, though, the shape of the
verb, i.e. *e-utz-i, makes one suspect that it was subject to synthetic inflection in
earlier times (see sections 6.2.2 of Chapter 1 and 6.4.1 of Chapter 3).
In any case, the first nine verbs in the ranking have attested synthetic forms.
These data, therefore, are a piece of evidence in favor of the hypothesis of
Conservation of high token-ffequency forms. We can conclude thus that the data do
not falsify the hypothesis.
However, among the 14 verbs in Table 4.9 that have documented synthetic
forms in historical Basque, only eight are subject to synthetic inflection in Modem
Basque. Moreover, some of the verbs that lost their synthetic forms in the interval
from Old to Modem Basque are among the most token-frequent in the list: e.g. erran
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
165
/ esan ‘to say’, egin ‘to do’, eman ‘to give’, ikusi ‘to see.’ Therefore, it appears that,
although token-ffequency is clearly a very important factor in the conservation of the
Synthetic Present (and the other synthetic forms of Basque), there may be other
factors conditioning this conservation (see discussion below). In any event, it seems
necessary to acknowledge that there may be an element of randomness in the
conservation of declining forms.
In the next section, I discuss the relation between the verbs subject to
synthetic inflection in Modem Basque and their lexical-semantics.
9.2. Lexical classes of verbs and the conservation of the synthetic forms.
It is not too far from the facts to state that in Modem Standard Basque the verbs
subject to synthetic inflection are the following: izan ‘to be’, *edun ‘to have’, egon
‘to be at’, eduki ‘to have’ ,j akin ‘to know\jo a n ‘to go’, etorri ‘to come’, ibili ‘to
wander’, ekarri ‘to bring’, eraman ‘to carry, to take to’, *io ‘to say’, and perhaps a
few others used in specific expressions, such as eritzi ‘to consider’, etzan ‘to lie’,
irudi ‘to seem.’ All these verbs but *io can be divided into two main lexical groups:
a) stative verbs: izan, *edun, egon, eduki, jakin, eritzi, etzan, irudi, and
b) motion verbs: joan, etorri, ibili, ekarri, eraman.
It appears difficult to come up with an account of these data only in terms of
lexical-semantics of predicates. From a synchronic perspective, one can readily find
many exceptions to a putative correspondence between lexical classes and the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
166
conservation of the Synthetic Present. Yet, it does seem that lexical-semantics of
verbs plays some role in the conservation of the synthetic forms of Basque (cf. Haase
1994:284). In what follows, I introduce some of the relevant issues in this discussion,
and outline some hypotheses towards an explanation.
1) There might be various factors playing a role in the conservation of the
Basque synthetic forms. These factors can furthermore interplay with one another.
2) Token-frequency is one of the major factors (as I have tried to prove in
section 9.1 above).
3) Lexical-semantics of predicates may be another factor. However, it cannot
be viewed from a purely synchronic perspective, but need also a diachronic account.
In other words, the situation of conservation we find with the synthetic forms of
Basque is partly a byproduct of the history of this language.
4) Thus, the current situation may partly depend on when a given verb was
created or borrowed. That is, whether a given verb appeared in the language before
or after the loss of the productivity of the synthetic rule, etc.
5) We can only very roughly explain why some inherited verbs got to have
synthetic forms in Pre-Basque, while some others did not.
6) Similarly, we can hardly explain why some inherited stative verbs got to
have proper synthetic verbs, while some others were only verbal locutions hosted by
the former.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
167
7) Among the verbs that got to have synthetic forms in Pre-Basque, only a
handful of them were stative. The bulk of the stative verbs were verbal locutions (as
we will see in the next section).
8) However, this handful of stative verbs subject to synthetic inflection has
rather consistently maintained their synthetic forms.
9) It would seem then that the handful of stative verbs that for one reason or
another happened to have synthetic forms have not experienced the same
competition that dynamic predicates have had with the periphrastic constructions.
10) We could tentatively propose that this small group of stative verbs did not
have to compete with the Periphrastic Present for the sphere of present tense (i.e. to
convey stative present meaning). This was so precisely because of their being stative;
i.e. for their being rather incompatible with a construction grammaticalized as a
present progressive.
11) Dynamic predicates, on the other hand, did experience the competition
with the periphrastic constructions and concretely with the Periphrastic Present. Most
of them were thus ousted from the sphere of strict present tense and were confined to
future time-reference contexts.
12) For some reason, some motion verbs (Joan, etorri, etc), the verb *io ‘to
say’, and a very small number of other dynamic verbs (e.g. irakin ‘to boil’) were not
ousted from pure present tense contexts.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
168
13) It appears extremely difficult and perhaps unnecessary to question why
high-frequency verbs such as egin ‘to do’, eman ‘to give’, or (moreover) esan ‘to
say’ (apparently with the same meaning as *io) were confined to future time-
reference contexts, while the verbs in 12 above were not. (It may be just a matter of
chance.)
14) One could tentatively contemplate the possibility that motion verbs (joan
‘to go’, etorri ‘to come’, etc), *io ‘to say’, etc. have some kind of special reading that
is not readily subsumable under a primarily present habitual meaning. This
hypothesis, though, appears as extremely speculative at this point.
15) In any event, the bulk of the dynamic verbs that happened to be subject to
synthetic inflection (probably a minority in terms of type-frequency after the massive
borrowing from Latin-Romance), which were already confined to future time-
reference contexts, started to lose their synthetic forms, mainly ousted by the Present
Subjunctive.
16) The gradual loss of the dynamic synthetic forms started with low token-
frequency verbs. The first texts of Basque give us the possibility of witnessing the
loss of the last dynamic synthetic forms (i.e. those belonging to the most frequent
verbs within this group: egin ‘to do’, esan ‘to say’, eman ‘to give’, *i ‘to give’, etc).
17) The handful o f stative verbs that happened to have synthetic forms,
however, did not lose them. (Surprisingly, motion verbs and *io ‘to say’ did not lose
their synthetic forms either.)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
169
18) In any case, the distribution of motion verbs subject to synthetic
inflection is not the same in all dialects. Synthetic forms of the verb etorri ‘to come’
are not common in eastern dialects (see Lafitte 1962:270). In the eastern dialects too,
the Synthetic Present of the verb joan ‘to go’ appears to suffer a strong competition
with the Periphrastic Present (cf. Lafitte 1962:256).
10. COMPETITION FOR STATIVE PRESENT MEANING: TYPE-
FREQUENCY.
In this section, I offer the data on the competition for stative present and discuss the
role o f type-frequency, touching upon the issue of prototypicality. (For more
discussion on prototypicality see section 2 of Chapter 7.) I already put forward in
section 3.4 above that the competition for stative present meaning (and other similar
stative contexts) takes place among three types of formations. These are the
Synthetic Present, the Periphrastic Present and what I called the Stative Present
Verbal Locutions (see section 6.5 of Chapter 3). I also mentioned in section 3.4
above that this competition is of a kind different to the other cases of competition we
have seen so far in this work: first, it is a competition for generality o f use; second,
the Stative Present Verbal Locutions are most possibly not to be analyzed as proper
TAM constructions. These locutions may constitute a lexical rather than grammatical
phenomenon. Therefore, the competition at issue appears quite peculiar.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
170
10.1. On the synchronic status of the Stative Present Verbal Locution.
The competition between the Periphrastic Present (e.g. maitatzen du ‘s/he loves’) and
the Stative Present Verbal Locutions (e.g. maite du ‘s/he loves’) is often analyzed as
a matter of choice between two synonymous lexical entries: e.g. the verb maitatu ‘to
love’ and the verbal locution maite *edun ‘to love. ’ According to this synchronic
analysis, the verb maitatu would be a typical verb in the Basque system: i.e. it would
not be subject to synthetic inflection. Thus, maitatu will generally use the
Periphrastic Present to convey stative present: e.g. maitatzen du. This synchronic
analysis would also propose that maite *edun would be a different lexical verb, of
the verbal-locution type. Its conjugation would lie on that of its light-verb *edun ‘to
have’, subject to synthetic inflection. Therefore, maite *edun will convey stative
present meaning by way of what would be analyzed as a Synthetic Present form (of
*edun): e.g. maite (nominal) + du ‘s/he has.’
Although there are important complications with this analysis, which I cannot
discuss in detail here, I would regard it as essentially valid. As a matter of fact, up to
this point in the discussion I have been including expressions such as maite du
among the Synthetic Present forms. However, from the perspective of the theory of
grammaticalization and of language design, we may want to understand the
processes that may have led to a system like the one we find in Basque. Towards this
end, in what follows I offer the data on the competition for stative present in the
database and discuss it. Then, I try to explain the lexical status of the Stative Present
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
171
Verbal Locution by considering it constitutes a case of construction only applicable
to a low type-frequency class of predicates.
10.2. The competition for stative present.
I give in Table 4.10 below the data on stative present meaning (i.e. continuous non
progressive present) in the Main Corpus. I only include in Table 4.10 core-stative
predicates; that is, I exclude primarily dynamic predicates that can occasionally
assimilate to stative present readings (e.g. ibili ‘to walk\/'ocm ‘to go’, etc.). In Table
4.10, as in the ensuing tables and discussion, the Stative Present Verbal Locutions
will be considered separately from the proper Synthetic Present forms.
TABLE 4.10: Stative present of core-stative predicates (Main Corpus).
LEIZAR H A R AN ITUN
TOKENS ITEMS TOKENS ITEMS TOKENS ITEMS
TOTAL 78 14 82 18 72 18
SYNTHETIC
PRESENT (proper)
65 5 64 4 55 6
STATIVE PRESENT
VERBAL LOCUTION
12 8 16 12 16 11
PERIPHRASTIC
PRESENT
1 1 2 2 1 1
The data in Table 4.10 show the competition for stative present between the three
formations at issue. All three texts display very similar data. The data reveal that the
Periphrastic Present is not commonly used for conveying stative present. There
might be a tendency in recent times, especially in the eastern dialects, towards using
the Periphrastic Present more often in these contexts: e.g. maitatzen du ‘s/he loves’,
merezitzen du ‘s/he deserves\faltatzen da ‘s/he is missing’, inportatzen du ‘it is
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
172
worth.’ However, this tendency, which HARAN seems to be pointing to (see also
Table 4.11 below), is not recorded in ITUN, a rather western text.
The other two formations in the competition for stative present display a very
interesting behavior. Regarding token-ffequency, the Synthetic Present is clearly the
most important. However, concerning type-frequency (i.e. number of different
lexical verbs), the Stative Present Verbal Locutions are preponderant in all three
texts. As a matter of fact, we have already seen in section 9 that the Synthetic Present
is a declining form only appearing for high token-ffequency verbs. That is, the
Synthetic Present is very common for expressing stative present. Yet, this only
occurs for a handful of verbs. Therefore, it appears that the ‘rule’ for conveying
stative present is to use a Verbal Locution.
10.3. Type-frequency vs. token-frequency data of stative predicates.
It seems worth verifying whether the high type-ffequency of the Stative Present
Verbal Locution holds for other contexts of stative predicates, and also for larger
amount of data. Thus, I present in Table 4.11 the distribution of all core-stative
predicates in any context within the sphere of present / non-past. (I only include in
the table the three constructions at issue in this section, and exclude other minor
constructions.)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
173
TABLE 4.11: Core-stative predicates in the cline of present (Main Corpus). (Incomplete).
LEIZAR HARAN ITUN
TOKENS ITEMS TOKENS ITEMS TOKENS ITEMS
TOTAL 142 22 156 36 118 28
SYNTHETIC
PRESENT (proper)
106 6 99 6 80 7
STATIVE PRESENT
VERBAL LOCUTION
34 14 42 20 33 17
PERIPHRASTIC
PRESENT
1 1 5 4 1 1
Table 4.11 corroborates the observation made above. Namely, while the Synthetic
Present is undoubtedly the most important construction in stative predicates in terms
of token-frequency, the most relevant in terms of type-frequency is certainly the
Stative Present Verbal Locution. In other words, the majority o f stative verbs of
Basque recur to this ‘rule’ to express stative present and related meanings.
10.4. Type-frequency, prototypicality and lexical vs. grammatical status.
The observation just presented would apparently seem to be at odds with the
synchronic analysis of the Stative Present Verbal Locution put forward in section
10.1. That is, it could perhaps appear surprising that the general rule for expressing
stative present in Basque is not actually to be analyzed as a grammatical
phenomenon. The solution seems to come from the comparison of the data on type-
frequency of stative and dynamic predicates. More specifically, it seems that in a
situation of two grammaticalizing constructions in complementary distribution along
the lexical-semantic dimension, the one with lower type-frequency will often be
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
174
analyzed as having a lexical status, while the one with high type-frequency is
analyzed as the grammatical ‘model.’
Thus, I am proposing the following scenario in the competition for the
domain of present tense in Basque. The Periphrastic Present and the Stative Present
Verbal Locution constitute a case of constructions grammaticalized in a
complementary distribution regarding the lexical-semantic dimension. While the
former had a grammaticalization source fitted for dynamic verbs (concretely, a
present progressive), the latter had a source fitted for stative predicates (see section
6.5 of Chapter 3). My hypothesis is then that one of the reasons for the lexical status
of the Stative Present Verbal Locution is the low type-frequency of the class of
stative verbs to which this construction is applicable.
I test this hypothesis below. To that end, I need to prove that the type-
frequency of the class of dynamic predicates in Basque is considerably higher than
the type-frequency of the class of stative predicates. (We would also expect that for
the domain of present tense, the Periphrastic Present might be the main rule for
expressing dynamic predicates, in the same way that the Stative Present Verbal
Locution appears to be the rule for expressing stative predicates. Note, however, that
in the case of dynamic predicates we would also expect more competition with other
constructions different from the Periphrastic Present: cf. Dahl 1985:28)
I offer in Table 4.12 below the data of dynamic predicates corresponding to
the data of stative predicates we saw in Table 4.11.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
175
TABLE 4.12: Dynamic (non-stative) predicates in the cline of present (Main Corpus). (Incomplete).
LEIZAR HARAN ITUN
TOKENS ITEMS TOKENS ITEMS TOKENS ITEMS
TOTAL 268 143 250 134 165 101
SYNTHETIC
PRESENT (proper)
53 12 34 9 38 6
VERBAL
LOCUTIONS
10 7 7 5 8 5
PERIPHRASTIC
PRESENT
132 64 130 59 94 57
Table 4.12 confirms the prediction on type-frequency. If we compare the figures on
type-frequency (i.e. number of verbal items) in Table 4.11 and in Table 4.12, we can
see that the type-frequency of dynamic predicates is much higher in all three texts:
22-36-28 vs. 143-134-101. (In the case of dynamic predicates, the main ‘rule’ for
this sphere of meaning is the Periphrastic Present, which is also significantly more
type-frequent in absolute terms than the Stative Present Verbal Locution: 14-20-17
vs. 64-59-57.)
It follows from the above discussion the prediction that sources of
grammaticalization primarily affecting stative predicates will hardly achieve high
generality of use. This is so because it will be very difficult for a primarily stati ve
source, with low type-frequency, to get to be applicable to non-states too. The
opposite extension, on the contrary, appears rather unproblematic. Stative predicates,
thus, could fluctuate between having their own specific markers and ‘borrowing’
those of the dynamic verbs. Yet, the whole verbal system will hardly be organized
upon a primarily stative construction. As Kemmer (1992:150) puts it, “the marking
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
patterns of stative events show a great deal of cross-linguistic variation since
major event prototypes all represent actions rather than states.”
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
177
CHAPTERS
THE GRAMMATICALIZATION OF PAST PERFECTIVE
ACROSS LANGUAGES AND IN BASQUE
1. INTRODUCTION.
This chapter and the next are devoted to the TAM domain of past perfective and
related meanings. I will be using the term ‘past non-imperfective’ to refer to this
sphere of TAM. The use of the term ‘past non-imperfective’ implies that past
imperfective will not be considered in this study. The TAM meanings that are
included in this term are resultatives, perfects, past perfective, completive, evidential,
experiential, etc. (See the Glossary for clarification on the terminology used in this
chapter.)
The present chapter examines, on the one hand, the sources and clines of
grammaticalization of past non-imperfective across languages, and, on the other, the
grammaticalization sources of the Basque constructions in this TAM sphere. (These
are: the Synthetic Past, the Old Basque Periphrastic Aorist, the Perfect-Perfectives,
the Double-compound Perfects, and the Resultatives. See Chapter 2 for descriptions
of these constructions.) Chapter 6, in turn, studies the domain of past non-
imperfective in Basque from a historical-grammaticalization perspective. In other
words, the contents and methodology we followed in Chapters 3 and 4 will be
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
178
paralleled now in Chapters 5 and 6, with the difference that the former were devoted
to the domain of present and the latter will examine the domain of past perfective.
As I did in Chapter 3, in the present chapter I will present and justify
unidirectional clines of grammaticalization in the TAM sphere under discussion. I
will also try to ascertain the source of grammaticalization of most of the Basque
constructions in this domain. These tasks are partly a preparation for the next
Chapter. That is, they provide the basis for distinguishing various scenarios of
competition, which will be used in different ways to organize, understand and
‘predict’ the behavior of our data. (See Chapter 1 for more details on the
methodology and hypotheses in these chapters, and for discussion on the term
‘prediction.’)
The division of this chapter is as follows. In section 2 ,1 present the clines of
grammaticalization starting with present resultative. While the main cline starting
with present resultative is well understood, there are two other less studied paths one
of which, concretely the path involving experiential perfect, would need further
research. Section 3 deals with the cline starting with past resultative. I also put
forward in this section the issue of pluperfects in narration sequences. Sections 4 and
5 examine other less common sources of grammaticalization of past non-
imperfective. Namely, dynamic sources (section 4) and what might be called
narrative-discourse sources (section 5). Section 6 tries to clarify the use of the
Double-compound Perfect constructions found in languages like German, French,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
179
Occitan and Basque. Finally, section 7 determines the source of grammaticalization
of the majority of the Basque constructions at issue, and presents hypotheses on the
most obscure.
2. THE CLINES STARTING WITH PRESENT RESULTATIVE.
2.1. The main cline.
The kind of Present Resultative construction we are mostly interested in as the
initiator of the cline is the one we will find in our Basque case (see sections 7.4 and
7.6 below): namely, the combination of the stative verbs or quasi-auxiliaries ‘to be’
and/or ‘to have’ in present tense plus a past participle. In any case, the general
evolution I will study in this section is probably applicable to most present
resultative constructions of any kind (cf. Bybee et al. 1994:64-65). In general terms,
the overall change can be defined as a gradual loss of the semantic relationship with
the present moment, and a consequent gain of the expression o f pure past
(perfective) meaning. The cline progresses approximately along the lines in (1)
below.
(1) present resultative -> present perfect -> hodiernal perfective -> perfective ->
(simple past)
The first step is the change from present resultative to present perfect (probably via
perfect of result). Bybee & Dahl (1989:70) explain this change in the following way:
“The shift represents a change in emphasis: whereas resultatives focus on the state
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
180
which is the result of a previous event, perfects focus on the event itself which leads
to the extension to non-resultative cases.” This change entails a big increase in
generality of use. That is to say, the construction undergoing the change shifts from
being lexically restricted to verbs of result towards lexical generalization.
(Unfortunately, this shift towards generality of use is very difficult to track in the
Basque data, and I will not deal in this case with issues related to lexical-semantics
of verbs.) Also, “the semantic change connected with the transition from resultatives
to perfect is accompanied by changes in grammatical properties” (Bybee & Dahl
1989:70).
The second step in (1) evolves from present perfect to hodiernal perfective; it
would be perhaps more precise to state that this change progresses from perfect of
result to hodiernal perfective, via perfect of recent past. Actually, all of these senses
are often conveyed by the same Present Perfect construction in different languages.
As Comrie (1976:60) says, “in many languages, the perfect may be used where the
present relevance of the past situation referred to is simply one of temporal
closeness, i.e. the past situation is very recent.” In fact, resultativity seems to be
favored by recentness, so that these two features often correlate. Therefore, an
explanation for the shift from perfect of result to recent past would be that a
construction “which is used in the sense of a ‘perfect of result’ will thus
automatically be used more frequently of recent events” (Dahl 1985:136).
“Moreover, the fact that recent past is often on the same day as the moment of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
181
speech leads to the further inference that the Present Perfect signals past events on
the same day” (Bybee et al. 1994:87, citing Schwenter 1993). After this last shift
towards hodiernal time-reference occurs, definite time-adverbials can be allowed
within the restricted hodiernal frame.
The third step in (1) progresses from hodiernal perfective to general
perfective. That is, it entails a total loss of the semantic link with the present moment
(including recentness and hodiernal-frame). As Bybee et al. (1994:86) put it, “the
specification of current relevance is lost.” Likewise, Comrie (1976:61) says:
“Gradual relaxation of the degree of recentness required for use of the Perfect seems
to have been a key part of the development of the Perfect in many Romance
languages to oust the Simple Past completely.” What is not totally clear, though, is
how this final relaxation takes place which involves the shift from hodiernal past
reference to any past time-reference, including narrative contexts. In other words, it
is not straightforward to me what the beachhead is that aids in trespassing the limits
of hodiernal reference. Bybee et al. (1994) cite again Schwenter (1993), for whom
this semantic extension does not seem so enigmatic: “Schwenter argues that, once
the Present Perfect is established as a hodiernal past rather than a past with current
relevance, it will gradually extend beyond the limits of the day and will eventually
become a general perfective” (1994:87).
The final step in (1), i.e. that from perfective to simple past, is not so well
documented as the previous ones. Moreover, it appears to importantly depend on the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
182
presence or absence of a construction with imperfective past meaning in the
language (Bybee et al. 1994:85). In any case, Bybee et al. (1994:92) propose the
hypothesis that in some cases pasts “are further developments from perfectives. [...]
Diachronically this would mean that, as perfectives develop into pasts, they
gradually come to be used in an [aspectually unmarked] past sense with all classes of
verbs, including stative verbs.”
As a conclusion to the discussion so far, we can propose the schema in Table
5.1, which repeats the diachronic cline in (1) with more detail.
TABLE 5.1: Diachronic semantic subtypes in the main cline starting with present resultative.
1. YOUNGEST 2. CENTRAL-YOUNG 3. CENTRAL-OLD 4. OLDEST
present resultative
perfect of result
perfect of recent past
[experiential perfect]
hodiernal perfective perfective
narrative
(simple past)
The main feature that defines the meanings grouped within the youngest subtype of
the cline is ‘present relevance.’ In the central-young subtype the sense of present
relevance has blurred into ‘hodiernal frame.’ Therefore, the main difference between
the youngest and the central-young subtypes is that the latter but not the former can
collocate with definite time-adverbials (within the restricted hodiernal frame). In the
central-old subtype, in turn, there is no sense of present relevance whatsoever, and
any kind of definite time-adverbial is acceptable, as it is applicable to punctual-like
narrative sequences. (As mentioned above, the change towards the oldest subtype is
not so well attested as the others.)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
183
2.2. Experiential perfect.
There is a label in Table 5.1 that has not been cited in the diachronic evolution
outlined above. This is the ‘experiential perfect.’ This meaning is probably the most
distinct of all the uses typically covered by a Present Perfect. Thus, some languages
have a special construction to express experiential perfect meaning. Comrie
(1976:59) cites the cases of Mandarin Chinese and Kpelle, both having similar
distinctions between sentences like ‘have you eaten the shark’s fin?’ vs. ‘have you
ever eaten shark’s fin?’ (Mandarin) or ‘I have fixed it (sc. and it works)’ vs. ‘I have
indeed on one or more occasions fixed it, but it keeps breaking down’ (Kpelle).
Dahl (1985:142-143) claims that not all the cases of seemingly ‘experiential
perfect’ constructions have in fact the same range of meaning. Thus, he uses the
label ‘experiential’ (EXPER), which is similar but apparently not necessarily
identical to ‘experiential perfect.’ He cites Inoue (1975) as providing “the fullest
description [...] of the semantics of an EXPER category in an individual language”
(1985:141): i.e. the Japanese -to koto ga aru construction. Following Inoue, Dahl
(1985: 141-142) argues that, when the ‘experiential’ construction is used,
she says, ‘it is implied that the activity, event or state obtained at one or more
points in a stretch of time extending from the past to the present’ and,
moreover, ‘that the activity, event or state is repeatable.’ The latter can be
reinterpreted as saying that the sentence must concern a generic activity,
state, etc., rather than an individual or specific one. [...] The use of the
EXPER seems to be disfavored even by the mere presence of a definite
temporal frame. [....) Inoue [...] says that as ‘the time span becomes shorter
... the sentence becomes increasingly unacceptable.’
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
184
Although this issue needs farther research, it seems we are dealing here with
a distinct cline of evolution; one which may start with present resultative, but which
progresses somewhat in parallel with the cline we saw in (1) above. If this were
correct, this cline would evolve along the lines in (2) below.
(2) a. present resultative experiential perfect -> frequentative perfect ->
(imperfective perfect)
b. present resultative -> experiential perfect -> remote perfect
The clines in (2) would depart from the main cline in (1) at the point of the
experiential perfect. Since the experiential perfect contrasts with the perfect of recent
past in that the former does not necessarily entail recentness, this seems to lead to the
inference that a construction conveying experiential perfect expresses actually a
sense of remote past with present relevance (i.e. a remote perfect). Hence the
unacceptability that Inoue mentions when the time span between the event-time and
the present-time becomes shorter.
More importantly, there seems to be another inference from experiential
perfect towards repeatability and from here to frequentativity, thus yielding a sort of
frequentative perfect. This cline of development seems to further lead to a meaning
that we may perhaps call ‘imperfective perfect.’ Recall that the main cline of
grammaticalization of perfects presented in (1) evolves towards perfectivity: recent
past, hodiernal perfective, perfective. In the sub-cline(s) I am proposing now in (2),
however, the repeatability of the experiential perfect seems to lead to a sense of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
185
imperfectivity combined with perfect meaning, where boundedness and definiteness
are avoided (cf., for instance, the Mandarin and Kpelle examples above). Thus, in
languages with no distinct Perfect, such as Russian, both the Perfective and the
Imperfective can express perfect meaning; in many cases, Russian “uses the
Perfective to correspond to explicitly perfect forms in languages that have such
forms” (Comrie 1976:63). However, “Russian will use the Imperfective where the
English Perfect refers to a habitual situation, as in ‘they continue to use the old
methods, which have already on many occasions justified themselves’” (1976:63).
Furthermore, Dahl (1985:143) mentions that the following example in his
questionnaire is seldom expressed by an Experiential construction: “(Context: A: I
want to give your brother a book to read, but I don’t know which. Is there any of
these books that he READ already?). Sentence: B: (Yes,) he READ this book.” This
sentence is almost always expressed by a Perfect construction cross-linguistically,
and hardly by means of an Experiential. Notice that the sentence at issue is not
“repeatable”, plus it is “individual” and “specific.” Thus, it would correlate with a
common Perfect, but less so with an Experiential, i.e. with a frequentative or
imperfective perfect.
2.3. Evidential.
There is one further cline of grammaticalization related to present resultative and
present perfect, which also seems to progress somewhat in parallel with the cline in
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
186
(1). This is the cline that involves “’ inferential” (Comrie 1976:108-110), “evidential”
(Chafe & Nichols eds. 1987), “quotative” (Dahl 1985:149-153) and “fictional”
meanings, among perhaps other similar semantic values.
This cline seems to also start with present resultative, like the paths in (1) and
(2) above. However, it bifurcates from the main cline in (1) and covers a quite
different sphere of TAM meaning. The first association leading towards this
bifurcating cline seems to be the implicature from present resultative or perfect of
result to “inference from results, which indicates that a past action is known or
inferred on the basis of a current state” (Bybee et al. 1994:96), as in the English
sentence it has been raining (Dahl 1985:152). Then, “it appears likely that from the
inference-from-results use [a construction grammaticalized as present resultative]
can generalize to include inference from reasoning (without direct results) and
reported evidence, that is, the full range of indirect evidence” (Bybee et al. 1994:97).
As Dahl (1985:153) puts it, “the quotative use [of the Present Perfect] is tertiary in
the sense that it has developed as an extension of the secondary inferential use.”
From evidential uses, “such a form is also used for the narration of folk tales,
demonstrating another path by which [perfects] can come to he used in narrating
sequences of events” (Bybee et al. 1994:97).
Thus, we may propose the following cline for the grammaticalization of
evidential meanings.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
187
(3) perfect of result -> inferential -> unwitnessed perfective -> fictional
narrative
As Bybee et al. (1994) point out, the outcome of the cline in (3) is partly similar to
that of the main cline in (1). Both paths evolve towards the loss of the semantic
relationship with the present moment. “In each case the change brings the
[construction] closer to signaling a simple past action” (1994:96; cf. also Dahl 152-
153 for the case of written Swedish). If fictional narrative could become simple
narrative, it may be perhaps the case that the cline in (3) constitutes a beachhead
alternative to the cline in (1) for Present Perfects to acquire pre-hodiemal perfective
meaning. This might be the case of some French-Basque varieties (cf. section 7.4
below; but this hypothesis would need further research. (See discussion in Chapter 6,
section 8.)
3. THE CLINE STARTING WITH PAST RESULTATIVE.
3.1. The main cline.
The cline I will study in this section starts with past resultative meaning, as opposed
to the present resultative that initiated the clines considered in the preceding section.
More concretely, I am mostly interested in constructions grammaticalized as the
combination of a past participle plus the stative verbs or quasi-auxiliaries ‘to be’
and/or ‘to have’ in past tense, and specifically past imperfective, if the language has
a perfective / imperfective opposition. The very first stages of this path of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
188
grammaticalization are parallel to what we saw for the cline examined in (1) above,
with the difference of past time-reference instead of the present time-reference of the
present resultative. Moreover, it can be said in general terms that the overall
evolution in (1) and that examined in the present section progress along similar lines
and ultimately according to the same motivation. Namely: the gradual loss of the link
with a reference-time point and consequently the gradual gain of absolute past
perfective meaning.
There are, however, several reasons that recommend a separate study of the
two clines. Yet, I will not pursue this issue any further here. (See Comrie 1985:77-
82, Squartini 1999, and the Glossary, for discussion.) In any event, we would have in
this case a cline practically parallel to (1) above, which may be outlined as in (4)
below. (I also repeat (1) for convenience.)
(1) present resultative -> present perfect -> hodiernal perfective -> perfective ->
(simple past)
(4) past resultative -> pluperfect -> remote perfective perfective (simple
past)
As in the preceding case of the cline in (1), we could refine the cline in (4) in
more detail, as in Table 5.2 below.
TABLE 5.2: Diachronic semantic subtypes in the cline starting with past resultative.
1. YOUNGEST 2. CENTRAL-YOUNG 3. CENTRAL-OLD 4. OLDEST
past resultative
perfect-in-the-past
past-in-the-past
past temporal frame
remote perfective
perfective
narrative
(simple past)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
189
The meanings grouped under the youngest subtype of the cline all share their being
‘anaphoric’, i.e. ‘absolute-relative.’ That is, they all “express anteriority with respect
to a reference time” (Squartini 1999:56). However, among them, only past-in-the-
past meaning is compatible with definite time-adverbials. Squartini’s (1999) study
seems to verify that perfect-in-the-past is an earlier evolution from past resultative
than past-in-the-past. It appears that the Zuritiiutsch Pluperfect, seemingly a rather
young construction, only covers perfect-in-the-past meaning (including “reversed
results” 1999:62-68). On the other hand, the Portuguese Synthetic Pluperfect,
apparently a very old form, is restricted to past-in-the-past meaning, with definite
event-time adverbials (1999:68-72). In any case, the most common synchronic
situation cross-linguistically seems to be for one construction to cover perfect-in-the-
past and past-in-the-past meanings (usually called Pluperfect).
Now, in the central-young subtype, the anaphoric sense linked to a reference
time-point has blurred into ‘past temporal frame’, i.e. a “relative remoteness”
(Squartini 1999:58, citing Hedin 1987:30-32). Then, the idea of ‘past temporal
frame’ suffers further relaxation into ‘remote perfective.’ Therefore, the main
difference between the youngest and the central-young subtypes is that the former is
anaphoric (i.e. absolute-relative) while the latter is deictic (i.e. absolute, though still
within a remote frame). Finally, in the central-old subtype there is no anaphoric or
remote sense, and in principle any definite time-adverbial is acceptable: even those
with hodiernal reference; although, as mentioned above, this depends on the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
190
evolution of other remoteness distinctions in a language. A construction in the
central-old stage of this cline could also be applicable to punctual-like narrative
sequences.
3.2. Perfective pluperfect (narrative pluperfect).
We turn now to narrative contexts (see the Glossary for terminological clarification).
In a narrative context there is a strong “tendency for linear order of clauses to follow
chronological order of events” (Comrie 1985:67). That is, the use of a sequence of
Perfectives following a chronological order seems the most economic means to
create a narrative discourse, as in veni, vidi, vici (cf. Dahl 1985:112) or John arrived;
Mary left, which “is most likely to be interpreted as meaning that John’s arrival took
place first, then Mary’s departure” (Comrie 1985:67). On the other hand, “if for
some reason it is desired to present events in other than chronological order, the
pluperfect is an ideal mechanism for indicating this, as [in] John arrived; Mary had
lefr (1985:67).
However, it is not uncommon to find that a narrative context where the linear
order of clauses keeps to the chronological order of events does have an extra
indication of this temporal ordering. That is, it is frequent (especially in the written
language) that the first clause of a series of two (or more) narrative sentences appears
subordinated to the last and is marked for anteriority. I will refer to the function
fulfilled by a subordinate clause o f this kind as ‘narrative anteriority.’ Economic
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
191
motivations seem to also lead to the common use of nonfinite clauses to express
narrative anteriority, as in crossing the street, I entered the supermarket or, less
economically and more explicitly, having crossed the street, I entered the
supermarket (Comrie 1985:62).
Yet, there are cases where finite clauses are also employed in this function.
Here again it appears that the most economic situation is to use a Perfective form in
the subordinate clause expressing narrative anteriority, as in the following example:
when the young man heard that saying, he went away sorrowful (Mt XIX, 22). There
is still a further possibility, apparently the least economic and most explicit: namely,
the use of a finite form indicating both past tense and anteriority, as in when Jesus
had entered Capernaum, a centurion came to him (Mt VIII, 5). This last type of form
is not infrequent in written texts.
The form had entered in the example above expresses a past-in-the-past; i.e.
“that there is a reference point in the past [e.g. the centurion’s arrival], and that the
situation in question is located prior to that reference point” (Comrie 1985:65). The
form at issue is therefore a pluperfect. However, the function of this pluperfect seems
different from the pluperfect uses we saw in section 3.1. Compare the following
examples from Comrie (1985:65-66): when John had left, Mary emerged from the
cupboard vs. John had already left when Mary emerged from the cupboard.
Although the truth meaning of these two sentences may be the same, it appears there
is an implicature in the first sentence indicating that Mary’s emerging from the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
192
cupboard occurred immediately after John’s leaving. This reading is not implied in
the second sentence. Moreover, the pluperfect in the second sentence has a
“backgrounding” function (Hopper 1979), while the pluperfect in the first sentence is
not backgrounding, but rather “foregrounding”, as much as any perfective can be.
The only difference with a normal perfective is the explicit indication of sequential
anteriority.
Furthermore, in languages such as Old Spanish and Old French we find that
the pluperfects expressing narrative anteriority are not identical to the pluperfects
conveying non-narrative meanings. The former are usually called Past Anteriors
(Preterito Anterior in Spanish, Passe Anterieur in French), and are construed by the
combination of the past participle and the past perfective of the stative auxiliaries. I
will refer to them as ‘perfective pluperfects.’ Recall that normal pluperfects are built
upon the past imperfective of the stative auxiliaries, whenever the language at issue
has a perfective / imperfective distinction. If the language does not have this
aspectual opposition, as in English, then the Pluperfect is construed with aspectually
unmarked past auxiliaries. In this case, thus, the Pluperfect is also suitable for the
function of narrative anteriority. On the other hand, the Pluperfects created upon
imperfective auxiliaries (e.g. Spanish habia salido) do not seem to fit for expressing
narrative anteriority (unless, perhaps, they have experienced a significant erosion of
their original backgrounding meaning).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
193
Therefore, it seems we need to refine the cline in (4) and hypothesize the
following tentative sub-clines proposed in (5) below.
(5) a. aspectually neutral past resultative pluperfect
b. aspectually neutral past resultative -> perfective pluperfect (narrative
anteriority)
c. perfective past resultative perfective pluperfect (narrative anteriority)
d. imperfective past resultative pluperfect
4. DYNAMIC-TELIC SOURCES.
The sources and clines of grammaticalization studied in the preceding sections are
well understood especially for their commonness in European languages, as well as
for their presumed young age. However, in the Basque case we find other types of
sources of grammaticalization which are not that well-known. I will therefore try and
look for cross-linguistic parallels to these less familiar constructions and clines. I
deal with this issue in sections 4 and 5.
The first kind of grammaticalization source I will consider is the past-tense
counterpart of the dynamic-achievement constructions I examined for present tense
in Chapter 3. Recall that these forms were built upon dynamic auxiliaries such as
‘become’, ‘get’ and ‘do.’ In the cline of present, we saw that these constructions, e.g.
the Basque Present Subjunctive, evolve towards future time-reference meanings (cf.
the German werden ‘become’ construction). In section 7.2, we will see that the past
counterpart, i.e. the so-called Old Basque Periphrastic Aorist, has perfective
meaning.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
194
It is not easy to find in the literature clear parallels to constructions formed
upon auxiliaries such as ‘become’ in past tense. Bybee et al. (1994:56) report a
number of dynamic verb sources for completives (cf. 1994:58) and perfects (cf.
1994:69-73), but none of them seems to include verbs like ‘become.’ The most
similar source is probably the verb ‘finish.’ Other constructions that may qualify as
comparable to the Basque Periphrastic Aorist are those with directional and
movement sources (e.g. ‘go to’)
4.1. The verb ‘finish’ as a source of grammaticalization.
The verb “to finish” or “to be finished, ready, complete” is in Bybee et al.’s data a
source of grammaticalization for both completives and anteriors. Bybee et al.
(1994:56) claim that “in such cases the relation between the lexical meaning and the
grammatical meaning is very clear and direct.” Indeed, it seems that a verb like
‘finish’ is, generally speaking, an obvious source for grammaticalizing a completive
sense, which is very close to a perfective past (see the Glossary for clarification).
However, in the cases reported, there seem to be obscure points about the actual
construction that enters into grammaticalization; i.e. it is not always clear what exact
form and function the verb ‘finish’ is grammaticalized with. Specifically, it seems
sometimes that the verb ‘finish’ appears ambiguous between a nonfinite and a finite
form. It is not always clear either whether the finite forms are grammaticalized in
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
195
past tense or not. In the Basque case, in turn, a verb like ‘become’ is clearly used as a
finite auxiliary in past tense.
4.2. The Slavic-type of perfective.
As I already pointed out in Chapter 3 for the case of present tense, the Basque
constructions with achievement auxiliaries ‘become’, ‘get’ and ‘do’ (i.e. the Present
Subjunctive and the Periphrastic Aorist) appear very similar in meaning to the Slavic
Perfectives. Unlike most of the cases o f ‘finish’ cited above, in both the Basque
achievement constructions and the Slavic Perfectives, we unequivocally know
whether they were grammaticalized as present or past. The present construction
evolves in both cases towards future time-reference meaning, whereas the past
construction is in both cases a past perfective. Hence, the semantic parallelism
between the Basque and the Slavic constructions is clear, and can perhaps be found
in other languages such as Ancient Greek (Comrie 1976:67)
On the other hand, the Slavic type of perfective is derivational in nature.
Thus, the exact source of grammaticalization in the perfectives from bounders is
formally very different from the Basque Periphrastic Aorist we are interested in. It
also diverges from the sources of completives cited above: i.e. the verb ‘finish’ and
directional sources. However, the semantic relation among all the sources of
grammaticalization presented in this section is, even if rather vague, ultimately clear.
All the sources are dynamic and imply some sort of achievement or attainment of a
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
196
limit. Moreover, movement sources are partly similar to the perfectives of the Slavic
type in that both imply directional prepositions. (We may even hypothesize some
diachronic relation between the two: see Bybee et al. 1994:105.)
4.3. Conclusion: dynamic-telic sources.
We can conclude this section by saying that there appears to be a somewhat
heterogeneous group of sources of grammaticalization of perfective past which
seems to be ultimately based on the idea of attainment of a limit. Although less
familiar than the clines we saw in sections 2 and 3, the hypothesis of Source
Determination seems to correctly predict the behavior we find in these cases. This
group includes sources such as dynamic verbs (finish, go, etc), achievement
auxiliaries (become, get, etc), directional prepositions (away, up, out, into, etc),
derivational verbs (also based on directional prepositions), etc. Whenever there is a
formal present / past opposition in the grammaticalization of this group of sources, it
seems they tend to evolve towards future meaning in the formally-present and
towards past perfective meaning in the formally-past. (Thus, when looked at from a
pure synchronic point of view, the evolutions in present and past seem to strangely
bifurcate. However, they are easy to explain based on the grammaticalization
perspective adopted in this work. The concept of attainment of a limit in the present
leads to future time-reference, while the same concept in the past leads directly to a
past completive or past perfective meaning. See also section 5 of Chapter 3.)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
197
5. NARRATIVE-DISCOURSE SOURCES.
The last type of source of grammaticalization of perfective meaning I will examine is
what might be called narrative-discourse sources. It seems that cross-linguistically
there exist perfective constructions which originated in discursive narrative contexts.
This does not mean, however, that all perfectives are grammaticalized from
discourse. As a matter of fact, it appears that most perfectives have their source of
grammaticalization in independent sentential constructions of the types we have seen
in the preceding sections.
On the other hand, the cline o f grammaticalization we are interested in now
would evolve in the following fashion.
(6) narrative -> perfective non-narrative -> (simple past)
Unfortunately, the narrative-discourse sources of perfective I am trying to present
here are not so familiar as the sentential sources examined so far. In fact, it does not
seem straightforward to define what a discourse source is.
5.1. Narrative anteriority and grammaticalization from discourse.
What I called in section 3.2 ‘narrative anteriority’ appears as a function that may be
interpreted as discursive. This involves a narrative context of at least two sentences,
which are lineally ordered according to the chronological sequentiality of events.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
198
The function of narrative anteriority is precisely to present the chronological priority
of one event over the other, and apparently only that. However, not all sources of
grammaticalization of narrative anteriority I presented qualify to the same extent as
discourse sources. The pluperfects (even the Past Anteriors of Old Spanish and Old
French) seem to be grammaticalized out of independent sentential constructions, not
necessarily linked to discourse; although their grammaticalization may start in
subordinate clauses (for discussion, see section 9 of Chapter 6 and the entry
‘sentential sources’ in the Glossary). On the other hand, two grammaticalization
sources of narrative anterior that really seem to qualify as discursive are, for
instance, the Basque endings -rik and, specially, -ta. The former has its origin in the
ablative-partitive case-marker. The latter, in turn, seems to originate from the enclisis
of the conjunction eta ‘and.’ The grammaticalization of these non-verbal markers as
expressions of narrative anteriority appears to be based on what Hopper (1979) calls
“the principle of strict chronological sequentiality.”
5.2. Narrative-perfective and grammaticalization from discourse.
As for discourse sources of pure narrative meaning (excluding narrative anteriority),
it is very little what the literature presents about them. Hopper (1979:40-48) reports
the case of the Malay particle -lah, which is apparently grammaticalized as a marker
of perfective meaning out of a “focusing function.” According to Hopper (1979:48),
in this case we see “a diachronic change from a focus marker to a tense marker”;
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
199
concretely, “focusing of the verb is associated with an ‘aorist’ or narrative function.”
The original function of the particle -lah in discourse would be, then, “to indicate
that the word to which it is a part, as well as any dependent clauses, are a central part
of the narrative, and are foregrounded” (1979:47).
Dahl (1985:113-114) reports several cases of special narrative forms which
“exhibit signs of being in some way subordinated or otherwise syntactically
connected to the preceding verb”: this occurs, for instance, in Karaboro, the Nguni
languages and Biblical Hebrew. In these languages, “a narrative discourse typically
starts with a verb in some non-narrative past form: all the subsequent verbs then
obtain narrative.” It is not evident, though, why these narrative sentences should look
like subordinate clauses (while synchronically the sentences “should probably be
treated as syntactically independent of each other”: 1985:114). Dahl points to the
idea that “in a pure narrative discourse, every sentence except the first... is
determined by the point in time at which the last event related in the preceding
context took place” (1985:112-113). Thus, “it is probable that the original function
of narrative verb forms was to serve as the second member of constructions of the
type ‘S and (then) S’” (1985:114).
Another possibility, similar to Dahl’s, may be that these subordinate-like
clauses in narrative contexts could sometimes be related to purpose markers (cf.
Haspelmath 1989). If this were the case, the use of purpose constructions to express
the non-first sentences of narrative sequences could perhaps originate in cases
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
200
parallel to he went home to find out that she was not there. Interestingly enough,
infinitives are also grammaticalized out of purpose markers (see Haspelmath 1989);
and uses of infinitives in narrative sequences are documented in languages like Early
Latin. This is known as the “historical infinitive”, although it is claimed to have a
past imperfective value (cf. Bennett 1910:419-423, Poultney 1957).
5.3. Conclusion: narrative-discourse sources.
To summarize this section, we may say that narrative-discourse sources of
perfectives do seem to exist, but their exact process of grammaticalization is not very
well understood. Despite the uncertainties in the mechanisms of grammaticalization,
the motivation under these processes appears relatively clear. In sequences of
narration, the non-first sentences of the sequence seem to be understood as following
from the first. Hence, allative or purposive markers may be used for marking them.1
The use of ablative markers to express the function of narrative anteriority (cf. the
Basque -rik case mentioned above) seems to be the exact counterpart of the use of
purposives to convey narrative-perfective.
1 The use of the verb ‘go’ to grammaticalize past perfective seems to originate sometimes from
discourse sources similar to the ones outlined above. Thus, Bybee et al. (1994:60) mention “the
English construction go and, which is usually used in the past to emphasize the deliberateness and
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
201
5.4. Conclusion: dynamic and discourse sources.
The TAM sources of grammaticalization examined in sections 4 and 5, i.e. dynamic
sources and narrative-discourse sources respectively, seem to present remarkable
parallelisms with one another. It appears as if the most important dynamic and
movement sources could be classified into those roughly indicating an idea of
telicity-motion-up-to, and those roughly expressing motion-from (the source of
“motion from” is appropriate “as the model for perfect meaning”: Comrie 1976:106;
see also Anderson 1973:28, Comrie 1985:94, Bybee et al. 1994:56). On the other
hand, the most important narrative sources appear to use similar metaphors based on
spatial and motion concepts to focus on the linear order of clauses in the narrative
discourse: i.e. purposive-motion-towards on the one hand, and motion-from on the
other.
Therefore, it sometimes appears difficult to exactly determine specific
sources of grammaticalization o f these types, because, for instance, when discourse
sources are proposed, parallel dynamic sources cannot be always discarded. Thus,
the appearance of allative and purposive sources in perfectives might be due, not
only to a narrative-discourse origin, but also to the grammaticalization of
“bounders”, similar in meaning to the case of perfectives of the Slavic type. In any
event, the existence of narrative-discourse sources, even if in need of research, seems
secure.
finality of an action.” (The Catalan go + infinitive construction with perfective meaning may perhaps
have a related source of grammaticalization.)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
202
It seems worth mentioning here too that purpose markers are also very
suitable sources for grammaticalizing imperatives and subjunctives (and in general
all types of subordinate clauses, as I mentioned following Haspelmath 1989). This
seems to explain the formal relation in some languages between the Subjunctive
mood and the Perfective (“Aorist”).
The following table tries to comprise some of the information in sections 4
and 5.
TABLE 5.3: Dynamic and directional sources of grammaticalization (including narrative-discourse).
SOURCES OF GRAMMATICALIZATION GRAMMATICALIZED TAM MEANINGS
present tense with achievement auxiliaries
Slavic-type Present Perfective (from bounders)
future time-reference meanings
past tense with achievement auxiliaries
Slavic-type Past Perfective (from bounders)
(completive past), perfective past
present tense of ‘go to’ + infinitive prospective, future
present tense of ‘come from’ + infinitive perfect of recent past
purposive, allative in narrative discourse
‘go to’ in narrative discourse
narrative-perfective past
ablative in narrative discourse
‘come from’ in narrative discourse
perfective pluperfect
narrative anteriority
6. DOUBLE-COMPOUND PERFECTS.
This section is devoted to examining the double-compound perfect constructions,
which are documented in some Germanic and Romance languages (cf. French temps
surcomposes). The double-compound perfects seem to be “exceedingly rare” across
languages (Comriel985:76). Moreover, “native speakers judgments on these forms
are not always unequivocal, and not all logical possibilities receive the expected
interpretation” (1985:76). In any case, double-compound perfects are reported at
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
203
least in German (including Swiss German, cf. Squartini 1999:60-63), French (cf.
Harris 1988:228-230) and Occitan (cf. Wheeler 1988:265).
Double-compound perfects are construed in a parallel fashion to perfects,
with the addition of an extra past participle of the auxiliaries ‘be’ or ‘have.’ Thus,
parallel to the Present Perfect he has done, we would have a Double-compound
Present Perfect of the shape he has had done. Similarly, parallel to the Pluperfect he
had done, we would have a Double-compound Past Perfect he had had done.
“German double-compound forms are often described as introducing an extra
degree of anteriority with respect to a past situation” (Squartini 1999:61). This
functional description is probably valid for all double-compound forms in European
languages. However, there seem to be some complications with this description.
First, the exact function of the double-compound constructions appears to depend on
the meaning of their corresponding simple construction. Thus, I mentioned in section
3.2 that Double-compound Perfects are often used in French and apparently in other
languages or varieties where the Perfect has evolved towards perfective meaning
(displacing the Simple Past). In Swiss German, for instance, where the Simple Past
and the Pluperfect are totally lost and the Present Perfect has replaced the former as a
perfective (besides conveying perfect meaning), the Double-compound Present
Perfect has mostly occupied the function of the Pluperfect (Squartini 1999:60-63).
Second, the “extra degree of anteriority” can be interpreted in many ways,
and in fact it appears that different languages and varieties have interpreted this
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
204
concept in very different fashions. Thus, Swiss German, for instance, seems to have
introduced a degree of anteriority in an absolute or absolute-relative temporal axis:
Present Perfect = present perfect & past perfective / Double-compound Present
Perfect = pluperfect (i.e. past perfect). However, other languages and varieties appear
to have interpreted the concept of anteriority along other axes. These can be, at least:
anteriority in the discourse-sequence (i.e. narrative anteriority); anteriority implying
relative remoteness, i.e. in a very similar fashion to the function of Pluperfects in
past temporal frames; anteriority in the sense of experiential perfects; etc. These
different readings of ‘anteriority’, nevertheless, appear related to each other in many
instances.
6.1. Double-compound perfects expressing narrative anteriority.
The function of anteriority in the narrative sequence seems to be one of the most
widespread of those fulfilled by double-compound perfects in all the languages
where they are attested. According to the first-expected possibility, in principle, both
the Double-compound Present Perfect and the Double-compound Past Perfect can
perform this function: the former would be anterior in discourse (and thus also in
time-reference) to a Present Perfect, and the latter anterior to a Pluperfect (and
probably anterior to a perfective too). In a French example from Stendhal cited in
Comrie (1985:76), for instance, “the verb of the first clause is one step further back
into the past than the pluperfect.”
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
205
However, since in French the Present Perfect can function not only as a
perfect but also as a perfective, the Double-compound Present Perfect can be used in
this language as anterior in discourse to both a perfect and a perfective. In the latter
case, the Double-compound Present Perfect appears in place of the Past Anterior of
Old French. According to Harris (1988:228-229), in French,
The loss offit [‘he did (perfective)’] necessarily entailed the loss of eut fait
[‘he had (perfective) done’], the ‘past anterior’, [...] and this was one reason
for the creation of the double compound form a eu fait [...] (literally ‘he has
had done’). The primary use of this paradigm is in place of the past anterior
(apres qu ’ il I ’ a eu fait, il a vu Pierre ‘after he had done it, he saw Peter’),
where eut fait has become a eu fa it just as vit has become a vu.
This narrative function of the double-compound perfects is mentioned for all
the languages involved. Similarly to the French case above, Wheeler (1988:265) says
that the “temps sobrecompausats” of Occitan have one use “in anterior relative time
contexts”: e.g. quand a agut ausit aqud, es sortida de I ’ ostal ‘when she has heard
that, she has left the house’, quand aguet agut ausit aqud, sortiguet de Postal ‘when
she had heard that, she left the house.’ Note in the Occitan examples provided by
Wheeler that the Simple Past has not been ousted by the Present Perfect here, and
thus we have the conservative stage of two double-compound perfects: the Double
compound Present Perfect, anterior to a Present Perfect, and the Double-compound
Past Perfect, which is anterior to the Simple Past.
In the same way, Squartini (1999:61) gives two German examples of double
compound perfects from literary texts of the XVIII and XIX centuries, where the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
206
Double-compound Present Perfect “refers to a situation anterior with respect to a
[Present] Perfect”, while the Double-compound Past Perfect “denotes a situation
anterior to a [Pluperfect], in order to underline the temporal sequence o f events”
(emphasis added). What is interesting in one of the examples provided by Squartini
is that the Double-compound Past Perfect employed “to underline the temporal
sequence of events” (i.e. in the function of narrative anteriority) appears in a main
clause and not in subordinate clauses, where we mostly find this use: Mignon hatte
sich versteckt gehabt, hatte ihn angefafit und ihn in den Arm gebissen ‘Mignon had
hidden (lit. had had hidden), had caught him and had bitten his arm.’ (See Chapter 6,
section 9.)
6.2. Double-compound perfects as remote and experiential perfects.
Apart from their function in narrative sequences, other uses of double-compound
perfects are difficult to predict or even to understand as following some regular
pattern. I hypothesized in section 2.2, based on cases such as the Japanese
experiential, that there might be a cline of grammaticalization involving experiential
and remote perfects, which would depart from the most common cline of evolution
of perfects. This putative cline would evolve according to (2), which I repeat below
for convenience.
(2) a. present resultative experiential perfect frequentative perfect
(imperfective perfect)
b. present resultative -> experiential perfect remote perfect
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
207
We might also hypothesize that, i f ‘anteriority’ of double-compound perfects is
understood in the sense of relative remoteness, then the Double-compound Present
Perfects could perhaps convey some sort of ‘remote perfect’, while the Double
compound Past Perfects could convey ‘remote pluperfect.’ In fact, these uses seem to
appear in the scanty literature on the topic. Thus, Harris (1988:229-230) reports:
One more regionally localized use of the Parfait Surcompose [of French] is
in the sense of a truly perfective present perfect; thus la vigne ga a eu paye,
mais ga paye plus ‘vines have paid (in the past), but they don’t pay any
longer.’
It is not perhaps totally clear what Harris means by a “perfective present perfect”, but
the example presented appears as a remote perfect or a past-temporal-frame perfect,
which in fact implies also a frequentative or imperfective sense. Similarly, Wheeler
(1988:265) mentions that
the [Occitan] sobrecompausat expresses present relevance of an event remote
in time, in contrast to the present perfect, which expresses present relevance
of a recently past event/state; for example:
[a] ai agut vist 'Le Dernier Tango a Paris ’ ‘I have seen [lit. I have had seen]
“Last Tango in Paris” (i.e. already / once upon a time / several times in the
past)’ vs.
[b] ai vist ‘ Le Dernier Tango a Paris ’ ‘I have seen [lit. I have seen] “Last
Tango in Paris” (i.e. recently this week)’;
[c] i siatz aguda estada en Arle, vos? ‘have you ever been [lit. are you had
been] to Arles?’
Clearly, the Double-compound Present Perfects, which are employed in the
first and last sentences cited by Wheeler, express an experiential perfect. This seems
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
208
to point again to a relation between experiential and remote (i.e. not-necessarily-
recent) perfect.
The following double-compound example given by Squartini (1999:62) for
German seems to also convey the idea of experiential (perhaps also of remote
perfect), but this time it is applied to the past; i.e. the Double-compound Past Perfect
appears to express here an experiential pluperfect: “Doctor Preindl had already
mentioned (lit. had had mentioned) the question ... in front of the Baron before.”
6.3. Double-compound perfects in Basque.
Double-compound perfects are also attested in Basque. (See section 7.5 below for
some remarks on their form, including the use of the extra participle izan ‘to be.’)
However, as in the languages examined above, their range of uses is not totally clear.
Generally, it can be said that the double-compound perfects are more common in the
eastern than in the western dialects, although they are not unknown in the latter. In
what follows, I provide a (somewhat interpreted) relation of the functions of the
Basque double-compound perfects appearing in the literature.
6.3.1. Narrative anteriority.
One function of the Basque double-compound perfects appearing in the literature is
that of narrative anteriority. Lafitte (1962:386) refers to this use in the following
way: The double-compound perfects of Basque (especially the Double-compound
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
209
Past Perfect) “marquent un simple moment de reflexion, de repit, entre Paction
indiquee dans la subordonnee et celle de la principale” [they mark a simple moment
of reflection, a pause, between the action expressed by the subordinate clause and the
main clause]. For instance: haren hitza irakurtu izan nuenean, ihardetsi nion ‘when I
had read [lit. I had been read] his/her word, I answered [lit. I had answered] to
him/her.’ Whatever the exact implication of Lafitte’s description, his example is
clear in showing a Double-compound Past Perfect fulfilling the function of narrative
anteriority in the first (i.e. subordinate) clause, while the second (i.e. main) clause
has a simple Past Perfect-Perfective, expressing perfective meaning.
6.3.2. Remote and experiential perfect.
A function of the Basque Double-compound Present Perfect cited in the grammars
may be related to remote and/or experiential perfect. Lafitte (1962:386) mentions
that the double-compound perfects of Basque “rejettent souvent Paction et son
accomplissement dans un passe revolu, lointain et indetermine” [transfer the action
and its accomplishment towards a completed, remote and undetermined past]. This
statement seems in principle a bit vague, but the example he provides appears more
revealing: irakurtu izan dut [lit. I have been read] ‘j ’ai lu jadis, quelque part, il m ’est
arrive de lire’ [I have long ago read, some part; I happen to have read]. The nuance I
understand from Lafitte’s explanation may not be too far from remote and
experiential perfect meanings, which I mean to associate to each other and to the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
210
clines in (2) above. The opposition to capture here seems to be parallel to that
provided by Wheeler for Occitan. Namely, while the Basque Present Perfect-
Perfective conveys often, though by no means exclusively, recent perfect and
hodiernal perfective meanings, the Double-compound Present Perfect seems to
express remote and experiential perfect. That is, a sense close to: ‘I happen to have
read at least once (but probably more than once) some part [e.g. of a book] at some
undetermined time in the past up to the present moment.’
Furthermore, Euskaltzaindia (1987:464, 479) reports a “frequentative” use of
the Double-compound Present Perfect. As in, for instance: nere ibillaldi geienak uste
gabeko egokieratik etorri izan dira (1987:464) ‘most of my excursions have come
[lit. are been come] from unexpected situations.’ This “frequentative” is clearly a
frequentative perfect, and this could as well be an implicature from the experiential
perfect, (as I argued in section 2.2 above). The ‘frequentative perfect’ sense is the
most common function of the Double-compound Present Perfect in the western
dialects of Basque, with which I am most familiar.
6.3.3. Pluperfect.
Another use of double compound perfects, concretely of the Double-compound Past
Perfect, seems to be pluperfect meaning. However, the examples provided in the
literature do not contribute much in clarifying when this pluperfect is actually used.
Lafitte (1962:386) gives the following example: irakurtu izan nuen [lit. I had been
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
211
read] ‘j ’avais anterieurment lu, il m ’etait arrive de lire’ [I had read before, I
happened to have read before].
6.3.4. Passive.
We should finally mention here that the passive voice of Basque is expressed by a
construction very similar to double-compound perfects. In reality, the Modem
Basque perfect passive bears an article in the past participle; therefore, it is not
identical to the double-compound perfects. Thus, Lafitte says: “II ne faut pas
confondre avec les temps surcomposes les constructions passives” [we should not
confuse the passive constructions with the double-compound perfects]. For instance:
irakurria izan da ‘it has been read [lit. is been read]’ vs. irakurri izan du approx.
‘s/he has read [lit. (s/he) has been read].’
On the other hand, in older texts the perfect passive did not bear any article in
the past participle. Yet, it could be generally distinguished by the kind of extra
participle and auxiliary verb used: izan ‘to be, been’ for the perfect passive (e.g.
irakurri izan da lit. ‘is been read’), ukhan ‘to have, had’ for the active transitive
double-compound perfect (e.g. irakurri ukhan du lit. ‘has had read’). (See section 7.5
below.) However, in verbs that can function as transitive and intransitive, there could
be ambiguity between the two functions at issue. Thus, sartu izan da [lit. is been
entered / inserted] could have both readings: perfect passive, i.e. ‘it has been
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
212
2
inserted’, and double-compound active perfect, i.e. approx. ‘he has entered.’ In any
case, the passive will not be object of study in the present work.
6.4. Conclusion: double-compound perfects.
Summarizing: as quoted above from Comrie (1985:76), the behavior of double
compound perfects is extremely difficult to predict in the languages where they are
attested. Moreover, they do not seem to be very widely used; and they appear to be
often optional rather than obligatory. In the case of Basque, some grammars mention
that the difference in meaning between double-compound perfects and their
corresponding simple perfects is frequently non-existent (Euskaltzaindia 1987:479).
However, it seems that some of the uses attributed in the literature to double
compound perfects tend to recur across languages. Thus, I have proposed above that
the functions of narrative anteriority, on the one hand, and of experiential and/or
remote perfect, on the other, are the most commonly expressed by double-compound
perfects. In the case of Basque, the literature seems to report these two uses, plus
some others that I have cited above. In Chapter 6 ,1 will examine the use of the
Double-compound Perfects in the Main Corpus database. We will see there that the
data confirm the use of the Double-compound Present Perfect as experiential
2 In the oldest texts of Basque, in turn, the perfect passive is mostly construed just by the combination
of the participle of the lexical verb plus a finite form of the intransitive auxiliary izan ‘to be.’ That is,
in the oldest passive there is no extra participle added. Thus, we can order chronologically three types
of perfect passives in the following way, from oldest to youngest: irakurri da lit. ‘it is read (no
article)’, irakurri izan da lit. ‘it is been read (no article)’, irakurria izan da lit. ‘it is been read (+
article)’; all three meaning ‘it has been read.’
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
213
(although the simple Present Perfect-Perfective can also express this meaning). On
the other hand, the other uses reported in the literature do not show up in this
database. To understand these uses, thus, other kinds of data should be collected: e.g.
spoken records.
7. SOURCES OF GRAMMATICALIZATION OF THE BASQUE
CONSTRUCTIONS IN THE DOMAIN OF PAST NON-IMPERFECTIVE.
7.1. Synthetic Past.
I will try to prove in Chapter 6 that the Synthetic Past (in a parallel fashion to the
Synthetic Present that I studied in Chapters 3 and 4) is a very old construction
undergoing a process of recession already from the earliest texts of Old Basque,
except for a handful of high-frequency stative verbs. Moreover, due to its highly
bound and opaque morphology, it is very difficult to come up with an etymology of
this form. In Aldai (2000), however, I outlined a hypothesis for the
grammaticalization source of the Synthetic Past. I proposed there, based on the
examples from Dahl (1985:113-114) cited in section 5, that the Synthetic Past of
Modem Basque could have had its origin in the reanalysis of past imperfective
suffixed-subordinate forms, (with the exception of the third-person forms which had
partly a different source). Thus, the modem past marker, i.e. the suffix -(e)n, would
have its origin in the homophonous subordinator ~(e)n, which appears in purpose
clauses, relative clauses and other subordinate clauses of Modem Basque.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
214
This hypothesis needs obviously further research. Tentatively, this may be a
case of grammaticalization of purposive constructions into narrative meaning, in
sequences of narration (see section 5.2 above). There is no formal complication with
this hypothesis. What is not clear, though, is why the base of these putative
purposive-in-origin forms should be a past imperfective. Dahl (1985:114), in any
case, presents a parallel from Biblical Hebrew. (The third-person forms of the
Basque Synthetic Past, on the other hand, need a partly different explanation.)3
If this hypothesis were correct, then the evolution of the Basque Synthetic
Past would have been: narrative perfective simple past past imperfective.
7.2. Periphrastic Aorist.
The source of grammaticalization of the Old Basque Periphrastic Aorist is
completely parallel to that of the Present Subjunctive (see Chapter 3), but with past,
instead of present, auxiliaries. Recall that the Present Subjunctive is a periphrastic
construction built upon the combination of the Radical with the Synthetic Present
forms of the achievement-dynamic auxiliaries *edin ‘to become’ and *ezan / egin ‘to
do, to achieve.’ Likewise, the Periphrastic Aorist is constructed by combining the
Radical and the Synthetic Past of these achievement auxiliaries.
3 1 also reconstructed in Aldai (2000) an older perfective construction of Pre-Basque. I proposed that
the third person form of this older construction would have been *etor and *ekar, for the verbs ‘to
come’ and ‘to bring’, respectively. This reconstructed perfective could perhaps have a discourse
origin too. It might be that the *e- prefix in the oldest reconstructed perfective would have the same
origin as the prefix in the Radical (an ancient infinitive) and in imperatives of Modem Basque. If this
were the case, we could hypothesize a purpose origin for the three types of constructions just
mentioned. (See section 5.2 above for discussion on these issues.)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
215
I have already mentioned that the Present Subjunctive and the Periphrastic
Aorist of Basque appear somewhat similar in their grammaticalization to the Slavic
Perfective Present and Perfective Past, respectively. In the Basque case, instead of a
derivational source of grammaticalization, we have an analytic construction which
combines a nonfmite form and a set of auxiliaries. However, since these auxiliaries
have a sort of achievement-telic meaning, the sense of “attainment of a limit” we
find in the Slavic case is also present in the Basque constructions. Therefore, the
hypothesis of Source Determination would claim that a construction grammaticalized
in the way of the Old Basque Periphrastic Aorist appears very suitable for conveying
perfective meaning. (This source of grammaticalization seems very appropriate too
for rendering past subjunctive meaning: see section 5.4.)
There is an important complication with the Periphrastic Aorist, though,
which ought to be mentioned here. To be precise, this complication affects all past
forms of Basque, but it is much more apparent in the Periphrastic Aorist, as I will try
to explain next.
I just said in the previous subsection that the Synthetic Past (e.g. nentorren ‘I
came’) was probably grammaticalized in Pre-Basque, with the exception of the third-
person forms, by way of the reanalysis of imperfective suffixed-subordinate forms
(e.g. *nentor+en). These imperfective forms in turn rendered the Old Basque
Synthetic Irrealis-Past Subjunctive (e.g. nentorren ‘so that I came’). Thus, the non-
third person forms of the Synthetic Past were identical to their suffixed Irrealis
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
216
counterparts. In other words, for non-third-person forms, there was no overt
opposition between past indicative and past subjunctive in the synthetic
constructions.
Now, this ambiguity was carried on to the constructions with Past
achievement-dynamic auxiliaries. (In fact, it was kept in all Past periphrastic
constructions of Old Basque. But, since most of the constructions with stative
auxiliaries are seldom documented in subjunctive contexts, the problem is only
evident in the constructions with dynamic auxiliaries, which were suitable for both
indicative and subjunctive meanings.) Thus, the non-third person forms of the
Periphrastic Aorist (e.g. etor nendin ‘I came’) were indistinguishable from their
Periphrastic Irrealis-Past Subjunctive counterparts (e.g. etor nendin ‘so that I came’).
The system has been disambiguated in Modem Basque, following different
directions for the synthetic forms and the dynamic periphrases respectively. The
Synthetic Past Subjunctive was lost, and a form like nentorren has only past
indicative meaning in Modem Basque. On the other hand, the Periphrastic Aorist
was lost too (as we will see in the data in Chapter 6), and thus a form like etor
nendin has only subjunctive meaning in Modem Basque.
This is the reason why the achievement-dynamic constructions are often
considered “subjunctive-only” forms in the literature. While this is true of Modem
Basque, it is not so for Old Basque. This is also the reason for the strangeness that
the Old Basque Periphrastic Aorist forms have traditionally caused to scholars
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
217
working on the issue (starting with Larramendi 1745:xxxvii). The
grammaticalization approach adopted in this work, however, provides us with a
natural explanation of the subjunctive / indicative “mystery.”
7.3. Past Perfect-Perfective
The Past Perfect-Perfective is, within European languages, a construction much more
familiar than the two forms I examined in the previous subsections. Concretely, its
source of grammaticalization is exactly parallel to that of the Pluperfect of most
Romance and Germanic languages. That is to say, the Past Perfect-Perfective is a
periphrastic construction consisting of the combination of the Perfect Participle (i.e.
the nonfinite form with -i, -n, -tu endings) and the Synthetic Past of the stative
auxiliaries izan ‘to be’ and *edun ‘to have’: e.g. etorri nintzen ‘I came [lit. I was
come]’, hartu nuen ‘I took [lit. I had taken].’ (See section 4.2 of Chapter 2.)
The hypothesis of Source Determination predicts that a form like the Basque
Past Perfect-Perfective would have been grammaticalized as past resultative, and the
hypothesis of Unidirectionality predicts that it would evolve along the path in (4)
above, (repeated for convenience).
(4) past resultative pluperfect -> remote perfective -> perfective (simple
past)
Thus, Lafitte (1962:374) says: “Le passe eloigne basque ressemble pour la
construction au plus-que-parfait fran^ais” [the Basque remote perfective resembles in
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 1 8
its construction the French Pluperfect]. The framework in this work readily explains
the differences in meaning between form ally-parallel constructions in the two
languages: While the French form, as the corresponding construction in Spanish and
most of the neighboring languages, has only evolved in (4) till acquiring the meaning
of pluperfect, the Basque counterpart has further developed a sense of pre-hodiemal
perfective, and is also used in narrative contexts. This evolution is quite rare in the
languages of Europe, but is well attested cross-linguistically (cf. Comrie 1985:69,
Dahl 1985:147).
It should be added, however, that, unlike the French and Spanish Pluperfects
and like the English Pluperfect, the Basque Past Perfect-Perfective was
grammaticalized upon aspectually-unmarked past auxiliaries. That is, the Basque
Synthetic Past (currently in high recession, as I mentioned in 7.1) was formerly a
simple past unmarked for aspect. Thus, the Synthetic Past forms of the auxiliaries
izan ‘to be’ and *edun ‘to have’ could convey both imperfective and perfective past
meanings, although the former was much more frequent. An example of the
Synthetic Past o f the verb *ednn ‘to have’ with perfective meaning can be found in
the Biscayan text by Mikoleta (1653): zeinbatan ebe berorrek eskritori-au? (“Cuanto
le costo este escritorio?”) ‘how much was this desk? [lit. in how much did you have
this desk?]’ Therefore, if the Basque Synthetic Past auxiliaries were unmarked for
aspect, then the Past Perfect-Perfective could evolve not only along the cline in (4)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 1 9
but also towards expressing narrative anteriority, as in (5b) above. (See Chapter 6,
section 9.)
7.4. Present Perfect-Perfective.
The Present Perfect-Perfective is completely parallel to the Past Perfect-Perfective,
but has present, instead of past, stative auxiliaries: e.g. etorri naiz ‘I have come [lit. I
am come]’, hartu dut ‘I have taken.’ (See Chapter 2.) The hypotheses of Source
Determination and Unidirectionality predict that a form like the Basque Present
Perfect-Perfective would have been grammaticalized as present resultative and
would evolve along the path in (1) above, (repeated for convenience).
(1) present resultative -> present perfect -> hodiernal perfective -> perfective ->
(simple past)
Like in Spanish, and unlike in French and other European languages, the Present
Perfect-Perfective in the western dialects has not evolved further than hodiernal
perfective. In some eastern (French-Basque) varieties, however, there has occurred a
development whereby the Present Perfect-Perfective is used in contexts with pre-
hodiemal past time-reference.
7.5. Double-compound Perfects.
The Double-compound Present Perfect and Double-compound Past Perfect of
Basque were already examined in section 6.3 above. As stated there, their meaning is
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
220
not straightforward, and will actually be one of the issues to check in the data in
Chapter 6. It might be worth mentioning here some formal remarks about their extra
participle. Recall that these forms are parallel to the Present Perfect-Perfective and
Past Perfect-Perfective, respectively, with the addition of an extra auxiliary-
participle. Now, the extra participle added is currently izan ‘been’ in all cases,
transitive and intransitive. Thus, etorri izan naiz lit. ‘I am been come’ and irakurri
izan dut lit. ‘I have been read’ both have the intransitive extra-auxiliary' izan ‘been.’
In the oldest texts, however, the extra participle was izan ‘been’ for intransitive
forms and ukhan ‘had’ for transitives.
Therefore, the appearance of izan ‘been’ in all types of verbs, which seems
hard to explain from the perspective of grammaticalization, is actually a regularizing
innovation. As a matter of fact, this is part of a more general phenomenon where the
non-finite form izan ‘been, to be’ has supplanted the non-finite form ukhan ‘had, to
have’, which in turn seems to have previously ousted *edun ‘had, to have’ (see Trask
1981:287; cf. also Chapter 2: note 1 and Table 2.14).
7.6. Present and Past Resultatives.
The last Basque constructions I will take into account in this sphere of meaning are
the Present and Past Resultatives. These constructions were presented in some depth
in Chapter 2. Recall that there is a great deal of variation with these forms, regarding
transitivity-related issues and also across dialects. Also, these constructions seem to
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
221
be subject to important lexical restrictions. The picture we seem to encounter for
these forms appears very similar to what we found for the Present Progressives. In
section 2 of Chapter 6 ,1 will try to prove that these constructions are the youngest in
the domain of past non-imperfective and that they constitute a case of same-source
layering when compared with the Perfect-Perfective constructions.
The most important type of Resultative constructions in Basque is formally
very similar to the Perfect-Perfectives, with the only difference of the presence of the
article -a in the past participle of the former. Thus, corresponding to the Present
Perfect-Perfective etorri naiz ‘I have come [lit. I am come (no article)]’, we have the
Present Resultative etorria naiz; and corresponding to hartu dut ‘I have taken’, we
find hartua dut, with the only formal difference of the presence of the article in the
participle. Likewise in past, parallel to the Past Perfect-Perfective etorri nintzen ‘I
came [lit. I was come]’, there is the Past Resultative etorria nintzen; and parallel to
hartu nuen ‘I took [lit. I had taken]’, we have hartua nuen. The source of
grammaticalization of the corresponding Perfect-Perfectives and Resultatives is thus
clearly the same. As for the appearance of the article in the participle of the latter,
this actually constitutes a piece of evidence of the relative recentness of this form, as
I will try to prove in section 2.3.2 of Chapter 6.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
222
CHAPTER 6
THE DOMAIN OF PAST PERFECTIVE
IN HISTORICAL BASQUE
1. INTRODUCTION.
The present chapter examines the TAM sphere of past non-imperfective in historical
Basque. It will be shown that the developments and the competition we find in
Basque can be readily explained based on the grammaticalization framework
outlined in Chapter 1 and particularized in Chapter 5 for non-imperfective past.
The goals and methodology of this chapter are the same 1 followed in Chapter
4 for the domain of present. On the one hand, the chapter accounts for the history of
the Basque constructions in the sphere of past during the last five centuries. On the
other hand, this subject will constitute an illustration of the hypotheses on
grammaticalization put forward in Chapter 1, which I will also test when possible.
(See Chapter 1 for details on the methodology and possible scenarios of
competition.)
The presentation of the chapter is as follows. In section 2 ,1 order
chronologically the eight constructions that appear in the domain of past non-
imperfective in historical Basque. These are the Synthetic Past, the Periphrastic
Aorist, the Present and Past Perfect-Perfectives, the Present and Past Double
compound Perfects, and the Present and Past Resultatives. The safest situation to
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
223
provide a relative chronological order for is the scenario of same-source layers. This
is also one of the most interesting cases from a theoretical perspective. In the data,
we find a case of same-source layers between the Perfect-Perfectives and the
Resultatives. In section 3 ,1 outline the general ‘predictions’ we can propose for the
TAM domain of past non-imperfective in Basque. Sections 4 and 5 are devoted to
predicting and testing the situation of the youngest domains in the clines starting
with present resultative and past resultative, respectively. These two sections are
those where we can predict (always to a certain extent) the synchronic and
diachronic competition of the Basque forms in this TAM area: concretely, the same-'
source-layer competition between Perfect-Perfectives and Resultatives.
Section 6 presents the competition for degrees of remoteness, about which
the database has, unfortunately, very few data. In section 7 ,1 examine the historical
evolution in the TAM area of (pre-hodiemal) perfective. This is probably the sphere
of TAM where the most noticeable changes have occurred in the last five centuries
of Basque. Section 8 presents a characteristic of Old Basque that, to my knowledge,
has not been reported in the literature: namely, the existence of special (albeit
optional) forms to convey unwitnessed and fictional perfective, i.e. evidential
contexts. Section 9 studies the function of narrative anteriority, which appears as a
beachhead in historical Basque for the change undergone by the Past Perfect-
Perfective from pluperfect to perfective. Section 10, finally, offers the conclusions
from this chapter, with special emphasis on clarifying the complex scenario in the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
224
domain of pre-hodiemal perfective and narrative. (See Chapter 5 and the Glossary
for clarification on the TAM terminology.)
2. CHRONOLOGICAL ORDERING.
This section is devoted to providing theory-independent evidence for a relative
chronological order of the Basque constructions I will examine in the TAM domain
of non-imperfective past. The chronological order I want to find substantiation for is
the following (where the symbol ‘>’ stands for ‘older than’):
(1) Synthetic Past > (Periphrastic Aorist) > Perfect-Perfectives > (Double
compound Perfects) > Resultatives
In reality, the only types of constructions I will confidently propose a relative order
for are those which I have already proved to have had the same source of
grammaticalization. Thus, I showed in Chapter 5 that the Perfect-Perfectives and the
Resultatives had the same kind o f source. In section 2.3 below, I will be able to
propose a relative chronological ordering between these constructions, as well as a
scenario where they constitute layers evolving along the same cline. The Double
compound Perfects, in turn, although presenting a source of grammaticalization
similar to those of Perfect-Perfectives and Resultatives, appear rather obscure at this
point. In section 2.4 below, I will try to propose a putative chronology of these forms
relative to the Perfect-Perfectives.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
225
2.1. The Synthetic Past.
I already provided in Chapter 1, section 6.2, and Chapter 4, section 2.1, evidence for
considering the synthetic forms of Basque as the oldest constructions in the TAM
system. Among the most important types of proof supporting the age of the synthetic
forms, we should mention 1) their high degree of boundness, 2) their unproductivity
(apparently resulting from a productive rule in earlier stages), and 3) the fact that the
periphrastic constructions are built upon erstwhile synthetic forms (see the sections
just cited for more details). Thus, these factors point to the synthetic forms of
Basque, and concretely to the Synthetic Past at issue, as very old declining
constructions, actually in an advanced process of regression.
The Synthetic Past was unmarked for aspect in Old Basque (see section 7.3
of Chapter 5): i.e. it could convey both perfective and imperfective past meanings. It
was nevertheless restricted to some specific verbs already in the oldest texts. These
were mostly high-frequency verbs, as in the case of the Synthetic Present (see
Chapter 4). It seems that at some point in time within the 17th century, with slight
variations across dialects, the Synthetic Past lost the possibility of expressing
perfective meaning. Thus, it remained confined to imperfective contexts. The
restrictions with classes o f predicates seemed to also increase with the narrowing in
meaning of this form. In Modem Basque, the Synthetic Past is practically limited to
a handful of stative verbs. In this chapter I will only study the Synthetic Past insofar
it conveys perfective meaning. Therefore, we will find instances of this form
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 2 6
expressing perfective only in the oldest text in the database (16th century). Moreover,
these examples are unfortunately rather scarce.
2.2. The Old Basque Periphrastic Aorist.
It is not easy to find evidence for locating the Periphrastic Aorist in a chronological
order relative to the Perfect-Perfective constructions. In any event, I will propose
three factors below, all of which constitute proof for positioning the Periphrastic
Aorist as older than the Perfect-Perfectives.
1) For ancient non-borrowed verbs, the nonfinite form in the Periphrastic
Aorist, i.e. the Radical, appears as the oldest of all the nonfinite forms known in
historical Basque. All other nonfinite forms, e.g. the Past Participle, the Imperfective
Participle, etc, are constructed by adding different endings to the Radical (cf. Trask
1997:211-216, 246-247). The fact that the Radical is the oldest nonfinite form of
historical Basque does not necessarily entail that the Periphrastic Aorist and the other
constructions construed with the Radical are older than those built upon younger
nonfinite forms. However, this fact does seem to constitute a small piece of evidence
for the age of the Periphrastic Aorist.
2) The relative order between nonfinite form and auxiliary is extremely
consistent in all the instances of the Periphrastic Aorist documented in Old Basque.
In the database, there are 128 examples of the Periphrastic Aorist in LEIZAR, and in
all of them the order is always [Radical + dynamic auxiliary]. On the other hand, we
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
227
can find in LEIZAR the following examples with the order [stative auxiliary +
nonfinite form]: three examples of the Periphrastic Future (Mt VII, 24; Mt XX, 19;
and Mt XXIII, 14), three examples of the Periphrastic Present (Mt IV, 5 and Mt IV,
8), and one example of the Present Perfect (Mt XVIII, 23). Likewise, in the Biscayan
book Refranes y Sentencias (1596), for instance, all the 16 examples of the
Periphrastic Aorist present the order [nonfinite form + dynamic auxiliary], whereas
out of the 17 examples of Present-Perfective constructions, there are two with the
order [stative auxiliary + nonfinite form]: RS 329,433. (Similarly, there are two
instances of the latter order within the 31 examples of the Periphrastic Present: RS
352.)
The former data constitute a piece of evidence pointing to a higher degree of
boundness and rigidity between nonfinite form and auxiliary in the Periphrastic
Aorist than in the constructions with stative auxiliaries, e.g. the Perfect-Perfectives.
Thus, this factor supports again the relative seniority of the Periphrastic Aorist over
the Perfect-Perfective constructions.
3) Finally, we may point to a different kind of evidence as substantiation for
the hypothesis that the Periphrastic Aorist is older than the Perfect-Perfective
constructions. Namely, the fact that the Periphrastic Aorist, and in general all the
Basque constructions with dynamic auxiliaries, are atypical of the geographical area,
while the constructions with stative auxiliaries had sources and paths of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
228
grammaticalization very similar to those of the corresponding forms in the
neighboring languages.
2.3. Perfect-Perfectives vs. Resultatives.
As mentioned above, Perfect-Perfectives and Resultatives of Basque constitute a
fairly clear case of layering: i.e. constructions with the same source but at different
degrees of grammaticalization. The Resultatives are among the constructions I
termed ‘loose periphrases’ in Chapters 1 and 2; (I also called them ‘peripheral
analytic forms’, due to their low frequency of use, at least in the written texts and
specifically in the database in this work). The Perfect-Perfectives, on the other hand,
were among the constructions dubbed ‘rigid periphrases’ (and ‘main analytic
forms’). There are two factors that make the task of chronologically ordering the
Perfect-Perfectives vs. the Resultatives relatively easy. These are the degree of
relative mobility of nonfinite form and auxiliary / quasi-auxiliary, and the absence /
presence of the article in the Past Participle. Another factor that can be taken into
account to order in diachrony the two types of constructions at issue is their degree
of combinability with other types of constructions.
2.3.1. Degree o f mobility of nonfinite form and auxiliary.
I already presented in Chapter 1 that the Resultative constructions of Modem Basque
are freer than the Perfect-Perfectives regarding the relative order between nonfinite
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
229
form and auxiliary. This is the reason why the former were included among the loose
periphrases, while the latter were considered rigid periphrases. In reality, the most
common order is in both types of constructions that of [participle + auxiliary / quasi -
auxiliary]: e.g. apurtu dut T have broken it’ (Perfect-Perfective) / apurtua dut ‘I have
it broken’ (Resultative). However, in wh-questions and in sentences with a focused
component, the inverse order [quasi-auxiliary + participle] is easy to find in
Resultative constructions. In the Perfect-Perfectives, on the other hand, this inverse
order [auxiliary + participle] is practically non-existent, especially in the western
dialects: e.g. zer egin duzu? ‘what have you done?’, *zer duzu egin? vs. zer duzu
egina? ‘what (e.g. part) is it that you have (already) done?; lehenengo kapitulua
amaitu dut ‘I have finished the first chapter’, *?lehenengo kapitulua dut amaitu vs.
lehenengo kapitulua dut amaitua ‘it is the first chapter that I have finished.’ (See
nevertheless Lafitte 1962:47-48, 336-337.)
2.3.2. Absence / presence o f the article in the participle.
As can be observed in the examples just given, the main formal difference between
the corresponding Perfect-Perfectives and Resultatives of Standard Basque is the
presence in the latter of the article -a (sg) / -ak (pi) in the Past Participle. (For other
dialectal possibilities of constructing the Resultatives, see Chapter 2.) Thus,
corresponding to the Present Perfect-Perfective apurtu dut ‘I have broken it’, we
have the Present Resultative apurtua dut; and formally similar to the Past Perfect-
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
230
Perfective apurtu nuen ‘I broke it’, the Past Resultative apurtua nuen. We will see
next that the absence / presence of the article constitutes evidence for
chronologically ordering the Basque Resultatives as younger than the Perfect-
Perfectives.
It is widely accepted among Bascologists that the article is in Basque a
relatively recent innovation. (See section 3.1 of Chapter 2 and section 6.2 of Chapter
3, for the contrast between the locative case-marker with and without the article.)
One of the contexts where the article seems to have been later introduced is in the
predicate of attributive sentences. As a matter of fact, the evolution of the Basque
predicative sentence from the structure [subject NP + adjective / noun + copula] to
[subject NP + adjective / noun + article + copula] has practically occurred within the
historical period.
Thus, in Old Basque, predicative constructions were most often construed
with no article on the adjective or noun. In Modem Basque, on the other hand, these
sentences bear the article on the adjective / noun. In the database, the simplest
predicative sentence consisting of [subject NP + adjective / noun + copula] is nearly
always constructed in LEIZAR (16th c.) with no article on the adjective / noun, while
in ITUN (20th c) the majority of attributive sentences of this type bear the article.
Some examples of this contrast between LEIZAR and ITUN are the following: justo-
0 bait-zen / gizon zuzen-a zen ‘being a just man’ (Mt 1,19); peskadore-0 ziraden /
arrantzale-ak bait-ziren ‘they were fishermen’ (Mt IV, 18); zuen saria handi-0 baita
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
231
/ zuen saria handi-a izango baita ‘great is your reward’ (Mt V,12); zareten zuek
perfekt-0 / izan zaitezte guztiz on-ak ‘you shall be perfect’ (Mt V,48); uzta handi-0
da / uzta ugari-a da ‘the harvest is plentiful’ (Mt IX,37).
Now, I already determined in Chapter 5 that the Basque Perfect-Perfectives
and Resultatives were both grammaticalized as the combination of stative auxiliaries
and past passive participles; i.e. both were grammaticalized as resultatives. But the
past participle in a resultative construction is after all a verbal adjective. In other
words, resultatives were grammaticalized in Basque, as in many languages, in a
parallel fashion to stative predicative sentences. It appears sound, then, to assume
that the formal contrast between Resultatives and Perfect-Perfectives (i.e. the
presence of the article in the former) is a consequence of the evolution of predicative
sentences in Basque. That is to say, it seems that Resultatives bear the article due to
their having been grammaticalized in a younger stage than Perfect-Perfectives: i.e. in
a stage of language where the article was already present in resultative constructions.
As a matter of fact, the presence of the article on participles was not yet
absolutely mandatory in Old Basque. It did appear in most cases of resultatives, but
it was not obligatory when the participle behaved similarly to an independent
adjective. As Nedjalkov & Jaxontov (1983/1988:6) put it, “sometimes the past
passive participle may be a stative, as in the village is surrounded by woods.'1 ' ’ In this
last case, i.e. when the participle behaves as a normal adjective / adverb and denotes
a stative situation (although it can also have passive and/or resultative implications),
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
232
the predicative sentence consisting of [subject NP + past passive participle + copula],
did not need to bear the article in Old Basque.
Clear examples of participial stative sentences in Old Basque with no article
on the participle are the following: enganatu zirela ‘that they were wrong’ (Etxepare
XIV, 11); Heuskaldunakpreziatu ziraden ‘the Basques were appreciated’ (Etxepare
XV,6); obligatu giren ‘that we be obliged’ (Etxepare, intro); den konplitu ‘may it be
granted’ (Etxepare 1,416); gomendatu duzun ‘take him as your recommended’
(Etxepare 1,450); arima den salbatu ‘may my soul be saved’ (Etxepare XIII,45); dira
ezaun (“son conocidos”) ‘they are known’ (RS 433); edozein da azarri (“cada uno es
atrevido”) ‘everybody is courageous’ (RS 383). In Modem Standard Basque, all the
above sentences need to bear the article on the past participle (or otherwise an ending
-rik / -ta; see Chapter 2 and also Euskaltzaindia 1987:82): enganatu-akzirela,
Euskaldunakpreziatu-ak ziren, obligatu-ak izan gaitezen, konplitu-a izan dadin,
gomendatu-a izan dezazun, arima salbatu-a izan dadin, ezaun~ak dira, edozein da
auzartu-a.
In sum, the absence / presence of the article in the Basque Perfect-Perfective /
Resultative constructions seems a consequence of the ways the language construed
resultative sentences in two different stages. The examples presented above appear
as evidence of a general evolution towards the obligatoriness of the article in all
types of predicative sentences (including resultatives). It seems that the article was
introduced in resultatives prior to its becoming obligatory in attributive sentences;
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
233
hence, the contrast between resultatives and statives in Old Basque. Yet, this contrast
appears just due to different chronologies of the same general evolution. Thus, in
Modem Basque both adjectives and participles bear the article in both resultative and
stative constructions. Therefore, these data constitute a piece of evidence in favor of
the relative age of the Perfect-Perfectives over the Resultatives.
2.3.3. Combinability with other types o f constructions.
A further piece of evidence of the relative seniority of the Perfect-Perfectives over
the Resultatives is the fact that the latter combine with more types of constructions
than the former in Modem Basque. In this work, I will only examine the Present
Perfect-Perfective and the Past Perfect-Perfective, and the Present Resultative and
the Past Resultative. However, there are other Perfect-Perfectives and other
Resultatives in Modem Basque. Now, while the Perfect-Perfective constmction is
hardly combinable (cf. “mutual exclusivity”, Bybee et al. 1994:7) with the
Periphrastic Future and the Periphrastic Imperfective (e.g. *?ikusi izango dut, *?ikusi
izaten dut), the Resultative can in principle combine with any type of constmction of
Modem Basque, as in ikusia izango dut ‘I will have seen it’, conveying future perfect
meaning.
The high combinability of the Resultative appears as a consequence of the
formal freedom of the items that constitute this type of constmction. More
concretely, it seems a consequence of the independent existence of the [Past
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
234
Participle + article] in Modem Basque. The fact that the Past Participle alone seldom
has independent existence in Modem Basque seems a reason for its being hardly
combinable with other analytic constructions, (while it is combinable with all of the
synthetic constructions of Basque).
However, since the Past Participle alone did have some independent
existence in stages of the language as close as the 16th century (see examples above),
we may hypothesize that some combinations among analytic constructions involving
the Participle alone may have been, at least partly, grammaticalized. This might be
the source of grammaticalization of the Double-compound Perfects, which I study
next.
2.4. Perfect-Perfectives vs. Double-compound Perfects.
As just mentioned, the source of grammaticalization of the obscure formations I have
been calling the Double-compound Perfects might be related to the independent
existence and combinability of the Past Participle alone (i.e. with no article or
ending) in earlier stages of Basque. If this hypothesis were correct, the Double
compound Perfects would be clearly younger formations than their corresponding
Perfect-Perfectives. (An alternative possibility could be that these constructions
would have been formed by analogy, instead of by the grammaticalization of
independent items.)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
235
Furthermore, if the above hypothesis is correct, then the addition of the extra
auxiliary-participle could have taken place in a stage where the Perfect-Perfectives
had already undergone a partial process of grammaticalization. Thus, it might be that
the apparent addition of the extra auxiliary-participle, occurred in reality as the
combination of a somewhat grammaticalized Perfect-Perfective of the stative
auxiliary with an independent Past Participle of the lexical verb.
According to this putative scenario, the combination of the independent
Participle of a lexical verb such as egin ‘done’ or erori ‘fallen’ with the Present
Perfect-Perfective of the stative auxiliary, i.e. ukan dut ‘I have had’ or izan naiz ‘I
have been’, appears intuitively as an appropriate source of grammaticalization for an
experiential perfect or a remote perfect, because the perfects of stative verbs seem to
often express experiential meaning (i.e. imperfective perfect). Recall that these
meanings are often expressed by Double-compound Perfects (cf. in Chapter 5:
section 6 and the clines in (2) presented in section 2.2). Concretely, the source of
grammaticalization I am referring to could be understood as expressing an original
meaning like ‘I have had (something) done before’ (i.e. ‘I have done it some time(s)
in the past’) or ‘I have been fallen before’ (i.e. ‘I have fallen some time(s) in the
past’).1 This issue, in any event, would need to be further studied.
1 Allative PPs (-ra case marker) constitute a very interesting way in which Basque has created some
verbal participles: e.g. ate-ra ‘gone out’ [lit. to-the-outside], argi-ta-ra ‘published’ [lit. to-light] (cf.
Trask 1995:210-211). Now, the following two examples from Etxepare (1545) with allative-
participles seem to provide a good illustration of the source of grammaticalization of the Double
compound Present Perfect vs. that of the Present Perfect-Perfective. The two examples are very lowly
grammaticalized, and may actually be analyzed as instances of allative PPs instead of verbal
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
236
3. GENERAL PREDICTIONS.
Although there are many points in the Basque case which are, strictly speaking,
unpredictable and language-specific (as is probably the case of the evolution of any
TAM construction in any language), there are still a good number of predictions and
explanations that the grammaticalization model can offer with a high degree of
confidence. The Double-compound Perfects are, for instance, constructions for
which we do not have solid predictions at this point. In principle, they will have a
behavior similar to that of their corresponding Perfect-Perfectives (see section 6 of
Chapter 5). On the other hand, among the most important predictions and
explanations of the Basque constructions in the domain of non-imperfective past, we
can mention the following.
3.1. Competition between Resultatives and Perfects.
The competition between Basque Resultatives and Basque Perfect-Perfectives
constitutes a good case to test the hypotheses of Source Determination, Layering and
Unidirectionality of Meaning. As we saw in Chapter 5 and in section 2.3 above, the
source of grammaticalization of these constructions is extremely similar. The
hypothesis of Source Determination, then, will predict that the corresponding forms
will evolve through very similar paths of grammaticalization. We also saw in
section 2.3, that the Resultatives are younger in the language than the Perfect-
participles: heuskara da kanpo-ra (Etx. XV, 1) ‘Basque has come forth’ [lit. Basque is to-the-outside]
vs. izan dira here goihen grado-ra (Etx. XIV, 20-21) ‘(other languages) have been to their zenith. ’
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
237
Perfectives. Thus, the prediction is that the Basque Present Resultative and the
Basque Present Perfect-Perfective are evolving along the same cline (broadly
speaking), the latter progressing before the former. This cline was studied in Chapter
5 and is repeated below for convenience.
(2) present resultative -> present perfect -> hodiernal perfective -> perfective
(simple past)
The prediction for the Basque Past Resultative and the Basque Past Perfect-
Perfective is totally parallel, their corresponding cline being the one in (3) below,
which was also examined in Chapter 5.
(3) past resultative -> pluperfect -> remote perfective perfective -> (simple
past)
As we will see below, the Basque data does not falsify these predictions. In section 4
I will examine the competition between the Present Resultative and the Present
Perfect-Perfective (plus the Double-compound Present Perfect) for the youngest
semantic subtypes in the cline in (2). In section 5 the competition examined will be
that between Past Resultative and Past Perfect-Perfective, for the youngest subtypes
in the path in (3). Section 6, in turn, will be devoted to studying the competition
between Present and Past Perfect-Perfective for the central-old subtypes in both
clines; that is, the competition for degrees of remoteness in perfective meaning.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
238
3.2. Degrees of remoteness.
The clines in (2) and (3) above provide also predictions for degrees of remoteness.
As I said in Chapter 5, there are different and partly language-specific options in this
evolution. However, there are several possibilities that are readily expected while
others are either non-expected or ruled out. The Basque case presents what might be
considered the most consistent evolution according to the clines in (2) and (3),
although this case is not common in European languages. Namely, the Present
Perfect-Perfective and the Past Perfect-Perfective contrast in a hodiernal / pre-
hodiemal opposition. I will examine this contrast in section 6.
3.3. The case of the Old Basque Periphrastic Aorist.
Although the Old Basque Periphrastic Aorist constitutes one of the most challenging
issues in the Basque data on the sphere o f non-imperfective past, the
grammaticalization framework assumed in this work seems to provide a convincing
account on this matter. It appears that the main problem the Periphrastic Aorist
presented for most scholars working on the field came from their approaching the
issue from a synchronic-only perspective. Most scholars have regarded the
auxiliaries in the Periphrastic Aorist as ‘subjunctive auxiliaries.’ Hence, it became
extremely problematic to explain why the subjunctive auxiliaries should appear in
perfective past meaning. These difficulties are overcome when approaching the
subject from a grammaticalization-diachronic point of view. While analyzing the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
239
auxiliaries at issue as ‘non-indicative auxiliaries’ is correct for a synchronic analysis,
there is nothing inherent to these auxiliaries that precludes them from having
appeared in meanings other than indicative in earlier stages of Basque.
As a matter of fact, the source of grammaticalization of the Old Basque
Periphrastic Aorist, which I studied in Chapter 5, although not very common cross-
linguistically, appears as a good candidate for grammaticalizing completive past or
perfective meaning. It is likewise a very suitable source of grammaticalization for
subjunctive meaning. This appears to be the reason why perfective and subjunctive
appear formally related not only in Basque but also in other languages (see sections
5.4 and 7.2 of Chapter 5 for discussion). What we should explain now is how the Old
Basque Aorist disappeared from the language, thus confining the auxiliaries at issue
to subjunctive contexts. I address the replacement and loss of the Old Basque Aorist
in section 7 below (see also sections 9 and 10).
4. COMPETITION FOR THE YOUNGEST SEMANTIC SUBTYPES IN THE
CLINE STARTING WITH PRESENT RESULTATIVE.
4.1. Predictions.
As we saw in section 2.1 of Chapter 5, the youngest semantic subtypes in the cline
starting with present resultative are: present resultative (including present predicative
implying some result of an anterior event), perfect of result, experiential perfect, and
perfect of recent past. The Basque constructions we find for the competition o f these
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
240
subtypes are the Present Resultative (PrR), the Double-compound Present Perfect-
Perfective (DcPr) and the Present Perfect-Perfective (PrP).
1 have already advanced in 3.1 above the prediction for the PrR and the PrP.
These constructions constitute layers with the same source, the former being younger
than the latter. The prediction is that both constructions evolve along the same cline
of grammaticalization: i.e. the path in (2). Thus, we can predict a competition
between old and young layers, especially for the very youngest subtypes in this area:
i.e. present resultative and perfect of result (which are, incidentally, not always easy
to distinguish from each other). In this competition, we can expect to find that the
PrR is diachronically gaining importance, at the expense of the PrP, starting from
these very youngest subtypes and towards experiential and later subtypes. As for the
Double-compound Present Perfect-Perfective, I already mentioned that its behavior
seems very difficult to predict cross-linguistically.
We ought to point out here that this sphere of meaning is significantly
affected by the lexical-semantics of the verbal predicates. Concretely, verbs that do
not have a patent result are not expected to appear conveying present resultative
meaning. Yet, as the cline progresses to the right, that is, for somewhat older
subtypes (e.g. perfect of recent past), we expect to find less lexical-semantic
restrictions. In other words, the process of grammaticalization gives rise to a
generalization in use, as we saw in Chapter 1, However, in this chapter I will not
address issues related to lexical-semantic of verbs. The interplay of resultatives and
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
241
lexical-semantics is, at any rate, a very interesting topic for further research. This
would probably need elicited material, besides the written data 1 use in the Main
Corpus of the present study.
4.2. Testing the predictions.
I present in Table 6.1 the data on the youngest semantic subtypes in the cline starting
with present resultative.
TABLE 6.1: Youngest subtypes in the cline starting with present resultative (Main Corpus).
SEMANTIC
SUBTYPES
LEIZAR HARAN ITUN
PrR DcPr PrP PrR DcPr PrP PrR DcPr PrP
present resultative 11 8 0 3 13 13 0 0 13 13 0 0
perfect of result 12 0 1 11 13 0 3 10 11 6 0 5
experiential perfect 17 0 9 8 16 0 1 12 5 0 0 5
perfect of recent past 6 0 3 3 6 0 1 5 5 0 0 5
other 3 0 1 2 5 0 0 3 4 0 0 4
TOTAL 49 8 14 27 53 13 5 30 38 19 0 19
4.2.1. Present resultative and perfect o f result
As predicted above, the Basque Present Resultative (PrR) and Present Perfect-
Perfective (PrP) compete only for the very youngest subtypes in the cline: i.e. present
resultative and perfect of result. The Present Resultative does not show up in any
other semantic subtypes. Also, the data seem to constitute evidence for the predicted
evolution whereby the Present Resultative is gaining presence to the detriment of the
Present Perfect-Perfective. Thus, while in LEIZAR and HARAN the Basque Present
Resultative seems confined to present resultative meaning, in IT UN it appears also in
perfect of result contexts (although it should be admitted that these two meanings are
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
242
often hard to distinguish). More importantly, while in LEIZAR we only find 8 cases
of Present Resultative forms out of 49, and in HARAN only 13 cases out of 53, in
ITUN the Present Resultative is used in half of the instances belonging to the
subtypes at issue: 19 out of 38.
Perfect of result, thus, is a meaning for which the Present Resultative and the
Present Perfect compete in Modem Basque (cf. ITUN: 6 instances of the PrR for 5 of
the PrP). There seem to be some differences in use within this competition. If no
time adverbial is used, either one between the PrP and the PrR appears to be
possible. (Perhaps the former is more common in negative and interrogative
sentences and in relative clauses.) On the other hand, if time adverbials of the kind a
while ago, some time ago, long time ago are used, then the PrP cannot co-occur with
those implying a time reference beyond the hodiernal scope. In this case, either the
Past Perfect-Perfective is used (therefore conveying perfective meaning) or
alternatively the Present Resultative can appear to express a sort o f remote perfect of
result: *?liburua aspaldian idatzi du ‘he has written the book long time ago’, liburua
aspaldian idatzi zuen ‘he wrote the book long time ago’, liburua aspaldian du
idatzia lit. ‘hei has the book written (by himi) since long ago’ (cf. Haase 1994:281).
This is nevertheless a topic that could be studied in more detail.
In any event, the Present Resultative / Present Perfect-Perfective differences
between LEIZAR and HARAN, on the one hand, and ITUN, on the other, appear in
cases such as: HARAN izatu dute (bere golardoa), errezibitu dute (bere saria) vs.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
243
ITUN hartua dute (beren ordaina) ‘they have gotten (their reward)’ (Mt VI,2; VI,5);
LEIZAR ez da hil (neskatxa) vs. ITUN (neskatxa) ez dago hilda ‘(the girl) is not
dead’ (Mt IX,24); LEIZAR barkhatu zaizkik (cure bekhatuak) vs. ITUN barkatuak
dituzu (bekatuak) ‘(your sins) are forgiven you’ (Mt IX,2; IX,5) ; HARAN hil dire
(ezen) vs. ITUN hilak baitira ‘for they are dead’ (Mt 11,20); LEIZAR ethorri naizela
vs. ITUN etorria naizenik ‘[do not think] that I came [to bring peace on earth]’ (Mt
X,34).
4.2.2. Experiential perfect and perfect o f recent past
As for the other two subtypes in Table 6.1, namely experiential perfect and perfect of
recent past, they are mostly expressed by means of the Present Perfect-Perfective
(PrP), while the Present Resultative (PrR) is absent from these contexts in the
database. It should be said, however, that it is possible in Modem Basque to express
experiential perfect by way of the Present Resultative. This is common in western
Basque, with which I am most familiar. Yet, this use of the PrR is apparently not as
frequent, at least in written texts, as the ones studied in 4.2.1. Moreover, it seems to
be restricted to some classes of predicates.
It ought to be mentioned too that the Double-compound Present Perfect-
Perfective (DcPr) has a notable presence in the corpus, especially in LEIZAR. The
data corroborate what was stated in section 6 of Chapter 5: the DcPr is particularly
2 This example shows another problem o f these semantic subtypes; namely, the interaction between
resultative and passive voice. This is unfortunately an issue that I cannot study in depth here.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
244
used to convey experiential perfect meaning (e.g. 9 out of 17 instances in LEIZAR,
all of them in affirmative sentences; while 6 of the 8 cases of the PrP are negative or
interrogative). HARAN has also three examples of experiential perfect expressed by
a double-compound construction formally parallel to the Present Resultative (which
does not appear in Tablel). But these are instances of the passive voice, which
should be studied separately (see note 2).
As I stated in section 6.3.2 of Chapter 5, experiential perfect is still currently
one of the main uses of the DcPr, though this is not always clear in the grammars of
Modem Basque. Unfortunately, the data in this work does not clarify the issue;
because the DcPr is absent from the 20th century ITUN text. Yet, the DcPr is not
unknown in current Basque, although it is probably not as common in the western
dialects (thus in ITUN) as it is in the eastern dialects.
Thus, the three constructions in competition for the TAM area at issue (PrP,
DcPr and PrR) appear to be possible conveying experiential meaning in Modem
Basque. However, there seem to be differences among these uses. The PrP appears to
be the most common construction for expressing experiential, and is especially used
in negative and interrogative sentences (e.g. ez zaituztet inoiz ezagutu ‘I never knew
you’ Mt VII, 23). In affirmative sentences, on the other hand, there is more
competition among the three constructions at hand. It would appear that the PrP is
more common when the experiential makes also reference to a recent past situation.
The other two constructions, in turn, may refer to a remote experiential. It seems that
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
245
what was stated in the previous section about the competition for perfect of result
between the PrP and the PrR holds also for experiential perfect. Namely: that the
PrP, unlike the PrR, cannot collocate with time adverbials entailing a time reference
prior to hodiernal. The DcPr, finally, seems to imply a frequentative experiential, as
was discussed in Chapter 5: e.g. egin izan dugu ‘we have done it (more than once at
some indeterminate point in the non-near past).’
5. COMPETITION FOR THE YOUNGEST SEMANTIC SUBTYPES IN THE
CLINE STARTING WITH PAST RESULTATIVE.
5.1. Predictions.
We saw in section 3.1 of Chapter 5 that the youngest semantic subtypes in the cline
starting with past resultative are (cited according to the cline-ordering): past
resultative (including past predicative implying some result of an anterior event),
perfect-in-the-past, past-in-the-past and past temporal frame. (I examine the function
of narrative anterior separately, in section 9 below.) The Basque constructions we
find for the competition of these subtypes are the Past Resultative (PsR), the Double-
compound Past Perfect (DcPs) and the Past Perfect-Perfective (PsP).
The predictions we can propose for this TAM area are parallel to those
offered in section 4, and were already put forward in section 3.1. The major
prediction regards the PsR and the PsP. These constructions constitute layers with
the same source of grammaticalization, where the latter is older than the former. The
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
246
hypotheses of Source Determination, Layering and Semantic Unidirectionality
predict for these two constructions to evolve along the same cline of
grammaticalization: i.e. the path in (3) above. In practice we may predict a main
competition between the PsR and the PsP, where the former conveys mostly the
youngest subtypes in this domain and the latter in turn the older. We can also expect
that the PsR is diachronically gaining importance, at the expense of the PsP, starting
with perfect-in-the-past and towards covering all the domain of pluperfect.
Unfortunately, perfect-in-the-past and past-in-the-past, in the absence of specific
time adverbials, are not always easy to distinguish from each other; thus, it is
difficult to test the direction of the cline in this specific interval. As for the Double
compound Past Perfect, its behavior seems very difficult to predict cross-
linguistically.
5.2. Testing the predictions.
Unfortunately, the data on this sphere of meaning are very scarce. For this reason, I
can hardly test the hypothesis of Unidirectionality of Meaning. In any case, as
expected, the few data at hand show that the very youngest subtypes in the cline, i.e.
past resultative (and past predicative), are conveyed by the Past Resultative. As for
the other semantic subtypes, it is really difficult to draw any conclusions from the
data, other than a main competition between PsR and PsP. Therefore, this
competition should be studied in another work, with other types of data.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 4 7
The Double-compound Past Perfect (DcPs), in its turn, is only frequent in
LEIZAR, and particularly to express ‘past temporal frame’ meaning. (Incidentally,
this latter meaning is in the specific examples I collected not easy to distinguish from
unwitnessed perfective, which will be studied in section 8.) In ITUN, as in Modem
Standard Basque, both past temporal frame and evidential perfective are indistinct
from the common pre-hodiemal perfective, expressed by the Past Perfect-Perfective.
TABLE 6.2: Youngest subtypes in the cline starting with past resultative (Main Corpus)-
SEMANTIC
SUBTYPES
LEIZAR HARAN ITUN
PsR DcPs PsP PsR DcPs PsP PsR DcPs PsP
past resultative 2 2 8 8 3 3
perfect-in-the-past 3 1 2 4 1 1 2 1 1
past-in-the-past 3 2 1 2 2 1 1
past temporal frame 6 5 1 5 3 1 0
Other 1 1 5 1 4 2 2
TOTAL 15 3 10 2 24 10 4 9 7 3 4
6. COMPETITION FOR DEGREES OF REMOTENESS.
As previously mentioned, the case of Basque represents a rather consistent evolution
of the two Perfect-Perfective constructions at issue towards the sphere of perfective
meaning. Thus, the Present Perfect-Perfective extended its meaning to cover
hodiernal perfective, following the cline in (2) above, while, in a parallel fashion, the
Past Perfect-Perfective evolved to a remote perfective, along the cline in (3). This
way, the Present Perfect-Perfective and the Past Perfect-Perfective opposed to each
other, within the domain of perfective, in a distinction based on degrees of
remoteness. Concretely, the opposition is one of hodiernal / pre-hodiemal perfective,
respectively. I also stated in section 7.3 of Chapter 5 that the evolution of the Past
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
248
Perfect-Perfective to a remote perfective is not very common in the European
languages, but attested elsewhere.
The data in the database cannot add more specifics to the already mentioned,
because unfortunately there are no clear instances of hodiernal perfective that could
be distinguished from a recent past. In other words, there are no unambiguous cases
in the database of perfective with definite time-adverbials referring to the ongoing
day or the ongoing unit of time. It seems that, in order to observe what time-point is
the relevant in delimiting the hodiernal / pre-hodiemal opposition of Basque, spoken
data would be needed.
What a written-text database does offer is a fair number of instances of pre-
hodiemal and narrative perfective. This sphere of meaning will be the object of study
of the next two sections. We will see that there have occurred important changes in
historical Basque in the domain of pre-hodiemal and narrative perfective (section 7).
We will also see that the Present Perfect-Perfective could be used, at least in the
eastern dialects of Old Basque, as an unwitnessed and fictional narrative (section 8).
This fictional function of the Present Perfect-Perfective could perhaps be related to
the atypical pre-hodiemal use we find in current varieties of the French-Basque
dialects (see sections 2.3 and 7.4 of Chapter 5). However, in a first sight, it may seem
there is no relation between the two phenomena.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
249
7. (HISTORICAL) COMPETITION FOR PERFECTIVE MEANING.
As I stated above, one of the most noticeable changes we need to explain in the
history of the Basque TAM system throughout the last five centuries is the loss of the
Periphrastic Aorist. We will see from the data below that the Periphrastic Aorist was
ousted by the Past Perfect-Perfective. After the disappearance of the Periphrastic
Aorist, the dynamic-achievement auxiliaries remained confined to non-indicative
contexts. Hence, in Modem Basque they are analyzed as subjunctive or non
indicative auxiliaries. The Synthetic Past, in its turn, also lost its use as a perfective,
apparently in a concurrent time with the loss of the Periphrastic Aorist. Thus, the
Past Perfect-Perfective stayed as the only form conveying general pre-hodiemal
perfective in Modem Basque (but see section 7.4 of Chapter 5 for the pre-hodiemal
perfective use of the Present Perfect-Perfective in Modem French-Basque dialects).
The loss of the Periphrastic Aorist was followed by a subsequent change that
seems worth mentioning, though it will not be studied in this work. The Old Basque
(Periphrastic) Past Subjunctive and the Periphrastic Aorist were extremely similar in
formal terms. They only opposed each other in the third-person forms: e.g. har zezan
‘he took it’ vs. har lezan ‘so that he took it.’ (See Aldai 2000 for discussion on these
issues.) After the disappearance of the Periphrastic Aorist, the third-person forms of
the Old Basque Past Subjunctive had their /- prefix replaced by the z- past prefix.
Therefore, the Past Subjunctive of Modem Basque is identical in form to the Old
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
250
Basque Periphrastic Aorist (and different in the third persons to the Old Basque Past
Subjunctive).
I present in Table 6.3 the data in the Main Corpus on the evolution of (pre-
hodiemal) perfective meaning in the last five centuries of Basque. The constructions
appearing in this evolution are the Synthetic Past, the Periphrastic Aorist, the Past
Perfect-Perfective and also the Double-compound Past Perfect. I exclude from the
data in Table 6.3 unwitnessed and fictional perfective (they are examined in section
8 below). I also exclude from Table 6.3 the examples of passive voice, which are
touched upon below.
TABLE 6.3: Pre-hodiemal perfective non-fictional non-passive (Main Corpus).
LEIZAR (16th c.) HARAN (18th c.) ITUN (20th c.)
TOTAL 126 150 161
SYNTHETIC
PAST
1 0 0
PERIPHRASTIC
AORIST
115
(91.3%)
0 0
PAST PERFECT-
PERFECTIVE
2 134
(89.3%)
161
(100%)
DOUBLE-COMPOUND
PAST PERFECT
7 13 0
OTHER
CONSTRUCTIONS
1 3
[Aorist+fe] = 2
0
Table 6.3 shows the loss of the Periphrastic Aorist and of the Synthetic Past with
perfective meaning, both ousted by the Past Perfect-Perfective (PsP) and the Double
compound Past Perfect (DcPs). The Periphrastic Aorist disappeared from practically
all the varieties of Basque by the end of the 17th century (see Lakarra 1996:177-
179). (HARAN presents two instances of the Periphrastic Aorist plus the futurity-
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
251
modality morpheme -ke, but these are cases of until-perfectives, and are therefore
somewhat different from normal perfective meaning.) As for the Synthetic Past with
perfective meaning, it seemed to experience an evolution parallel to that of the
Periphrastic Aorist. That is, it appears that the loss of the latter occurred concurrently
with the confinement of the former to imperfective contexts, mostly in a handful of
stative verbs. While the Synthetic Past with perfective meaning appears seldom in
LEIZAR (only once in Table 3: concretely, a form of the verb egin ‘to do’), it is not
infrequent in other texts of Old Basque (cf. Lakarra 1996:177-179).
An interesting point that we would like to uncover from the data is the exact
beachhead by means of which the PsP started its penetration into the sphere of pre-
hodiemal perfective from pure pluperfect meaning. In other words, it appears worth
looking for the contexts where the PsP (and its similar the DcPs) appears in LEIZAR
expressing meanings other than past resultative or pluperfect. One of the most
important o f these uses occurs in past temporal frames. This beachhead was
predicted in Chapter 5: cf. Dahl (1985:147), Squartini (1999:55-56). In the data at
hand, past temporal frames are hard to distinguish from unwitnessed contexts. Thus,
I examine both of them when dealing with evidentiality in Table 6.4 (section 8).
Another use of the PsP and the DcPs in LEIZAR is in the function of narrative
anteriority, which I will study separately in section 9 below.
Now, besides the above-mentioned uses of the PsP and the DcPs in LEIZAR,
there are in Table 6.3 seven instances of the DcPs and two of the PsP that might be
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
252
worth mentioning. These are examples of the verbs j'aio ‘be bom ’, nahi ukhan
‘want’, and particularly several cases of intransitive verbs with plural subject and
dative agreement, as in jarreiki izan zaizkan ‘they followed him’, hurbildu izan
zaizkan ‘they came to him.’ In these cases, the DcPs or the PsP appear more common
than the Periphrastic Aorist. It would seem that stative verbal locutions, such as
‘want’, and passive-like verbs, such as ‘be bom’, would be among the most plausible
predicates to have stative-auxiliary perfectives, instead of dynamic-auxiliary
perfectives. Thus, it does not appear strange for these predicates to be diachronically
among the earliest showing a stative-auxiliary perfective, i.e. a PsP or a DcPs. As for
the dative-agreement intransitive verbs just mentioned, I do not find at this point any
reason for their having stative-auxiliary perfectives instead of the otherwise more
common in LEIZAR Periphrastic Aorist.
Not included in Table 6.3 are some instances of the DcPs with perfective
meaning in passive sentences. Again, the correlation between passive perfective and
stative-auxiliaries does not seem surprising, although passives with dynamic
auxiliaries (i.e. in a Periphrastic Aorist form) are not unknown in Old Basque: cf.
non egin zidi? (“donde fue hecho?”) ‘where was it done?’ (Mikoleta 1653). In
LEIZAR, the most frequent form to express passive-voice sentences with perfective
meaning is the Double-compound Past Perfect, while the Periphrastic Aorist is
seldom used in this function. As a matter of fact, one of these passive examples is a
Triple-compound Past Perfect: erran-izan izan zaien (lehenagokoei) ‘it was said (to
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
253
those of old)’ [lit. ‘it was been been said to them’] (Mt V, 21). This example, besides
being a passive, expresses a past temporal frame or an evidential.
8. EVIDENTIAL.
One of the most interesting TAM areas to examine in the domain of past in historical
Basque is that concerning unwitnessed and related meanings. Recall that in section
2.3 of Chapter 5 a sub-cline starting with present resultative was studied which
covered the area of evidentiality. I repeat that cline below for convenience.
(4) perfect of result -> inferential -> unwitnessed perfective fictional
narrative
There are two possibilities for this specific area of TAM meaning. Either
there is a special form to express these meanings (or to express some of them), or
alternatively these subtypes of perfective meaning are covered by the construction
that generally conveys perfective and narrative. The former alternative, i.e. the
existence of a specific evidential form, is reported for languages such as Turkish,
Bulgarian, Georgian, Macedonian, Estonian, Tibetan languages, Udmurt, Inuit,
Tucano, and a number of others (cf. Connie 1976:108-110; Bybee et al. 1994:95-97).
Notice that, according to the cline in (4) above, the special form expressing
unwitnessed perfective and related meanings can be formally identical to the one
expressing present perfect. Therefore, this scenario may lead to somewhat
schizophrenic TAM systems, such as the following: First, a given language can have
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
254
a specific construction to convey pre-hodiemal perfective and narrative, and an
opposing construction, e.g. the Present Perfect, to convey present perfect and
hodiernal perfective. Now, it may be the case that the latter acquires also the
meanings of unwitnessed perfective and fictional narrative. As just mentioned, this
situation appears as a split one, because the construction expressing hodiernal
perfective (and thus opposing another form with pre-hodiemal perfective meaning)
can also convey unwitnessed perfective, a meaning which often refers to situations
with time-references more remote than those expressed by the construction
conveying pre-hodiemal perfective. This is exactly the scenario we find in Old
Basque. More exactly, we should say this is documented for the eastern dialects of
Old Basque. As for the western dialects, the attestation of narrative contexts is so
scanty that we can hardly extract any conclusions regarding these issues.
I present in Table 6.4 below the data in the Main Corpus concerning the
meanings of unwitnessed, quotative and fictional narrative. I also consider a sub
division which takes into account the construction used in the first sentence of a
fictional narrative sequence. The reason for doing so is the fact that the first verb in a
sequence of fictional narrative appears to be most often marked with the special
fictional form (e.g. the Present Perfect-Perfective), while the ensuing verbs in the
fictional sequence seem more likely to be expressed by the construction conveying
general narrative perfective (cf. Dahl 1985:114). The Basque constructions that can
appear in the competition for these contexts are the following: a) special evidential
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
255
forms: Present Perfect-Perfective (PrP) and Double-compound Present Perfect
(DcPr), and b) general pre-hodiemal perfective forms (i.e. those studied in the
previous section: cf. Table 6.3): Periphrastic Aorist, Past Perfect-Perfective, Double
compound Past Perfect and Synthetic Past (the latter two hardly appear in the
database in these specific contexts). Recall that the main construction conveying
general perfective in LEIZAR is the Periphrastic Aorist (A), while in HARAN and
ITUN it is the Past Perfect-Perfective (PsP) that mainly expresses general perfective.
TABLE 6.4: Evidential: unwitnessed, quotation and fictional narrative contexts (Main Corpus).
SEMANTIC
SUBTYPES
LEIZAR HARAN ITUN
PrP DcPr A PrP DcPr PsP PrP PsP
unwitnessed 8 2 2 4 7 3 2 2 3 2 1
quotative 6 6 6 2 4 5 2 3
fiction narrative 37 1 19 16 1 34 7 4 18 5 25 25
first verb 12 6 2 3 1 9 1 4 3+
1
11 8 3
TOTAL 63 9 29 19 6 56 9 12 24 11 44 2 38 4
% 14 46 30 10 16 21 43 20 5 86 9
Table 6.4 shows a correlation between the contexts at issue (i.e. unwitnessed,
quotation, fictional) and the Present Perfect-Perfective (PrP) or its similar the
Double-compound Present Perfect (DcPr), both of which are constructions with
Present stative auxiliaries. This correlation is especially clear in LEIZAR, but it is
also appreciable in HARAN. In ITUN, on the other hand, there is no special
evidential form.
Looking at Table 6.4 in more detail, in LEIZAR the PrP and the DcPr
together sum a total of 60% of the evidential contexts, (the Periphrastic Aorist, a
30%; and the Past Perfect-Perfective and other Past-auxiliary forms, a 10%). In
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
256
HARAN, on the other hand, the correlation between evidential contexts and Present-
auxiliary forms is lower, but still significant: 37% of the total of instances are given
in either the PrP or the DcPr, 43% belong to the PsP, and 20% to the Double
compound Past Perfect or other constructions. Among the latter, it is worth
mentioning the existence of 3 instances, i.e. a 5% of the total, of the PrP plus the
potentiality marker -ke. These three examples of PrP+ke show up always in first-verb
instances of fictional narrative sequences. If we sum these last forms to the 37% of
PrP and DcPr, we have a total of 42% of evidential contexts expressed in HARAN
by means of Present-auxiliary forms.
Notice that in the two older texts it is the DcPr that is more often used as
evidential. However, in other Old Basque texts of the eastern dialects it is the PrP
that is mostly used: In Etxepare (1545), for example, the PrP is used in all but one of
the 9 instances of evidential perfective (the only exception is a Periphrastic Aorist).
Oihenart (1657), in turn, uses both the PrP and the DcPr as special evidential forms,
along with general perfective constructions (cf. his 20th poem); Tartas (1666)
employs the PrP when using a special fictional form (cf. his Iraccurtgaliari
introduction). Therefore, we must conclude that either the Present Perfect-Perfective
or the Double-compound Present Perfect (both with Present auxiliaries) could be
facultatively used as evidential, at least in the eastern dialects of Old Basque (and no
less than until the 18th century).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
257
Note that, when I refer generally to the Old Basque evidential, 1 include also
fictional narrative contexts. As a matter of fact, fictional narrative is among the
contexts where one can most frequently find the special evidential forms; i.e. the PrP
or the DcPr. This is most especially true of the first verbs in a fictional sequence.
Thus, in LEIZAR, 20 out of 37 cases of fictional narrative (54.0%) are given in one
of the special forms with Present auxiliaries. In the case of the first verbs in the
fictional sequence, the percentage increases to 8 out of 12: a 66.7%. In HARAN,
only 11 cases of 34 instances of fictional narrative (32.3%) are rendered by means of
special Present-auxiliary forms. In first-position verbs, in turn, the percentage of
Present-auxiliary forms is 4 out of 9: a 44.4%.
A complication with the above data, particularly with what I strictly termed
in Table 6.4 “unwitnessed” contexts, is related to the nature of the texts used in the
Main Corpus. Recall that the Main Corpus texts are versions of the Gospel according
to St Matthew. In this text, the normal narrative contexts are those referring to the
events in Jesus’ life; the “fictional narrative” contexts refer to the parables narrated
by Jesus to his disciples; the “quotative” contexts are mostly quotes of sayings by the
old prophets; and the “unwitnessed” contexts are mostly parts of the main text
referring to the prophets or to other protagonists of the Old Testament. It is in these
latter contexts where the doubts arise, for one does not know with certainty whether
these parts of the main body of the text should be interpreted as evidential or as past-
temporal-frame contexts. If evidential meaning is assumed, then the Present Perfect-
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2 5 8
Perfective is the most expected form. Yet, if the interpretation were one of a past-
temporal-frame, then the expected construction would be the Past Perfect-Perfective.
In reality, both types of forms appear in LEIZAR and HARAN. The Periphrastic
Aorist, in turn, does not show up in these contexts.
A question I will not pursue here is whether this use of the Basque Present-
auxiliary Perfect-Perfective forms in pre-hodiemal contexts could be related with the
employment of parallel forms as pre-hodiemal perfective in French. At first glance,
it would seem that the French-Basque and French evolutions are independent to each
other. In the case of the French-Basque dialects, it appears clear from the texts that
the Present-auxiliary Perfect-Perfective forms were only used in Old Basque as
unwitnessed and fictional, and not in any pre-hodiemal context. The current use of
the Present Perfect-Perfective as general pre-hodiemal perfective in some varieties of
the French-Basque dialects seems a more recent evolution. Yet, this issue should be
studied separately in closer detail.
9. NARRATIVE ANTERIORITY.
The last context I will examine in this chapter is that concerning the function I
termed ‘narrative anteriority’ in section 3.2 of Chapter 5. As discussed there,
narrative anteriority is often expressed by means of a nonfinite form, but, in written
texts, finite forms are not uncommon in this function. The purpose of the present
section is to examine what finite forms have fulfilled the function of narrative
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
259
anteriority in historical Basque. As mentioned in Chapter 5, the most expected
constructions to fulfill this function are those generally expressing perfective and
pluperfect meanings. In case the pluperfect is grammaticalized with past stative
auxiliaries having a perfective / imperfective opposition, then it seems it is the
perfective pluperfect that expresses narrative anteriority.
In the case of Basque, however, we saw in section 7.3 of Chapter 5 that the
stative auxiliaries of the Past Perfect-Perfective were unmarked for aspect.
Therefore, we may expect that this construction can express narrative anteriority. On
the other hand, it should be bom in mind that the Past Perfect-Perfective is the form
that conveys perfective meaning in Modem Basque, while it generally did not
express perfective in Old Basque. Thus, one may think that the change in meaning
undergone by the Past Perfect-Perfective (i.e. roughly speaking, from pluperfect to
perfective) could have triggered a change of the form fulfilling the function of
narrative anteriority. In fact, I mentioned in section 6.3 of Chapter 5, following
Lafitte (1962), that the Double-compound Past Perfect seems to express narrative
anteriority in Modem Basque.
However, the data in the Main Corpus database, even though scanty, do not
support Lafitte’s observation as being highly significant. I present these data in Table
6.5 below.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
260
TABLE 6.5: Finite forms in the function of narrative anteriority (Main Corpus).
LEIZAR HARAN ITUN
TOTAL 15 23 8
SYNTHETIC
PAST
0 0 0
PERIPHRASTIC
AORIST
0 0 0
PAST PERFECT-
PERFECTIVE
12 22 8
DOUBLE-COMPOUND
PAST PERFECT
3 1 0
As can be observed in Table 6.5, there is hardly any appreciable changes in historical
Basque regarding the finite form used in narrative anteriority contexts. Concretely, it
is the Past Perfect-Perfective that consistently appears as the major construction for
expressing the value under consideration throughout the whole period. Thus, the
Double-compound Past Perfect seems to be just a secondary form in narrative
anteriority contexts. As a matter of fact, it does not even show up in the
(unfortunately few) finite examples in ITUN. It ought to be conceded, nevertheless,
that we may be facing here one of the main general problems with the database in
this work, already mentioned in Chapter 1. Namely, the different dialectal source of
the three texts used. More specifically, it may be the case that the Double-compound
Past Perfect is more common as narrative anterior in Modem Navarro-Labourdin
(the object of study of Lafitte’s work) than it is in other dialects and in Standard
Basque (thus in ITUN). Yet, this ought to be dealt with in another study. As for the
Periphrastic Aorist, it is remarkable to notice that it was not used as narrative anterior
in LEIZAR, despite the frequent use of this construction (as perfective) in the oldest
text.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
261
In sum, the major construction used in narrative anteriority contexts has
apparently remained unchanged throughout the last five centuries of Basque. It is the
Past Perfect-Perfective that has been mostly used in this function, (with the Double
compound Past Perfect seemingly employed as a secondary form). On the other
hand, the function of narrative anteriority may have served as a beachhead in the
change undergone by the Basque Past Perfect-Perfective from pluperfect to
perfective. Recall that other beachheads in this passage seemed to be past-temporal-
frames, stative predicates, passives, passive-like predicates and apparently dative-
agreement intransitive verbs (see section 7 above).
It may seem that the path which leads from narrative anteriority towards pure
perfective meaning can progress through a stage where narrative anteriority applies
not only to subordinate clauses but also to sequences of main sentences “in order to
underline the temporal sequence of events” (Squartini 1999:61; see section 6.1 of
Chapter 5, and the Old German example given there). A case in point can be found
in the Old Biscayan dialect Refranes y Sentencias book (1596:513): eznea, guria ta
odola errorean atera neban, ta ezer irabazi ez nezan, ta ene beitxua gal nezan ‘I got
[lit. had gotten] milk, fat and blood out of the udder, and I didn’t earn anything, and I
lost my little cow’, where the first verb in the sequence of three is in the Past Perfect-
Perfective, while the other two are in the Periphrastic Aorist (cf. Aldai 1998: 383-
384).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
262
It seems worth mentioning that, if the proposed beachhead were correct, the
putative passage would progress starting with subordinate contexts and towards main
sentences. As a matter of fact, it does appear that the Past Perfect-Perfective was in
Old Basque more common in subordinate contexts than in main-clause contexts, and
that it only generalized as perfective and narrative in main sentences at a later stage.
(In the database, only 8 examples out of the 44 instances of Past Perfect-Perfective in
LEIZAR appear in main clauses, while there are 165 cases of main-clause Past
Perfect-Perfectives in ITUN out of a total of 184.) This evolution, thus, seems to
constitute an exception to the otherwise well-supported hypothesis that subordinate
clauses tend to be conservative because new grammaticalizations tend to occur in
main clauses (cf. Givon 1979:85; Bybee et al. 1994:231; Bybee to appear; and the
entry ‘sentential sources’ in the Glossary).
10. CONCLUSIONS.
The domain of non-imperfective past I have examined for historical Basque in this
chapter appears, on the one hand, as rather complex, regarding the number of
changes and constructions in competition during the last five centuries. On the other
hand, the picture we get from the data seems easy to explain according to the
grammaticalization framework assumed in this work and developed in Chapter 5 for
this sphere of meaning.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
263
1) The competition for the earliest semantic subtypes in the clines in (2) and
(3) above, which were put forward and explained in Chapter 5, shows up in Basque
as two fairly clear cases of layering: The Present Resultative and the Present Perfect-
Perfective are layers (the former being younger than the latter) in competition for the
earliest subtypes in the cline in (2). The Double-compound Present Perfect also
competes for some of these meanings, most especially for experiential perfect. In
turn, the Past Resultative and the Past Perfect-Perfective constitute layers in
competition for the earliest meanings in the cline in (3). Despite the existence of
some problems with the data (particularly, the paucity of data for some specific
subtypes, and the fact that the Resultative constructions have not evolved much in
their corresponding path of grammaticalization), the two cases of layering just
mentioned appear as verification of the hypotheses of Source Determination,
Layering and Semantic Unidirectionality.
2) As for older semantic subtypes in the clines o f (2) and (3), more concretely
for the domain of perfective, an opposition occurs between constructions of the same
origin and same age in the language. Specifically, the Present Perfect-Perfective and
the Past Perfect-Perfective compete for degrees of remoteness. The former conveys
hodiernal perfective, whereas the latter conveys pre-hodiemal perfective. This
scenario is again among those possible that can be (if not predicted) at least
explained a posteriori against the framework put forward in the clines in (2) and (3)
above.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
264
3) Finally, within pre-hodiemal perfective and narrative, there has occurred a
competition which appears as one of the most complex in the historical period of
Basque. In Old Basque, the major construction to express (non-fictional) pre-
hodiemal perfective was the Periphrastic Aorist. However, from the very first texts
the Past Perfect-Perfective and, to a lesser extent, the Double-compound Past Perfect
(both with past stative auxiliaries) compete also for this meaning. (I have proposed
above some of the contexts which seemingly served as beachhead for the Past
Perfect-Perfective to evolve from pluperfect to perfective. Among these beachheads,
we must apparently include the function of narrative anteriority.) It should be
acknowledged that the Past Perfect-Perfective seemed to be quite generalized to most
perfective contexts already in the 16th century (see discussion below). The Synthetic
Past was another form also in competition for this sphere of meaning in Old Basque.
It was, though, restricted to a small number of verbs. In any event, by the end of the
17th century, the Periphrastic Aorist and the Synthetic Past lost the possibility of
expressing perfective meaning. The Past Perfect-Perfective remained then as the
major construction to express perfective and narrative in Modem Basque (with a
divergent situation, apparently recent, in the French-Basque dialects).
A complication to the above scenario is the existence, at least in the eastern
dialects of Old Basque, of a distinct form that could be used optionally as pre-
3 Another context that may be worth examining as a putative beachhead of the passage from
pluperfect to perfective is that of negative sentences. Thus, in some of the Old Basque texts, most
importantly in Refranes y Sentencias (1596), there seems to be some correlation between affirmative
sentences and the Periphrastic Aorist vs. negative sentences and the Past Perfect-Perfective.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
265
hodiernal perfective in fictional and unwitnessed contexts. This function was
expressed by the Present Perfect-Perfective and also by the Double-compound
Present Perfect (both of which are constructed with Present stative auxiliaries). Thus,
in some authors, the picture is extremely complicated (and somewhat schizophrenic,
as explained above): the Synthetic Past, the Periphrastic Aorist, the Past Perfect-
Perfective, the Double-compound Past Perfect, the Present Perfect-Perfective and the
Double-compound Present Perfect could appear in competition for very similar
meanings. (A study that seems worth undertaking is the evolution in the 18th , 19th
and 20th centuries of the constructions with stative Present auxiliaries as pre-
hodiemal fictional perfective in the French-Basque dialects.)
The complex situation just described is readily explained based on the
framework put forward in Chapter 5. Although specific and detailed evolutions are
arguably impossible to predict, all the developments that occurred in Basque are
among those that perfectly fit with the clines explained in the previous chapter. Thus,
the use of the constructions with Past stative auxiliaries as remote perfective is
accounted for by the cline in (3) above. Likewise, the use of the constructions with
Present stative auxiliaries as inferential, unwitnessed and fictional narrative is
explained by the cline in (4) above.
The Synthetic Past and the Periphrastic Aorist, in turn, are constructions
whose source of grammaticalization is not as well understood as that of the perfect
constructions with stative auxiliaries. Thus, it is more difficult for us to apply the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
266
hypotheses on grammaticalization to the Synthetic Past and the Periphrastic Aorist.
However, for what we know about the source of these forms (see sections 7.1 and
7.2 of Chapter 5), they seemed to be constructions very appropriate for conveying
perfective meaning. The Periphrastic Aorist, concretely, was grammaticalized upon
dynamic-achievement auxiliaries in past tense. Hence, it does not appear strange to
find the Periphrastic Aorist expressing indicative meaning in Old Basque. In other
words, it appears unnecessary to doubt the reality and actual use of the Periphrastic
Aorist as a ‘normal’ perfective in earlier stages of Basque (cf. Krutwig, cited in
Irigoien 1985:7-8; or Trask 1997:238).
A different question is whether the Periphrastic Aorist was already in
recession in the spoken language by the 16th century (cf. Mitxelena 1960:55, Lakarra
1996:178). Clearly, any discussion on this issue will be partly speculative. However,
the fact that Etxepare (1545) uses most often the Periphrastic Aorist in his
autobiographic XIHth poem (Mosen Bemat Etxaparere kantuia ‘the song of
Monsieur Bernard d ’Etchepare’) and less so in the religious poems, appears as a
small piece of evidence for assuming that the Periphrastic Aorist was rather normal
in colloquial speech in the first half of the 16th century4. This could happen at least in
the most extreme dialects of Basque: Biscayan in the far west and Souletin and
Lower Navarrese in the far east. It seems that the loss of the Periphrastic Aorist
started earlier in the central dialects and occurred later in the farthest varieties. In the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
267
first half of the 17th century, the Periphrastic Aorist is still documented for Biscayan
(Mikoleta 1653, Kapanaga 1656; cf. Lakarra 1996:177) and Souletin (Oihenart
1657). On the other hand, it is not used in Labourdin, Navarrese and Alavese by this
time. Thus, the generalization of the Past Perfect-Perfective as pre-hodiemal
perfective and the consequent loss of the Periphrastic Aorist seem to have occurred
following one of the most common ways of change across dialects: starting from the
central area and later extending to the extremes.
4 Similarly, Leizarraga (1571), a meticulous Calvinist, uses the Periphrastic Aorist not only in the
passages referring to the life of Jesus (where incidentally he employs colloquial allocutive treatment
too) but also in his introduction, which is not of religious nature.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
268
CHAPTER 7
CONCLUSIONS
1. INTRODUCTION.
In the present chapter, I wrap up some of the most relevant questions examined
throughout this work. I put forward several conclusions, regarding a) general
characteristics of TAM systems (section 2), b) the hypotheses on grammaticalization
(section 3), and c) general principles or motivations behind these characteristics and
processes (section 4).
The most important conclusions may be presented as follows: The general
characteristics of TAM systems are due to the fact that TAM systems are created in
language use through a process of grammaticalization which is very similar across
languages, (though it can also vary importantly, specially from linguistic area to
linguistic area). The ultimate principles behind this process of grammar creation are
a number of competing motivations of diverse nature. To claim that TAM systems
are created in language use implies very different things, depending on what kind of
motivation is in operation in a given process. What these motivations have in
common is their being non-autonomous, i.e. non-specific to the language faculty, but
general to other cognitive abilities (Clark & Malt 1984:211).
I will discuss below motivations in the grammaticalization ofTAM systems
which are based on mental traits such as cognition and prototypicality, processing
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
269
(including frequency-sensitive processing), leamability, social fashions and
expressiveness, economy of communication, inertia vs. change, expansion-
generalization-extension, routinization-automation, differentiation, semantic and
pragmatic associations, perceptual optimality of form, etc. (See Hawkins ed. 1988,
for different types of explanations).
2. GENERAL CHARACTERISTICS OF T.A.M. SYSTEMS.
The data in this work provide evidence for a model of TAM systems which would be
importantly based on the following hypotheses.
1) TAM systems are created (and re-created) in language use.
2) Different competing motivations and conventions shape (design) the
structure of TAM systems.
3) In Basque, as in most languages of the world, TAM constructions are
subject to a specific process of grammaticalization (in a narrow sense)
that generally leads from lexical analytic expressions to more abstract
morphological markers. This process has been a major object of study in
this work. (See the next section for specific conclusions on
grammaticalization.)
4) On the one hand, synchronically, TAM systems show a competition of
various forms for each meaning (one-meaning-many-forms).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
270
5) On the other hand, there is a tendency towards one-meaning-one-form.
(This tendency is one possible motivation, among others.)
6) On the one hand, formal organization of TAM markers need not
correspond with semantic categorization.
7) On the other hand, there is a tendency (one possible motivation) towards
making formal organization correspond with semantic categorization.
8) Prototypicality of TAM meanings appears as a central concept in the
shaping of TAM systems (although more research is needed on this
issue). Prototypicality may affect the degree of the one-meaning-one-
form tendency, the speed of the diffusion of the changes, and in general
the whole design of the system, including the (un)markedness of TAM
constructions. Prototypicality of TAM meanings seems to correlate with
their (type) frequency.
9) The conservation of very old TAM forms, declining in the system,
correlates with the token frequency of specific items.
2.1. TAM systems are created (and re-created) in language use.
In this work I have been able to establish the source of grammaticalization of the
great majority of the Basque TAM forms under consideration. I have also shown
how the corresponding sources of grammaticalization gave rise to the TAM clines
that the Basque forms are following. Moreover, I have proved that these sources of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
271
grammaticalization have clear and common parallels cross-linguistically. Now,
proposing the source of grammaticalization of a given TAM morphological form
entails claiming that this morphological form was once created out of a more
syntactic construction. In other words, it implies maintaining that the TAM forms
were created (i.e. grammaticalized) in language use (i.e. in performance).
Obviously, if we are to base our claims on empirical grounds, it is not
possible to ensure that all TAM forms are created in performance. This is so because
it is simply impossible to establish the source of grammaticalization of all the TAM
constructions in all the languages of the world. However, it appears safe to assume in
general that (most of the grammatical) TAM constructions have been historically
grammaticalized from language use. Note that TAM systems that are lowly
grammaticalized do not constitute counter-evidence to this last claim.
In other words, determining sources of grammaticalization in the way this
work has been able to do is a piece of evidence in favor of the hypothesis that “all
synchronic states are the outcome of diachronic processes” (Greenberg 1966:10).
Thus, the theory of grammaticalization, which I assumed and tested in this work, is
part of the more general hypothesis maintaining that grammars are
conventionalizations of performance (see Hawkins 1994:19ff, 1999, 2001 and related
work).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
272
2.1.1. Morphology vs. syntax.
The above statement that grammars are conventionalizations of performance means
quite different things for morphology and for syntax, e.g. constituent ordering. This
is so because the factors that play a role in each of these components are also rather
different.
In the case of TAM systems and other morphological components, we need
to specify how a whole organization of semantic oppositions is created (designed).
Cognition and conceptual categorization of the real world plays a very important role
in this organization, (as we will see below). Moreover, the design of these
morphological systems is very indirect; that is, it takes a relatively long span of time
and a complex set of changes (i.e. a ‘process’ of grammaticalization with lexical
input) for the conventionalization and morphologization to get completed.
On the other hand, in the case of constituent order and syntax, the
conventionalization of motivating factors seems to be more direct. There is hardly
any historical process of formal reduction (morphologization), and not too many
overt changes seem to occur, other than the conventionalization itself. In addition,
there is no need to create a system of semantic oppositions. The factors in force
compete for the placing of syntactic constituents in a relative order.
2.2. Competing motivations and conventionalization.
I will discuss these issues in section 4.2.4 below.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
273
2.3. Competition among TAM forms.
The concept of ‘competition’ proves crucial for describing the data in this work in
both its synchronic and diachronic perspectives. For every stage of language, there is
nearly always more than one form to express one and the same TAM meaning. As a
matter of fact, in the whole database there is only one exception to this one-meaning-
various-forms situation. Namely, for the case of the past perfective in the 20th
century text, only one form (concretely the Past Perfect-Perfective) appears in the
database in all of the collected tokens (see Table 6.3). (In truth, perfective meaning
can also be expressed in Modem Basque by a ‘historical present’, but I did not
include this possibility in the above counting.)
A point that merits discussion (see below) is whether data such as the present
one, so clearly indicating that each synchronic stage of language has an important
component of overlap and optionality of forms for every TAM meaning, can be
accurately represented by symbolic discrete models. (There seems to be always the
possibility of assuming that the competition of forms is due to “coexisting
grammars” rather than to “optionality” (see Kroch 1989, Lightfoot 1999). Yet, this
may seem just an artificial gambit to try to fit the data into the theory.)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
274
2.4. The tendency towards one-form-one-meaning revisited.
Despite the generalized existence of competition among forms in all stages of the
language, the data in this work provide evidence for a one-form-one-meaning
tendency. As Rosch (1978:35) puts it, “cognitive economy dictates that categories
tend to be viewed as being as separate from each other and as clear-cut as possible.
[Although] to argue that basic [...] categories follow clusters of perceived attributes
is not to say that such attribute clusters are necessarily discontinuous.” Therefore, it
seems important to revise the one-form-one-meaning correspondence and to
formulate it quite differently from the way formal structuralist approaches have
stated it so far.
1) First, according to the data, the one-form-one-meaning correspondence
(i.e. the assumption of discreteness) does not seem to qualify for a principle or
primitive of the theory, but is rather a tendency or motivating factor; i.e. one factor
among others possibly competing with it.
2) Second, the tendency towards one-form-one-meaning is better observed in
terms of frequency rather than in absolute terms. That is, while in absolute terms the
database practically always shows more than one form for every TAM meaning (see
section 2.3 above), in statistical terms it is most often the case that one among those
forms is considerably more important than the rest, with percentages of around 80-
90% of the total of verbs (i.e. in type-frequency) and/or items (i.e. in token-
frequency).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
275
3) Third, the degree of the tendency towards a one-form-one-meaning
correspondence seems to correlate with the prototypicality of the TAM meanings
involved (see section 2.6 below). The issue of prototypicality, though, is a very
complex one to which I turn next.
2.5. Prototypicality of TAM meanings & TAM design.
In this section I touch upon some of the issues that the concept of ‘prototypical TAM
meaning’ may entail. Prototypicality appears as a very multifaceted matter, which I
cannot examine in detail here. Further research should be carried out to better
understand the relationship between prototypicality and (un)markedness (cf.
Greenberg 1966, Comrie 1976:111-122), basicness, relevance (cf. Bybee 1985),
accessibility,presupposition (cf. Givon 1979:48-49), frequency (cf. Greenberg 1966,
Comrie 1976, Bybee 1985), recoverability, (non)-complexity, etc, and also between
these concepts and cross-linguistic commonness (cf. Bybee & Dahl 1989, Kemmer
1992, Bybee et al. 1994:300-301). In addition, we need a non-circular cognition-
based definition of prototypicality, which ought to be independent from cross-
linguistic data (cf. Givon 1995:16-17).
In what follows, I provisionally assume a gradient concept of ‘prototypicality
of TAM meanings’ which would be based on the degree of cognitive salience of
each of the TAM meanings. Thus, I will talk about ‘prototypical TAM meanings’
and ‘non-prototypical or less prototypical TAM meanings.’ According to this point
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
276
of view, prototypicality of TAM meanings should be founded on psycholinguistic
evidence of speakers’ perception. That is, “perception of typicality differences is, in
the first place, an empirical fact of people’s judgments” (Rosch 1978:36). (The
primitive concepts in this tentative theory of prototypicality would be the TAM
meanings themselves, rather than the categorization of TAM systems: see the
Glossary for clarification.) Since I am unaware of such psycholinguistic evidence at
this moment, I will just present some introspective reflections on the topic, always
confronting the hypotheses with the Basque data.
The concept of prototype can also be applied to lexical-semantic differences.
We can state that dynamic predicates are more prototypical verbal predicates than
stative predicates are. In this case, prototypicality correlates with type frequency (i.e.
dynamic predicates are more type-frequent than stative ones). This fact has
implications for verbal design, since the competition for generalization of use (i.e. a
‘winner-take-alT marking tendency) is based on type frequency. Within each lexical-
semantic class too, we could talk about specific prototypical TAM meanings. Thus,
the most prototypical TAM meaning for stative predicates seems to be stative present
(i.e. continuous non-progressive present), whereas the most prototypical TAM
meaning for dynamic predicates might be past perfective. These speculative remarks,
though, are empirical questions which need further confirmation.
There are at least two different issues related to the grammatical marking of
prototypical vs. non-prototypical TAM meanings. One question that arises is the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
277
putative relationship across languages between prototypicality of TAM meanings
and grammatical marking (i.e. a commonly grammaticalized distinction) vs. non
marking (i.e. a seldom grammaticalized distinction). (See, for instance, Kemmer
1992). A different question is whether there is any correspondence between TAM
prototypicality and formal (un)markedness (cf. Greenberg 1966). I will discuss these
issues separately and confront them with the Basque data.
2.5.1. Prototypicality (vs. non-prototypicality) & grammatical marking (vs. non
marking or non-obligatoriness).
The hypothesis I present in this section may seem practically obvious, (if non
autonomy of grammar is accepted). Namely, that the most prototypical or cognitively
salient of TAM meanings will be most often grammaticalized across languages.
However, there are at least two evident practical problems with this
hypothesis. First, as mentioned above, it appears mandatory that we provide a
definition of prototype independent from cross-linguistic data. This independent
definition does no seem straightforward at the present moment. Second, cognitive
salience (and therefore prototypicality in the way it is intuitively interpreted here)
may change with areal, cultural, and temporal conventions. Thus, a cross-linguistic
and panchronic (“universal”) concept of prototypicality of TAM meanings appears
as virtually non-existent. (See Carstairs-McCarthy’s 1992:177-179 discussion on
Bybee’s 1985 ‘relevance.’)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
278
Yet again, we may concede that the top most prototypical of TAM meanings
may have something of universal across cultures, languages and times. If this were
correct, we could at least try to test the above hypothesis against the data of TAM
meanings such as past perfective, which intuitively appears as one of the most
salient. (Past perfective is also among the meanings which are most commonly
grammaticalized in the languages of the world: see Bybee & Dahl 1989; Bybee et al.
1994:15,300.)
An interesting example in the Basque data is provided by the comparison of
past perfective and evidential past, on the one hand, and general past and past
perfective, on the other. Among these TAM meanings, it appears that past perfective
is the most prototypical in the domain of past. (We could talk about past perfective
being the “basic-level prototype” in terms of Roschean categorization. General past
would then belong to a super-ordinate level, and evidential past to a subordinate
level.) Also, past perfective is the most frequent TAM meaning attested in the
database, in both type-frequency and token-frequency.
Now, Old Basque had a special form to express evidential past (the Present
Perfect-Perfective), though this form was only facultative, not obligatory, from the
very first attested texts. The forms conveying unmarked past perfective could also be
used in evidential contexts. Eventually, the Past Perfect-Perfective, which became
the main form to express past perfective, ousted the Present Perfect-Perfective from
evidential contexts, so that the evidential meaning was no longer grammaticalized in
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
279
Basque, not even optionally. In other words, what seems to be a very prototypical
TAM meaning (past perfective) was taken as central in the categorization of past
tense in Basque, while a non-prototypical meaning (evidential) was simply not
distinguished and left unmarked. (See section 4.2.4. below for possible areal factors
in this evolutionary ‘unconscious decision.’)
On the other hand, there was too a special form in Old Basque expressing
general past (namely, the Synthetic Past). The Synthetic Past could be used as both
perfective and imperfective past, while we could state that the specific forms for
perfective past were only optionally used at first. In this case, however, it was the
inclusive form (the Synthetic Past) which was practically lost, so that the meaning of
general past was no longer grammaticalized (as so) in Basque. The more specific
form (finally the Past Perfect-Perfective) was the winner in this competition, and
thus the sphere of past tense was organized around past perfective meaning.
The above data constitute an illustration of an evolutionary trail towards
probably the most common and prototypical situation across languages in the domain
of past tense. That is, the data show a specific linkage between cognition and
grammars (cf. Clark & Malt 1984:198-199), i.e. how cognitive salience is actually
implemented. Thus, the Basque data provide a small piece of evidence in favor of the
hypothesis that grammars reflect cognition to an important extent. Hence, we would
have further evidence for the general hypothesis that grammars are
conventionalizations of language use, which in this case implies that the design of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
280
TAM systems is significantly structured around the most prototypical TAM
meanings and not around the less prototypical. As Kemmer (1992:148) puts it,
prototypes “tend, across languages, to be associated with a particular
morphosyntactic form.” Or, in Greenberg’s (1957:89) famous words, “the relative
rarity of one of a number of alternatives on a world-wise basis is the resultant of two
factors, one of origin, the other of survival.”
2.5.2. Prototypicality and formal (un)markedness.
The topic of prototypicality and (un)markedness is even more complicated than what
I have discussed on prototypes so far. This issue should be studied at least from the
points of view of: 1) the general evolution of language and its complexity, 2)
stability of forms vs. renewal and the creation of new grammatical layers, 3) the
possibility of inhibiting and/or dropping morphological material and its conditions
(i.e. recoverability, constructional meaning, contextual interpretation).
It seems the most common case across languages is that where the most
prototypical TAM distinctions tend to end up being formally unmarked when
compared with the least prototypical. The result appears as a sort of economy of
marking. From the point of view of the theory of grammaticalization, an important
topic is to establish how this marking is actually implemented. One of the reasons
why TAM prototypes tend to correlate with unmarkedness is with all probability the
stability of prototype situations (Kemmer 1992:150). The relative stability of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
281
prototypes with respect to non-prototypes implies in the long ran that non-
prototypical meanings usually end up being formally secondary in relation to the
marking of the prototype. This leads to a situation where the prototypes are
unmarked and the non-prototypes marked.
From a synchronic perspective, what we see is a kind of “distinguishing
function” (Hopper & Thompson 1980:291) of grammatical markers, apparently with
an “economic motivation” (cf. Haiman 1985): i.e. “don’t bother to mark semantic
properties that are inherent in the semantics of an event” (Kemmer 1992:161). In
general, this seems to be the most ordinary scenario of formal marking. Therefore,
most commonly, prototypes will correlate with unmarkedness (cf. the Basque
examples below). (Yet, it might seem that prototypicality could also correlate with
markedness if/when grammatical markers had an “indexing function” (Hopper &
Thompson 1980:291) and an “expressive motivation” (Kemmer 1992:161).)
In Basque, most stative verbs are expressed by Stative Locutions of the type
bizi izan ‘to live’, nahi *edun ‘to want’, behar *edun ‘to need’ (see Chapter 3,
section 6.5). The prototypical meanings in stative predicates seem to be stative
present (e.g. I live, he wants, you need) and stative (imperfective) past. Past
perfective seems to be a non-prototypical meaning of stative predicates. The formal
marking o f Basque follows the predictions just mentioned above. Namely, the most
prototypical meanings are unmarked and the less prototypical marked: e.g. bizi naiz
‘I live’, bizi nintzen ‘I used to live’ vs. bizi izan nintzen ‘I lived (perfective).’
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
282
On the other hand, the marking of the normal verbs of Basque, the majority
of which are dynamic, follows almost an opposite pattern. In this case, it is past
perfective and perfect that are expressed by the most unmarked forms: the Past and
Present Perfect-Perfective, respectively; e.g.joan nintzen ‘I went (perfective)’, joan
naiz ‘I have gone.’ Recall that the most prototypical situation of dynamic verbs
seems to be past perfective. (For a detailed account of the differences in the
morphology of the Basque verbal forms because of TAM distinctions, see Chapter 2;
also Euskaltzaindia 1987: 404.)
2.6. Prototypicality of TAM meanings & the one-form-one-meaning
correspondence.
In section 2.5.1,1 proposed that the most prototypical TAM meanings tend to be
central in the organization (i.e. in the grammatical categorization) of TAM systems.
Prototypical meanings tend to be grammaticalized, i.e. assigned a specific TAM
form, while non-prototypical meanings tend to be expressed secondarily by (various)
constructions of those identified with prototypical meanings (Kemmer 1992:147-
148). Another way of looking at this topic would be to hypothesize that the most
prototypical TAM meanings (e.g. past perfective) will be among the meanings
showing the lowest degree of competition among forms, while there will be more
competition of forms to convey non-prototypical meanings. In other words, I predict
that, for instance, past perfective will be expressed by a small number of TAM
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
283
constructions; in the best case, perfective will be expressed only by one form (i.e.
one-form for one-meaning).
The Basque data appear to be in favor of this hypothesis. The data presented
in section 2.3 point to a higher one-form-one-meaning tendency for perfective than
for other meanings (although, as mentioned above, perfective meaning can also be
expressed by a historical present, not included in that counting).
2.7. Gradual and abrupt changes.
It has been common in the literature to relate, on the one hand, formal symbolic
discrete approaches with abrupt changes, and, on the other hand, more fluid
synchronic analyses (e.g. those based on the theory of grammaticalization) with
gradual changes. However, while symbolic discrete analyses do seem to always
imply abrupt changes, more fluid analyses do not necessarily entail gradual changes
in all cases. In other words, symbolic analyses are very constraining, and practically
rule out the possibility of synchronic variation and optionality (unless we adopt ad-
hoc theoretical assumptions such as analyzing variation as “coexisting grammars”, as
I mentioned above). Thus, the diachronic counterpart of a synchronic symbolic
model implies that all changes have to be abrupt and catastrophic (Lightfoot
1991:157ff; 1999:77ff, 148, and related work).
Conversely, there does not seem to be any necessary implication of
gradualness in the assumptions on grammaticalization. The theory of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
284
grammaticalization has room for both more gradual and more abrupt changes. Also,
a given change can be more gradual in one respect and more abrupt in another.
Changes entailing TAM categorial reanalysis seem to imply some abruptness. Yet,
changes involving lexical diffusion may be mostly gradual (hence the complexity of
lexical-semantic synchronic studies, as we have seen in this work). (See McMahon
1994, Harris & Campbell 1995, Campbell 1998 for discussion.)
Another possible factor in the gradual / abrupt controversy is the
prototypicality (basicness, salience) of the TAM meaning implicated in the change. It
seems that changes involving prototypical TAM meanings tend in general to be more
abrupt than changes involving less prototypical meanings. As a matter of fact, if the
above synchronic hypothesis that the most prototypical TAM meanings tend to have
a lower degree of competition of forms than the non-prototypical meanings is correct
(section 2.6), then this seems to imply a diachronic counterpart making correlate
prototypical meanings with more abrupt changes and non-prototypical meanings
with gradual changes.
In other words, as I will discuss in section 4, it seems we ought to propose
that, since prototypical meanings are central to the organization of TAM systems,
speakers tend to assign one form to them. That is, speakers tend to identify that form
with the prototypical meaning it expresses. It would seem as if speakers were more
‘conscious’ about prototypical meanings. Hence, prototypical meanings would imply
less competition of forms, synchronically, and more abruptness, diachronically. A
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
285
related point often mentioned in the literature is the presumed existence of
“clustering points” in the semantic dines of grammaticalization (Hopper & Traugott
1993:105-106).
Coming back to the past perfective of Basque, the change involving
perfective meaning that historical Basque underwent appears remarkably as a very
abrupt one. While 16lh century texts use typically the Periphrastic Aorist to convey
perfective meaning (even though the Past Perfect-Perfective is sometimes used,
starting from the very first texts), there is no trace of the Periphrastic Aorist
employed as a perfective in most of the 17th century texts. We may say that the way
perfective meaning was expressed in Basque was completely changed almost in a
time span of around 50 years. (See section 2.10 below for another putative factor in
this change: namely, a reanalysis towards a correlation between formal organization
and semantic categorization.)
2.8. Beachheads.
This work provides evidence that the semantic changes which a TAM construction
experiences start in specific contexts and are only generalized to the majority of
environments in later stages. I have been referring to the contexts where the semantic
changes start as ‘beachheads.’ Beachheads are neither mechanisms nor causes of
change. They are innovative environments where the mechanisms of semantic
change find appropriate land for growing. Regarding TAM systems, the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
286
environments that can function as beachheads can be either very specific semantic
meanings or syntactic contexts such as main clauses, specific subordinate clauses,
affirmative / negative clauses, etc.
When a form extends its meaning to a beachhead, it usually creates in the
beginning a situation of overlapping and optionality, i.e. a competition between an
old and a new form. The situation of optionality between the old and the young TAM
variants may stabilize in the beachhead contexts for some time. Also, a different
TAM marking between beachheads and more conservative contexts may stabilize for
a period of time. Since beachheads tend to recur cross-linguistically, the contrast
between beachheads and more conservative contexts may show up synchronically in
different grammatical behavior in these contexts across languages.
The above facts concerning the existence of beachheads seem to contradict
Kroch’s (1994:180) hypothesis that “change proceeds via competition between
grammatically incompatible options which substitute for one another in usage.”
Unfortunately, as in much work on language change elaborated within the generative
framework, Kroch’s main conclusions are often drawn by emphasizing to the
extreme observations which are, if mildly stated, relatively uncontroversial.
However, it is important to notice that the existence of recurrent patterns regarding
beachheads cannot be accounted for just by a mere substitution of forms. These
patterns demand theoretical solutions which have to contemplate the possibility of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
287
differences in the synchronic processing of different contexts, e.g. innovative vs.
conservative contexts.
2.9. Compositionality & semantic interpretation.
In this work I have not focused on the synchronic analysis of the formal
compositionality of the constructions at issue. It would seem that, for a semantic
study of TAM forms, analyzing the exact compositionality of the constructions (e.g.
whether they are synchronically biclausal or monoclausal) is not necessarily relevant.
In fact, in this work, I have dealt with both putative biclausal and putative
monoclausal constructions, and yet their semantic study has been satisfactorily
carried out in the same way.
If we were to base the meaning of a construction only on its compositionality,
then its semantic interpretation would be always understood as just the sum of the
interpretations of the expressions it is built from (cf. Keenan 2001). However, as I
touched upon in section 2.5.2 and will discuss in section 4, it appears we need to
leave room for other modes of semantic interpretation, such as “constructional
meaning” (cf. Haspelmath 1997), single-unit representation of high-frequency
complex constructions (cf. Bybee 1995), or new semantic associations occurring
during the process of grammaticalization. In this case, the idea of compositionality
would not seem so basic for a semantic study.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
288
The semantic interpretation of a TAM construction might be chiefly based on
compositionality in the first stages of the grammaticalization process. That is, when
TAM constructions are young, and its meaning is rather literal or ‘iconic.’ Yet, in
subsequent stages of the process, compositionality need not be the main factor in the
semantic interpretation of the constructions. (See section 4 for discussion.)
Nevertheless, from the kind of data collected in this work, we can hardly find
any direct evidence in favor or against compositionality as the main or the only
factor in the semantic interpretation of TAM forms. In any event, the semantic study
of the TAM constructions at hand seems satisfactory without focusing on their
compositionality.
2.10. Formal organization & semantic categorization.
Related to the interpretation of a given TAM construction is the issue of the
interpretation of semantic contrasts, i.e. of TAM categorization. In the latter case,
though, the data in this work do provide evidence against any approach which would
maintain that the interpretation of semantic contrasts of TAM in a given language is
solely based on formal compositionality, i.e. only based on the distributional
oppositions that TAM morphemes make up. In other words, the data go against the
common formalist assumption that equates formal organization with semantic
categorization. (See Bybee 1986 for discussion on this issue.)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
289
On the other hand, as in the case of the interpretation of a given TAM
construction (i.e. parallel to the tendency towards one-form-one-meaning), there is a
tendency towards a correspondence between formal organization and semantic
categorization. This tendency, though, probably has even more exceptions than the
one-form-one-meaning correspondence. These tendencies show up statistically when
looking at large databases of languages. However, they can hardly hold for any given
language randomly chosen. (This is a reflection of the historical ‘indirect’ adaptive
way grammars are created in language use as a compromise of competing
motivations.)
Thus, formalists’ analyses have equated the compositional organization of the
Basque TAM system with a purported universal categorization of TAM (cf.
Euskaltzaindia 1987, and most of the formal work on the topic). As just argued,
assuming that formal categorization exactly corresponds with semantic
categorization is hardly tenable for any language randomly chosen. Concretely, in
this case, proposing that the Modem Basque Future belongs in a super-category of
“Aspect” just on the basis of distributional parallelism with other aspectual
constructions (Euskaltzaindia 1987:404) is clearly incorrect. This is an over
application of the correlation between formal organization and semantic
categorization. (For details, see Chapter 2: section 2.1 and footnote 4.)
Conversely, there seems to be in the Basque data evidence pointing out the
alleged tendency towards making formal organization correspond with semantic
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
290
contrasts. Namely, the historical process that Basque underwent whereby the
constructions with dynamic auxiliaries ( [*edin ‘to become’, *ezan ‘to achieve’, egin
‘to do’) were confined to non-indicative contexts could be understood as being partly
motivated by a tendency towards having only stative auxiliaries (izan ‘to be’, *edun
‘to have’) in the indicative. Obviously, this hypothesis is hard to prove, but it does
seem plausible to assume that formal organization was one factor in this historical
process. Another factor may have been the fact that the contrast indicative / non
indicative seems to be a very relevant and general semantic distinction. In other
words, it appears sound to understand that there was a reanalysis towards correlating
type of auxiliary (stative vs. dynamic) with mood (indicative vs. non-indicative,
respectively). (Cf. Aldai 1994. See also section 4, for more discussion on these
issues.)
A partly similar example in the data may be the tendency towards making the
conservative Synthetic Present forms of Basque correspond with stative present
meaning (cf. some seemingly analogical late forms, such as donhetsa ‘he loves’,
perhaps dazagut ‘I know’). While the conservation of these forms may be only in
part (if at all) due to semantic factors, there seems to be some tendency in Modem
Basque to make them correlate with stative present. (A parallel reanalysis may be the
correlation between the Synthetic Past and past imperfective meaning.)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
291
2.11. Conservation of declining forms.
The data in this work corroborate the observation, which has often been raised in the
literature, that “relics are particularly likely to be retained in the most common words
of a language” (Meillet 1964, cited in A. Harris 1985:26). (See the Hypothesis of
Conservation of high token-frequency forms, put forward in Chapter 1 and tested in
Chapter 4.)
This correlation appears as an indicator that both the mental representation
and, more indirectly, the design of morphology seem to be somewhat related to
probabilistic frequency-based processing. (See Bybee 1985, 1995 for a proposal of a
frequency-based model of morphological processing.) Parts of syntax too may
perhaps have a component of frequency-based processing. (See Haspelmath 1997,
Bybee to appear, and section 4 below for discussion on these issues.)
In contrast, the formalist literature does not often address the issue of
conservativeness. Conservative structures are usually included within common
(generative, descriptive, stipulating) rules of synchronic grammar, with no specific
treatment differentiating conservative forms from more productive ones. When there
is a reference to historical conservation, it is just seen as a random irregularity and a
theoretically irrelevant characteristic (Lighfoot 1999:12-17). Thus, Keenan (2001)
accounts for conservative forms just by means of a Principle of Inertia. However, the
clear correlation between conservative forms and frequency suggests that these
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
292
formal approaches are missing a fundamental generalization of synchronic
processing.
3. HYPOTHESES ON GRAMMATICALIZATION.
In this section, I summarize the conclusions on the hypotheses on grammaticalization
I have arrived at throughout this work. Much of chapters 1,3-4 and 5-6 was devoted
to these hypotheses.
3.1. Variables in the process(es) of grammaticalization.
I proposed in Chapter 1 four variables that would be central in the process of
grammaticalization of lexical items. These are:
1. Formal reduction: (formal unidirectionality)
2. Semantic changes: (semantic unidirectionality, clines of grammaticalization,
mostly towards abstraction and generalization)
3. Generality of use: (the lexical-semantic dimension, mostly towards a
‘winner-take-alF situation)
4. The renewal of the system
As I will discuss in section 4 below, it seems it may be interesting to group these
variables into two general trends of change which make up the process of
grammaticalization. In other words, it m aybe worth distinguishing two main
processes within what I have been calling lexical grammaticalization.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
293
1) Process of Generalization:
1 a) Generalization of use
lb) Meaning generalization
lc) Fonnal reduction
2) Process of Renewal
In section 4.2.2,1 will contemplate the possibility of two different sets of general
motivations of change behind these two groups of processes (i.e. generalization vs.
renewal).
3.2. Scenarios of competition.
The framework presented in this work is importantly based on the concept of
‘competition’, which is valid for both synchrony and diachrony. I have distinguished
four kinds of scenarios o f competition between TAM forms. Each of these scenarios
gives rise either to different hypotheses or to different ways of testing hypotheses.
1) Competition among same-source layers
2) Competition among different-source constructions (layers)
3) Competition for generality of use (lexical-semantic classes)
4) Conservation of declining forms
In this work I have mainly focused on the competition among same-source layers.
Within the scenario of competition among different-source constructions, it is more
difficult to find generalizations and to test them. This is the reason why it has not
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
294
been as important as the former in this work. The competition for generality of use is
a hard-to-observe process, taking place among young constructions, which 1 have
only touched upon in several places throughout the work. Finally, the conservation
of declining forms is not part of grammaticalization properly. Nonetheless, it has
been generalized and tested in the Hypothesis of Conservativeness of high token-
frequency forms. The general implications of this latter question will be discussed in
section 4.2.2 below.
3.3. Grammaticalization hypotheses.
Focusing on the scenario of competition among same-source layers (though not
necessarily limited to it), there arise several generalizations which need to be
captured. The most important of these generalizations are the following hypotheses
that I have presented and tested against Basque data.
- Hypothesis of Source Determination
- Hypothesis of Layering-Displacement (unidirectionality)
Hypothesis of Form-Meaning Covariation
These hypotheses are enounced in section 3 of Chapter 1, to which the reader is
referred for discussion and details. Notice once again that the hypotheses are not
stated in terms of strict prediction (we will come back to this question in section
4.2.4 below). First, they are understood in statistical terms rather than obligatorily,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
295
i.e. as strong tendencies instead of universal laws. (This is, in any event, an empirical
issue which can be tested against data of other languages.)
Second, the hypotheses are always conditioned to the possibility of actual
change vs. non-change or inertia. In other words, the unidirectionality of the clines of
grammaticalization and the displacement of layers can only be observed when there
is change in the language. Yet, inertia (cf. Keenan 2001) is a very common situation
of TAM systems, and thus, if there is no change, the above hypotheses cannot be
tested. Put it differently, the clines of grammaticalization need no necessarily
‘move.’ Only when there is ‘movement’ of layers can we test the hypothesis of
Layering-Displacement. A different empirical question is whether the most common
situation is for TAM layers to remain unchanged and stable, or whether change is,
even though no necessary, very common. We can also think of the possibility of
different behaviors in this matter across languages and/or ages.
3.4. Clines of grammaticalization.
This work has only examined two semantic domains of TAM. Namely: the domain
of present (non-past), and the domain of past perfective (past non-imperfective).
Within these domains, we have seen several semantic clines of grammaticalization.
Yet, we can consider four clines as the most important in these spheres o f TAM
meaning. These four clines, on which this work has focused the most, are given in
(l)-(4) below.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
296
3.4.1. Domain o f non-past.
(1) present progressive -> present iterative present frequentative present
habitual gnomic, generic
(2) subordinate present habitual -> subordinate generic -> subordinate less-
dynamic future (if-clauses, when-clauses) (until-clauses) -> subordinate
dynamic future (complement clauses to ordering & wanting verbs, purpose
clauses)
3.4.2. Domain o f past non-imperfective.
(3) present resultative -> present perfect hodiernal perfective perfective
(simple past)
(4) past resultative -> pluperfect -> remote perfective -> perfective -> (simple
past)
3.5. Conclusions: the grammaticalization of TAM in Basque.
Below, I present the specific conclusions in this work, regarding the hypotheses on
grammaticalization as applied to the TAM system of historical Basque.
3.5.1. Main constructions studied in this work.
TABLE 7.1: Main constructions studied in this work.
LOOSE
PERIPHRASES
OF THE
INDICATIVE
RIGID
PERIPHRASES
OF THE
INDICATIVE
“SUBJUNCTIVE”
PERIPHRASES
(TELIC-
DYNAMIC)
SYNTHETIC
FORMS
DOMAIN OF
NON-PAST
Present
Progressive:
egiten ari zara
‘you are doing’
Periphrastic
Present:
egiten duzu
‘you do’
Present Subjunctive:
egin -dezazu-
(‘so that you do’)
Synthetic
Present:
-dagizu-
DOMAIN OF
PAST NON-
IMPERFECTIVE
Present
Resultative:
egina duzu
Present Perfect-
Perfective:
egin duzu
‘you have done’
Past
Resultative:
egina zenuen
‘you had done’
Past Perfect-
Perfective:
egin zenuen
‘you did’
Periphrastic Aorist
(Old Basque):
egin zenezan
‘you did’
Synthetic
Past:
zenegien
‘you did’
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
297
3.5.2. Competition between same-source layers in historical Basque.
The competition among same-source layers is the one I have focused the most in this
work. In historical Basque, this competition takes place between the loose
periphrases of the indicative and the rigid periphrases of the indicative. That is, we
have three pairs of same-source constructions (layers) in competition. These are the
following:
1. Present Progressive vs. Periphrastic Present
2. Present Resultative vs. Present Perfect-Perfective
3. Past Resultative vs. Past Perfect-Perfective
As mentioned above, these situations of competition provide us with the possibility
of testing the hypotheses of Layering-Displacement and Form-Meaning Covariation.
Unfortunately, the data on the loose periphrases of the indicative is very scanty and
present very little change in the historical period of Basque. Thus, in some of the
cases the hypothesis of Layering-Displacement cannot be confidently tested against
the Basque data. Yet, the Form-Meaning Covariation hypothesis can be tested based
on what I called ‘synchronic data of change’ (see the Glossary).
In any event, the Basque data fit very well with the grammaticalization model
proposed, and do not falsify any of the hypotheses, in the cases we can test them. On
the other hand, to better test the hypothesis of Layering-Displacement we should
recur to data of other languages with longer attestation than Basque.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
298
3.5.3. Competition between different-source layers in historical Basque.
Within the data of historical Basque, we can identify a case of competition among
different-source constructions. This occurs between the rigid periphrases of the
indicative and the “subjunctive” periphrases. Thus, we have two pairs of different-
source constructions in competition.
1. Periphrastic Present vs. (Periphrastic) Present Subjunctive
2. Past Perfect-Perfective vs. Old Basque Periphrastic Aorist
These scenarios of competition are not so easy to understand as the ones
among same-source layers. In any case, I proposed in Chapter 1 that the competition
among different-source constructions would tend statistically towards a one-winner
situation, similar to the competition among same-source layers. That is, once the
inertia of a stable situation has been broken, the system would evolve towards a new
situation where one form would be the main means to express each prototypical
meaning, (though practically always with exceptions such as the ones I discuss in the
next sub-section).
As a matter of fact, there might be some evidence to hypothesize that the
changes occurring in a situation of competition between different-source
constructions may be more abrupt than those taking place for same-source layers.
The evidence lies on the fact that different-source constructions are often formally
more different to each other. This is at least the case of the Basque “subjunctive”
periphrases I am dealing with, which have dynamic auxiliaries that make them
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
299
formally very different from the constructions with stative auxiliaries. As I
mentioned in section 2.10, in this case there seems to be a special factor conditioning
the abruptness of the change. Namely: a reanalysis towards a correlation between
formal organization (type of auxiliary) and semantic categorization (mood: indicative
vs. non-indicative). (See the discussion on diachronic vs. synchronic principles in
section 4.2.3 below.)
3.5.4. Conservation o f declining forms.
The case I have identified and studied in this work as a conservation of otherwise
declining forms is that of the Basque Synthetic Present. In Chapter 4 ,1 tested the
hypothesis of Conservativeness of high token-frequency forms against the data of the
Synthetic Present. As mentioned there, the correlation between high toke-frequency
items and conservation is highly significant, (regardless of other possible factors
playing a role in this conservation, such as a reanalysis of the Synthetic Present as a
specifically stative form). The correlation between token-frequency and conservation
seems to have important theoretical implications for the processing of grammatical
material, as I will discuss in section 4 below.
3.5.5. Competition for generality o f use.
The competition for generality of use is more difficult to observe in the data than the
above-presented situations. This is so because this last scenario of competition
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
300
appears mainly among very young TAM constructions, which are not well attested in
the database. We saw in Chapters 2 and 4 that present progressive, for instance, is
expressed by a number of formally similar but partly different constructions. The
choice of one among these constructions may seem to correlate to some extent with
lexical-semantic classes of predicates. Yet it appears that the ari Present Progressive
is resulting winner in this competition, at least in Modem Standard Basque.
Unfortunately, the data on this topic is very scanty. Nevertheless, the generalization
to capture seems to be similar to other cases of competition: the evolution progresses
statistically towards a one-winner situation. As a matter of fact, in this case we could
talk about a ‘winner-take-all’ competition, although even here there are exceptions
which resist regularization (see below for Basque examples).
In the competition for generality of use, the competing constructions cut
across lexical-semantic classes o f predicates, i.e. along the lexical-semantic
dimension. For this reason, this competition, which tends towards a one-winner
construction applicable to all classes of predicates, is based on type frequency. That
is, the winner tends to be the constmction originally applicable to (i.e.
grammaticalized for) the class of predicates with the highest type-frequency (we may
call it the ‘best winner’).
On the other hand, if there is a construction grammaticalized for a low type-
frequency class of predicates, the prediction is that this constmction will hardly
attain high applicability. This is the case of the Stative Present Verbal Locutions of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
301
Basque (see section 6.5 of Chapter 3), which were grammaticalized only for stative
predicates. Note that stative predicates have low type-frequency comparing with
dynamic predicates. Thus, the consensus opinion among Basque linguists is not to
consider the Stative Present Verbal Locutions as TAM grammatical constructions in
the system of Basque. They are usually analyzed in synchrony as a lexical
phenomenon. (This is a synchronic analysis which in principle I accept; but see
section 10 of Chapter 4 for more details.)
4. TOWARDS GENERAL MOTIVATIONS OR PRINCIPLES.
I conclude with a general discussion on some of the ultimate motivations or
principles that seem to play a role in the design of TAM systems. Many of these
factors are not yet totally understood. Therefore, the present discussion can only be
tentative at this point.
4.1. “Functional” motivations.
4.1.1. On the concept o f “ functional” motivation.
As has often been pointed out in the functional literature, when we talk about
motivations, we are not necessarily referring to functional needs, even though there
might be times when new needs appear in the system. (The idea o f functional needs
as the main motor of change has frequently been called “naive functionalism.”)
Thus, rather than needs we should talk about different motivations in competition. In
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
302
addition, we know that changes are not consciously sought by speakers (“teleology
of purpose”). Yet, speakers are in principle sensitive to the motivations I will discuss
below, even if unconsciously. Thus, it is actually the interplay, i.e. the competition,
among these forces that drives the unconscious ‘decisions’ of speakers; i.e. what has
been called the “hidden hand.” (For discussion on these issue see Bybee 1985: ch.9;
McMahon 1994: ch. 12; Bybee et al.1994: ch.8; Kirby 1997, 1999; Haspelmath
1999.)
Then again, some of these factors are not strictly functional, and many of
them are as unpredictable as fashions can be. Some of the motivations I present
below seem to be just reflections of the way the human mind processes TAM
systems, which need not be necessarily ‘functional’, in the sense of ‘optimal’
(specially if we compare them with discrete symbolic artificial languages). Another
main reason behind linguistic change is the search for expressiveness, which in
principle is neither functional nor predictable. In conclusion, it seems better to talk
about psycholinguistic and sociolinguistic motivations, non-autonomous to other
mental capacities, rather than to talk about functional pressures.
4.1.2. Types o f motivations.
One can think of at least the following types o f motivations or principles playing a
role in the design of TAM systems (see Clark & Malt 1984:192-195 for another
classification of possible “psychological constraints”):
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
303
1) expressiveness, fashions, and specific conventionalizations
2) cognition and conceptual categorization of the real world
3) functional motivations towards optimizing communication and learning
4) (other) constraints and/or reflections of the capacity and nature of the human
processor
In truth, it is hard to tell whether the motivations we are looking for belong to just
one of the types of principles just mentioned. It seems it is often a cluster o f things
that appear behind a given motivation.
The distinction that is possibly the hardest to draw is the one I present
between points 3 and 4 above. That is, the one distinguishing functional motivations
vs. constraints of the human mind. As a matter of fact, it is not easy to tell whether
there is actually any difference between these two ideas. The reason why I prefer to
keep them apart is because the concept of ‘functional’ applied to grammatical
markers can be understood as responding to a one-form-one-meaning (isomorphic)
discrete symbolic (or iconic) communicative language. (This is the way I will be
using it.) Thus, it has been sometimes assumed (mostly in the formal literature) that
for natural language to be ‘functional’ it ought to be close to the isomorphic perfect
and communicatively optimal language. As Givon (1995:7) puts it, “Chomsky began
in his writing from the early 1960s to license a blatantly functionalist idea - the
isomorphism between deep syntactic structure and propositional meaning.” Since it
was later proved that this was not the case, then, for the generativists, “language
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
304
design as such appears to be in many respects “dysfunctional”, yielding properties
that are not well adapted to the functions language is called upon to perform”:
Chomsky (1995:162).
However, it is not at all obvious that the human mind works always in a
discrete symbolic (or iconic) mode. This is one reason for making a distinction
between functional motivations, i.e. mostly those towards a one-form-one-meaning
isomorphic system, and (other) reflections of the human processor, which do not lead
to a discrete symbolic system.
4.1.3. Some possible motivations.
I will tentatively propose (with no intention of being exhaustive) the following
possible motivations, which may play a role in the design of TAM systems. I will
also try to assign them to one of the four types just mentioned.
1) specific conventionalizations:
- fashions, areal features
- expressiveness
- specific sources of grammaticalization
- cognitive relevance specific to cultural or areal conventions
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
305
2) cognition and conceptual categorization o f the real world:
- prototypical meanings, basicness, salience
- accessibility
- semantic and pragmatic associations, metaphors, contextual inferences
3) junctional motivations in communication and learning:
- one-form-one-meaning
- formal organization mirroring semantic categorization
- analogical extension, generalization
- economy of marking
- analogical leveling, formal regularization
- optimality in recognition, easy to learn rules, periphrases
- semantic differentiation
- adjacency, chunking, placing close together what belongs together mentally
- new functional needs (created by expressiveness, fashion, evolutionary
complexity?, etc)
4) other constraints and/or reflections o f the capacity and nature o f the human
processor:
- sensitivity to frequency / repetition / probabilistic patterns
- accessibility, connectionism
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
306
- tendency towards creating semantic associations and reanalysis
- routinization, automation
- formal reduction, phonetic erosion
- non-compositional interpretation, constructional meaning, context-based
interpretation
The possible motivations in the list above are not always mutually exclusive.
Some of them just reflect different sides or different levels of the same process. It
might be the case that some among them are the ultimate motivations, while the
others are just by-products of the former. However, at our present understanding, it is
difficult to distinguish between causal factors and superficial phenomena. For
instance, is economy always a causal factor behind linguistic change? Or is it
sometimes a by-product of other processes? It would also seem that some among the
putative motivations above may cluster together as a sort of ‘conspiracy’, where
apparently some of them trigger others in a sort of pull chain.
I discuss below some of these issues in more detail. These are, though,
questions that need further research.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
307
4.2. Discussion.
In this concluding section, I discuss several issues that arise from the list of possible
psycholinguistic and sociolinguistic motivations given above. I will just focus on a
number of topics that I consider important, with no intention of being exhaustive.
4.2.1. Prototypicality and types o f processing: (non)-discreteness and (non)-
compositionality.
The literature on grammaticalization and language change has often called attention
to the existence of evidence pointing out that grammars are not totally discrete. Now,
this rather diachronic observation has to have a synchronic counterpart which allows
for types of processing different from the one based on a discrete symbolic-iconic
compositional interpretation. We may thus propose that the synchronic processing of
TAM constructions and contrasts has to have a dual nature. In other words, we need
to postulate “both categoriality and non-discreteness”; that is, “flux at the margins
and solidity at the core of categories” (Givon 1995:12). However, even though this
statement seems evident, the actual way in which the discrete / non-discrete duality
is handled and implemented by speakers is far from being straightforward.
First, discreteness holds especially in the “core of categories”, i.e. regarding
prototypes. Once the one-form-one-meaning tendency is implemented in the
prototype situations, it seems to entail a kind of processing which might probably be
more “automated” (Givon 1995:13). Yet, the fact that prototypes are associated with
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3 0 8
one particular form seems to imply a more focal consciousness in the first place. Put
it differently: on the one hand, prototypical meanings seem to be more ‘conscious’,
i.e. more salient, accessible, central, relevant, etc.; hence they tend to be associated
with one particular form. On the other hand, prototypical meanings appear also more
‘automated’, because they are more often shared by speakers in the act of
communication, and they are quite predictable. In conclusion, the attended /
automated factor regarding prototypes (if relevant or if correctly captured here)
appears as a kind of paradox which is difficult to understand. (This issue could be
remotely related to a sort of speaker-oriented / hearer-oriented dichotomy. Yet, in
this case we should perhaps talk about a grammar-creation / grammar-sharing
dichotomy. In any event, the role of the hearer appears relevant in the latter.)
Second, types of synchronic processing other than the discrete symbolic-
iconic compositional model seem mandatory to explain phenomena such as the
graduality of categories, the “inhibition” of redundant morphological marking (cf.
Haspelmath 1999:236-239), or the frequency effects (see below). Several proposals
have been made in the literature: constructional meaning, analogy and similarity,
context-based interpretation, frequency-sensitive types o f processing, single-unit
autonomous representation, etc. However, here again it is not straightforward to
understand whether these kinds of processing may have any correlation with
prototypicality. It may seem that non-prototypical meanings are the most readily
subject to non-discrete and context-based processing, because of their gradualness
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
309
and because of their contextual dependency relative to the predictability of
prototypes. On the other hand, some prototypes may be processed by non-
compositional means, probably based on their high frequency (see discussion
below).
4.2.2. The role o f frequency-increase and routinization.
In the last decades, a number of scholars have called attention to the effects of
frequency and repetition in language use and grammars (see, for instance, Bybee
1985 and related work, Bybee & Hopper eds. 2001, etc.). Massive evidence has been
collected supporting the idea that frequency plays an important role in grammar and
performance, as it does in other human and non-human abilities (cf. Kelly & Martin
1994). In the present work, I have put forward evidence that the conservation of
specific items of old declining constructions (concretely, the Synthetic Present of
Basque) is very importantly conditioned by token-frequency, i.e. absolute frequency
of each particular item (see section 9.1 of Chapter 4, and section 2.11 above). I have
also discussed the role of (type)-frequency in the perception of prototypical
situations. In the discussion below, I present some reflections on how frequency may
be a factor in different stages of the process of grammaticalization of TAM
meanings.
The role of probabilistic patterns in grammars has been studied taking as
point of departure different phenomena where the effect of frequency is apparent.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
310
These phenomena are not readily connected to one another, and most of them belong
to the sphere of morphology. Yet, we could tentatively propose that it is throughout
the whole process of grammaticalization that speakers are sensitive to frequency
patterns. Pushing this idea a bit further, we might say that the process of
grammaticalization itself is a process of frequency increase. In other words, if the
speakers’ processing is sensitive to probabilistic patterns to the extent of treating
high-frequency material very differently from low-frequency material, then an abrupt
increase in the frequency of a given lexical or grammatical item (or even in that of a
given syntactic construction) may trigger an enormous change in performance and
eventually in grammar. Note that for the above to be correct we need to assume a
view of grammar as “emergent” (Hopper 1987 and related work), and a view of
processing notably based on the notion o f ‘accessibility.’
However, just by claiming that the process of grammaticalization is to a very
important extent a process of frequency increase, we are not explaining, or even less
predicting, everything about it. What we need to explain now is why frequency is
increased in the very first place. I will turn to this question below. Before addressing
this issue, it may be worth examining in more detail the effects of frequency
throughout the process of grammaticalization of TAM markers.
Roughly speaking, grammaticalization of TAM forms is the process whereby
different lexical-analytic expressions become TAM grammatical markers. As Harris
& Campbell (1995:54) put it, this process starts when these expressions “catch on”,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
311
and “only when the expression is used in additional contexts and is generalized may
we speak of a grammatical change having taken place.” Clearly, when this catching-
on occurs, the frequency of the expression is raised considerably. The moment the
frequency of the expression is highly increased, we hypothesize that the processing
of that expression starts being somewhat different from that of other expressions
with lower frequency. Eventually, when the expression becomes very frequent and
familiar among speakers, we can talk about a grammatical ‘rule’ rather than a lexical
expression. This process is usually accompanied by reanalysis of the lexical
expression, and eventually by phonological reduction. Whether frequency-increase is
the ultimate reason that triggers other changes is perhaps not totally relevant, since,
as I said, we should first explain why frequency is raised. In any event, there seems
to be a cluster of changes in the process. I will use the term ‘routinization’ to refer to
this process (cf. Labov 1990, Haiman 1992, Bybee et al. 1994:299, Givon 1995:10,
Bybee & Hopper 2001:13). What is important in the present discussion is that
frequency-increase is a necessary part in the process of routinization.
Although more research would be needed on this topic, it would seem that
routinization might be either responsible for or alternatively caused by (see below)
the other changes which cluster together in the process of grammaticalization. One
can tentatively propose that, once the frequency of a given expression is increased,
speakers become habituated to that expression. This may trigger more automated and
arbitrary processing, and this in turn may bring about phonological reduction and
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
312
semantic associations. It is difficult to understand exactly how or why automation-
routinization-habituation ought to be related with the creation of metaphors and other
semantic associations, but there might be a relation among them. That phonological
reduction may be related with automation does appear easier to grasp. In any case,
for phonological reduction to occur, a type of constructional-meaning processing
seems to also be necessary. Semantic associations, in their turn, create fuzziness and
polysemy, which seems to be processed by means of similarity. Polysemy, finally,
can eventually give place to semantic splits, where the original association between
the meanings of a given form is totally blurred. For semantic splits to occur, we
might need to postulate some kind of context-based interpretation.
Alternatively, one may think that, instead of frequency and routinization, the
trigger of the above cluster of changes may be the human tendency towards
expansion. Thus, semantic associations would be just caused by this expansionist
force. Associations in turn would trigger a “generalizing tendency” of grammars
(Kemmer 1992:161). The generalizing tendency brings about a kind of many-
meanings to one-form marking type. The many-to-one marking, in turn, may be a
new factor in the increase of the frequency of grammatical markers.
What appears clear is that the process of lexical grammaticalization is
semantic-wise mainly a process of generalization, and phonologically a process of
erosion, which causes opacity. Now, it seems that both of these processes tend to be
compensated in the long run by the appearance of new grammatical material. The
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
313
renewal of the system, thus, is basically an opposite process of “differentiation”
(Kemmer 1992:162) in both senses semantic and phonological. Therefore, we can
see here an equilibrium of opposite forces generating the design of grammar.
Whether there are optimal points in this equilibrium or whether there is a limit to the
generalization tendency beyond which the differentiating tendency has to start by
means of renewal are topics that m aybe worth exploring. (Cf. Lightfoot’s 1979
Transparency Principle, which raises issues similar to the above discussed.)
Below, I present two schemes that try to summarize the cluster of motivations
and changes in the processes of renewal and generalization, which have been
introduced in the tentative discussion above. As just mentioned, these two processes
are quite different, often opposite to each other in their motivations.
a) The process o f renewal and new grammar creation.
- Differentiating motivations
- Fashions: “catching on”
- Expressiveness
- Accessibility to basic metaphors
- Formal optimality in the recognition of constructions / rules
- Compositional interpretation, literal-lexical meaning, periphrases, rule or
combination processing
- Discreteness: necessary and sufficient conditions, no polysemy
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
314
- Iconic representation = non-arbitrary (McMahon 1994:171-172)
- Lower frequency, less routinization
b) The process ofgeneralization and formal reduction.
- Generalizing motivations (expansion, extension)
- Competition for generality of use (“winner-take-all”): rise of type-frequency
- Frequency increase
- Routinization, automation
- Reanalysis: lexical -> grammatical -> more grammatical
- Chunking, semantic dependency among constituents
- Formal reduction
- Semantic change, associations, metaphors, abstraction -> polysemy
- Non-compositional semantic interpretation, constructional meaning, similarity-
based processing, degrees, non-discreteness, fuzziness
- On the other hand, the most prototypical meanings tend to have one marker o f their
own: i.e. less competition of forms (one-meaning-^ one-form).
- Symbolic representation = arbitrary (McMahon 1994:171-172)
4.2.2.1- The role o f frequency & the morphology/syntax distinction.
By assuming that frequency-increase of phonological material is central in the
process of grammaticalization, as I have done above, we can perhaps find an
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
315
explanation for some of the differences between syntax and morphology, and
between regular and irregular morphology. In syntax, if there are repetitions, these
are repetitions of patterns, but it is hardly the same phonological material that is
repeated. In regular morphology, on the other hand, the expressions that are repeated
(e.g. verbal auxiliaries in the case of TAM forms) have the same or very similar
phonological material from one utterance to another. Thus, this material which is so
frequently repeated is what may become a function word (in our case, a TAM
marker), and may eventually suffer reanalysis, fusion and phonetic erosion. As I said
above, these markers may be considered as highly applicable rules in the speaker’s
mind. (We may speak of high type-frequency in this case.)
The case of irregular morphology might be the most complex to understand,
even though it is the easiest to observe. As has been pointed out in the literature,
irregular morphology is mainly an (indirect, i.e. conservative) by-product of the
processing of very high token-frequency items. (Recall also the Basque evidence for
a correlation between conservation of old forms and token-frequency of specific
items.) In this case, it is not only the marker that is highly repeated. It is the whole
expression, including the lexical item which the marker accompanies, that is subject
to frequent repetition. Thus, these items seem to be stored with a relatively
autonomous single-unit mental representation. (For discussion on these issues, see
Bybee 1985, 1995 and related work.)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
316
4.2.3. Are the motivations or principles synchronic or diachronic?
A long-standing debate in the literature is whether the ultimate generalizations to
capture regarding morphological structure (i.e. those parts of grammars typically
subject to grammaticalization of lexical items) should be stated in synchronic or in
diachronic terms. While the answer to this question is open to discussion, it may
seem that both possibilities are in fact equally legitimate, for they respond to
different subjects of study.
On the one hand, if morphological structure is created in the historical
process of grammaticalization, then it seems natural to raise diachronic
generalizations to capture the main characteristics of this process, (even though it
ultimately leads to the synchronic states we see in specific languages). This may
seem especially important in order to capture cross-linguistic generalizations
appearing between languages at different stages of the grammaticalization process.
On the other hand, since the ultimate motivations triggering the process of
grammaticalization are psycholinguistic in nature, it may appear an equally
legitimate object of study to try to capture (as I did above) the ‘panchronic’
motivations in the speakers’ mind which drive the historical process and
subsequently the synchronic design o f morphological material. (Which has not been
proved to be legitimate is to try to capture “universal” formal generalizations of
synchronic structure, with no relation to the grammar-external psycholinguistic
motivations that trigger them.)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
317
4.2.3.1- Semantic substance vs. contrast.
If the synchronic-generalizations vs. diachronic-generalizations debate may not be
particularly significant, there is still a related distinction among psycholinguistic
motivations which may turn out to be more meaningful. This regards Bybee’s
(1988a) distinction of “semantic substance” vs. “contrast”, both of which seem to
make up the meaning of grammatical forms, specially in the case of TAM systems.
This distinction draws a line between, on the one hand, psycholinguistic motivations
only affecting individual grammatical forms (i.e. internal to their substance) and, on
the other, those affecting relations and oppositions among competing forms.
Much of the process of grammaticalization of a given form can be explained
without reference to its putative competitors: e.g. the Hypothesis of Source
Determination, the Form-Meaning Correlation, the semantic clines, etc. However,
there is also the possibility that the meaning (and even the form) of a given
construction may be influenced by that of other constructions competing with it.
The structuralist tradition has mostly focused on contrast between forms. The
literature on grammaticalization, on the other hand, has paid prime attention to the
internal substance of individual forms. It is in fact the two types of psycholinguistic
factors that are relevant for the complete understanding of the diachronic evolution
of TAM systems, and thus of their synchronic structure. It would seem that the
characteristics related to the internal evolution of individual forms are more
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
318
amenable to generalizations and predictions (of the kind I have proposed in this
work) than the evolution due to reorganizations of formal oppositions.
The latter appears as (even) more language-internal, random and
unpredictable. The only generalization we may be able to capture concerning the
changes due to contrast of forms seems to be the tendency towards a correspondence
between formal organization and semantic categorization. (For an example, see in
section 2.10 the Old to Modem Basque indicative vs. non-indicative reanalysis.)
Flowever, this tendency appears as very loose and practically impossible to detail any
further (unlike the generalizations due to the internal substance o f forms, which can
be captured in the hypotheses on grammaticalization: see section 3 above.)
4.2.4. Grammaticalization, conventionalization and prediction.
I conclude with a discussion on grammaticalization as conventionalization. The term
‘conventionalization’, as employed in this work, entails that the process of grammar
creation may include fashions, borrowing, aborted or reverted changes, secondary or
specialized evolutions (cf. the evidential meaning of perfects), random changes
and/or areal conventions, perhaps not totally understood yet. In addition, we should
take into account changes due to reorganizations in the system of contrasts (see
sections 2.10 and 4.2.3 above). Moreover, even without taking into account these
less-predictable changes, it is difficult to tell, at least at our present understanding,
whether there is a hierarchy in the relative importance of the motivations I have
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
319
discussed throughout this chapter. Rather, what it seems is that, generally speaking,
languages rank these motivations based on a relatively random convention.
Therefore, even if we isolated all the competing variables in force in a given
phenomenon, it seems we could hardly come up with an algebraic expression which
would give us the universally optimal result.
Regarding areal features, I could not leave without mentioning here two
points of the Basque data. The first point to draw attention to is that double
compound perfects have been reported for German (specially Swiss German), French
and Occitan, besides Basque (specially in the Eastern dialects). One could not help
but noticing that these languages are located in a contiguous area, which points to the
possibility of a shared areal feature. (This appears as a fascinating topic for further
research.) The second point I will mention is the loss of a distinct evidential
construction in the evolution from Old to Modem Basque. As is well known,
evidentials are attested, among other languages, “in Turkish, Bulgarian, Georgian
and Estonian. To some extent this is an areal feature” (Comrie 1976:108). It may
extend over the areas of the Otoman Empire, Balto-Finnic, and Uralic (Comrie
1976:110, Dahl 1985:152). There does not seem to be any areal relationship between
the Old Basque evidential and the area just described. Yet it appears interesting to
note that Modem Basque might have lost the evidential distinction in accordance to
the general situation in western European languages.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
320
For the reasons just mentioned, no strict prediction in advance has been
proposed in this work, (as has not been proposed, to my knowledge, in the major
literature on grammaticalization on which this work relies). Thus, I totally agree with
Lightfoot (1999:39) when he claims: “Although we might not have been able to
predict the change in advance, we do have some kind of understanding of what
happened, some explanation in a weaker sense.” (I cannot agree with him, though,
when he says (1999:210): “The typologists remained faithful to the methods of the
nineteenth century. They retained the same kind of determinism.”)
The grammaticalization framework maintains that there are obvious patterns
of change which are strikingly unidirectional across languages. Denying this fact is
being blind to an overwhelming amount of data from a good number of different
languages of various areas and families. However, this does not imply that we can
predict whether the change will occur at all, or whether other possible competitors
may appear due to borrowing or just due to chance. The important contribution o f the
grammaticalization framework is precisely what the generativists tend to neglect the
most: namely, proposing explanations for why those recurring patterns of change
exist and why they should be so unidirectional (even with the exceptions). This is
what I have tried to contribute to in this work.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
321
GLOSSARY
NOTE: Within the definitions in this glossary, I write some terms in ALL CAPITALS
IN SMALLER FONT to cross-refer with other entries in the glossary. I specially use
ALL CAPITALS IN SMALLER FONT for the terms that are most closely related with the
entry I am explaining, so that the reader can compare all entries that refer to a given
topic. (For the use of Initial Capitals, see the entry INITIAL CAPITAL.)
ABSOLUTE-RELATIVE T.A.M. MEANINGS: (Taken from Comrie 1985:64ff.)
Alternatively ANAPHORIC TAM meanings. ABSOLUTE-RELATIVE TAM MEANINGS
combine “absolute time location of a reference point with relative time location of a
situation” with respect to that reference point. That is, “a reference point is situated
at, before, or after the present moment, and in addition a situation is located at,
before, or after that reference point” (Comrie 1985:65). See PERFECTS. See also
ABSOLUTE TAM MEANINGS.
ABSOLUTE T.A.M MEANINGS: (Taken from Comrie 1985:36ff.) Alternatively
DEICTIC TAM meanings. ABSOLUTE TAM MEANINGS “take the present moment as
their deictic center” (Comrie 1985:36). That is, in TAM meanings “with absolute
time reference, the situation is located at, before, or after the present moment”
(Comrie 1985:64). See PERFECTIVE. See also ABSOLUTE-RELATIVE TAM
MEANINGS.
ABSTRACTION: Alternatively BLEACHING, (GENERALIZATION). Many SEMANTIC
CHANGES occurring in the process of grammaticalization of lexical items evolve
towards ABSTRACTION or loss of specific lexical content. That is, the semantic
content of the forms undergoing grammaticalization becomes more ABSTRACT, more
relational, and more grammatical. This is especially true of the first stages in the
process of grammaticalization. In the present work, no special attention is paid to
distinguishing between the concepts of ABSTRACTION, BLEACHING, etc. (For
discussion, see Heine et al. 1991:108-113, Hopper & Traugott 1993:96-100.) The
prime attention of this work is rather on the ASSOCIATIONS which cause the
DIRECTIONALITY of specific TAM SEMANTIC CHANGES. See Chapter 1 section 2.3.
See also GENERALIZATION, LEVEL OF ABSTRACTION.
ACCOMPLISHMENT PREDICATES: See CLASSES OF PREDICATES, TELIC
PREDICATES. Vendler (1967:107) provides the following examples of
ACCOMPLISHMENTS: paint a picture, make a chair, build a house, write a novel,
read a novel, deliver a sermon, give a class, attend a class, play a game o f chess,
grow up, recover from illness, and get ready fo r something.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
322
ACHIEVEMENT AUXILIARIES of Basque: See DYNAMIC AUXILIARIES.
ACHIEVEMENT PREDICATES: See CLASSES OF PREDICATES, TELIC
PREDICATES. Vendler (1967:107) provides the following examples of
ACHIEVEMENTS: recognize, realize, spot, identify, lose, find, reach the summit, win
the race, cross the border, start, stop, resume, be born, and die.
ACTIVE-INACTIVE CASE-MARKING: Alternatively, AGENTIVE-
NONAGENTIVE CASE-MARKING. ACTIVE-INACTIVE CASE-MARKING is a term often
employed to dub the various case-marking systems which align, on the one hand,
active (or agentive) intransitive subjects with transitive subjects, and on the other
hand, inactive (or non-agentive) intransitive subjects with transitive objects. It
appears that this split-intransitive case-marking type is becoming general in the
western dialects of Basque. See Chapter 2 note 2. See ERGATIVE-ABSOLUTIVE CASE-
MARKING, INTRANSITIVE-ERGATIVE PREDICATES.
ACTIVITY PREDICATES: See CLASSES OF PREDICATES. Vendler (1967:107)
provides the following examples of ACTIVITIES: run, walk, swim, push something,
and pull something. Many ACTIVITY predicates that are intransitive in the
neighboring languages are constructed in Basque as verbal LOCUTIONS hosted by the
LIGHT-VERB egin ‘to do’: e.g. korrika egin, laster egin ‘to run’, igeri egin ‘to swim’,
hegal egin ‘to fly’, dantza egin ‘to dance’, lan egin ‘to work’, etc. All of them have
ERGATIVE subjects. See INTRANSITIVE-ERGATIVE PREDICATES.
AGE: See FORMAL AGE, SEMANTIC AGE. It seems convenient to distinguish three
general ages of TAM FORMS: young, mature and old FORMS. A fourth age could be
represented by very old forms in decline in the system. Due to the HYPOTHESIS OF
FORM-MEANING COVARIATION, formal age and semantic age tend to correlate, so
that each of the general ages (i.e. young, mature and old) tend to correspond not only
with a degree of BOUNDNESS but also with a SEMANTIC SUBTYPE.
AGENTIVE-N ON AGENTIVE CASE-MARKING: See ACTIVE-INACTIVE CASE-
MARKING.
ANALYTIC FORM. Alternatively PERIPHRASTIC FORM, PERIPHRASIS, (VERBAL
COMPLEX). I use these terms with no difference among them. ANALYTIC or
PERIPHRASTIC verbal forms are those FINITE forms which are made up of more than
one word. In some grammars of Basque, a difference is made between ANALYTIC
FORMS and PERIPHRASES. I use the terms RIGID PERIPHRASES and LOOSE
PERIPHRASES, respectively, to capture this distinction in the degree of BOUNDNESS.
See also AUXILIARY, QUASI-AUXILIARY.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
323
ANAPHORIC T.A.M. MEANINGS: (Taken from Squartini 1999:56.) See
ABSOLUTE-RELATIVE TAM MEANINGS.
AORIST, (OLD BASQUE PERIPHRASTIC AORIST): (Examples: etor zedin
‘s/he came’, ekar zezan ‘he brought’, ekar zegian ‘he brought’ Biscayan dialect.) The
Old Basque Periphrastic Aorist was formed by the combination of the RADICAL and
a Past form of the DYNAMIC AUXILIARIES. The Aorist conveyed PERFECTIVE
meaning in Old Basque. Due to several changes in the system, a form identical to the
Aorist expresses Past SUBJUNCTIVE in Modem Basque.
AREAS OF COMPETITION: Alternatively SEMANTIC SUBTYPES. This work
examines the COMPETITION of forms in the DOMAINS of PRESENT and PAST
PERFECTIVE in historical Basque. Within these wide domains, more specific AREAS
OF COMPETITION among forms are also distinguished. AREAS OF COMPETITION can
refer to both semantic and syntactic CONTEXTS. The most important TAM areas of
competition in the Basque system, which are object of study in this work, are the
following: 1) (imperfective) present (for both dynamic and stative verbs), 2)
subordinate future time-reference, 3) present resultative / present perfect, 4)
pluperfect, and 5) perfective.
ASSOCIATIONS, (SEMANTIC ASSOCIATIONS): (Taken from Haiman 1985:
39ff.) I use ASSOCIATIONS as a cover term to refer to the mental processes which are
responsible for the semantic and pragmatic changes in grammaticalization.
ASSOCIATIONS can be seen as little transgressions in the use of TAM constructions.
These can be due to metaphors, contextual inferences / implicatures, time-reference
‘relaxations’, generalizations, etc. ASSOCIATIONS are the ultimate reason behind the
existence of CLINES OF GRAMMATICALIZATION and their (UNI)DIRECTIONALITY.
This is so because the psycholinguistic ASSOCIATIONS that may occur with similar
SOURCES OF GRAMMATICALIZATION have been proved to be strikingly similar
across languages. See CLINE OF GRAMMATICALIZATION, SEMANTIC CHANGE,
SEMANTIC UNIDIRECTIONALITY. See also PROTOTYPE.
ATTAINMENT OF A LIMIT: (Taken from Dahl 1985:76.) See TELIC SOURCES.
For the differences between the concept of ATTAINMENT OF A LIMIT and TELIC
SOURCES in PRESENT (i.e. non-past) and PAST, see Chapter 3 note 3.
AUTOMATION: See ROUTINIZATION.
AUXILIARY: The great majority of the Basque verbal system is ANALYTIC. The
main constructions in this system (dubbed RIGID PERIPHRASES in this work) consist
o f a NON-FINITE FORM of the lexical verb and a FINITE FORM of an AUXILIARY verb.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
324
There are two main types of auxiliaries in Basque: STATIVE AUXILIARIES and
DYNAMIC AUXILIARIES. See also QUASI-AUXILIARY, LOOSE PERIPHRASES.
BEACHHEAD: (Taken from Givon 1979.) The SEMANTIC CHANGES which a TAM
construction experiences start usually in specific CONTEXTS and are generalized to
the majority of environments only in later stages. BEACHHEADS are these innovative
contexts where the mechanisms of semantic change find appropriate land for
growing. Regarding TAM systems, the environments that can function as
beachheads can be either very specific TAM MEANINGS or syntactic CONTEXTS such
as main clauses, certain subordinate clauses, affirmative / negative clauses, etc. See
Chapter 7 section 2.8.
BLEACHING: See ABSTRACTION, SEMANTIC CHANGE.
BOUNDEDNESS: I use the term BOUNDEDNESS to refer to all types of SITUATIONS
that are confined by any kind of limits. Thus, I refer to inherent BOUNDEDNESS of
the lexical predicate, i.e. shortness, punctuality; to BOUNDEDNESS of the participants,
i.e. definiteness; or to temporal BOUNDEDNESS, i.e. that provided by the occurrence
of definite time-adverbials. BOUNDEDNESS (in all its variants) appears as one of the
most relevant features in the definition o f PERFECTIVE meaning. It may seem that
BOUNDEDNESS correlates in the default case with totality, i.e. the idea of ‘single
whole’ or lack of differentiated phases. On the other hand, the existence of non
identical phases (one of the DOMINANT FEATURES in the definition of imperfeetive
aspect, and especially in PROGRESSIVE meaning) seems to correlate in the default
case with non-BOUNDEDNESS, i.e. with lack of limits. (However, it ought to be
mentioned that BOUNDEDNESS and phases can coexist. In these cases, some
languages use a special form that may be termed imperfeetive perfective, as in the
Spanish Progressive Perfective estuvo escribiendo toda la noche ‘s/he stayed up
writing all night long.’) The concept of TELICITY, although related with limits, puts
the emphasis on the end-point, and thus it is not considered in this work as identical
to BOUNDEDNESS. Telicity is more related, in the case of the past tense, with
completion (i.e. a completed action). See PERFECTIVES OF THE SLAVIC TYPE. (Note
the term BOUNDNESS, which is used here in a very different sense.)
BOUNDERS: (Taken from Bybee & Dahl 1989:85-86.) See PERFECTIVES OF THE
SLAVIC TYPE.
BOUNDNESS, (DEGREE OF BOUNDNESS): Alternatively, FORMAL AGE,
DEGREE OF FORMAL GRAMMATICALIZATION. Also, DEGREE OF RIGIDITY /
MOBILITY, DEGREE OF FUSION. The DEGREE OF BOUNDNESS of a given construction
is an indicator of the formal interdependency between the components that make up
the construction. Based on the FORMAL UNIDIRECTIONALITY of the process of
grammaticalization, DEGREE OF BOUNDNESS correlates with FORMAL AGE. In the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
325
case of PERIPHRASES, I usually talk about DEGREE OF RIGIDITY / MOBILITY between
the different parts (words) of the periphrasis. In the case of SYNTHETIC FORMS, I use
the terms DEGREE OF BOUNDNESS or DEGREE OF FUSION. (Notice the term
BOUNDEDNESS, which I use in a very different context; namely, in relation with
TELICITY and PERFECTIVE meaning.)
CATEGORY: One focus of this work is on the COMPETITION of forms for similar
(often overlapping) TAM MEANINGS, within specific AREAS OF COMPETITION. Thus,
the prime attention is not on maximal CONTRASTS between CATEGORIES or super
categories such as tense, aspect or mood. Also, rather than focusing on delimiting
boundaries between TAM CATEGORIES, by way of necessary-and-sufficient
conditions, this work tries to provide the features that prototypically correlate with
and define individual TAM MEANINGS (cf. Rosch 1978:35-36, Comrie 1985:18-19).
Therefore, the term CATEGORY is seldom employed in this work. However, this does
not entail that maximal CONTRASTS do not play a role in the DESIGN of TAM
systems. See Chapter 7 section 4.2.3.1. See also PROTOTYPE, DISCRETENESS.
CHARACTERISTIC BEHAVIOR: (Taken from Fleischman 1982, cited in Bybee
1988b:373.) CHARACTERISTIC BEHAVIOR is a LABEL which might be interesting to
use, in order to capture a TAM MEANING that would comprise PRESENT HABITUAL,
GNOMIC and STATIVE PRESENT (cf. Bybee et al. 1994:151-152). In this work,
however, these three TAM MEANINGS are kept apart. As a matter of fact, in the
Basque case, STATIVE PRESENT is usually expressed by FORMS different from that
conveying PRESENT HABITUAL and (dynamic) GNOMIC. See Chapter 4 sections 5
and 10.
CLASSES OF PREDICATES, (CLASSES OF VERBS): Alternatively VERBAL
CLASSES. In this work, I consider two classifications along the LEXICAL-SEMANTIC
DIMENSION, i.e. regarding CLASSES OF PREDICATES:
1) Vendler’s classification into four classes: ACTIVITIES, ACHIEVEMENTS,
ACCOMPLISHMENTS and STATES or STATIVE PREDICATES. (See Vendler
1967 and Dowty 1979:54-71, for examples and criteria and syntactic tests for
classification.)
2) A reduced classification into two classes: STATIVE PREDICATES vs. NON-
STATIVE or DYNAMIC PREDICATES. (See Lakoff 1965, cited in Dowty
1979:55, and Comrie 1976:48-51 for definitions.)
CLINE OF GRAMMATICALIZATION. Alternatively PATH OF
GRAMMATICALIZATION. The CLINE or PATH of grammaticalization is the semantic
route through which constructions travel in the process of grammaticalization.
CLINES of grammaticalization are created by means of semantic and pragmatic
ASSOCIATIONS, i.e. metaphors, inferences, etc., which lead from one meaning to the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
326
next. See SEMANTIC AGE, SEMANTIC CHANGE, SEMANTIC UNIDIRECTIONALITY. See
Chapter 1 section 2.3.
COMPETITION: The term COMPETITION is used in this work to refer to the
relation of those TAM forms that, rather than making up maximal CONTRASTS or
OPPOSITIONS (e.g. present vs. past), “share or compete for overlapping territories” or
AREAS within one given DOMAIN of TAM (Bybee et al. 1994:21). I distinguish in
this work four types of scenarios of COMPETITION: 1) SAME-SOURCE LAYERING, 2)
DIFFERENT-SOURCE LAYERING, 3) the CONSERVATION of very old declining forms,
and 4) the competition for GENERALITY OF USE (most specially the competition for
the expression of STATIVE PREDICATES). See also AREAS OF COMPETITION,
DISPLACEMENT, LAYERING. See Chapter 1 section 1.1, section 4; Chapter 7 sections
2 and 3.
COMPLETIVE: (Taken from Bybee et al. 1994:54, 57-61.). COMPLETIVE meaning
indicates the idea of “to do something thoroughly and to completion” (Bybee et al.
1994:54). There may be some overlapping between this definition and that of
RESULTATIVES. Yet, a distinct label for COMPLETIVE seems to be justified on the
basis of the differences in meaning and in grammaticalization sources that
COMPLETIVES and RESULT ATIVES have. The lexical SOURCES of completives “are all
dynamic verbs or directionals; that is, all suggest action or movement” (1994:59). On
the other hand, it may seem that COMPLETIVES could be similar in meaning to
PERFECTIVES, since both can express completion, telicity and attainment of a limit.
More specifically, COMPLETIVES would appear comparable in meaning (though not
exactly in formal expression) to PERFECTIVES OF THE SLAVIC TYPE.
CONSERVATIVENESS of old declining forms: The CONSERVATION of an old
declining TAM construction correlates with the TOKEN-FREQUENCY of specific
items of that form. Therefore, in this case the variable of GENERALITY OF USE does
not necessarily correlate with lexical-semantic classes of predicates. One could thus
talk about the LEXICAL DIMENSION in this case, but not about the LEXICAL-
SEMANTIC DIMENSION. See Chapter 1 sections 2.4, 3.5 and 4.3, Chapter 4 section
9.1, Chapter 7 sections 2.11 and 4.2.2. See also HYPOTHESIS OF CONSERVATION OF
HIGH TOKEN-FREQUENCY FORMS.
CONSTRUCTION: See FORM.
CONTEXT: I use CONTEXT as a cover term to refer either 1) to a group of similar
TAM MEANINGS (alternatively SEMANTIC SUBTYPE) or, 2) more often, to finer
divisions within TAM MEANINGS (alternatively SENSE, READING, VALUE, USE,
FUNCTION). It is difficult to make clear distinctions among these terms. Tentatively, I
use READING, SENSE and VALUE when I want to convey a more semantic flavor. In
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
327
turn, FUNCTION, USE, VALUE and CONTEXT itself have a more discourse-pragmatic
nuance. Finally, CONTEXT is most often used while referring to a syntactic
environment. See MEANING, DOMAIN, BEACHHEAD.
CONTINUOUS NON-PROGRESSIVE PRESENT. Alternatively STATIVE
PRESENT. I follow Comrie (1976:25) to use the label CONTINUOUS NON
PROGRESSIVE PRESENT to refer to the TAM MEANING which is most characteristic
of STATIVE PREDICATES in the present tense, as in I have, you posses, he desires, she
wants, we like, you hate, they know, or we believe. (The term CONTINUOUS as used in
this work means to confer an idea of ‘constant’, ‘uninterrupted’. Note, however, that
some authors use CONTINUOUS in the sense of PROGRESSIVE. See Comrie 1976:12.)
CONTRAST: Alternatively, OPPOSITION. The term CONTRAST or OPPOSITION is
used, following the structuralist tradition, in the sense of maximal and/or clear-cut
(DISCRETE) semantic differentiation between CATEGORIES. Also, from the
structuralist perspective, forms “are assigned a value by the OPPOSITIONS that they
enter into” (Bybee 1988:247). On the other hand, the theory of grammaticalization
maintains that forms have also an inherent semantic SUBSTANCE o f their own. In
fact, the inherent content of TAM forms and the changes it experiences is one of the
prime concerns of this work. Thus, instead of CATEGORIES, borders, and
CONTRASTS, this work focuses on TAM MEANINGS, PROTOTYPES, and
COMPETITION, respectively. This is not to say that CONTRASTS do not play any role
in the DESIGN of TAM systems: cf. PRESENT HABITUAL VS. PRESENT PROGRESSIVE.
See also, for the case of Basque, the colloquial reshaping of Synthetic Past forms on
the basis of Synthetic Present forms (Chapter 2 section 4.2), or the historical
reanalysis of the dynamic-auxiliary forms as the non-indicative (“subjunctive”)
mood (Chapter 5 section 7.2, Chapter 6 section 3.3, Chapter 7 section 2.10). See
Chapter 7 section 4.2.3.1, for discussion on these issues.
CONVENTIONALIZATION: I use the term CONVENTIONALIZATION to refer to
the process whereby major patterns of language use or performance become
rigidified or obligatory in grammars, often implying the exclusion of other minor
patterns. Thus, CONVENTIONALIZATION is equivalent to GRAMMATICALIZATION in
its widest sense. The process of grammatical CONVENTIONALIZATION could be
paralleled with other non-linguistic trends of human culture, such as the creation of
rules out of tradition (cf. consuetudinary law). See discussion on Chapter 7.
DEGREE OF FORMAL GRAMMATICALIZATION: See BOUNDNESS.
DEGREE OF FUSION: See BOUNDNESS.
DEGREE OF RIGIDITY / MOBILITY: See BOUNDNESS.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
328
DEGREES OF REMOTENESS: See HODIERNAL PERFECTIVE, PRE-HODIERNAL
PERFECTIVE.
DEICTIC T.A.M. MEANINGS: (Taken from Squartini 1999:56.) See ABSOLUTE
TAM MEANINGS.
DESIGN: Alternatively STRUCTURE. I use the term DESIGN specially to refer to the
relationship between form and meaning in grammars, as in the DESIGN of TAM
systems. One of the most important assumptions in this work is that different
competing motivations operating in performance are responsible for the DESIGN of
TAM systems, since grammar is basically a CONVENTIONALIZATION of language
use. See Chapter 7 for discussion on these issues.
DIFFERENT-SOURCE LAYERING: See SAME-SOURCE LAYERING.
DIRECTIONALITY: See (UNI)DIRECTIONALITY.
DISCOURSE-PRAGMATIC USES of a TAM form: Alternatively NARRATIVE
USES. See SENTENTIAL USES. See also MEANING.
DISCOURSE SOURCES of grammaticalization: Alternatively NARRATIVE-
DISCOURSE SOURCES. In this work, I try to contribute to the understanding that both
SENTENTIAL and DISCOURSE SOURCES of grammaticalization are possible, even
though the former are more familiar and apparently more common. Thus, I present
and discuss some sources that seem to qualify as DISCOURSE SOURCES. These are
sources of NARRATIVE ANTERIORITY (see Chapter 5 section 5.1) and of NARRATIVE-
PERFECTIVE (see Chapter 5 section 5.2). See also Chapter 5 sections 5.3 and 5.4.
DISCRETE MODELS, DISCRETENESS: The assumption of DISCRETENESS, or
clear-cut divisions between categories, entails an ISOMORPHIC relation between form
and meaning. Conversely, NON-DISCRETE models leave room to polysemy of various
similar meanings for one form, and/or to overlapping of various forms for one
meaning. See Chapter 7 section 2. See also SYMBOLIC MODELS, CONTRAST,
CATEGORY, PROTOTYPE.
DISPLACEMENT: I use the term DISPLACEMENT to refer to the tendency whereby
a new TAM construction results winner in the COMPETITION for a given TAM
MEANING, thus displacing a previous construction from that meaning. The term
DISPLACEMENT is intended to leave room for a (temporary) overlapping of the forms
in competition to express the meaning at hand. Thus, the synchronic result is usually
a NON-DISCRETE system. Yet, one form is statistically preponderant. See Chapter 7
section 2. See LAYERING, RENEWAL, HYPOTHESIS OF LAYERING-DISPLACEMENT.
See also SUBSTITUTION.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
329
DOMAIN: Alternatively SPHERE. This work is primarily concerned with examining
in detail the COMPETITION of forms for similar (often overlapping) TAM MEANINGS.
This competition takes place within DOMAINS or SPHERES of TAM, which are
mostly defined by the CLINES OF GRAMMATICALIZATION that the forms at issue
follow. Therefore, this work does not focus on comparing maximal CONTRASTS, e.g.
present vs. past. Rather, each of these is taken as a different object of study and
examined ‘from the inside’. The two wider DOMAINS studied in this work are
PRESENT and (PAST) PERFECTIVE. I also call them NON-PAST and PASTNON-
IMPERFECTIVE, respectively. Within PRESENT and PERFECTIVE, the competition
among forms occurs for more specific AREAS OF COMPETITION or SEMANTIC
SUBTYPES. Hence, we may very roughly classify the following terms from more
inclusive to less inclusive, as follows: DOMAIN, SPHERE > AREA OF COMPETITION,
SEMANTIC SUBTYPE > TAM MEANING > SENSE, READING, VALUE / USE, FUNCTION.
See also CONTEXT, SUBSTANCE.
DOMINANT FEATURE: I use the term DOMINANT to refer to the FEATURE that
most significantly defines or singles out a given TAM MEANING and its PROTOTYPE.
In a sense, DOMINANT FEATURES could be taken as a rough equivalent in NON
DISCRETE MODELS to necessary-and-sufficient conditions in DISCRETE MODELS. It
should be understood, though, that NON-DISCRETE models are based on
PROTOTYPES, and these are in principle defined by means of several features, thus
leaving room for a core and a periphery, i.e. for not-always clear-cut boundaries.
DOUBLE-COMPOUND PERFECTS: Double-compound Perfects are constructed
in a parallel fashion to Perfects with the addition o f an extra participle of the
auxiliaries ‘be’ or ‘have.’ Double-compound Perfects are cross-linguistically rare
and their uses are not very well understood. Thus, I devote a lengthy discussion to
these constructions. Two of the most common functions of the Double-compound
Perfects seem to be NARRATIVE ANTERIORITY and EXPERIENTIAL PERFECT. See
Chapter 5 section 6. See also DOUBLE-COMPOUND PRESENT PERFECT of Basque, and
DOUBLE-COMPOUND PAST PERFECT of Basque.
DOUBLE-COMPOUND PAST PERFECT of Basque (DcPs): (Examples: etorri
izan ginen literally ‘we were been come’, egin izan genuen literally ‘we had been
done.’) See DOUBLE-COMPOUND PERFECTS. One of the most common functions of
the Double-compound Past Perfect seems to be NARRATIVE ANTERIORITY. For a
discussion on the meaning and functions of this form, see Chapter 5 section 6.3, and
Chapter 6 sections 5, 7, 9 and 10. For a description and discussion on its form, see
Chapter 2 section 5.3, Chapter 5 section 7.5, and Chapter 6 section 2.4.
DOUBLE-COMPOUND PRESENT PERFECT of Basque (DcPr): (Examples:
etorri izan gara literally ‘we are been come’, egin izan dugu literally ‘we have been
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
330
done.’) See DOUBLE-COMPOUND PERFECTS. Two of the most common functions of
the Double-compound Past Perfect seem to be EXPERIENTIAL PERFECT and
EVIDENTIAL. For a discussion on the meaning and functions of this form, see
Chapter 5 section 6.3, and Chapter 6 sections 4, 8 and 10. For a description and
discussion on its form, see Chapter 2 section 5.3, Chapter 5 section 7.5, and Chapter
6 section 2.4.
DYNAMIC AUXILIARIES of Basque. Alternatively, ACHIEVEMENT (TELIC)
AUXILIARIES. The DYNAMIC AUXILIARIES of Modem Standard Basque are *edin ‘to
become’, which functions as INTRANSITIVE-ABSOLUTIVE auxiliary, and *ezan (no
attested meaning as independent verb), which functions as the ERGATIVE (transitive)
counterpart. In the Western dialects, the intransitive auxiliary is still *edin, but the
transitive auxiliary is egin ‘to do, to achieve’. The dynamic auxiliaries combine only
with the Radical (see NONFINITE FORMS), and only have NON-INDICATIVE uses in
Modem Basque, unlike in Old Basque. See Chapter 2 section 7, Chapter 3 section 5
and section 6.4.2, Chapter 5 section 4 and section 7.2, Chapter 7 section 2.10. See
SUBJUNCTIVE, STATIVE AUXILIARIES, ERGATIVE-ABSOLUTIVE MARKING.
DYNAMIC FUTURE TIME-REFERENCE SUBORDINATE CONTEXTS: I
use this term to refer to the subordinate contexts which are most commonly grouped
under the label SUBJUNCTIVE, (but see also the subgroup I dub LESS-DYNAMIC
CONTEXTS). The most important contexts in this DYNAMIC subgroup are: 1)
complement clauses to verbs of ordering and wanting, and 2) purpose clauses. Both
LESS-DYNAMIC and DYNAMIC contexts are non-assertive and have future time-
reference. They differ in that the DYNAMIC contexts have a more TELIC sense. They
are also more clearly bounded, punctual, definite, and specific. The DYNAMIC
contexts are partly similar to imperatives: in both cases their dynamic and telic
nature is one of their most defining features. See Chapter 4 section 7.
DYNAMIC PREDICATES: Alternatively NON-STATTVE PREDICATES. See
CLASSES OF PREDICATES. Dynamic predicates include in principle all three classes of
ACTIVITIES, ACCOMPLISHMENTS and ACHIEVEMENTS. In a DYNAMIC situation, “the
situation will only continue if it is continually subject to a new input of energy”
(Comrie 1976:49). DYNAMIC predicates are the most characteristic or prototypical of
verbal predicates. See Chapter 4 section 10.4.
DYNAMIC SOURCES: See TELIC SOURCES.
ERGATIVE-ABSOLUTIVE CASE-MARKING: An issue I cannot address in this
work is whether the morphological marking of Basque is predominantly ERGATIVE-
ABSOLUTIVE or ACTIVE-INACTIVE (AGENTIVE-NONAGENTIVE). For the most part, I
use the traditional terminology, which considers Basque as ERGATIVE-ABSOLUTIVE.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
331
Therefore, the two main formal cases of Basque are called ABSOLUTIVE and
ERGATIVE. However, I do refer sporadically to the split of intransitive predicates.
See INTRANSITIVE-ERGATIVE PREDICATES and Chapter 2 note 2. Bear in mind that
there are significant cross-dialectal differences regarding this question.
EVIDENTIAL: EVIDENTIALS “indicate something about the source of the
information in the speaker’s assertion” (Bybee et al. 1994:95). In the study of the
evidential uses of the PRESENT PERFECT-PERFECTIVE in Old Basque, I include
within the label EVIDENTIAL ‘unwitnessed perfective’, ‘quotative’ or quotation
perfective, and ‘fictional narrative’. I also distinguish the first verb in a sequence of
fictional narratives from the subsequent verbs. See Chapter 5 section 2.3, and
Chapter 6 section 8.
EXPERIENTIAL PERFECT: “The experiential perfect indicates that a given
situation has held at least once during some time in the past leading up to the
present” (Comrie 1976:58). For example: Bill has been to America is an experiential
perfect, while Bill has gone to America is a perfect of result. The EXPERIENTIAL
PERFECT is the most differentiated of the uses of Perfects. As a matter of fact, I
propose in this work that the EXPERIENTIAL PERFECT may be related in diachrony
(and perhaps also in synchrony) with meanings such as ‘frequentative perfect’ and
‘imperfeetive perfect’. (These meanings would partly be out of the scope of the
DOMAIN of PAST NON-IMPERFECTIVE.) See Chapter 5 section 2.2, and Chapter 6
section 4.2.2. EXPERIENTIAL PERFECT is one of the most important meanings that the
D o u b l e -c o m p o u n d P r e s e n t P e r f e c t of Basque conveys.
EXPRESSION: See FORM.
EXTENSION: See GENERALIZATION.
FINITE FORM: Conjugated, inflected form. In the Basque case at least, FINITE
verbal forms are those with personal agreement. See FORM, SYNTHETIC FORM,
ANALYTIC FORM.
FORM, (T.A.M. FORM, VERBAL FORM, VERB FORM): Alternatively,
CONSTRUCTION, MARKER, EXPRESSION, FORMATION. I use all these terms,
especially FORM and CONSTRUCTION, almost interchangeably to refer generally to
the formal material which expresses grammatical meaning. (In the case of TAM
meaning, I use TAM FORM, TAM CONSTRUCTION, etc. and also VERBAL FORM.)
The only subtle difference among the above terms may appear when trying to
distinguish degrees of BOUNDNESS. In this case, MARKER has an implication of
higher degree of boundness than FORM, and the latter term, in turn, refers usually to
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
332
materia] more bound than that of a CONSTRUCTION. See FINITE FORM, NONFINITE
FORM. See also (VERBAL) LOCUTION.
FORMAL AGE: See BOUNDNESS. See also FORMAL UNIDIRECTIONALITY.
FORMAL (UNI)DIRECTIONALITY: The formal side in the process of
grammaticalization evolves overwhelmingly in the direction of formal reduction,
rather than in the opposite direction.
FORMATION: See FORM.
FREQUENTATIVE: Unlike ITERATIVE, the label FREQUENTATIVE is used in this
work to indicate that the action is repeated in multiple occasions, which include the
reference time. In practice, as Bybee et al (1994:160) acknowledge, “the matter of
distinguishing a single occasion from multiple occasions is, of course, not always
straightforward.” PRESENT FREQUENTATIVE is also temporary or contingent (recent),
e.g. lately, he is working on his book everyday, as opposed to PRESENT HABITUAL
which is more permanent and characteristic.
FRAME, (TEMPORAL FRAME): (Taken from Dahl 1985:146-147.) See PAST
TEMPORAL FRAME, HODIERNAL PERFECTIVE, PRE-HODIERNAL PERFECTIVE.
FUNCTION: See CONTEXT, MEANING, DOMAIN.
FUTURE TIME-REFERENCE SUBORDINATE CONTEXTS: See
SUBJUNCTIVE CONTEXTS.
GENERAL PAST: See SIMPLE PAST.
GENERALITY OF USE: (Taken from Bybee et al. 1994.) Alternatively, TYPE-
FREQUENCY, LEXICAL GENERALITY (Bybee 1985:16). GENERALITY OF USE indicates
the degree of lexical applicability that a given TAM construction has. Therefore, this
variable in the process of grammaticalization applies in the LEXICAL DIMENSION.
Since I refer to GENERALITY OF USE only when examining the process of
GENERALIZATION, and since in this case GENERALITY OF USE progresses through
lexical-semantic classes of predicates, then in this work the terms GENERALITY OF
USE and LEXICAL-SEMANTIC DIMENSION actually make reference to the same
phenomena. See Chapter 1 section 2.4, section 5, note 3; Chapter 4 section 10;
Chapter 7 section 3.5.5, section 2.5. See also CONSERVATIVENESS.
GENERALIZATION IN TIME-REFERENCE: The SEMANTIC CHANGE
experienced by a given TAM form will be regarded as a GENERALIZATION IN TIME-
REFERENCE when the time-reference conveyed by the form evolves from less to
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
333
more INCLUSIVE. For instance, the passage from recent past and hodiernal past to
perfective is, roughly speaking, a generalization in time-reference. (The general trend
in the cline of present perfect, though, is better defined as a loss of the relevance of
present time-reference: see Chapter 5 section 2.) Nevertheless, in the present work
the prime attention is on the ASSOCIATIONS responsible for the DIRECTIONALITY of
the SEMANTIC CHANGES, rather than on whether the changes are always
GENERALIZATIONS or not. See Chapter 1 section 2.3. See also GENERALIZATION,
(PROCESS OF GENERALIZATION), ABSTRACTION.
GENERALIZATION, (PROCESS OF GENERALIZATION): 1) The process of
ABSTRACTION or BLEACHING, which is characteristic in grammaticalization
(especially in its first stages) can also be considered a semantic GENERALIZATION or
weakening of semantic content. 2) Another relationship between ABSTRACTION and
GENERALIZATION comes from the fact that higher abstraction (i.e. lower semantic
content) triggers higher lexical applicability (i.e. higher GENERALITY OF USE). Thus,
the first stages in grammaticalization can be globally seen as a process of
GENERALIZATION. Sometimes I refer to this PROCESS OF GENERALIZATION (together
with the usually co-occurring process of FORMAL REDUCTION) as
GRAMMATICALIZATION PROPERLY. See Chapter 1 section 2.3, section 2.4; Chapter 7
section 3.1. See also GENERALIZATION IN TIME-REFERENCE, ABSTRACTION.
GENERIC: See GNOMIC.
GNOMIC: Alternatively GENERIC. GNOMIC meaning might be viewed as
CHARACTERISTIC BEHAVIOR in its highest degree: that which “has held, holds and
will hold for the class of entities named by the subject” (Bybee et al. 1994:319).
Apparently, there is a small difference between PRESENT HABITUAL and GNOMIC:
“the generic statement dogs pant to cool o ff differs from the so-called habitual one
my dog pants to cool o ff only by having a generic rather than specific subject”
(Bybee et al. 1994:152). However, this small difference implies an important
divergence in time reference: GENERICS are “timeless”; they can be applied to the
future as well as to the present. There is also dissimilarity between them in their
LEVEL OF ABSTRACTION: GENERICS apply to KINDS. Moreover, if compared with
PRESENT PROGRESSIVE, GENERIC shows a remarkable difference regarding internal
structure: even if everlasting, GENERICS are conceived as a single whole, with no
phases. In other words, GNOMIC is extended but not internally differentiated.
TABLE G .l: Prototypical features of GNOMIC or GENERIC meaning.
Timeless, no time reference, holds for all time
In effect at the moment of speech ___________________ '___________
Totality, single whole, no phases, unity, continuousness _________________
Permanent, everlasting, extended________________________ ______________
Indefinite
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
334
TABLE G.l - continued___________________________
Descriptive, characteristic; but implies some dynamicity
Abstract, induction, empirical prediction_____________
GRAMMATICALIZATION: (Some scholars use GRAMMATICIZATION.) I
distinguish the following uses of the term GRAMMATICALIZATION. (Cf. Traugott &
Heine 1991: 2-3.)
1) GRAMMATICALIZATION as CONVENTIONALIZATION. This is the
broadest use of the term. It encompasses all the possibilities in
considering grammar as a conventionalization of performance. For
instance, Hawkins (1994) examines the GRAMMATICALIZATION of
preferred performance word orders, and Comrie (1998) studies not only
the GRAMMATICALIZATION OF LEXICAL ITEMS but also that of pragmatic
and semantic structures.
2) GRAMMATICALIZATION OF LEXICAL ITEMS. Alternatively, LEXICAL
GRAMMATICALIZATION, GRAMMATICALIZATION WITH LEXICAL INPUT.
If not otherwise stated, when I use the term GRAMMATICALIZATION, I
refer to GRAMMATICALIZATION OF LEXICAL ITEMS. See Chapter 1 section
2.1. For definitions of GRAMMATICALIZATION OF LEXICAL ITEMS, see
Meillet 1912/1948: 132, Kurylowiz 1965/1975: 52. (Cf. Campbell 1998:
238.)
3) GRAMMATICALIZATION PROPERLY. This is the narrowest use I make of
the term GRAMMATICALIZATION. I use the expression
GRAMMATICALIZATION PROPERLY to refer to the core part of the process
of GRAMMATICALIZATION OF LEXICAL ITEMS; namely, the process of
GENERALIZATION and FORMAL REDUCTION. See Chapter 1 section 2.4,
Chapter 7 section 3.1.
HABITUAL: See PRESENT HABITUAL. (PRESENT HABITUAL is the only habitual
meaning studied in this work.)
HABITUATION: See ROUTINIZATION.
HODIERNAL PERFECTIVE: HODIERNAL PERFECTIVE meaning indicates “that a
certain kind of event has taken place within a certain temporal FRAME [that] must not
be ended before the point of speech” (Dahl 1984b: 114). “Thus, a hodiernal past
would refer to ‘the current unit of time’, variously interpreted as ‘today’, ‘this week’,
‘this year’, etc. [...] The day as a unit of time would then presumably represent the
default value” (Dahl 1985:136). HODIERNAL PERFECTIVE contrasts with PRE-
HODIERNAL PERFECTIVE. We can thus consider the former as a specific type of
RECENT PERFECTIVE, with the specification of referring to a CURRENT TEMPORAL
FRAME, and the latter as a type of REMOTE PERFECTIVE. It is common across
languages for PRESENT PERFECT and HODIERNAL PERFECTIVE to be non-distinct.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
335
However, I consider in this work that these meanings are diachronically related but
synchronically separate. HODIERNAL PERFECTIVE meaning, unlike PERFECT
meaning, can co-occur with definite time-adverbials referring to the HODIERNAL
TEMPORAL FRAME. The Basque TAM system has a construction that conveys the
two meanings at hand. I dub it the PRESENT PERFECT-PERFECTIVE, due to its perfect
and perfective dual nature.
HYPOTHESIS OF CONSERVATION OF HIGH TOKEN-FREQUENCY
FORMS: See Chapter 1 sections 3.5, 4.3, and 8; Chapter 4 section 9.1.
HYPOTHESIS OF FORM-MEANING COVARIATION: See Chapter 1 sections
3.4, 4.1.2, and 8.
HYPOTHESIS OF LAYERING-DISPLACEMENT: See Chapter 1 sections 3.3,
4.1.1, and 8.
HYPOTHESIS OF SOURCE DETERMINATION: See Chapter 1 section 3.1,
and section 8.
ICONIC relation between form and meaning: A given TAM construction will
have an ICONIC relation with the meaning it conveys if there is a non-arbitrary link
between them. (Cf. McMahon 1994:171-172.)
IMPERFECTIVE PARTICIPLE of Basque: Alternatively PRESENT PARTICIPLE,
GERUND. (Examples: esaten ‘saying’, ikusten ‘seeing’, joasatzen ‘passing.’) See
NONFINITE FORMS.
INCLUSIVENESS, (LEVEL OF INCLUSIVENESS): See GENERALIZATION IN
TIME-REFERENCE, ABSTRACTION, LEVEL OF ABSTRACTION.
INDICATIVE: See RIGID PERIPHRASES and LOOSE PERIPHRASES.
INDIVIDUAL-LEVEL PREDICATES: INDIVIDUAL-LEVEL PREDICATES are the
predicates that apply to INDIVIDUALS. For a distinction between STAGE-LEVEL
PREDICATES and INDIVIDUAL-LEVEL PREDICATES, see OBJECTS.
INDIVIDUALS: “INDIVIDUALS are entities that serve to ‘tie together’ a series of
STAGES to make them stages of the same individual” (G. Carlson 1979:54). In other
words, INDIVIDUALS are usually abstractions from the level of STAGES. There are
two LEVELS OF ABSTRACTION that are considered INDIVIDUALS: the level of
OBJECTS and the level of KINDS.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
336
INITIAL CAPITAL: I follow in this work Comrie’s (1976:10) policy “of using an
initial capital for the names of language-particular categories, [...] while not using
initial capitals for language-independent semantic distinctions.” (C£, for example,
‘perfective’ meaning vs. ‘Perfectives’ of the Slavic type, or ‘perfect’ meaning vs. the
‘Present Perfect-Perfective’ of Basque.)
INTRANSITIVE-ABSOLUTIVE PREDICATES of Basque: See INTRANSITIVE-
ERGATIVE PREDICATES.
INTRANSITIVE-ERGATIVE PREDICATES of Basque: When I refer (in
passing) to the split of intransitive predicates of Basque (especially relevant in the
western dialects), I use a slightly schizophrenic terminology. On the one hand, I
employ the traditional names for the two core formal cases of Basque; namely,
ABSOLUTIVE and ERGATIVE, which in principle presuppose an ERGATIVE-
ABSOLUTIVE CASE-MARKING. On the other hand, if one divides intransitive
predicates in two groups, one is proposing an ACTIVE-INACTIVE (AGENTIVE-
NONAGENTIVE) CASE-MARKING. Within this schizophrenic scenario, I call
INTRANSITIVE-ABSOLUTIVE PREDICATES the intransitive verbs whose subject is
marked with the ABSOLUTIVE case, i.e. as the object of transitive predicates. (These
are called ‘unaccusatives’ in formal grammars.) Conversely, I call INTRANSITIVE-
ERGATIVE PREDICATES the intransitive verbs whose subject is marked with the
ERGATIVE case, i.e. as the subject of transitive predicates. (These are called
‘unergatives’ in formal grammars; a term which is confusing for Basque.) See also
ERGATIVE-ABSOLUTIVE CASE-MARKING.
ISOMORPHISM between form and meaning: A given TAM construction will
have an ISOMORPHIC relation with the meaning it conveys if there is a one-to-one
correspondence between them. (Cf. Haiman 1985: 2Iff.) Isomorphic relations can
always be stated in terms of neeessary-and-sufficient conditions. See also
DISCRETENESS, CATEGORY, CONTRAST, PROTOTYPE.
ITERATIVE: The terms ITERATIVE, FREQUENTATIVE and HABITUAL are not
always used in the same sense in the literature. I adopt Bybee et al.’s (1994:159)
definition of ITERATIVE meaning. This states that “in the case o f iterative, the action
is repeated on the same occasion”, which is ongoing at the reference time. Therefore,
an example of PRESENT ITERATIVE would be the bell is tolling.
KINDS: Alternatively TYPES. KINDS constitute the high LEVEL OF ABSTRACTION
within INDIVIDUALS. Thus, KINDS are INDIVIDUALS, but these individuals are
(possible) kinds of things, instead of OBJECTS, which are more familiar individuals.
Therefore, “here, KINDS of things are looked upon not as being sets of OBJECTS, as is
commonly supposed, but rather as being INDIVIDUALS themselves. [...] Just as
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
337
‘John’ names an individual [i.e. an OBJECT], ‘cats’ names an individual as well [i.e. a
KIND]” (G. Carlson 1979:54).
LABELS: See TAM MEANING LABELS.
LAYERING, LAYERS: Many cases of COMPETITION among TAM constructions
occur in the way of LAYERING, i.e. “an array of grammaticized and grammaticizing
constructions of different AGES” (Bybee et al. 1994:21). Thus, “layering is the
synchronic result of successive grammaticalization of forms which contribute to the
same DOMAIN” (Hopper & Traugott 1993:124). “The term LAYERING ... allows ...
for multiple origins of a grammatical form” (1993:106). Therefore, it seems
important to distinguish between competition among layers with the SAME SOURCE
of grammaticalization and competition among forms with DIFFERENT SOURCES. See
Chapter 1 section 2.5. See RENEWAL, DISPLACEMENT, HYPOTHESIS OF LAYERING-
DISPLACEMENT.
LESS-DYNAMIC FUTURE TIME-REFERENCE SUBORDINATE
CONTEXTS: I use this term to refer to some subordinate contexts which are
sometimes included in the label SUBJUNCTIVE, (along with the LESS-DYNAMIC
CONTEXTS, which are always included in that label). The most important contexts in
this LESS-DYNAMIC subgroup are: open conditionals, future time-reference temporal
clauses, generic relative clauses, concessive clauses, and until-clmses. The time-
reference of the LESS-DYNAMIC contexts can be future or gnomic. Sometimes they
refer to indefinite, unbounded, repeated SITUATIONS (similar to HABITUALS or
GENERICS). Other times they denote a unique, specific, bounded, singular, punctual
situation, even if unreal or hypothetical. In the latter case, they are compatible with
definite time adverbials expressing future time-reference, as in open conditionals
such as i f you come tomorrow, you will see it, or temporal clauses such as when you
come to school tomorrow, you will see it. See Chapter 4 section 6.
LEVEL OF ABSTRACTION: “The term LEVEL OF ABSTRACTION within a
taxonomy refers to a particular level of INCLUSIVENESS” (Rosch 1978:30). See
GENERALIZATION IN TIME-REFERENCE, STAGE-LEVEL, INDIVIDUAL-LEVEL
PREDICATES.
LEXICAL DIMENSION: See GENERALITY OF USE, CONSERVATIVENESS. See
Chapter 1 section 2.4.
LEXICAL-SEMANTIC DIMENSION: See CLASSES OF PREDICATES,
GENERALITY OF USE. See Chapter 1 section 5.
LIGHT-VERB: The term LIGHT-VERB refers to verbs with low degree of specific
lexical content, which are very appropriate to enter into GRAMMATICALIZATION as
part of a TAM ANALYTIC CONSTRUCTION. These are verbs such as izan ‘to be, to be
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
338
at’, egon ‘to be at’, *edun ‘to have’, eduki ‘to have’, egin ‘to do’, ibili ‘to walk, to
wander, to be doing something’, ari izan ‘to be engaged doing something’, joan ‘to
go’, etc. The difference between the terms LIGHT-VERB and QUASI-AUXILIARY is
small. However, I use QUASI-AUXILIARY whenever I refer to the conjugated verb in
the LOOSE PERIPHRASES, to indicate that these constructions have already started a
process of GRAMMATICALIZATION. On the other hand, LIGHT-VERB is used to refer
to lexical verbs alone, or as hosts of verbal LOCUTIONS.
LOCUTION, (VERBAL LOCUTION) of Basque: (Examples: nahi dut ‘I want’,
bizi dira ‘they live’, zin dagit ‘I promise’.) I borrow the term VERBAL LOCUTION
from Lafitte (1962), to refer to highly lexicalized PERIPHRASTIC verbal EXPRESSIONS
with very low degree of GENERALITY OF USE. That is, they appear only for some
specific lexical verbs. In the case o f Basque, these FORMATIONS are traditionally
considered a lexical phenomenon; i.e. their lexical entry is made up of a noun /
adjective / adverb plus a host LIGHT-VERB. This is an analysis that I accept in this
work. For the VERBAL LOCUTIONS of Basque, see Chapter 1 section 5; for the
STATIVE PRESENT VERBAL LOCUTION, see Chapter 3 section 6.5, Chapter 4 section
10. It is not always easy to draw a distinction between VERBAL LOCUTIONS and
some LOOSE PERIPHRASES of Basque. In principle, VERBAL LOCUTIONS have a
higher degree of SOUNDNESS but a lower degree of GENERALITY OF USE than LOOSE
PERIPHRASES have.
LOOSE PERIPHRASES: (Examples: egiten ari naiz ‘I am doing’, egiten nago ‘I
am doing’, joana da ‘s/he is gone’, joanda dago ‘s/he is gone’.) See ANALYTIC
FORM, QUASI-AUXILIARY. For the LOOSE PERIPHRASES OF BASQUE, or PERIPHERAL
ANALYTIC FORMS OF THE INDICATIVE, see Chapter 1 section 6, Chapter 2 section 6,
Chapter 7 section 3. See also LOCUTION.
MAIN ANALYTIC FORMS OF THE INDICATIVE: See RIGID PERIPHRASES.
MAIN CORPUS: The name MAIN CORPUS refers to the most important corpus of
data I employ, which was especially collected for this work. The MAIN CORPUS
consists of three databases extracted from three versions of the Gospel according to
Saint Matthew. I call these databases LEIZAR, HARAN and ITUN, respectively
from the 16th century author Leizarraga, the 18th author Haraneder, and the 20th
century Itunberria, version of the New Testament in Modem Standard Basque. See
Chapter 1 section 7.
MARKER: See FORM.
MEANING, T.A.M. MEANING: Alternatively, SEMANTIC SUBTYPE. [Also,
READING, SENSE, VALUE, USE (semantic); FUNCTION, USE, VALUE, CONTEXT
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
339
(discourse-pragmatic); CONTEXT (syntactic). See CONTEXT, DOMAIN, SENTENTIAL
USES.] I use the term TAM MEANING as a cover term for any kind of semantic or
pragmatic distinction regarding temporal (i.e. T.A.M.) concepts. TAM MEANINGS
make reference to specific conceptual labels, rather than to formally combinable
grammatical markers. Also, the focus is not on delimiting the borders / boundaries of
CATEGORIES, but rather on positively defining individual MEANINGS, which may
sometimes overlap. Thus, these TAM MEANINGS correspond to LABELS such as
PRESENT HABITUAL, PRESENT PROGRESSIVE, (PAST) PERFECTIVE, PRESENT
RESULTATIVE, PAST RESULTATIVE, (PRESENT) PERFECT, PLUPERFECT, etc, or to
even finer divisions such as perfect of result, perfect of recent past, experiential
perfect, etc. See also PROTOTYPE, T.A.M., CATEGORY, COMPETITION, CONTRAST.
NARRATIVE ANTERIORITY: I dub NARRATIVE ANTERIORITY the function
fulfilled by a finite or nonfinite clause which: 1) appears in first position in a
sequence of clauses in NARRATIVE DISCOURSE, 2) is subordinated to the subsequent
clause(s), and 3) expresses anterior chronological order relative to the subsequent
clause(s). Thus, in the example when Jesus had entered Capernaum, a centurion
came to him (Mt VIII, 5), the clause when Jesus had entered Capernaum has a
function of NARRATIVE ANTERIORITY. As in this example, narrative anteriority is
sometimes expressed by a finite verb form overtly marked for anteriority. In this
case, it seems it is not normal Pluperfects but rather PERFECTIVE PLUPERFECTS or
DOUBLE-COMPOUND PAST PERFECTS that express narrative anteriority. See Chapter
5 sections 3.2 and 6.1, and Chapter 6 section 9.
NARRATIVE CONTEXT: (Taken from Dahl 1985.) “A sentence occurs in a
NARRATIVE CONTEXT if the temporal point of reference is determined by the point at
which the last event related in the preceding context took place. Thus, the event
referred to by vidi in veni, vidi, vici is understood to have taken place directly after
that referred to by veni. Basically, this means that in a pure NARRATIVE DISCOURSE,
every sentence except the first one is in a NARRATIVE CONTEXT” (Dahl 1985:112-
113). See NARRATIVE SEQUENCES.
NARRATIVE-DISCOURSE SOURCES of grammaticalization: See DISCOURSE
SOURCES.
NARRATIVE (FORMS): Alternatively NARRATIVE-PERFECTIVE FORMS, or
N a r r a t iv e s . Since n a r r a t iv e and s e n t e n t ia l (non-narrative) u s e s of
PERFECTIVES are partly different, we may expect to encounter languages that have
different constructions to express each of them. In fact, this situation does occur in
several languages: see Dahl (1985:139,113-114, 119-120). However, it appears that
the cross-linguistically most common scenario is for narrative and non-narrative
PERFECTIVE to be non-distinct; i.e. both are most often expressed by the same
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
340
construction. Thus, we may hypothesize that Perfectives with SENTENTIAL SOURCES
of grammaticalization further extend to cover narrative contexts, whereas Narratives
with DISCOURSE SOURCES generalize to cover perfective non-narrative.
NARRATIVE-PERFECTIVE: See NARRATIVE (FORMS), and PERFECTIVE.
NARRATIVE SEQUENCES, NARRATIVE DISCOURSE: “A NARRATIVE
DISCOURSE is one where the speaker relates a series of real or Active events in the
order they are supposed to have taken place” (Dahl 1985:112). Example: veni, vidi,
vici ‘I came, I saw, I conquered.’
NARRATIVE USES of a TAM form: Alternatively DISCOURSE-PRAGMATIC USES.
See SENTENTIAL USES. See also MEANING.
NON-DISCRETE MODELS, NON-DISCRETENESS: See DISCRETE MODELS,
DISCRETENESS.
NONFINITE FORM: Non-conjugated, non-inflected form. In the Basque case at
least, NONFINITE verbal forms are those with no personal agreement. For the
NONFINITE FORMS OF BASQUE, see Chapter 2 section 3. For the IMPERFECTIVE
PARTICIPLE specifically, see Chapter 3 section 6.2. For the RADICAL and its origin,
see Chapter 3 section 6.4.1.
NON-INDICATIVE: See SUBJUNCTIVE.
NON-PAST, (DOMAIN): Alternatively DOMAIN OF PRESENT. The DOMAIN OF
NON-PAST is one of the two SPHERES of TAM examined in this work. This DOMAIN
is mostly defined by the CLINES OF GRAMMATICALIZATION that the forms at issue
follow. Thus, since the cline starting with PRESENT PROGRESSIVE evolves to
PRESENT HABITUAL and GNOMIC, and finally to FUTURE-TIME REFERENCE
CONTEXTS, all of these meanings are considered as pertaining to the same DOMAIN.
The central meanings in this domain are PRESENT PROGRESSIVE and PRESENT
HABITUAL. Note that the term NON-PAST is (albeit more cumbersome) more accurate
than PRESENT to define this domain, since FUTURE TIME-REFERENCE CONTEXTS are
included in it.
NON-STATTVE PREDICATES: See DYNAMIC PREDICATES.
OBJECTS: OBJECTS constitute the low LEVEL OF ABSTRACTION within
INDIVIDUALS. “ OBJECTS are the most familiar, for these are what are most readily
thought of as constituting entities, or as corresponding to the set of (possible)
INDIVIDUALS. OBJECTS, then, are things like Jimmy Carter, the chair I now occupy,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
341
or the world’s fattest magician” (G. Carlson 1979:53). The difference between
predicates that apply to the level of OBJECTS and to the level of STAGES can be
understood as follows: “If the predicate speaks of general characteristics, or
dispositions, we represent it as applying to a set of OBJECTS. If something more
fleeting is intended, somehow more temporary, and in some sense less intrinsic to the
nature of a given INDIVIDUAL, the predicate is represented as denoting a set of
STAGES. This distinction is intended to correspond to the basically atemporal nature
of INDIVIDUALS as opposed to their time-bound STAGES. [...] In general, verbs that
take a PROGRESSIVE form apply to STAGES, and the remainder (generally) apply to
OBJECTS” (G. Carlson 1979:57). See also KINDS, for the distinction between the
latter and OBJECTS.
OPPOSITION: See CONTRAST.
PAST: 1) See SIMPLE PAST. 2) See PAST NON-IMPERFECTIVE, (DOMAIN).
PAST NON-IMPERFECTIVE, (DOMAIN): Alternatively DOMAIN OF PAST,
DOMAIN OF PAST PERFECTIVE. The DOMAIN OF PAST NON-IMPERFECTIVE is one of
the two TAM SPHERES examined in this work. This DOMAIN is mostly defined by
the CLINES OF GRAMMATICALIZATION that the forms at issue follow. Thus, since the
clines starting with RESULTATIVES evolve to PERFECTS and finally to PERFECTIVES,
all of these meanings are considered as pertaining to the same DOMAIN. The central
meaning in this domain is PERFECTIVE. On the other hand, past imperfective
meanings are excluded from this domain, and hence from the study in this work. The
only partial exceptions to the exclusion of imperfective past meanings concern to
SIMPLE PAST and to EXPERIENTIAL PERFECT. Note that the term PAST NON-
IMPERFECTIVE is (albeit cumbersome) the most accurate to define this domain.
PAST PERFECT: See PLUPERFECT.
PAST PERFECTIVE: See PERFECTIVE.
PAST PERFECT-PERFECTIVE of Basque (PsP): (Examples: pasatu zen ‘he,
she, it passed’ [INTRANSITIVE-ABSOLUTIVE];pasatu zuen ‘s/he passed it’
[transitive].) The PAST PERFECT-PERFECTIVE of Basque is formally parallel to the
English, French or Spanish Pluperfects, or even more to the Italian Pluperfect, in that
both have auxiliary selection. Regarding meaning, the PAST PERFECT-PERFECTIVE is
a PRE-HODIERNAL PERFECTIVE in all dialects. It can also be used as a PLUPERFECT,
especially in subordinate clauses. Due to this dual character and for consistency with
its Present counterpart, I dub it PAST PERFECT-PERFECTIVE. See Chapter 2 section
5.2, Chapter 5 section 7.3, Chapter 6 sections 5, 6, 7, 9 and 10.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
342
PAST RESULTATIVE: See RESULTATIVES.
PAST RESULT ATIVE of Basque (PsR): (Examples: joana zen ‘s/he was gone’,
joanda zegoen ‘s/he was gone’, eskutitza idatzia neukan ‘I had the letter written (by
me)’, eskutitza idatzita neukan ‘I had the letter written (by me)’.) See Chapter 2
section 6.1, Chapter 5 section 7.6, Chapter 6 section 5.
PAST TEMPORAL FRAME: (Taken from Dahl 1985:147.) The term PAST
TEMPORAL FRAME “signals that the situation belongs to a temporal section now
concluded and not including the speech time” (Squartini 1999:55), as in asking
“Who BUILD this house?” looking at a picture of a house which has been tom down
(taken from Dahl’s 1985 questionnaire). “It appears that the use of Pluperfects to
signal PAST TEMPORAL FRAMES is the first step towards a situation where they are
used as a general REMOTE past” (Dahl 1985:147); that is, “an intermediate step in the
gradual evolution of the Pluperfect from an originally ABSOLUTE-RELATIVE to an
ABSOLUTE tense” (Squartini 1999:56).
PATH OF GRAMMATICALIZATION: See CLINE OF GRAMMATICALIZATION.
PERFECT: 1) See PERFECTS. 2) See PRESENT PERFECT.
PERFECTIVE: 1) See PAST NON-IMPERFECTIVE, (DOMAIN). 2) I use the TAM
meaning label PERFECTIVE always as PAST PERFECTIVE; that is, confined to past
time-reference (for clarification, see Chapter 3 note 2; see also PERFECTIVES OF THE
SLAVIC TYPE). Perfective appears as the central meaning within the semantic
DOMAIN I dub PAST NON-IMPERFECTIVE. It does not seem easy to give a definition
of perfective, and there is some disagreement in the literature regarding this
definition. In order to clarify things, it may be of interest to distinguish between
SENTENTIAL and NARRATIVE uses of Perfective constructions. In NARRATIVE
SEQUENCES, a situation expressed by a Perfective is mostly understood as punctual
(i.e. point-like), with short and limited duration. It is also understood as completed.
That is to say, the sequence of narration gives the impression of punctual definite
actions each one occurring right after the previous one has been completed. As a
matter of fact, actions that are extended in time or indefinite are often introduced in a
NARRATIVE CONTEXT by means of the Perfective of a light-verb like start, become,
etc, (or otherwise expressed by a specific form that combines perfective and
imperfective meanings). In these cases, the use of the normal Perfective of the lexical
verb seems uncommon in many languages. On the other hand, in non-narrative
SENTENTIAL USES, the features just mentioned (i.e. punctuality, short and limited
duration, completion) are not necessarily appropriate to characterize all appearances
of Perfectives (see Comrie 1976:16-21). Nevertheless, although these features are
not necessary-and-sufficient conditions for the appearance of Perfectives, they do
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
343
seem to correlate with them. Therefore, we could tentatively propose the following
list of features, which would define most cases of narrative contexts, but would just
correlate with perfective non-narrative. See also BOUNDEDNESS, TELICITY.
TABLE G.2: Prototypical features correlating with PERFECTIVE meaning.___________
BOUNDEDNESS of the lexical predicate: shortness, punctuality_____________________
BOUNDEDNESS of the participants: definiteness_________________________________
Completion (completedness), TELICITY, attainment of a limit_______________________
temporal BOUNDEDNESS: occurrence of definite time-adverbials___________________
totality, single whole, no phases, unity (cf. Comrie 1976:16, 21)______________________
PERFECTIVE PLUPERFECT: Alternatively ‘narrative pluperfect.’ The forms I
call PERFECTIVE PLUPERFECTS are different from the more common Pluperfect
constructions in that the former are constructed with the past PERFECTIVE of the
STATIVE AUXILIARIES. Two of the best known Perfective Pluperfects are the Old
Spanish Preterito Anterior (hubo amado) and the Old French Passe Anterieur (// eut
aime). Perfective Pluperfects are used to express NARRATIVE ANTERIORITY. See
Chapter 5 section 3.2, and Chapter 6 section 9.
PERFECTIVES OF THE SLAVIC TYPE: Alternatively Perfectives from
BOUNDERS. By the term PERFECTIVES OF THE SLAVIC TYPE I refer to those
derivational Perfectives whose SOURCE of grammaticalization is formed by “adverbs,
comparable to English up, down, over and through [...] paired with verbs to create a
sense of completion” (Bybee et al. 1994:87). This source for Perfectives is common
in the Slavic languages. Due to their SOURCE of grammaticalization, PERFECTIVES
OF THE SLAVIC TYPE are based on the idea of TELICITY, and thus, in the past tense,
they “emphasize that a limit has been attained” (Bybee et al. 1994:89). Yet, this
derivational type of Perfective “occurs with all tenses and is not necessarily
restricted to the past” (Bybee et al. 1994:89). Note, however, that in this work I use
the label PERFECTIVE as a conceptual meaning pertaining only to the past tense.
Thus, the Present Perfectives of the Slavic type are not considered here as conveying
perfective meaning. (Notice the use of initial capitals in ‘Present Perfectives of the
Slavic type’ vs. the non-use of the initial capital in the conceptual meaning
‘perfective.’) See Chapter 3 note 2.
PERFECTS: (Some scholars use ANTERIORS: cf. Bybee et al. 1994:55.) PERFECTS
are “relational” (Bybee et al. 1994:54) TAM MEANINGS. That is, they have an
“absolute-relative” (Comrie 1985:64ff) or “anaphoric” (Squartini 1999:56) function.
It could also be stated that PERFECTS imply a reference point (Reichenbach’s R),
which is in principle different from the moment of speech. However, in the case of
the PRESENT PERFECT, reference point and moment of speech would coincide, and
thus its absolute-relative meaning is not as clear as in other PERFECTS. The two
PERFECTS that are studied in this work, namely the (PRESENT) PERFECT and the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
344
PLUPERFECT, have therefore several disparities from one another (unlike the
PRESENT RESULTATIVE and the PAST RESULT ATIVE, which have very similar
definitions with the only difference being their time reference). In any case, a general
characterization of both PERFECTS may be that they indicate “the continuing
relevance of a previous situation” (Comrie 1976:56). (For discussion on the
differences and similarities between the PERFECT and the PLUPERFECT, see Comrie
1985:77-82, Squartini 1999. See also Chapter 5 section 3.1.) Note that PERFECTS
differ from PERFECTIVE meaning in that, in the latter, “the situation is reported for its
own sake and independent of its relevance to any other situation” (Bybee et al.
1994:318).
PLUPERFECT: Alternatively PAST PERFECT. PLUPERFECT is an ABSOLUTE-
RELATIVE meaning that indicates the relevance of a past situation (which occurred at
an event time) at another point in the more recent past (i.e. at a reference time,
Reichenbach’s R). PLUPERFECT meaning is usually divided into perfect-in-the-past
and past-in-the-past (see Comrie 1976:56). This distinction is based on whether
definite time-adverbials indicate the reference time or the event time, respectively.
REVERSED RESULT and PAST TEMPORAL FRAME are also two uses often conveyed by
Pluperfect constructions (cf. Squartini 1999:55-60), although it is difficult to know
whether the latter should be considered as part of pluperfect meaning in synchrony.
Rather, I would prefer to consider the value of PAST TEMPORAL FRAME as part of a
REMOTE PERFECTIVE, where the relevance of the reference time-point is practically
lost. (See nevertheless the diachronic link between PLUPERFECT and REMOTE
PERFECTIVE in Chapter 5 section 3.1, and throughout Chapter 6. See also Dahl
1984:114-115, Comrie 1985:68-69, Squartini 1999.)
PREDICTION: See discussion in Chapter 1 section 3.1 and Chapter 7 section 4.2.4.
PRE-HODIERNAL PERFECTIVE: “In most languages with oppositions of
temporal distance in the past, one of the cut-off points is between today and
yesterday” (Comrie 1985:93), thus giving rise to HODIERNAL and PRE-HODIERNAL
temporal FRAMES, respectively. This is a distinction that occurs in the TAM system
of Basque. In Modem Standard Basque, the PRESENT PERFECT-PERFECTIVE
conveys HODIERNAL PERFECTIVE meaning, i.e. it is used to refer to any past
situation within today’s time reference. In turn, the PAST PERFECT-PERFECTIVE
expresses PRE-HODIERNAL PERFECTIVE meaning, i.e. it is used to refer to any past
situation not included within today’s time reference. PRE-HODIERNAL PERFECTIVE,
therefore, is a kind of REMOTE PERFECTIVE, but with a very specific (and not
necessarily too remote in absolute terms) temporal FRAME. PRE-HODIERNAL
PERFECTIVE meaning can be expressed by a construction conveying PLUPERFECT
meaning. This is the case of the Basque PAST PERFECT-PERFECTIVE. Yet, I assume
that synchronically these two meanings are distinct. The difference between PRE-
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
345
HODIERNAL PERFECTIVE and PLUPERFECT is the same that holds between all
PERFECTS and PERFECTIVES; namely, that the former has an ABSOLUTE or DEICTIC
meaning while the latter has an ABSOLUTE-RELATIVE or ANAPHORIC meaning.
Diachronically, however, there is a CLINE of grammaticalization that leads from
PLUPERFECT to PRE-HODIERNAL and/or REMOTE PERFECTIVE. (For degrees of
remoteness, see Comrie 1985:83ff, Bybee et al. 1994:98ff. For the diachronic
relationship between PLUPERFECT and REMOTE PERFECTIVE, see Comrie 1985:68-
69, Dahl 1985:146ff, Squartini 1999.) See also PAST TEMPORAL FRAME.
PRESENT: See NON-PAST, (DOMAIN). See also PRESENT PROGRESSIVE, PRESENT
HABITUAL.
PRESENT HABITUAL: The DOMINANT FEATURE in defining prototypical
PRESENT HABITUALS is that of CHARACTERISTIC BEHAVIOR rather than ‘repetition’;
i.e. habituals “describe a situation which is characteristic of an extended period of
time, so extended in fact that the situation referred to is viewed not as an incidental
property of the moment but, precisely, as a characteristic feature of a whole period”
(Comrie 1976:27-28). On the other hand, it ought to be acknowledged that in most of
the real-world situations, especially of NON-STATIVE PREDICATES, a habit implies
also the repetition of a behavior. (For discussion on the mixed nature of habituals,
see L. Carlson 1981:42-43, Givon 1984, Comrie 1985:40, Fleischman 1995:537.) In
any event, PRESENT HABITUAL seems to be most often singled out on the basis of its
CONTRAST with PRESENT PROGRESSIVE. Present progressive and present habitual
can be viewed as the focal points in a putative hierarchy of LEVELS OF
ABSTRACTION, which would be ordered as follows: 1) present PROGRESSIVE (i.e. one
dynamic situation occurring in one STAGE), 2) present ITERATIVE (i.e. repetition of a
situation within one stage), 3) present FREQUENTATIVE (i.e. repetition of stages), 4)
present HABITUAL (i.e. a habit or characteristic behavior of an INDIVIDUAL), 5)
GNOMIC (i.e. a characteristic behavior of KINDS). Therefore, while PROGRESSIVE
applies to the level of STAGES, HABITUAL applies to the level of INDIVIDUALS (more
specifically to OBJECTS). The chart below puts forward a list of features that
correlate with the PROTOTYPE of PRESENT HABITUAL. See PRESENT PROGRESSIVE
and Table G.4.
TABLE G.3: Prototypical features of PRESENT HABITUAL meaning._______________
Refers to a period which includes the moment of speech____________________________
Totality, continuousness; but often entails repetition: (i.e. multiplicity fr unity)_________
Extended, more permanent_____________________________________________________
Descriptive, characteristic; but often implies dynam icity___________________________
More abstract, inductive___________________________ __________ __________________
Applies to individuals_________________________________________________________
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
346
PRESENT PERFECT: Alternatively PERFECT. The (PRESENT) PERFECT conveys
the idea that a past situation is relevant at the moment of speech. Its meaning is
therefore dual: the PERFECT refers to a past situation and also to its present relevance.
Hence, I assume in this work that restricted PERFECT meaning may not be used
together with specification of the time of the past situation. In other words, I assume
that the possibility of combining with definite time-adverbials is what distinguishes
an HODIERNAL PERFECTIVE from a restricted PRESENT PERFECT. (See nevertheless
the diachronic link between these meanings in Chapter 5 section 2.1.) Comrie
(1976:56-61) gives four situations where a past action can have relevance in present;
that is, what may be considered four types of PRESENT PERFECT. These are: perfect
of result, experiential perfect, perfect of persistent situation, and perfect of recent
past. It is not always easy to distinguish among these uses. Specifically, the perfect
of result and perfect of recent past seem to co-occur often: “a recent event is more
likely to have a persistent result than a distant one” (Dahl 1985:136). Moreover, the
perfect of result is frequently similar to the PRESENT RESULTATIVE. Also, the
distinction among the four types of perfect might be partly based on lexical-semantic
factors. In any case, the most distinct of the types of perfects seems to be the
EXPERIENTIAL PERFECT. See also PERFECTS.
PRESENT PERFECT-PERFECTIVE of Basque (PrP): (Examples: pasatu da
‘he, she, it has passed’ [INTRANSITIVE-ABSOLUTIVE]; pasatu du ‘s/he has passed it’
[transitive].) The PRESENT PERFECT-PERFECTIVE of Basque is formally parallel to
the English, French or Spanish Perfects, or even more to the Italian Perfect, in that
both have auxiliary selection. Regarding its meaning, in the Spanish-Basque dialects
the use of this construction is very similar to that of the Spanish Perfect, while its use
in some of the French-Basque varieties is more similar to the French Perfect. I dub it
Pr e s e n t P e r f e c t -P e r f e c t iv e in accordance with the P a s t P e r f e c t -P e r f e c t iv e ,
since both of them, and especially the latter, are not only PERFECTS but also
PERFECTIVES: the former is an HODIERNAL PERFECTIVE in the Western dialects and
a general PERFECTIVE in some Eastern varieties; the latter is a PRE-HODIERNAL
PERFECTIVE in all dialects. See Chapter 2 section 5.2, Chapter 5 section 7.4, Chapter
6 sections 4, 6, 8 and 10.
PRESENT PROGRESSIVE: Bybee et al. (1994:126) give the following definition
of PROGRESSIVE: “Progressive views an action as ongoing at reference time. [...] It
applies typically to DYNAMIC PREDICATES and not to STATTVE ones. Thus the
progressive is typically used for actions that require a constant input of energy to be
sustained.” They also propose “that the original meaning of the progressive
construction is ‘the subject is located in the midst of doing something’. [...] Thus the
construction contains either explicitly or implicitly the following elements of
meaning: a) An agent, b) is located spatially, c) in the midst of, d) an activity, e) at
reference time” (Bybee et al. 1994:136). We can therefore propose the following
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
347
interconnected features, as constituting the multidimensional PROTOTYPE of
restricted PRESENT PROGRESSIVE meaning. See also PRESENT HABITUAL and Table
G.3.
TABLE G.4: Prototypical features of PRESENT PROGRESSIVE meaning.__________
Ongoing at the moment of speech_____________________________________________
Internal differentiation, gradualness, structure divided into non-identical phases______
Durative; but temporary, contingent;
incomplete (cf. Dowty’s 1977 “progressive paradox”)____________________________
Dynamic__________________________________________________________________
Specific, non-abstract_______________________________________________________
Applies to stages___________________________________________________________
PRESENT RESULT ATIVE: See RESULTATIVES.
PRESENT RESULT ATIVE of Basque (PrR): (Examples: joana da ‘s/he is gone’,
joanda dago ‘s/he is gone’, eskutitza idatzia dut ‘I have the letter written (by me)’,
eskutitza idatzita daukat ‘1 have the letter written (by me)’.) See Chapter 2 section
6.1, Chapter 5 section 7.6, Chapter 6 section 4.
PROGRESSIVE: See PRESENT PROGRESSIVE. (PRESENT PROGRESSIVE is the only
progressive meaning studied in this work.)
PROTOTYPE OF A CATEGORY: The concept of PROTOTYPE, although not
often overtly used in this work, is basic in the spirit of it (cf. Comrie 1985:18ff,
Givon 1995:12ft). PROTOTYPES are implicit in all classifications in this work: e.g.
lexical-semantics CLASSES OF PREDICATES (see Chapter 4 section 10.4), STAGE
LEVEL vs. INDIVIDUAL LEVEL predicates, and especially in the definitions of each of
the TAM MEANINGS. In these definitions, I intend to capture the features which
correlate with the PROTOTYPE of each TAM meaning, regardless of whether one
among those features is DOMINANT over the others in distinguishing that given TAM
meaning. PROTOTYPES OF CATEGORIES can be defined as “the clearest cases of
category membership defined operationally by people’s judgments of goodness of
membership in the category” (Rosch 1978:36). The concept of PROTOTYPE is the
basis for NON-DISCRETE MODELS of categorization. The most important validation
behind these models is the fact that “most, if not all, categories do not have clear-cut
boundaries” (Rosch 1978:35). Therefore, instead of proposing necessary-and-
sufficient criteria for category membership, “another way to achieve separateness
and clarity of actually continuous categories is by conceiving of each category in
terms of its clear cases rather than its boundaries” (Rosch 1978:35-36). Thus, the
assumption of PROTOTYPES leaves room for fuzziness and overlapping in the
periphery, i.e. where differences are smaller (as in much of the subject matter in this
work, which seldom deals with maximal CONTRASTS). This gives rise to fluidness
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
348
and ultimately to change, unlike the static nature of DISCRETE MODELS. In addition,
prototype-based models imply a kind of processing and learning importantly rooted
in real-world perception and similarity cues, which are very hard to formalize
concepts. These concepts are in the line of a relationship between grammar and
cognition, and ultimately in favor of non-autonomous (non-innate) grammar and
learning. See Chapter 7 for discussion on these issues.
PROTOTYPICAL T.A.M. MEANINGS, PROTOTYPICALITY: The definition
of PROTOTYPE OF A CATEGORY presupposes the existence of CATEGORIES as theory
primitives. Another possible perspective of prototypes would be to assume a degree
of PROTOTYPICALITY, a gradient concept that would distinguish prototypical
situations from less prototypical and non-prototypical ones. In the case of TAM
systems, PROTOTYPICAL TAM MEANINGS would be the clearest, most salient, most
basic MEANINGS that speakers distinguish within the sphere of TAM. In this case, the
primitive concepts would be the TAM meanings themselves, rather than the
categorization of TAM systems. Then, grammatical categories would tend to be
constructed around a PROTOTYPICAL MEANING. In other words, PROTOTYPICAL
situations would be “natural targets for distinctive and characteristic marking”
(Kemmer 1996:239). See the discussion on prototypicality and grammatical marking
in Chapter 7 section 2.5.
QUASI-AUXILIARY: I use the term QUASI-AUXILIARY in this work to refer to the
conjugated verbal form (e.g. ari naiz, nago) which is part of a LOOSE PERIPHRASIS
(e.g. egiten ari naiz ‘I am doing’, egiten nago ‘I am doing’). That is, QUASI
AUXILIARY is the conjugated item in those ANALYTIC FORMS which are only lowly
grammaticalized. (The difference between the terms LIGHT-VERB and QUASI
AUXILIARY is small. See LIGHT-VERB.) In general, I do not provide a synchronic
analysis of the compositionality o f the LOOSE PERIPHRASES, i.e. whether they should
be considered monoclausal or biclausal. See, nevertheless, Chapter 1 section 6.2,
Chapter 2 sections 2.2 and 6, Chapter 4 section 2, Chapter 6 section 2.
RADICAL, (VERBAL RADICAL) of Basque: (Examples: esan ‘say’, ikus ‘see’,
pasa ‘pass.’) See NONFINITE FORMS.
READING: See CONTEXT, MEANING, DOMAIN.
REMOTE PERFECTIVE: See PRE-HODIERNAL PERFECTIVE, PAST TEMPORAL
FRAME, PLUPERFECT.
RENEWAL: The term RENEWAL makes reference to the emergence of new TAM
constructions which catch on and start to grammaticalize. The term RENEWAL does
not imply SUBSTITUTION of old forms by new forms. Rather, the young forms
COMPETE with the old ones, although the tendency in the long run is usually the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
349
DISPLACEMENT of old forms. The term RENEWAL does not either entail that
functional needs are driving the change in the system. See Chapter 1 section 2.5,
Chapter 7 section 4.
REPLACEMENT: See SUBSTITUTION.
RESULTATIVES: Resultative meaning “denotes a state that was brought about by
some action in the past”, and it further “consistently signals that the state [result of
the previous action] persists at reference time” (Bybee et al. 1994:63). The following
English examples may illustrate the notion of RESULTATIVE meaning and its
difference with PERFECT meaning: he is gone ‘present resultative’ vs. he has gone
‘present perfect’ (B. Comrie, p.c.). Resultative meaning is subject to LEXICAL-
SEMANTIC restrictions, and thus the constructions conveying resultative are usually
not very GENERALIZED IN USE. The most significant restriction these constructions
have is that they only combine with predicates expressing a direct result, usually
verbs involving some kind of change. The main CLINES of grammaticalization
examined in this work in the DOMAIN of PAST NON-IMPERFECTIVE start with
RESULT ATIVES; concretely with either PRESENT RESULTATIVE or PAST
RESULTATIVE. Resultative constructions are often formally similar to passives and/or
to temporary predicative stative sentences. (A reference work for resultatives is
Nedjalkov (ed.) 1983/1988. A typology of resultative constructions, including the
basic terminology, is offered in Nedjalkov & Jaxontov 1983/1988. Other remarks on
resultatives, with special emphasis on the distinction RESULTATIVES / PERFECTS are
given in Dahl 1985:133-135 and Bybee & Dahl 1988:68-69. See also Bybee et al.
1994:54, 63-68.)
REVERSED RESULT: (Taken from Squartini 1999:57.) The term REVERSED
RESULT signals that “the result of a past action is no longer valid, thus triggering the
effect of what has traditionally been called a two-way action, [...] as in he did
promise, but now he pretends not to remember i f (Squartini 1999:57).
RIGID PERIPHRASES: (Examples: egin dut ‘I have done’, egin nuen ‘I did’,
egiten dut ‘I do’, egiten nuen ‘I used to do’, egingo dut ‘I will do’.) See ANALYTIC
FORM. For the RIGID PERIPHRASES OF BASQUE, or MAIN ANALYTIC FORMS OF THE
INDICATIVE, see Chapter 1 section 7, Chapter 2 sections 2.1 and 5, Chapter 3
sections 6.2 and 6.3, Chapter 5 sections 7.3 and 7.4, Chapter 7 section 3.
ROUTINIZATION: Alternatively, HABITUATION, AUTOMATION. The process
“wherein an organism ceases to respond at the same level to a repeated stimulus”
(Bybee & Hopper 2001:13).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
350
SAME-SOURCE LAYERING: SAME-SOURCE LAYERING is one of the scenarios of
COMPETITION I distinguish in this work. The term SAME-SOURCE LAYERING refers to
the situation made up by two (or more) LAYERS of constructions that have the same
SOURCE of grammaticalization. I argue that the competition between SAME-SOURCE
LAYERS, unlike the other scenarios of competition including that of DIFFERENT-
SOURCE constructions, is a good situation to test the HYPOTHESES of Layering-
Displacement, and Form-Meaning Covariation (not to mention the Hypothesis of
Source Determination). Thus, if the Hypothesis of Source Determination holds for
two layers with the same source, and if there is no inertia, then the tendency
(‘prediction’) stated in the Hypothesis of Layering-Displacement is for the young
layer to displace the old layer, both following the same PATH of grammaticalization.
This hypothesis has to be tested against STRICTLY DIACHRONIC DATA. On the other
hand, the Hypothesis o f Form-Meaning Covariation can be tested in the case of
SAME-SOURCE LAYERING even in the lack of strictly diachronic data; namely,
against SYNCHRONIC DATA OF CHANGE. See Chapter 1 sections 3.5 and 4.
SEMANTIC AGE: Degree of semantic grammaticalization. The SEMANTIC AGE of
a form f within a PATH OF GRAMMATICALIZATION A-> B C -> D -> E ... might
be defined as the point (e.g. A, B, C, ...) form f has progressed to the right in the
path. See also SEMANTIC UNIDIRECTIONALITY.
SEMANTIC CHANGE: Many, probably most, changes in meaning in the process
of grammaticalization (either semantic or pragmatic changes) progress towards
GENERALIZATION (in its different senses) and ABSTRACTION. However, this work is
primarily concerned with the SEMANTIC (UNI)DIRECTIONALITY of specific TAM
CLINES and its reasons (i.e. the ASSOCIATIONS between one TAM meaning and the
next in the CLINE), rather than with labels such as ABSTRACTION and
GENERALIZATION. See Chapter 1 section 2.3.
SEMANTIC SUBTYPE: 1) See MEANING. 2) See AREAS OF COMPETITION.
SEMANTIC (UNI)DIRECTIONALITY: The PATHS OF GRAMMATICALIZATION
which TAM forms follow are mostly unidirectional, and are importantly determined
by the SOURCE OF GRAMMATICALIZATION of the forms. SEMANTIC
UNIDIRECTIONALITY is a byproduct o f the general cross-linguistic tendency to create
very similar psycholinguistic ASSOCIATIONS (i.e. metaphors, implicatures, etc.)
departing from similar sources of grammaticalization. However, this does not imply
that strict PREDICTION in advance is possible. The development of one specific form
in a given language may be aborted (by inertia, etc), it may bifurcate, or it may even
be partly altered from the expected path, due to the competition with other forms of
difference sources, or to language contact. See Chapter 1 section 2.3, Chapter 7
section 4.2.4.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
351
SENSE: See CONTEXT, MEANING, DOMAIN.
SENTENTIAL SOURCES of grammaticalization: SENTENTIAL SOURCES OF
GRAMMATICALIZATION, i.e. those grammaticalizing out of independent non-
discursive sentences, appear to be more common than DISCOURSE SOURCES. This
seems to be part of the explanation of why main clauses are innovative and
subordinate clauses conservative, as most of the SENTENTIAL SOURCES are
grammaticalized in main clauses. There are however some sentential sources which
start in subordinate clauses. This is the case of the PERFECTIVE PLUPERFECTS, which
have the function of NARRATIVE ANTERIORITY in their first stages of
grammaticalization (see Chapter 5 sections 3.2 and 5.1). The Basque PAST PERFECT-
PERFECTIVE, which in later stages evolved to a PRE-HODIERNAL PERFECTIVE and
extended to main clauses, is thus a counter-example of the non-innovativeness of
subordinate clauses, regarding sources of grammaticalization (see Chapter 6 sections
7, 9 and 10). See also SENTENTIAL USES.
SENTENTIAL USES of a TAM form: The term SENTENTIAL USE refers to the
properly semantic VALUE of a given TAM form in independent sentences, as
opposed to its DISCOURSE-PRAGMATIC USES, i.e. those fulfilled in NARRATIVE
SEQUENCES. (Cf. the distinction sentence-semantics vs. discourse-pragmatics:
Hopper 1982:5-6, 15-16; Givon 1995:18.) See also MEANING.
SIMPLE PAST: Alternatively GENERAL PAST. The label with the widest
INCLUSIVENESS within the TAM SPHERE of PAST is that of SIMPLE PAST or
GENERAL PAST. As a matter of fact, the SIMPLE PAST surpasses the scope I aim to
cover in the DOMAIN of PAST NON-IMPERFECTIVE. “The simple past is semantically
more general [than the perfective] since it can also be used to signal past time for
situations viewed imperfectly” (Bybee et al. 1994:84). In other words, the SIMPLE
PAST can convey both perfective and imperfective past meanings, as in the case of
the English Preterit. The readings of either perfective or imperfective past seem
partly conditioned by the LEXICAL-SEMANTICS of the predicates. Thus, it appears
that many of the simple past cases of DYNAMIC PREDICATES would have a perfective
reading, while many of the simple past cases of ST ATIVE PREDICATES would express
imperfective past.
SITUATION: (Taken from Comrie 1976:13). The term SITUATION is often used
here as a cover-term to subsume STATES, events and processes.
SOURCE OF GRAMMATICALIZATION: The term SOURCE in
grammaticalization of lexical items refers to the lexical or literal meaning that the
entire TAM construction (e.g. not only the AUXILIARIES) originally had at the
moment the process of grammaticalization started. See HYPOTHESIS OF SOURCE
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
352
DETERMINATION, SAME-SOURCE LAYERING. I distinguish at least two types of
SOURCES in this work: 1) SENTENTIAL SOURCES, and 2) DISCOURSE SOURCES.
SPHERE: See DOMAIN.
STAGE-LEVEL PREDICATES: STAGE-LEVEL PREDICATES are the predicates that
apply to STAGES.
STAGES: STAGES constitute the lowest LEVEL OF ABSTRACTION within entities.
STAGES “are essentially time-space slices of INDIVIDUALS. [...] So, for instance, two
distinct time-space slices of Jimmy Carter, though quite dissimilar in appearance, are
thought of as being the same thing in that they are STAGES of the same INDIVIDUAL.
[...] STAGES [...] are never named. Only INDIVIDUALS may bear names” (G. Carlson
1979:54).
ST ATIVE AUXILIARIES of Basque: The STATIVE AUXILIARIES of Basque are
izan ‘to be’, which functions as INTRANSITIVE-ABSOLUTIVE auxiliary, and *edun ‘to
have’, which functions as the ERGATIVE (transitive) counterpart. (For ditransitive
forms, there is a great amount of cross-dialectal variation in the auxiliary used.)
STATIVE AUXILIARIES are part of all of the RIGID PERIPHRASES or MAIN ANALYTIC
FORMS OF THE INDICATIVE. See Chapter 2 sections 2.1 and 5, Chapter 7 section 2.10.
See DYNAMIC AUXILIARIES, ERGATIVE-ABSOLUTIVE MARKING.
STATIVE PREDICATES, (STATES): See CLASSES OF PREDICATES. Vendler
(1967:107) provides the following examples of STATES: have, posses, desire, want,
like, dislike, love, hate, rule, dominate, know, and believe. See also STATIVE
PRESENT VERBAL LOCUTION of Basque.
STATIVE PRESENT: Alternatively CONTINUOUS NON-PROGRESSIVE PRESENT. I
use the label STATIVE PRESENT to refer to the TAM MEANING which is most
characteristic of STATIVE PREDICATES in the present tense, as in I have, you posses,
he desires, she wants, we like, you hate, they know, or we believe. (Note that this
TAM meaning is not readily paralleled in other CLASSES OF PREDICATES.) STATIVE
PRESENT meaning can be divided into ‘temporary stative present’ and ‘permanent
stative present’. The marking of this temporary / permanent distinction significantly
correlates across languages with the PRESENT PROGRESSIVE / PRESENT HABITUAL
formal opposition, respectively (see Comrie 1976:37). On the other hand, an
opposite tendency towards grouping together all STATES (or at least the clearest
cases) can also be encountered. In any case, it seems the use of a primarily Present
Habitual construction is one of the most representative for conveying STATIVE
PRESENT meaning. See also STATIVE PRESENT VERBAL LOCUTION OF BASQUE.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
353
STATIVE PRESENT VERBAL LOCUTION of Basque: (Examples: nahi dut ‘I
want’, bizi dira ‘they live’, behar duzu ‘you need’, maite du ‘s/he loves’, uste dut ‘I
believe’.) See Chapter 3 section 6.5, Chapter 4 section 10. (Cf. Dahl 1985:28.)
STRICTLY DIACHRONIC DATA: I call STRICTLY DIACHRONIC DATA those
obtained from the comparison of the very same construction in different stages of a
language. In the case of historical Basque, STRICTLY DIACHRONIC DATA are
unfortunately not always very revealing, due to the shortness of the attestation in
Basque. The semantic changes witnessed in the historical period are not abundant;
the formal changes are even less clear. See SAME-SOURCE LAYERING, SYNCHRONIC
DATA OF CHANGE.
SUBJUNCTIVE FORMS of Basque: (Examples: etor dadi-n ‘so that s/he comes’,
egin dezazu-n ‘so that you do it’, egin dagizu-n ‘so that you do it’.) The
SUBJUNCTIVE FORMS of Basque are the only NON-INDICATIVE forms I study in this
work. They are made up of the NONFINITE FORM called the RADICAL, and a
conjugated form of a DYNAMIC AUXILIARY. Although in Modem Basque these
forms are only used in subordinate SUBJUNCTIVE CONTEXTS, it is important to
mention that they had indicative uses in Old Basque (e.g. the Old Basque
Periphrastic AORIST, which is formally identical to the Modem Basque Past
Subjunctive). See Chapter 2 section 7, Chapter 3 section 6.4, Chapter 5 section 7.2,
Chapter 7 section 2.10.
SUBJUNCTIVE CONTEXTS, SUBJUNCTIVE MOOD: The label SUBJUNCTIVE
or SUBJUNCTIVE MOOD includes usually what I term here DYNAMIC FUTURE TIME-
REFERENCE SUBORDINATE CONTEXTS. However, sometimes the LESS-DYNAMIC
FUTURE TIME-REFERENCE SUBORDINATE CONTEXTS are also included in this label.
SUBORDINATE FUTURE TIME-REFERENCE CONTEXTS: See
SUBJUNCTIVE CONTEXTS.
SUBSTANCE, SEMANTIC SUBSTANCE: (Taken from Bybee 1988.) The theory
of grammaticalization maintains that, besides the meaning based on CONTRAST or
OPPOSITION of forms, TAM constructions have also an inherent semantic
SUBSTANCE or content o f their own. The inherent SUBSTANCE of TAM forms and
especially their changes are the primary concern in this work. That is, the CLINES OF
GRAMMATICALIZATION, main object of study of this work, are mostly internal
changes in the SUBSTANCE of forms. It should not be forgotten, though, that
CONTRASTS, and especially the COMPETITION of forms, also play a role in the design
of TAM systems. See Chapter 7 Section 4.2.3.1.
SUBSTITUTION: The term SUBSTITUTION or REPLACEMENT of one form by
another is seldom used in this work. In its place, the term DISPLACEMENT is used,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
354
which is intended to leave room for OVERLAPPING of forms and NON
DISCRETENESS.
SUBTYPE: See SEMANTIC SUBTYPE.
SYMBOLIC MODELS, SYMBOLIC relation between form and meaning: A
given TAM construction will have a SYMBOLIC relation with the meaning it conveys
if there is a totally arbitrary link and an ISOMORPHIC (i.e. one-to-one)
correspondence between them. (Cf. McMahon 1994:171-172.)
SYNCHRONIC DATA OF CHANGE: I call SYNCHRONIC DATA OF CHANGE
those obtained from the comparison in a specific stage of a language of two
constructions that constitute SAME-SOURCE LAYERS. See STRICTLY DIACHRONIC
DATA. See Chapter 1 section 4.1.
SYNTHETIC FORM: (Examples: daukat ‘I have’, dator ‘s/he comes’, neukan ‘I
had’, zetorren ‘s/he was coming’.) SYNTHETIC verbal forms are those FINITE or
conjugated forms which are made up of only one word.
T.A.M.: The abbreviation TAM (from Tense-Aspect-Mood) is used as a cover term
for any temporal concept. In general, no adscription is made for any given TAM
MEANING (e.g. present habitual, future, etc) into one super-category out of the three,
i.e. tense, aspect or mood. As a matter of fact, providing a definition for the
categories of tense, aspect or mood, is not a prime concern of this work. See
CATEGORY, MEANING, SUBSTANCE, CONTRAST.
T.A.M. MEANING: See MEANING.
T.A.M. MEANING LABELS: The great majority of the LABELS used in this work
to refer to TAM MEANINGS are borrowed from Comrie (1976), Comrie (1985), Dahl
(1985), and Bybee, Perkins & Pagliuca (1994). See also Nedjalkov & Jaxontov
(1983/1988), Bybee & Dahl (1989) and Squartini (1999). Some interesting
terminological notes are given in Comrie (1976:11-13). A useful glossary of
Meaning Labels is offered in Bybee et al. (1994:316-324).
TELIC PREDICATES, TELICITY, TELIC SOURCES: The term TELIC refers to
those SITUATIONS, predicates, TAM meanings or grammaticalization SOURCES
which imply the idea of leading to an end-point or limit. Thus, TELIC PREDICATES
include ACCOMPLISHMENTS, but can also refer to punctual ACHIEVEMENTS and the
immediately preceding process. (See Comrie 1976:44-48.) Similarly, TELIC SOURCES
are those conveying the idea of “attainment of a limit” (Dahl 1985:76), e.g. sources
with ACHIEVEMENT AUXILIARIES, directional sources (motion verbs and
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
355
prepositions), PERFECTIVES OF THE SLAVIC TYPE, etc. See Chapter 3 section 5,
Chapter 5 section 4. For the differences between telic sources in PRESENT (i.e. non
past) and PAST, see Chapter 3 note 3. See also BOUNDEDNESS.
TOKEN-FREQUENCY: Token-frequency is measured by the number of tokens,
i.e. absolute number of forms. For instance, the token-frequency o f a construction
that only appears in the database 5 times with the verb esan ‘to say’, 4 times with the
verb edan ‘to drink’, and 1 time with the verb jan ‘to eat’, would be 10, i.e. the
absolute number of tokens. (See also TYPE-FREQUENCY.)
TYPE-FREQUENCY: T ype-fr equency is measured by means of the number of
items, i.e. lexical verbs. For instance, the type-ffequency of a construction that only
appears in the database 5 times with the verb esan ‘to say’, 4 times with the verb
edan ‘to drink’, and 1 time with the verb jan ‘to eat’, would be 3, i.e. the number of
lexical verbs. See GENERALITY OF USE, LEXICAL-SEMANTIC DIMENSION. (See also
TOKEN-FREQUENCY.)
(UNIDIRECTIONALITY: See FORMAL (UNI)DIRECTIONALITY, SEMANTIC
(UNI)DIRECTIONALITY, HYPOTHESIS OF FORM-MEANING COVARIATION.
USE: See CONTEXT, MEANING, DOMAIN, SENTENTIAL USES.
VALUE: See CONTEXT, MEANING, DOMAIN.
VERBAL CLASSES: See CLASSES OF PREDICATES.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
356
BIBLIOGRAPHY
Aldai, Gontzal. 1994. “Euskal aditz sistema zaharraz: aspektu oposaketatik modu
oposaketara” [On the verbal system of Old Basque: from aspect opposition to mood
opposition]. Enseiukarrean 10: 67-82.
---------- - 1998. “A proposito del Aoristo vasco.” Fontes Linguae Vasconum 30: 377-
386.
---------- - 2000. “Split ergativity in Basque: The pre-Basque antipassive-imperfective
hypothesis.” Folia Linguistica Historica 21: 31-97.
Altuna, Patxi. 1979. Etxepareren hiztegia / Lexicon dechepariano. Bilbao:
Mensajero.
Anderson, John. 1973. An essay concerning aspect: some considerations of a
general character arising from the Abbe Darrigol’s analysis of the Basque verb. The
Hague / Paris: Mouton.
Bennett, Charles E. 1910. Syntax o f Early Latin. Reprinted in 1966 in Hildesheim:
Olms.
Bybee, Joan. 1985. Morphology: A study of the relation between meaning and form.
Amsterdam / Philadelphia: John Benjamins.
-............. 1986. “On the nature of grammatical categories: a diachronic perspective.”
Eastern States Conference on Linguistics 2: 17-34.
-. 1988a. “Semantic substance vs. contrast in the development of grammatical
meaning.” Berkeley Linguistic Society 14: 247-264.
----------- 1988b. “The diachronic dimension in explanation.” In John A. Hawkins
(ed.), 350-379.
----------- 1995. “Diachronic and typological properties of morphology and their
implications for representation.” In Laurie B. Feldman (ed.), Morphological aspects
of language processing, 225-246. Hillsdale, New Jersey: Lawrence Erlbaum
Associates.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
357
---------- “Main clauses are innovative, subordinate clauses are conservative:
consequences for the nature of constructions.” To appear in Joan Bybee & Michael
Noonan (eds.), Complex sentences in grammar and discourse. Essays in honor of
Sandra A. Thompson. Amsterdam / Philadelphia: John Benjamins.
Bybee, Joan & Osten Dahl. 1989. “The creation of tense and aspect systems in the
languages of the world.” Studies in Language 13: 51-103.
Bybee, Joan & Paul Hopper. 2001. “Introduction to frequency and the emergence of
linguistic structure.” In Joan Bybee & Paul Hopper (eds.), 1-24.
-........— (eds.) 2001. Frequency and the emergence of linguistic structure.
Amsterdam / Philadelphia: John Benjamins.
Bybee, Joan; Revere Perkins & William Pagliuca. 1994. The evolution of grammar:
Tense, aspect and modality in the languages of the world. Chicago / London:
University of Chicago Press.
Campbell, Lyle. 1998. Historical linguistics: An introduction. Edinburgh: Edinburgh
University Press. Reprinted in Cambridge, Massachusetts: MIT Press.
Carlson, Greg. 1979. “Generics and atemporal when." Linguistics and Philosophy 3:
49-98.
Carlson, Lauri. 1981. “Aspect and quantification.” In Philip Tedeschi & Annie
Zaenen (eds.), Tense and Aspect. (Volume 14 of ‘Syntax and Semantics.’) New
York: Academic Press.
Carstairs-McCarthy, Andrew. 1992. Current morphology. London / New York:
Routledge.
Chomsky, Noam. 1995. The minimalist program. Cambridge, Massachusetts: MIT
Press.
Clark, Herbert & Barbara Malt. 1984. “Psychological constraints on language: a
commentary on Bresnan & Kaplan and on Givon.” In Walter Kintsch, James Miller
& Peter Poison (eds.), Method and tactics in cognitive science, 191-214. Hillsdale,
New Jersey: Lawrence Erlbaum Associates.
Comrie, Bernard. 1976. Aspect: An introduction to the study o f verbal aspect and
related problems. Cambridge, UK: Cambridge University Press.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
358
----------- 1982. “Future time reference in the conditional protasis.” Australian
Journal of Linguistics 2: 143-152.
............. . 1985. Tense. Cambridge, UK: Cambridge University Press.
---------- - 1989. Language universals and linguistic typology: syntax and
morphology. 2n d edition. Chicago: University of Chicago Press.
---------- - 1998. “Perspectives on grammaticalization.” In Toshio Ohori (ed.), Studies
in Japanese grammaticalization: cognitive and discourse perspectives, 7-24. Tokyo:
Kurosio Publishers.
Dahl, Osten. 1984a. “Perfectivity in Slavonic and other languages.” In Casper de
Groot & Hannu Tommola (eds.), Aspect bound: A voyage into the realm of
Germanic, Slavonic and Finno-Ugrian aspectology, 3-22. Dordrecht, Holland: Foris.
----------- 1984b. “Temporal distance: remoteness distinctions in tense-aspect
systems.” In Brian Butterworth, Bernard Comrie & Osten Dahl (eds.), Explanations
for language universals, 105-122. Berlin: Mouton.
............. . 1985. Tense and aspect systems. Oxford / New York: Blackwell.
Darrigol, J-P. 1829. Dissertation critique et apologetique sur la langue basque.
Bayonne: Duhart-Fauvet.
Declerck, Renaat. 1979. “On the progressive and the imperfective paradox.”
Linguistics and Philosophy 3: 267-272.
Dowty, David. 1977. “Toward a semantic analysis of verb aspect and the English
imperfective progressive.” Linguistics and Philosophy 1: 45-77.
---------- - 1979. Word meaning and Montague grammar. Dordrecht: Reidel.
Euskaltzaindia [Royal Academy of the Basque language]. 1987. Euskal gramatika.
Lehen urratsak - II [Basque grammar. First steps - II]. Bilbao: Euskaltzaindia.
Fleischman, Suzanne. 1982. The future in thought and language. Cambridge, UK:
Cambridge University Press.
---------- - 1995. “Imperfective and irrealis.” In Joan Bybee & Suzanne Fleischman
(eds.), Modality in grammar and discourse, 519-551. Amsterdam / Philadelphia:
John Benjamins.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
359
Givon, Talmy. 1979. On understanding grammar. New York: Academic Press.
---------- 1984. Syntax. Afunctional-typological introduction. Volume 1. Amsterdam
/ Philadelphia: John Benjamins.
---------- 1995. Functionalism and grammar. Amsterdam / Philadelphia: John
Benjamins.
Greenberg, Joseph. 1957. Essays in Linguistics. Chicago: University of Chicago
Press.
---------- 1966. Language universals, with special reference to feature hierarchies.
The Hague / Paris: Mouton.
Haase, Martin. 1994. ‘Tense and aspect in Basque.” In Rolf Thieroff & Joachim
Ballweg (eds.), Tense systems in European languages, 279-292. Tubingen: Max
Niemeyer Verlag.
Haiman, John. 1985. Natural syntax: iconicity and erosion. Cambridge, UK:
Cambridge University Press.
----------- 1994. “Ritualization and the development of language.” In William
Pagliuca (ed.), Perspectives on grammaticalization, 3-28. Amsterdam: John
Benjamins.
Harris, Alice. 1985. Diachronic syntax: The Kartvelian case. (Volume 18 of ‘Syntax
and Semantics.’) Orlando: Academic Press.
Harris, Alice & Lyle Campbell. 1995. Historical syntax in cross-linguistic
perspective. Cambridge, UK: Cambridge University Press.
Harris, Martin. 1988. “French.” In Martin Harris & Nigel Vincent (eds.), The
Romance Languages, 209-245. New York: Oxford University Press.
Haspelmath, Martin. 1989. “From purposive to infinitive: a universal path of
grammaticalization.” Folia Linguistica Historica 10: 287-310.
----------- 1997. Explaining article-possessor incompatibility: economic motivation in
noun phrase syntax. Manuscript. Free University of Berlin / University of Bamberg.
............. . 1999. “Explaining article-possessor complementarity: economic motivation
in Noun Phrase syntax.” Language 75: 227-243.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
360
Hawkins, John A. 1994. A performance theory of order and constituency.
Cambridge, UK: Cambridge University Press.
---------- 1999. “Processing complexity and filler-gap dependencies across
grammars.” Language 75: 244-285.
---------- 2001. “Why are categories adjacent?” Journal of Linguistics 37: 1-34.
-----------(ed.) 1988. Explaining language universals. Oxford, UK / Cambridge,
USA: Blackwell.
Heath, Jeffrey. 1972. “Genitivization in Northern Basque complement clauses.”
Anuario del Seminario Julio de Urquijo (ASJU) 6: 46-66.
---------- 1981. “The role of Basque in modem linguistic theory.” In Euskaltzaindia
(ed.), Iker 1. Bascologists’ International Meetings, 433-444. Bilbao: Euskaltzaindia.
Hedin, Eva. 1987. On the use of the Perfect and Pluperfect in Modem Greek.
Stockholm: Almqvist & Wiksell.
Heine, Bemd. 1993. Auxiliaries: Cognitive forces and Grammaticalization. New
York / Oxford: Oxford University Press.
Heine, Bemd; Ulrike Claudi & Friederike Hiinnemeyer. 1991. Grammaticalization:
A conceptual framework. Chicago / London: University of Chicago Press.
Hopper, Paul. 1979. “Some observations on the typology of focus and aspect in
narrative language.” Studies in Language 3: 37-64.
— . 1982. “Aspect between discourse and grammar: An introductory essay for
the volume.” In Paul Hopper (ed.), Tense-Aspect: between semantics & pragmatics,
3-18. Amsterdam / Philadelphia: John Benjamins.
---------- 1987. “Emergent grammar.” Berkeley Linguistic Society 13: 139-157.
Hopper, Paul & Sandra Thompson. 1980. “Transitivity in grammar and discourse.”
Language 56: 251-299.
Hopper, Paul & Elizabeth Traugott. 1993. Grammaticalization. Cambridge, UK:
Cambridge University Press.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
361
Hualde, Jose Ignacio; Joseba A. Lakarra & R.L. Trask (eds.) 1995. Towards a
history of the Basque language. Amsterdam / Philadelphia: John Benjamins.
Inoue, K.1975. Studies in the perfect. Unpublished dissertation. University of
Michigan.
Irigoien, Alfonso. 1985. En torno a la evolucion y desarrollo del sistema verbal
vasco. Bilbao: Pracor.
Irizar, Pedro. 1991. “Sobre la extension del subjuntivo y del potencial vizcainos.” In
Joseba A. Lakarra & Inigo Ruiz Arzalluz (eds.), Memoriae L. Mitxelena magistri
sacrum, 1093-1104. San Sebastian: Diputacion Foral de Guipuzcoa.
Keenan, Edward. 2001. “Explaining the creation of reflexive pronouns in English.”
To appear in D. Minkova & R. Stockwell (eds.), Studies in the history of English: a
millennial perspective. Berlin / New York: Mouton de Gruyter.
Kelly, Michael & Susanne Martin. 1994. “Domain-general abilities applied to
domain-specific tasks: Sensitivity to probabilities in perception, cognition and
language.” Lingua 92: 105-140.
Kemmer, Suzanne. 1992. “Grammatical prototypes and competing motivations in a
theory of linguistic change.” In Garry Davis & Gregory Iverson (eds.), Explanation
in historical linguistics, 145-166.
...... . 1996. “Collective and distributive marking, or: where unity meets
multiplicity.” LACUS Forum 23: 231-249.
Kirby, Simon. 1997. “Competing motivations and emergence: Explaining
implicational hierarchies.” Linguistic Typology 1: 5-31.
---------- 1999. Function, selection and innateness: The emergence o f language
universals. Oxford: Oxford University Press.
Kratzer, Angelika. 1988. Stage-level and individual-level predicates. Manuscript.
University of Massachusetts at Amherst.
Kroch, Anthony. 1989. “Reflexes of grammar in patterns of language change.”
Language Variation and Change 1: 199-244.
Krutwig, Federico. 1985. Esbozo gramatical de la lengua vasca literaria con base en
el labortano clasico. Manuscript. Euskaltzaindia.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
362
Kurylowicz, Jerzy. 1965. “The evolution of grammatical categories.” Reprinted in
1975 in Esquisses linguistiques II, 38-54. Munich: Fink.
Labov, William. 1990. “On the adequacy of natural languages: I. The development
of tense.” In John V. Singler (ed.), Pidgin and creole tense-mood-aspect systems, 1-
58. Amsterdam / Philadelphia: John Benjamins.
Lafitte, Pierre. 1962. Grammaire Basque: Navarro-Labour din Litteraire. 2n d edition.
Reprinted in 1979 in San Sebastian / Bayonne: Elkar / Ikas.
Lafon, Rene. 1943. Le systeme du verbe Basque au XVI siecle. 2 volumes. Reprinted
in 1980 (one volume) in San Sebastian: Elkar.
Lakarra, Joseba A. 1996. Refranes y Sentencias (1596): ikerketak eta edizioa.
Bilbao: Euskaltzaindia.
Langacker, Ronald. 1997. “Generics and habituals.” In Angeliki Athanasiadou &
Rene Dirven (eds.), On conditionals again, 191-222. Amsterdam / Philadelphia:
John Benjamins.
Larramendi, Manuel. 1745. Diccionario trilingue del castellano, bascuence y latin.
San Sebastian: Riesgo y Montero.
Lehmann, Christian. 1995. Thoughts on grammaticalization. Munich: Lincom
Europa.
Lightfoot, David. 1975. Natural logic and the Greek moods: the nature of the
Subjunctive and Optative in Classical Greek. The Hague / Paris: Mouton.
----------- 1979. Principles o f diachronic syntax. Cambridge, UK: Cambridge
University Press.
----------- 1991. How to set parameters: arguments from language change.
Cambridge, Massachusetts: MIT Press.
----------- 1999. The development o f language: Acquisition, change, and evolution.
Malden / Oxford: Blackwell.
McMahon, April. 1994. Understanding language change. Cambridge, UK:
Cambridge University Press.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
363
Meillet, Antoine. 1912. “L’evolution des formes grammaticales.” Reprinted in 1948
in Linguistique general et linguistique historique, 130-148. Paris: Champion.
Mitxelena, Luis. 1960. Historia de la literatura vasca. Reprinted in 1988 in San
Sebastian: Erein.
............... 1964. Sobre el pasado de la lengua vasca. Reprinted in 1988 in Sobre
Historia de la Lengua Vasca, 1-73. San Sebastian: Diputacion Foral de Guipuzcoa.
............. . 1977. “Notas sobre compuestos verbales vascos.” Reprinted in 1987 in
Palabras y textos, 311-335. Vitoria-Gasteiz: University of the Basque Country.
---------- - 1979. “Vizcaino a- con temas demostrativos.” Reprinted in 1987 in
Palabras y textos, 437-454. Vitoria-Gasteiz: University of the Basque Country.
Mpnnesland, Svein. 1984. “The Slavonic frequentative habitual.” In Casper de Groot
& Hannu Tommola (eds.), Aspect bound: A voyage into the realm o f Germanic,
Slavonic and Finno-Ugrian aspectology, 53-76. Dordrecht, Holland: Foris.
N’Diaye, Genevieve. 1970. Structure du dialecte Basque de Maya. The Hague /
Paris: Mouton.
Nedjalkov, Vladimir P. (ed.) 1983. Typology o f resultative constructions. English
translation (1988) edited by Bernard Comrie. Amsterdam / Philadelphia: John
Benjamins.
Nedjalkov, Vladimir & Sergej Jaxontov. 1983. “The typology of resultative
constructions.” In Vladimir P. Nedjalkov (ed.), 3-62.
Ortiz de Urbina, Jon & Myriam Uribe-Etxebarria. 1991. “Participial predication in
Basque.” In Joseba A. Lakarra & Inigo Ruiz Arzalluz (eds.), Memoriae L. Mitxelena
magistri sacrum, 993-1012. San Sebastian: Diputacion Foral de Guipuzcoa.
Phillips, Colin. 1996. Order and structure. Doctoral dissertation. MIT.
Pikabea, Josu. 1993. Lapurtera idatzia (XVII-XIX): bilakaera baten urratsak [The
evolution of literary Labourdin between the XVII-XIX centuries]. San Sebastian:
University of the Basque Country.
Poultney, James. 1957. “Imperfect indicative and historical infinitive.” In Ernst
Pulgram (ed.), Studies presented to Joshua Whatmough on his sixtieth birthday, 219-
223. The Hague: Mouton.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
364
Rebuschi, Georges. 1979. “Sur les deux passifs et quelques phenomenes connexes en
Basque d’Onate (biscayen oriental).” Verbum 2: 211-231.
de Rijk, Rudolf P.G. 1995. “’Nunc’ in Old Basque.” In Jose Ignacio Hualde, Joseba
A. Lakarra & R.L. Trask (eds.), 295-311.
Rosch, Eleanor. 1978. “Principles of categorization.” In Eleanor Rosch & Barbara
Lloyd (eds.), Cognition and categorization, 27-48. Hillsdale, New Jersey: Lawrence
Erlbaum Associates.
Saltarelli, Mario. 1988. (With Miren Azkarate, David Farwell, Jon Ortiz de Urbina &
Lourdes Onederra.) Basque. New York: Croom Helm. Reprinted in London / New
York: Routledge.
Schwenter, Scott. 1993. “The grammaticalization of an anterior in progress:
Evidence from peninsular Spanish dialect.” Studies in Language 17.
Squartini, Mario. 1999. “On the semantics of the Pluperfect: Evidence from
Germanic and Romance.” Linguistic Typology 3: 51-89.
Trask, R.L. 1977. “Historical syntax and Basque verbal morphology: two
hypotheses.” In William Douglass, Richard Etulain & William Jacobsen (eds.),
Anglo-American contributions to Basque studies: Essays in honor of Jon Bilbao,
203-217. Reno, Nevada: Desert Research Institute Publications on the Social
Sciences.
----------- 1981. “Basque verbal morphology.” In Euskaltzaindia (ed.), Iker 1.
Bascologists ’ International Meetings, 285-304. Bilbao: Euskaltzaindia.
— .......... . 1990. “The -n class of verbs in Basque.” Transactions of the Philological
Society 88: 111-128.
............. . 1995. “On the history of the non-finite verb forms in Basque.” In Jose
Ignacio Hualde, Joseba A. Lakarra & R.L. Trask (eds.), 207-234.
----------- 1997. The history o f Basque. London / New York: Routledge.
Traugott, Elizabeth C. & Bemd Heine. 1991. “Introduction.” In Elizabeth Closs
Traugott & Bernd Heine (eds.), Approaches to Grammaticalization. Volume I. Focus
on theoretical and methodological issues, 1-14. Amsterdam / Philadelphia: John
Benjamins.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
365
Vendler, Zeno. 1967. “Verbs and times.” In Linguistics in Philosophy, 97-121.
Ithaca, New York: Cornell University Press.
Wheeler, Max W. 1988. “Occitan.” In Martin Harris & Nigel Vincent (eds.), The
Romance Languages, 245-278. New York: Oxford University Press.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Linguistic attitudes in the Basque Country: The social acceptance of a new variety

PDF

Grammaticalization and the development of functional categories in Chinese

PDF

The independence of binding and intensification

PDF

On the identification of null arguments

PDF

The beginnings of language loss in discourse: A study of American Lithuanian

PDF

The fine structure of negative polarity items in Chinese

PDF

Two types of dependency

PDF

The effect of non-native speaker status on the use of linguistic accommodation by native speakers of Japanese: Implications for communication accommodation theory

PDF

The modal preterite phenomenon (MPP) in colloquial American English: A diachronic and synchronic analysis

PDF

Overt focus movement and minimal information marking

PDF

Representational richness in phonological development

PDF

Syntactic and non -syntactic scope

PDF

Second language acquisition of Spanish morpho -syntax by Quechua -speaking children

PDF

Form and meaning: Negation and question in Chinese

PDF

The face of others, the taste of things: Photogenie and sensation in silent cinema

PDF

Natan Altman and the problem of Jewish art in Russia in the 1910s

PDF

Ellipsis constructions in Chinese

PDF

Writing the crusade, writing the present: Vernacular prose histories as cultural markers of transformations in thirteenth-century lay society

PDF

Asymmetry of scope taking in wh -questions

PDF

At wit's end: The rhetoric of humor and the ends of talk

Asset Metadata

Creator Aldai, Gontzal (author)

Core Title The grammaticalization of present and past in Basque

Degree Doctor of Philosophy

Degree Program Linguistics

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag language, linguistics,OAI-PMH Harvest

Language English

Contributor Digitized by ProQuest (provenance)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c16-243143

Unique identifier UC11334974

Identifier 3093727.pdf (filename),usctheses-c16-243143 (legacy record id)

Legacy Identifier 3093727.pdf

Dmrecord 243143

Document Type Dissertation

Rights Aldai, Gontzal

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA