University of Southern California Dissertations and Theses

Borrowing and borrowability

(USC Thesis Other)

Borrowing and borrowability

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content INFORMATION TO USERS
This manuscript has been reproduced from the microfilm master. UMI
films the text directly from the original or copy submitted. Thus, some
thesis and dissertation copies are in typewriter free, while others may be
from any type o f computer printer.
The quality of this reproduction is dependent upon the quality of the
copy submitted. Broken or indistinct print, colored or poor quality
illustrations and photographs, print bleedthrough, substandard margins,
and improper alignment can adversely afreet reproduction.
In the unlikely event that the author did not send UMI a complete
manuscript and there are missing pages, these will be noted. Also, if
unauthorized copyright material had to be removed, a note will indicate
the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by
sectioning the original, beginning at the upper left-hand comer and
continuing from left to right in equal sections with small overlaps. Each
original is also photographed in one exposure and is included in reduced
form at the back of the book.
Photographs included in the original manuscript have been reproduced
xerographically in this copy. Higher quality 6” x 9” black and white
photographic prints are available for any photographs or illustrations
appearing in this copy for an additional charge. Contact UMI directly to
order.
UMI
A Bell & Howell Information Company
300 North Zed) Road, Ann Arbor MI 4S106-1346 USA
313/761-4700 800/521-0600
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
B o r r o w in g a n d Bo r r o w a b u jt y
by
Fredric W. Field
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)
August 1998
© 1998 Fredric W. Field
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UMI Number: 9919036
UMI Microform 9919036
Copyright 1999, by UMI Company. All rights reserved.
This microform edition is protected against unauthorized
copying under Title 17, United States Code.
UMI
300 North Zeeb Road
Ann Arbor, MI 48103
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UNIVERSITY OF SOUTHERN CALIFORNIA
THE GRADUATE SCHOOL
u n iv e r s it y p a r k
LOS ANGELES. CALIFORNIA 90007
This dissertation, written by
Fr e d r i c W i l l i a m F i e l d ......................................................
under the direction of h.te. Dissertation
Committee, and approved by all its members,
has been presented to and accepted by The
Graduate School, in partial fulfillm ent of re
quirements for the degree of
DOCTOR OF PHILOSOPHY
'or Graduate Studies
Date
DISSERTATION COMMITTEE
‘A ' •
Chairperson
✓
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ACKNOWLEDGEMENTS
I would like to acknowledge my committee chair, Professor John Hawkins, for his
vision, professional support, and personal encouragement during the time it took to
complete this work.
I wish also to express appreciation to Professors Bernard Comrie and David Eskey
for their participation as members of the committee. I am especially grateful to Professor
Comrie, whom I have considered an unofficial co-chair, for his rigorous reading of my
drafts and detailed comments, and for spending countless hours providing personal and
intellectual support. I owe much to his hands-on assistance. I have also greatly
appreciated the opportunity to work alongside Professor Eskey, whose encouragement and
support made much of my stay at the University of Southern California possible.
In addition, I would like to take this opportunity to acknowledge those former
professors who have had a considerable impact on me academically and intellectually. For
support at key times in my doctoral work, I am indebted to Professor William Rutherford.
I am also very grateful to Professor Maria Polinsky, whose insightful comments have
played an important role in the development of my work; to Professor Joseph Aoun, who,
from the very outset of my graduate work, offered his constant support; and to Professor
Jean-Roger Vergnaud, who provided continual intellectual stimulation. I am also indebted
to Professor Edward Finegan for trying to help me grasp the importance of paying attention
to detail and for taking a personal interest in my work. Finally, I would like to express
special thanks to Kenneth and Jane Hill of the University of Arizona for making available
their extensive corpus of Mexicano and to Peter Bakker, now of the University of Aarhus,
for his comments and continual support
ii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CONTENTS
Acknowledgements page ii
List of Figures vi
List of Tables vii
Abstract viii
1.0 Chapter 1—Introduction 1
1.0.1 Social factors 4
1.0.2 Linguistic factors 6
1.1 Borrowing as bilingual performance. 10
1.2 Mixed languages—language intertwining or extensive
borrowing? 13
1.2.1 The concept of mixed language 19
1.2.2 The correlation of transfer, substrate, and matrix 22
1.3 The plan of this book 27
2.0 Chapter 2—Morphological structuring and system compatibility 31
2.1 Scales, indices, hierarchies, and clines: continua of forms
and meanings 33
2.1.1 The indices of synthesis and fusion 34
2.1.2 The cline of grammaticality 37
2.1.3 The cline of lexicality 40
2.1.4 A scale (or continuum) of morpheme types 43
2.2 Hierarchies of borrowability 44
2.2.1 Borrowing hierarchies: lexical items 46
2.2.2 Borrowing hierarchies: grammatical items 47
2.2.3 Summary of hierarchies 48
2.3 The Principle of System Incompatibility 52
2.3.1 Predictions within general classes 55
2.3.2 The occurrence of anomalies: reanalysis 57
2.3.3 Predictions within subclasses 59
2.3.4 The relative timing of borrowed elements 60
3.0 Chapter 3—Form classes and semantic types 63
3.0.1 The relationship of word class and semantic type 64
3.0.2 Form-meaning sets and semantic types 67
3.0.3 Organization of this chapter 68
3.1 Notions of word and word class 69
3.2 Contrasting points on a continuum 73
3.2.1 Formal characteristics: grammatical affixes versus content
items 74
3.2.2 Semantic characteristics: grammatical affixes versus content
items 77
3.2.3 Function words: formal and semantic characteristics 80
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.2.4 Function words versus content words 84
3.2.5 Function words versus inflectional affixes and elements in
between 85
3.2.6 Derivational affixes: between content and inflection 88
3.3 Semantic types: groupings of morphemes according to
meanings 91
3.3.1 Semantic types and subtypes of N, V, and Adj 96
3.3.2 Person 98
3.3.3 Object 100
3.3.4 Activity 101
3.3.5 Space 103
3.3.6 Time 104
3.3.7 Quality 104
3.4 Summary and comments 107
4.0 Chapter 4—The identification of form-meaning sets 109
4.0.1 The spread and integration of borrowed form-meaning sets 111
4.0.2 Identifying clusters of properties 113
4.1 Form identification characteristics: the role of salience 116
4.1.1 Transparency versus opacity 117
4.1.2 The opacity of fusional affixes 124
4.1.3 Borrowing continua of forms 128
4.2 Semantic characteristics 132
4.2.1 Inflectional meanings 134
4.2.2 Borrowing continua of meanings 138
4.2.3 Linking form and meaning 146
4.3 Issues of semantic complexity 149
4.4. Summary and general predictions 154
5.0 Chapter 5—Borrowing patterns in Modern Mexicano 160
5.0.1 Bilingual phenomena. 164
5.0.2 The corpus 166
5.0.3 Organization of the chapter 168
5.1 Overview of the participants: Mexicano and Spanish 171
5.1.1 Morphological structuring: the words of each language 173
5.1.2 The borrowing hierarchy of Mexicano 181
5.2 The role of form-meaning interpretation characteristics
(FMICs) 186
5.2.1 FMICs pertaining to form 188
5.2.2 FMICs pertaining to meaning 190
5.3 The effects of borrowing 197
5.3.1 A chronology of borrowing 198
5.3.2 Phrasal & clausal organization 203
5.4 Discussion: the roles of form and meaning in borrowing 205
5.4.1 Issues of form 207
5.4.2 Issues of meaning 209
iv
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6 .0 Chapter 6—Discussion 212
6.0.1 The relevance of form-meaning interpretation characteristics
(FMICs) 213
6.0.2 Organization of this chapter 216
6.1 The PS I, FMICs, and other contact situations 218
6.1.1 Typologically similar languages 219
6.1.2 Along the index of fusion: agglutinating versus fusional
types 221
6.1.3 Along the index of synthesis: isolating-analytical versus
synthetic types 222
6.1.4. th e analysis of apparent exceptions 223
6.2 The borrowability of inflectional categories 225
6.2.1 The emergence of category values 226
6.2.2 The inheritance of categories and category values 229
6.2.3 Limits on types of borrowable meanings/concepts 231
6.3 Connecting borrowing and various contact phenomena 235
6.3.1 Distinguishing code-switching and borrowing 239
6.3.2 Convergence and a composite matrix 243
6.4 Conclusions 246
Appendix A: Additional Mexicano text 250
Appendix B: Spanish borrowings in the data 253
References 269
v
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
FIGURES
Chapter 2
Figure 2.1. Language types and allowable morphemes 52
Figure 2.2. Compatibility and incompatibility 53
Chapter 4
Figure 4.1. Mapping patterns of grammatical affixes 126
Figure 4.2. The mappings of Spanish verbal -o 127
Figure 4.3. Calculating mapping possibilities (present tense) 128
vi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
TABLES
Chapter 4
Table 4.1. Form-meaning interpretation characteristics (FMIQ: forms 123
Table 4.2. Inflectional categories associated with verbs 135
Table 4.3. Form-meaning interpretation characteristics (FMICs): meaning 137
Table 4.4. Content items according to semantic and syntactic complexity 152
Table 4.5. Compatibility versus incompatibility 155
Table 4.6. FMICs: form-meaning sets 157
Table 4.7. Summary of predictions regarding FMICs 158
Chapter 5
Table 5.1. The occurrence of Spanish content items 182
Table 5.2. The occurrence of Spanish function items 182
Table 5.3. The occurrence of Spanish form-meaning sets in Mexicano 183
Table 5.4. The occurrence of Spanish form types based on selection 189
Table 5.5. The occurrence of Spanish concept types 192
Table 5.6. Borrowed Spanish nouns 194
Table 5.7. Borrowed nouns: semantic (sub)types 195
Table 5.8. Borrowed nouns: ratios of hyperonyms to hyponyms 196
Table 5.9. Spanish content items according to Karttunen & Lockhart (1976) 200
vii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Ab st r a c t
This work investigates the phenomenon of linguistic borrowing, the adoption of
form-meaning sets from one language into another. The approach is primarily typological.
It begins with notions of morphological structuring along the lines of Sapir (1921) and
Comrie (1989) and postulates two general principles of morphological compatibility and
incompatibility: The Principle of System Compatibility (PSQ and its correlate, the
Principle of System Incompatibility (PSI). These principles establish basic predictions of
borrowability that are inviolable. They stipulate, for example, that an agglutinating
language can borrow any element that is compatible with ways in which it constructs
words—content words, function words, and, agglutinating-type affixes. It cannot borrow
fusional-type affixes from a language higher along the index of fusion without violating the
integrity of its morphological system.
Irrespective of possible morphological constraints, however, content items are
always preferred in any borrowing situation over function words and/or inflectional
affixes. To account for this, two heretofore distinct lines of inquiry are merged involving
grammaticalization clines and various borrowing hierarchies, synthesized in the general
Hierarchy of Borrowability. Discussion continues to the roles of particular characteristics
of form and kinds of meaning, or Form-Meaning Interpretation Characteristics (FMICs)
(e.g., sufficient, segmentable form versus minimal or zero phonetic shape; concrete versus
abstract concepts; and so on).
As a case in point, loans from Spanish into Modem Mexicano (Ndhuatl) are
examined. First, every type of compatible morpheme is, in fact, borrowed, including the
agglutinating-type inflectional affix -s (indicating plural noun). Nevertheless, none of the
viii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
fusional morphology of the Spanish verb is taken, as predicted by the PSI (Mexicano is
primarily of the agglutinating-type). Second, preferences accord with the Hierarchy of
Borrowability. However, in terms of frequency, the impressive numbers of borrowed
function words surely demonstrate that borrowing has greatly exceeded what is expected
garden-variety borrowing situations. Clearly, more is appropriated from Spanish than a
mere list of words. Finally, FMICs provide a comprehensive rationale for the actual
borrowing patterns, including the allocation of borrowed content items into particular
semantic groupings.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1.0. C h a p t e r l— in t r o d u c t io n
A number of attempts have been made to establish systematic approaches to the
studies of lexical borrowing, code-switching, contact-induced language change, language
attrition and convergence, and so on, with some proposing various links among these
seemingly disparate phenomena. Recent developments have focused attention on the social
and linguistic factors they hold in common and apparent similarities found in underlying
processes. Examples of this growing trend are Thomason & Kaufman (1988), who stress
that the transmission of languages within differing social contexts by other than normal
(i.e„ “parental”) means shapes their fundamental characters (hence, their distinction
between genetic and non-genetic origins), and Myers-Scotton (forthcoming, 1995,1993a,
1993b), who consistently advances the argument that the similar characteristics found in
contact phenomena are traceable to underlying cognitive processes operating in the heads of
individual (bilingual) speakers that collaborate to form what appears to be a matrix language
system in performance, and, in turn, determines the nature of these phenomena (cf. 1995,
p. 239). She also suggests correspondences between language transfer (and the
development of interlanguages) in second language acquisition and substrate influence (or
matrix language) in the emergence of new speech varieties, especially with respect to those
evincing degrees and types of language mixing1 .
In many of these studies, one can find a number of common linguistic threads,
especially regarding the roles of the languages involved. Specifically, when speakers of
two distinct languages (representing two individual cultures) come into intense, day-to-day
1 Some linguists include pidgins, creoles, and other mixed languages under a general heading of “mixed
languages” (Thomason & Kaufman, 1988, p. 3). Lines between these types may appear blurry owing to
the particular and varied circumstances of their origins. Nevertheless, to avoid the obvious terminological
confusion, in the present work, the term “mixed language” will refer specifically to a variety that clearly and
overtly shows relationships to two (or more) distinct languages and does not include pidgins or creoles.
1
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
contact with each other, degrees of bilingualism are to be expected; their respective
languages, then, are said to be in contact when they are both spoken (alternately) by the
same persons (Weinreich, 1953, p. 1). The languages in this kind of intimate contact often
undergo a number of resultant changes. In cases where one language is clearly dominant in
a number of social domains, the dominant (or superordinate) will usually exert greater
influence on the recessive (or subordinate) than the recessive does on the dominant
(Thomason & Kaufman, 1988, pp. 67-68). The dynamic relationships established among
speakers and between linguistic systems have the potential to induce (perhaps, precipitate) a
number of possible outcomes. For instance, the dominant language may assume the role of
lexical donor, providing certain kinds of words or morphemes to be selected by speakers of
the recessive language for adoption while the recessive language system becomes the
recipient of the “donated” words and morphemes, acting as a kind of morphosyntactic
matrix into which these elements are grafted. In the most extreme cases, borrowed
elements have replaced native ones to such an extent that a new variety emerges (Thomason
& Kaufman, 1988, pp. 48, 76ff).
The present study is primarily concerned with the processes by which forms (i.e.,
form-meaning sets) from a lexical donor language, language Y, are imported and integrated
into a recipient language, X—X being the original language spoken by a speech
community. It is assumed that speakers of X initially attempt to reproduce in their own
speech (perhaps by some sort of imitation) forms that previously existed only in Y
(Haugen, 1950, p. 212); these forms may or may not be fully accepted by speakers of X as
subsequently belonging to X. Consequently, the term “borrowing” will be used primarily
to refer to the integration of forms into a recipient language. As discussed in later sections
of this work, the importation of foreign words or morphemes into one’s native language
may include various degrees of phonological adaptation, one possibility being the direct
2
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
borrowing of a foreign phoneme. However, to point out the obvious, phonological
processes which may have applied to a particular phoneme in its source language are not
normally borrowed along with the morpheme (or phonetic string) in which it appears.
Borrowed morphemes, including those with non-native sound segments, generally become
subject to the phonological processes of the new linguistic environment.
While it is possible that a borrowing language will adopt certain phonological and
structural characteristics from another independently of lexical borrowing, extensive
borrowing from an individual source may gradually lead to phonological and other
structural changes in the recipient (Haugen, 1950, p. 225). It is also safe to say that a
significant amount of lexical borrowing is to be expected before one finds evidence of other
“interferences”, i.e., actual structural borrowings (Thomason & Kaufman, 1988, pp. 20-
21). Regarding morphology, it has been proposed that free morphemes are more easily
borrowed than bound, and that the more highly bound the morpheme (e.g., inflectional
affixes versus function words), the less likely it will be borrowed (Weinreich, 1953, pp.
29-37). Generally speaking, the more closely associated elements are to the particular
grammar (morphosyntax) of the potential donor, the more difficult they will be to borrow
(Haugen, 1950, pp. 224-225). Consequently, syntactic characteristics are often
considered to be the least easily diffused aspects of language (Romaine, 1995, p. 64) and
the very last to be borrowed2.
On the one hand, when there is casual contact between languages, lexical items may
be borrowed where there is little or no extensive bilingualism; for instance, American
2 This is with the likely exception of word order (Thomason & Kaufman, 1988, p. 54f.). Contact induced
word order changes have been observed in a number of instances, for example, in U.S. versions of Spanish
(Sdnchez, 1982, p.34ff.) and Low German (Thomason & Kaufman, 1988, pp. 81-83), Asia Minor Greek
(Thomason & Kaufman, 1988, pp. 18,220-222), and so on.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
English has borrowed many cultural items from immigrant groups, e.g. kosher from
Yiddish, pizza from Italian, sauerkraut from German, tortilla from Mexican Spanish, sushi
from Japanese, and so on. On the other hand, many studies of extensive borrowing
assume that the requisite starting point is a subset of the total number of native speakers of
the recipient variety who are also relatively proficient in the donor3, who act as a kind of
conduit for the diffusion of lexical items and other properties of the donor language
(Thomason & Kaufman, 1988, p. 66). In this regard, Grosjean (1982) distinguishes
between “language borrowing”, in which words from one language have been borrowed
by another (and may be used by monolingual speakers of the recipient language), and
“speech borrowing”, in which an individual speaker spontaneously uses a form from
another language within an utterance (perhaps adapting this borrowing phonologically and
morphologically to varying degrees). There is a clear connection between the two:
languages borrow words because individual speakers have at one time borrowed them.
1.0.1. Social factors
A number of social factors have been discussed to account for the amount and types
of borrowing. Thomason & Kaufman (1988, p. 65ff.), for example, discuss (a) the
intensity and length of contact; (b) the relative number of speakers of each variety; (c)
cultural and political (therefore, economic) dominance of one group of speakers, and so on.
In situations where these factors conflict, e.g., when a politically dominant group is
numerically inferior to the subordinate group, patterns of borrowing may differ. The
3 In many respects, degrees and types of bilingualism are always relative and difficult to assess, especially
regarding competence and patterns of usage in each language (Hoffmann, 1991, pp. 17-32, Grosjean, 1982,
230ff.). As a consequence, many scholars posit the existence of proficiency continua in all varieties
represented in a particular community (e.g., Silva-Corvaldn, 1994, p. 11; Campbell & Muntzel, 1988, p.
185).
4
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
cultural pressure of a politically and numerically dominant group on a subordinate
population is also offered as an explanation for why speakers of a minority language often
learn a majority, presage variety, while members of the dominant group do not, as a rule,
seek to become bilingual by mastering the minority language.
Below is a brief list summarizing reasons for borrowing that have been posited by
researchers in recent years:
(a) as a result of the cultural dominance of the donor language (Watson, 1989, pp.
49-51; Mougeon & Beniak, 1989, p. 303-307; Hill & Hill, 1986, p. 4; cf.
Gal, 1989, p. 318;);
(b) to be associated with speakers of the dominant language (and gain socially from
its prestige) (Mertz, 1989, p. 112; Hill & Hill, 1986, p. 103ff.; Thomason &
Kaufman, 1988, p. 44ff; Grosjean, 1982, pp. 336-337);
(c) to fill lexical gaps in a recessive language well along in the process of shift
(Myers-Scotton, 1993a, p. 167; Huffines, 1989, p. 212; Bavin, 1989, p.
270ff.; Haugen, 1989, p.65; Grosjean, 1982, p. 336; Karttunen & Lockhart,
1976, p. 16ff.);
(d) to facilitate understanding with younger speakers who are no longer familiar
with original forms of the recessive language (Bavin, 1989, p. 277; Haugen,
1989, p. 67);
(e) for affect or convenience (Hoffmann, 1991, pp. 102-103; Grosjean, 1982, pp.
311-313).
Grosjean (1982) also points out that borrowing of specific words may occur because only
one language has the desired word, or because an individual is not equally familiar with the
words of both languages and chooses the most available word (p. 311). Whatever the
actual reasons may be, patterns of borrowing remain nonetheless fairly predictable with
respect to the formal characteristics of borrowed elements.
5
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1.0.2. Linguistic factors
Two linguistic factors often cited as playing promoting and inhibiting roles in
borrowing are frequency and (formal) equivalence (Van Hout & Muysken, 1994, p. 42;
Weinreich, 1953, p. 61). The first, frequency, refers to how often specific items occur in a
donor language. Frequently occurring items may have a pushing effect on a borrowing
language: on the one hand, the more frequent an item is in the donor, the better it is as a
candidate for borrowing; on the other, the more frequent an item is in the recipient
language, the more of an inhibiting affect it will exert, thereby resisting or blocking the
borrowing and subsequent usage of a corresponding lexical item from the donor. The
second of these factors, equivalence, pertains to word classes, i.e., whether or not a
particular form finds a structural or formal equivalent (usually defined as an equivalent
form class such as N, V, Adj, and so on), which will either facilitate (if the answer is in the
affirmative) or inhibit its inherent borrowability.
There are three points of caution when considering frequency as a main cause.
One, if frequency has a significant statistical impact on borrowing, its effects appear
primarily with respect to certain morpheme types, i.e., those constituting content items
such as nouns, verbs, and adjectives, whether free-standing or bound roots or bases. For
example, content morphemes in the Romance languages normally occur with obligatory
inflections for gender, number, and so on. Nevertheless, when a particular lexical item is
borrowed, only the content morpheme, as perceived by speakers of the borrowing
language, is incorporated into that language4. Grammatical morphemes, which may consist
4 This implies morphological reanalysis. For example, there are a number of Spanish borrowings into
U.S. varieties of English, for instance, the word taco, the name for a popular Mexican food item. In
Spanish, the -o ending is an inflectional affix indicating grammatical gender. English, which does not have
grammatical gender, has borrowed the entire word as one unanalyzable unit—a content item. It has not
borrowed an affix or the inflectional category of gender (which would apply to the entire lexicon).
6
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
of independent function words, roots, or affixes, though they are among the most
frequently occurring forms in any language, are clearly not borrowed on this basis (if at
all). Two, it remains to be demonstrated how core vocabulary items (e.g., words for
certain basic body parts, kinship relationships, everyday activities, and the like), which are
particularly resistant to change (replacement or loss), correlate with frequency counts: if
they are, indeed, as frequent as one might think, why are they almost never borrowed or
replaced?
Three, while frequency may be a factor in the integration of particular content items
into a recipient language, its overall effect may depend on other linguistic factors, for
example, semantic transparency, relevance, and so on (Van Hout & Muysken, 1994, pp.
52— 54). Moreover, the frequency of a particular word in a language (e.g., in corpora
formally obtained from a wide range of native speakers or written texts) does not
necessarily determine an individual or identifiable group of borrowers’ relative exposure to
that form. For example, a typical native speaker of, say, Spanish in Mexico City may not
be exposed to agricultural or other terms from specific semantic fields (e.g., occupational
nomenclatures) to the same degree as a bilingual speaker (of relative proficiency) of
Spanish and Mexicano in the Malinche region of central Mexico who may have more
intense exposure to such terms as a result of his/her expertise in a particular occupation. In
addition, many bilinguals are likely to have access to a somewhat narrower range of
registers of speech in one or both of their languages (hence, fewer semantic types) as a
result of socioeconomic conditions, especially if they are systematically restricted as a
consequence of subservient or subordinate social status.
With respect to the possible effects of frequency on lexicon, we can contrast
borrowing and the processes of pidgjnization or creolization. In the emergence of a pidgin,
7
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
there is only the pull of the emerging pidgin to establish a rudimentary, core vocabulary and
the complicit, uninhibited push from the lexifier (source) language. No possibility exists of
an overt blocking affect from an original (recipient) lexicon because there is no ostensible
(or tangible) competition among lexical items given the separate linguistic identity and
function of a pidgin against that of a native language. There is only one target, the lexicon
of the donor/superstrate. However, if frequency is a main force, it remains to be seen why
the most frequent items (function words and various affixes) are noticeably absent In fact
their absence is even more conspicuous in the beginning stages when frequency would
seem to have its strongest potential affect. Processes of equal or greater force must be
present to over-ride its effects. We do, however, acknowledge that frequency may be one
of a number of factors that figure in the long term in the lexical expansion of a pidgin or
creole. It is also important to note that sufficient evidence comes to us from studies of
creole languages such as Berbice Dutch, Tok Pisin, Kikongo, Lingala, Haitian Creole, and
others, to strongly suggest that there is, in fact, competition among grammatical categories
from substrate languages—acting in relatively covert fashion as morphosyntactic
matrices—that manifests itself in various ways, in some cases along with recognizable
lexical contributions from substrate and adstrate sources.
Due to the multitude of ways lexical and grammatical meanings are represented in
the languages of the world, any notion of formal equivalence would have to be established
according to some kind of gradient scale. The morphological character of each language
will vary predictably, but a scale for each language must be proposed in order to identify
where the two languages may have potential correspondences or mismatches among their
diverse form-meaning sets. Insofar as nouns are consistently reported to be the first and
most frequently borrowed items (followed by verbs or adjectives), perhaps we should
conclude that it is easier for Y nominals to match up with X nominals both semantically and
8
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
formally because nominal classes appear to be more homogeneous across languages. This
leads us to ask how this relatively easy task might be accomplished given the potential
differences of morphological structuring in each participant language (donor and recipient).
Considering variable levels of proficiency in each language, identifying corresponding
nominals would seem to require a relatively low level of proficiency in either (or, perhaps
both) languages. Conversely, in specific areas where we find a greatly decreased
likelihood of semantic and formal correspondence, borrowers would have the greatest
difficulty finding equivalences. The most obvious example would be situations in which
one language encodes a particular relatively opaque grammatical concept with an affix (or,
even a zero or unmarked form) and the other with an individual function word. In such
cases, both semantic and formal correspondence would be difficult to find, though
probably not impossible—especially if some kind of paraphrase or translation is a
possibility.
More sophisticated ability in translation (seeking equivalent expressions) obviously
requires a much higher degree of proficiency in both languages. It is also clear that
bilingual proficiency will fluctuate among members of a given community, producing a
diluting affect that might skew borrowing to areas of greater possible equivalence, reducing
language borrowing to the lowest common denominator, and, perhaps, obfuscating
structural borrowing (one aspect of convergence). Any number of individual forms from
Y, however, will diffuse even to monolingual speakers of X through the agency of more
proficient bilingual members of the community (the most likely conduits of lexical
innovation). As a logical consequence, any and all members of a speech community in
which borrowing is a productive process can actively participate, irrespective of bilingual
ability.
9
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1.1. BORROWING AS BILINGUAL PERFORMANCE
Two key figures in the study of language contact are Einar Haugen and Uriel
Weinreich5. It is to their credit that much of their work still stands as the basis for current
approaches. Although certainly not the first to do so, Haugen pointed out the obvious
difficulties in the use of the term “borrowing”. The recipient language is not expected to
pay back the word; neither can the process be called “stealing”, in that nothing is actually
taken or actually removed from the donor. Despite the inherent inadequacies of such
analogies, a more recent one may better illustrate the character of the processes involved.
Taking a concept from the realm of computers, lexical borrowing can be seen as the
copying of a form from one language system (the lexicon of Y) into another (X)
(Johanson, 1992), with or without all the associated meanings or concepts it typically
expresses in its source language6.
In one of his most cited works, Haugen (1950, pp. 211-220), in an attempt to
clarify then current terminology, divided borrowed elements into a number of classes
depending on phonological and semantic characteristics. For example, he made
distinctions among (a) loanwords—which show the importation of form and meaning with
degrees of phonological integration (all, none, or partial); (b) loanblends—hybrids or
combinations of foreign and native forms, e.g., co-worker (Hartmann & Stork, 1972, p.
133); and (c) loanshifts—in which a foreign concept (meaning) is represented by a native
form. This last term includes “loan translations” (caiques), e.g., English superman from
German Ubermensch (Crystal, 1991, p. 205), and “semantic loans” (semantic extensions),
in which the range of meanings expressed by a native form is extended to include a new,
5 See, for instance, Haugen, 1950,1953.1989 and Weinreich 1953.
6 Obviously, concepts can be imported without their associated labels, as well, in what Haugen (1950)
termed “loanshifts”—discussed in the following paragraph.
10
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
usually related concept, e.g., U.S. Spanish grados “degrees” extended to include the
meaning of English “grades” (Spanish notas) (Silva-Corval£n, 1994, p. 170)7. Relevant to
the present work, loanwords can be further classified into (a) additions, those that provide
labels for objects and concepts newly introduced to the culture8, and (b) substitutions,
those for which forms are already available in the recipient language (relexification) (Albo,
1970). Much attention is paid to (b) because the question naturally follows as to why
speakers of one language would select forms from another when corresponding forms
already exist in theirs. Speakers of a recipient variety must derive sufficient benefit to
warrant the selection and usage of competing forms from a lexical donor.
It is apparent that the phonetic shape of only one morphological unit is taken in
cases of “drastic allomorphy” such as in so-called strong (radical changing) verbs (e.g.,
Spanish tengo < tener) or suppletion (Spencer, 1991, p. 8). This may produce the
appearance of simplification in the recipient language, though interpreting borrowing as a
form of simplification can only be made from the perspective of the donor language and its
speakers and not from that of the borrower. This applies in cases of relexification, as well.
When only the form or label is borrowed, the semantic content is assumed to be more or
less the same as the native word it replaces. However, inherent in this is the possibility of
further semantic splits where both native and borrowed forms exist but their meanings
become more specialized (Sanchez, 1982, pp. 37-40).
Regarding the actual starting point of the borrowing process, for example, whether
or not spontaneous borrowings in the speech of proficient bilinguals are better termed
7 Silva-Corvaldn, 1994 (p. 170ff) refers to these semantic extensions as “single-word caiques”.
8 See, for example, Karttunen & Lockhart (1976), Hill & Hill (1986) and Hill (1988) for reference to
Spanish loans into Ndhuatl (Mexicano), Bavin (1989) for English loans in Warlpiri, and Sdnchez (1982) for
English loans in Chicano Spanish (especially, p. 37ff.).
11
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
nonce borrowings or instances of code-switching (cf. Myers-Scotton, 1993a, 1993b,
1995; van Hout & Muysken, 1994; Muysken, 1995; Poplack & Meechan, 1995), it is
assumed here, that “...every loan starts as an innovation...” (Haugen, 1950, p. 212). It
seems reasonable to assume that some sort of progression exists from speech borrowing to
language borrowing (in Grosjean’s terminology). For the present purposes, forcing
distinctions among terms like “innovation”, “nonce borrowing”, “lexical interference”, and
“single-item code-switeh” does not significantly afreet the course of our discussion (cf. van
Hout & Muysken, 1994, p. 40).
Weinreich (1953) brought to the fore the possible roles that individual members of a
bilingual community play in borrowing by defining “contact” as the situation in which two
or more languages are used alternately by the same speakers (p. 1). As a consequence, the
focus shifts to the mental processes (or “interferences”) that can be inferred to operate.
Accordingly, those showing greater proficiency in the two (or more) languages are
assumed to have a heightened ability and opportunity to draw upon the resources of either
(or all) language system(s). Specifically, Weinreich made general distinctions among
Types A (coordinate), B (compound), and C (subordinate) bilinguals (pp. 9-11). Type A
bilinguals have, in effect, acquired their languages in such a way (in separate
environments) that they appear to possess two distinct linguistic systems. In Saussurean
terms, each language has its own set of signifiers (forms) and signifieds (meanings); viz.,
the forms of each language remain separate with their own associated meanings. Type B
bilinguals have learned their languages in such a way that only one set of meanings
underlies two sets of forms; this may occur when both languages are acquired in the same
contexts. Type C bilinguals, in contrast to both Type A and Type B, can only access
meanings of weaker language forms through their stronger one, effectively succeeding in
certain (limited) communicative functions only when engaging in continuous translation.
12
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
More recent work suggests, as Weinreich was quick to note, that the form recognition
abilities (word memory) of individual bilinguals cannot be accurately described in such
strict terms, i.e., as exclusively A, B, or C. An individual’s representational system(s) that
can affect lexical access may be situated anywhere on a continuum between extremes (i.e.,
from types A to C), determined by such factors as bilingual acquisition history (e.g.,
sequential or simultaneous), levels of proficiency in each language, form type (e.g., certain
types of content items), and so on (de Groot, 1993, p. 46).
These distinctions and the conclusions that can be drawn with respect to the ways in
which the bilingual lexicon might be organized become important when attempting to
understand how such things as code-switching, borrowing, simultaneous translation, and
other abilities that only proficient bilinguals possess can actually occur. A number of
issues broached by Weinreich remain the focus of much current psychoiinguistic research
into bilingualism: (a) Just how closely associated is lexical material from each language
stored—is there one lexical system or two (Hoffmann, 1991, pp. 75-79, Romaine, 1995,
pp. 205-210)? (b) To what degree is each language activated during performance
(Grosjean, 1995, 270-272; Romaine, 1995, p. 98ff.)? (c) How do morpheme types figure
into differential access (Myers-Scotton, 1995, pp. 235-237; de Groot, 1993, pp. 37-41) in
bilingual performance?
l .2. m ix e d L a n g u a g es— l a n g u a g e In t e r tw in in g o r e x t e n s iv e b o r r o w in g ?
Because of their special status in the literature on language contact phenomena, we
include here a brief discussion of so-called mixed languages or bilingual mixtures (c.f.
Thomason, 1997a), for example, Media Lengua, Michif, Copper Island Aleut, and so on.
13
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
These languages appear to consist mainly of the lexicon (or lexical items) of one language
held together and acted upon by the grammatical elements of another. That is, based on the
lexicon, a mixed language might be classified as belonging to one family, and on the basis
of grammar as belonging to another (Bakker & Mous, 1994b, p. 5). The intensity of
contact (both participant languages spoken in the same home and in the same
neighborhood, by members of a single social group) and the relatively small size of the
speech communities undoubtedly play influential roles in the apparent speed with which
such varieties are said to emerge, claimed to be within a single generation in some very
acute cases.
The grammatical system into which lexical items are grafted (the recipient) acts as a
grammatical matrix of the mixture, retaining its original grammatical categories and so on in
some fashion. For example, it may keep its original verbal tense-aspect system and
number, gender/class, and case distinctions on nominals. Imported lexical items appear to
be inserted into syntactic frames generated by the underlying morphosyntactic system, thus
preserving both its grammatical categories and the patterning of lexical and grammatical
elements. Accordingly, the basic character of the recipient language grammar remains more
or less the same while the vast majority of its original lexicon is replaced by the newly
appropriated elements, a process sometimes referred to as relexification9.
In an attempt to define the term mixed language, Bakker & Mous (1994b, p. 5)
offer the following (which is repeated in Golovko, 1994, p. 119 and Bakker, 1997, p.
213):
9 The degree of similarity of the new, mixed system to the original may depend on a number of factors,
e.g., the amount of time speakers of the languages that have mixed have been in contact and the knowledge
these speakers have of the respective languages, the relative status of each language and community, the
types and amount of borrowed material (structural and lexical), and the degree to which grammatical
elements of the languages have been integrated (or intertwined).
14
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(a) bound morphemes (always of a grammatical nature) arc in language A.
(b) free lexical morphemes arc in language B.
(c) free grammatical morphemes can be in either language.
(d) syntax is that of language A.
These four criteria, however, need to be modified somewhat to describe more
accurately many of the the actual results of mixing. First, it is certainly not the case that all
free-standing lexical items can come from language B (the donor). Even taking into
consideration the difficulty of distinguishing among lexical and grammatical elements
crosslinguistically, it still remains to be demonstrated that every content item of a language
can indeed be replaced. This would mean that every single topic, including situations,
events, and all participants, is identified and expressed by a non-native (i.e., not original)
form despite the fact that it is encased in native morphology. Second, given the fact that the
originators of a mixed language were members of a bilingual community exhibiting various
acquisition scenarios and continua of proficiencies in the relevant varieties, we can
anticipate a number of significant effects of contact on the matrix (recipient) system itself.
Hence, we should be somewhat circumspect when accepting both (a) and (d) in their
extreme forms. Only (c), which leads us to expect forms from either language, allows for
the assumption of variation and implies that a mixed language may be mixed at any (or all)
level(s) of grammar.
There have been other attempts to define the term mixed language. For instance,
Weinreich (1953) refers to the work of Rosetti (1945/49), who used “the interpenetration
of two morphologies as the criterion for defining a langue mixte, which he contrasts with a
langue m6lang£e, containing but isolated borrowings” (p. 29). Bakker & Mous (1994b)
speak of a quantitative measure, i.e., in terms of the percentages of borrowed lexical items,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
which, they claim, may be as high as 90% (p. 5ff.)1 0 . For the present purposes, we
propose the following definition, which represents a synthesis of sorts of those above: A
MIXED LANGUAGE is one whose morphosyntactic matrix—including the vast majority of its
most highly grammaticalized form-meaning sets—has essentially survived from an
identifiable progenitor and whose total lexicon (e.g., content items and function words) has
been substantially augmented and replaced (Le., relexified) by a lexical donor, with
borrowings of a more grammatical and/or structural nature to be expected. That is, its
characteristic morphosyntax is inherited more or less intact from an original language
system (indigenous to the specific speech community) while significant numbers of its
content items and, perhaps to a somewhat lesser extent, free-standing function items are
drawn from another, heretofore foreign source.
This definition can be made considerably narrower by adding quantitative criteria,
for example, by proposing that a majority of the native content items must be replaced
along with many function items and some inflectional morphology. However, such factors
can be used only on a case-by-case basis, if for no other reason than one or both languages
may have no inflectional morphology at all (in the case of highly analytical languages, for
instance). It accords with Rosetti’s definition of langue mixte, which points to the
likelihood that true mixing involves grammatical (whether free-standing or not) as well as
lexical elements. It can also be inferred that such items as function words, which occupy
midpoints along continua of forms, can be from either language. In fact, we might suspect
that free-standing function words that have a high degree of semantic independence and
identifiability (i.e., they are not obligatorily selected in response to a syntactic requirement)
10 Thomason (1997a) cites the need for societal levels of bilingualism (to which reference is made later in
Chapters 5 and 6) as a precondition for mixing of this sort. At this point, however, strictly linguistic
criteria are discussed despite the obvious importance of the speakers' perspective.
16
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
may be integrated into the mix in ways similar to content items but with important
implications regarding the syntactic organization of the variety that emerges.
The above may serve as a descriptive account of the essential or necessary
characteristics of a mixed language but, as such, still cannot fully account for the
systematicity displayed by mixed languages in general, which is undoubtedly influenced by
the interplay of the morphological structuring of each participant language relative to each
other (perhaps systematically limiting what form types can or cannot be part of an eventual
mix) and the gradient qualities of the form-meaning sets that are actually borrowed.
Regarding the latter, some words and morpheme types are obviously easier to borrow than
others and are, therefore, borrowed more frequently and in greater numbers (e.g., nouns
relative to, say, adpositions). The current study proposes a set of principles, the Principle
of System Compatibility (or PSQ and its correlate, the Principle of System Incompatibility
(PSI), which set basic limits on borrowable and unborrowable morpheme types (Section
2.3, below); it also argues that the relative numbers of items borrowed from within
individual word classes and morpheme types are predictable on the basis of their formal
and semantic identification characteristics, i.e., how their forms and meanings are identified
and picked out of the speech stream by speakers of the participant languages (Sections 4.1
and 4.2.).
To reflect for a moment on previous work in mixed languages, many studies have
certainly provided much interesting data and thought provoking discussion, especially
regarding the social circumstances in which these varieties have arisen1 1 . However, while
1 1 For example, mixed languages such as Michif, Javindo, Pecu' (Petjo), Island Carib, and possibly
Mednyj Aleut reportedly emerged from mixed marriages of indigenous people and foreign, in many cases
European, settlers (Bakker & Mous, 1994b, p. 8). For more detailed information concerning these
varieties, see Bakker & Mous, 1994a.
17
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
adhering to a rather strict dichotomy between lexicon and underlying grammatical (matrix)
system, they have yet to provide sufficient evidence that the primary mechanism by which
such varieties originate is anything other than extensive lexical and/or structural borrowing.
Consequently, mixed languages may be more accurately seen as cases of extreme
borrowing—a claim that is, in principle, falsifiable. This is not to deny that there are mixed
languages; it merely points to the fact that many sorts of borrowing are, indeed, instances
of language mixing. They clearly offer a unique window into the limits of lexical
borrowing and the nature of a morphosyntactic matrix system due to the clarity with which
the relationships between grammar and lexicon can be examined1 2 . However, it may not be
necessary to posit additional mechanisms or processes. Perhaps, what is actually needed is
a better understanding of the effects of borrowing. It seems amply clear that mixed
languages result from natural linguistic processes that normally occur gradually and over
long periods of time, though these processes can be accelerated by a number of social
factors, e.g., intense contact, simultaneous acquisition, a relatively small speech
community, etc. (see, also, Field, 1998).
Claims that intra-sentential code-switching plays a direct role in the formation of a
mixed language (Bakker, 1997, p. 21ff.; cf. Slabbert & Myers-Scotton, 1997), in our
view, require careful research to explain just how code-switching, which generally
assumes degrees of proficiency in both or all languages involved and the relative autonomy
of each, can congeal into a single, coherent code that operates under only one predictable
set of morphosyntactic procedures (processes or rules) and constraints, however mixed it
might appear to be (e.g., as a result of borrowing and convergence). Positing a separate
process (or amalgam of processes) needs to be carefully done. The appearance of lexical
items with co-occurrence patterns (i.e., word order) of two languages, along with a
12 See Field, 1997a, for a review of Bakker & Mous 1994a.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
blending (or alternation) of phonological and syntactic characteristics attributable to both
linguistic systems is easily traceable to intense lexical and structural borrowing. The
environment in which such extensive borrowing occurs is certainly likely to include code
switching, but that does not necessarily imply that code-switching is a cause—it may, in
fact, facilitate borrowing or merely be another manifestation of similar underlying
conditions.
1.2.1. The concept of matrix language
The term “matrix” is also used in code-switching models that view the switching
process as insertional. That is, one language is the main or base language into which
elements from the other(s) are embedded (Myers-Scotton, 1993a, pp. 3, 20,75). Thus,
the matrix plays the greater role in determining such matters as word order and the
appearance of grammatical (i.e., “system”) morphemes (Myers-Scotton, 1995, p. 235).
Such models stand in opposition to those which assert that code-switching is altemational
and symmetrical, with each language equally taking part in the shaping of the utterance
(Haugen, 1950, p. 211; Muysken, 1995, 177ff.). In this respect, many researchers
consider types of code-switching and borrowing as constituting a kind of continuum of
phenomena, from clear instances of inter-sentential code-switching to various kinds of
intra-sentential switching and from one-time borrowings (nonce borrowings, lexical
interference, and so forth) to fully adapted loanwords (e.g., Poplack, Sankoff, & Miller,
1988, p. 52f; Poplack, 1982, p. 231ff.; Pfaff, 1979, pp. 295-297). As a consequence,
whether or not a particular phenomenon is strictly altemational or insertional is probably
best viewed as situated along a similar, corresponding scale.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The concept of “matrix” is quite easily applied to borrowing, which is clearly
insertional. However, in addition to its function as a morphosyntactic matrix, the recipient
(or borrowing) system will likely operate in phonological and semantic aspects of
language, as well (cf. Haugen, 1950, p. 217 and Weinreich, 1953, p. 39). Emphasizing
the role of the recipient language, Haugen (1950) divides the phonological aspects of
borrowing into two complementary processes: one termed “importation”, and the other
“substitution”, both of which originally referred primarily to the phonemic shape of a
borrowed form. Importation signifies the seizing of a linguistic expression (a morpheme or
aggregate of morphemes) from a donor system by an individual speaker or group of
speakers and its crossing over into the borrowing system (it becomes a member of the
borrowing lexicon). If all we were concerned with were one-time borrowings (which may
retain their original phonological forms), then the process is basically complete and perhaps
transitory. Reaching broader currency requires a form’s diffusability, that is, its general
adoptability (including semantic transparency) to the borrowing linguistic community.
Because precise phonemic correspondence (i.e., equivalence) is rarely (if ever) the
case between two individual languages, the borrowing system must adapt the borrowed
form/label to be consistent with its native phonemic inventory by substitution, at least in the
vast majority of cases (barring the direct borrowing of the donor phoneme). In other
words, individual phonemes constituting the borrowed morpheme(s) are interpreted
according to the matrix phonology and occur in production as substitutions for (or
alterations of) the original segments. Some phonemes are replaced relatively easily as a
result of their more or less exact correspondence in place and manner of articulation with
native phonemes. Others may require perceptually more distant substitutions (for which
examples abound in the literature) that may render the borrowed string (the entire form)
unrecognizable to speakers of the original donor language, e.g., Spanish virgen in Taos
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
pronounced as [m’ ilxina] (Haugen, 1950, p. 215ff). This scenario is consistent with
current views of relexification in mixed languages (cf. Muysken, 1988,1994, and 1997;
van Rheeden, 1994; Bakker & Mous, 1994b; de Gruiter, 1994; and so on).
Loanwords, by definition, are fully integrated into the matrix language system.
That is to say, all fully adapted borrowings must be tagged and organized as belonging to
matrix language form-classes for participation in phonological and other grammatical
processes. Perhaps analogous to the ways in which borrowed elements are phonologically
adapted, the morphological character of borrowed items appears to be analyzed (at least
initially) in accordance with the nearest matrix language equivalents—Y nominals as X
nominals; Y verbals as X verbals; and so on. One question to be addressed, therefore, is
as follows: to what degree must foreign elements be identified with matrix form classes? It
is not unusual to find Y items from one form class reanalyzed as belonging to a different
class in X. For example, some Spanish adjectives (e.g., loco and rico) are borrowed into
Chiricahua as verbs (Haugen, 1950, p. 217). One must first ask whether or not an
equivalent form class exists in X (for the borrowed Y form). An additional (confounding)
factor is semantic: perhaps in Chiricahua, such concepts are traditionally (linguistically)
represented in verbs. This reflects the general possibility that forms belonging to one
semantic sub-type in Y (in this case, belonging to the form class Adj in Spanish) may need
to be reanalyzed as belonging more properly to a corresponding sub-type in X which may
belong to an entirely different form class in Y (in this case, V in Chiricahua).
The current work concludes that morphological equivalence must be interpreted at
the level of morpheme type and not at the level of language particular word classes, which
is significantly complicated by the fact that individual languages construct words in various
and contrasting ways, as any crossiinguistic survey will attest. By casting a wider net, we
21
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
get a broader view of the kinds of forms that are actually borrowed (which in many
instances may be whole words), and, by implication, what forms are borrowable. Again,
mixed language data illustrate the kinds of limits or constraints that can be set on
borrowing. For instance. Media Lengua, with a clear Quechua matrix, allows Spanish
prepositions (the case-marking functions of which are accomplished through suffixation in
Quechua, which has no adpositions). This also holds true modem forms of Mexicano (see
subsection 5.3.2). Neither original grammatical matrix possessed classes of words
properly called adpositions of any kind prior to the incorporation of Spanish prepositions.
Both marked such relational concepts as location, direction, and so on by means of
suffixation or other adverbial elements (Field, 1997b; Hill & Hill, 1986, p. 186). In
addition, the integration of these foreign elements was according to donor word order
patterns (i.e., as prepositions), not according to the position of the affix in the recipient
Perhaps the most plausible explanation is that all three already had general classes of
function words whose members consisted of free-standing grammatical morphemes. As a
consequence, it appears that semantic characteristics (in a broad sense) are more relevant to
any case for equivalence than word class or, perhaps, any other formal characteristics.
1.2.2. The correlation of transfer, substrate, and matrix
Whenever bilingual performance data come under careful scrutiny, the concept of
an underlying (matrix, base, or substrate) linguistic system is always a “reasonable
inference” (Myers-Scotton, 1997a). For this reason (i.e„ for its potential for explanation),
the present work pursues a line of inquiry that assumes links between processes of
bilingual language acquisition and bilingual performance phenomena (e.g., language
transfer, interlanguage, language mixing, etc.). A good deal of recent research in second
22
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
language acquisition (SLA) points to the persistent residual effects of an underlying (native
or first) linguistic system which leaves an unmistakable imprint on the language (second or
secondary) being learned subsequently. The most likely things to be transferred to a
second, non-native language are (a) phonology, which has a physical basis (Gass, 1995,
p. 31; Corder, 1993, p. 19), and (b) deeply ingrained grammatical concepts that can
influence all areas of production (e.g., Selinker & Lakshmanan, 1993, pp. 197-216;
Schachter, 1993, pp. 32— 46), for example, the ways in which reference to past or present
time are expressed. In addition, transfer may involve the classification of words in an alien
language (the target) into syntactic and semantic classes according to classificatory
principles already known (i.e., based on the learner’s LI) and include grammatical
processes (when the learner assumes that native processes and those of the target are the
same or similar) and the transference into the target language of linguistic elements, forms,
rules, and strategies (Gass & Selinker, 1993, p. 234).
In either pidgin or creole genesis, members of a community, which at one time
possessed its own indigenous language(s), were brought into direct contact with a foreign
community and its language(s). Creators (and learners) of pidgin varieties (which, by
definition, have no native speakers) are all in the process of learning a second language
and, therefore, becoming bi- or multilingual. Consequently, the effects of an original
language system (the substrate) are unequivocal; transfer is obvious and visible. Based on
the careful investigation of historical records, many current approaches to creolistics
conclude that the emergence of many creole languages is the consequence of adult SLA,
perhaps along with native language acquisition (NLA) by children of relatively small
numbers—that is, via simultaneous bilingual acquisition with the language(s) of their
parents (Singler, 1993; cf. Field, 1998 and Bakker, 1998). In either case, learner/acquirers
may be a range of possible ages. Discussion of such creole languages as Haitian Creole
23
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
French (HQ (Lefebvre & Lumsden, 1989; Lefebvre, 1986, 1993; Lumsden, 1994) and
Tok Pisin (Siegel, 1997; Reesink, 1990; Miihlhausler, 1990,1980) often focus on the
interaction of an underlying substrate and superstrate ( le x ic a l donor), especially with
respect to phonological, syntactic, and lexical (semantic) characteristics (cf. Holm, 1986,
p. 261-4; Mufwene, 1986; Alleyne, 1986, p. 303ff.)13.
When historical investigation reveals that the variety forming the basis (i.e.,
grammatical matrix) of a creole most likely emerged as the consequence of L2 acquisition
(perhaps a composite of LI and L2), we would also expect evidence of language transfer
from an LI or aggregate of underlying, substrate languages1 4 . Linguistically, the end
product under all possible circumstances will, nevertheless, depend on the amount and
quality of exposure. When a creole has completely displaced the original language(s) of the
community, descendants of the founding population may then become monolingual in the
creole, and they will have learned the creole natively. Perhaps to state the obvious, in cases
where the first language(s) of creole speakers have disappeared completely from their
linguistic repertoires (individually and collectively), shift has taken place, probably quite
rapidly and under a good deal of social pressure (often under very adverse conditions).
The visible presence of a matrix in a mixed language suggests a number of things.
First, if the mixed language emerged as the result of the sequential acquisition of two (or
13 Overt morphosyntactic features attributable to specific substrate languages often appear to dissipate
completely, leaving behind only subtle (covert) traces whose actual origins can only be inferred. As a
consequence, substrate identification may be rendered quite problematic. In fact, the lack of clear and
unambiguous evidence for particular substrate languages has generated much controversy in creolistics,
providing considerable grist for important alternative theories of their origins, for instance, those defended
by proponents of so-called universalist positions (e.g., the language bioprogram hypothesis of Bickerton,
1981,1984,1987, and so on).
14 The fact that creole lexicons are typically composed of donor content items (including core vocabulary)
and free-standing function words despite the (near) complete lack of donor inflectional morphology appears
to support the idea that full acquisition was not attained, which is precisely what we would expect in
processes of secondary acquisition.
24
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
more) languages (on a community, hence, individual level), the one acquired first is more
likely to form the grammatical matrix. It also implies that this first language was acquired
through normal processes of transmission—in the home, among family members, and so
on. The language learned subsequently very likely was (or was becoming) culturally
dominant within the community at large, and was most likely learned under an entirely
different set of circumstances. Each language in such a bilingual’s repertoire is typically
acquired and used in complementary ways, that is, with different people, in different social
domains, for different purposes (Grosjean in press). It is also a possibility that the
participant languages were acquired more or less simultaneously, but, even in this case,
one is the language of the home and primary in some sense (see Cutier, 1994).
Mixed languages may also be the consequences of processes operating during
language shift, a situation in which a community is shifting away from an original, ethnic
language to the use of a culturally dominant one, and, therefore, represent attempts at
preserving an original language and the ethnic identity associated with it. However, shift
obviously cuts two ways, each showing the effects of an underlying matrix or substrate
system. For instance, in the language that was acquired first, we expect borrowings and
the effects of attrition in terms of numbers of speakers, functions, and forms (see, for
example, Dorian, 1989). In the language being adopted, the culturally dominant variety
that has been acquired as a second language, we anticipate evidence of transfer, perhaps to
the extent that the adopted language resembles an individual interlanguage1 5 . The evidence
clearly suggests that transfer can have community-wide effects, especially when that
15 See Lipski (1994) for a comprehensive discussion of contemporary Latin American varieties of
Spanish—some of which he terms interlanguage varieties. Significant substrate influence, unquestionably
the result of transfer from original indigenous Amerindian languages, is quite transparent This is also
likely in the case of particular varieties of English, for instance, Hiberno-English (see, for example, Harris,
1985; Thomason & Kaufman, 1988, pp. 42-43,47-48), Indian English (Hock & Joseph, 1996, pp. 375-
380), Chicano English (cf. Penfield & Omstein-Galicia, 1985, pp. 34-36) and so on.
25
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
community is relatively homogeneous (i.e., when all members speak the same first
language).
Each of the above possibilities (transfer in processes of SLA, substrate in pidgin
and creole genesis, and matrix system in various mixing phenomena) reinforces our
original premise: there are significant connections among different acquisition scenarios,
the nature of the bilingual lexicon (including representation and lexical access), and various
bilingual phenomena—which, we shall show, includes the emergence of entirely new
varieties (whether or not they can be truly classified according to a family tree model). As
an important consequence, all have the potential to shed light on the language faculty, itself.
In any situation of intense contact, borrowings of various kinds and conversational (intra-
sentential) code-switching may be quite common in the speech of community members
who are under similar social and linguistic circumstances (see subsection 6.3). And, just
as performance errors and other apparent anomalies by native language acquirers illustrate
how language is constructed anew by each child (cf. Slobin, 1985, p. 1158ff), the
integration of features and elements of two entirely different language systems into one
(that systematically diverges from standard usage in either participant variety) may show
the ways language is constructed anew in communities characterized by such intense
cultural and linguistic contact. This, in turn, has the potential to shed light on the ways
bilingual competence (e.g., during acquisition) can be represented.
26
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1.3. t h e Pl a n o f T his Bo o k
A number of studies have proposed hierarchies of borrowability (nouns arc
borrowed more frequently than and before verbs, for instance), including some whose
applicability have been restricted to the specific cases from which they were derived
(Whitney, 1881; Haugen, 1950; Singh, 1981; Muysken, 1981; van Hout & Muysken,
1994, p. 41). The many counter-examples present in a wider range of data have made the
generalizability of individual hierarchies somewhat problematic, though clearly there must
be something underlying their fundamental similarities. Moreover, hierarchies denoting
borrowability bear striking resemblances to those proposed in work on morphological
typology (the synchronic analysis of word and morpheme types), e.g. Sapir (1921),
Greenberg (1974), and others, and later applied to grammaticalization (the diachronic),
e.g., Hopper & Traugott (1993), Heine etal (1991), C. Lehmann (1986), and so on. The
similarities among the various indices, scales, clines, or hierarchies suggest
correspondences that go beyond mere coincidence.
A major goal of this study is to account for these correspondences. To this end, the
approach taken here is based on the following: the similarities among the various scales
and hierarchies are likely to be the result of the general nature of language—i.e., the
morphological types that actually occur and their characteristics. Individual differences are
likely to be the result of language particular formal and semantic factors. A viable theory of
borrowability, therefore, must be constructed to include both the general and particular,
along both formal and semantic lines. As a first step, Chapter 2 discusses morphological
structure and its role in comparative linguistics, the diachronic study of grammaticalization,
and the establishment of scales of borrowability. While hierarchies representing
quantitative and/or temporal claims to borrowability may elucidate which forms are
27
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
borrowed (and, therefore, made compatible with the borrowing morphosyntactic matrix),
they, nevertheless, fail to explain why other forms are not. We propose an absolute cutoff
point along a scale of morpheme types relative to the typology of both or all the languages
involved beyond which forms cannot be borrowed (as embodied in the Principle of System
Incompatibility); hence, they are not borrowable. Truly borrowable forms can only be seen
in an absolute sense against forms that are (absolutely) unborrowable.
Hierarchies also accurately illustrate observations that members of certain
borrowable form classes are borrowed more frequently than those of other form classes
(e.g., nouns more than either verbs or adjectives) and that members of particular semantic
sub-types within a general class are borrowed more frequently than those of other sub-
types (e.g., concrete versus abstract nouns). Hopper & Traugott’s (1993) cline of
grammaticalization represents a path along which a content item travels in its evolution to
grammatical element (function word, then affix). One directional aspect of this cline
involves the diminution of form and the other loss or change of semantic content (so-called
“semantic bleaching”). This, in turn, provides the motivation for discussion (in Chapter 3)
of both form classes and semantic types and their possible roles in borrowing. Potential
points of conflict are, therefore, indicated, suggesting that form and content conspire in
processes of borrowing.
Chapter 4 continues this line of reasoning and proposes that explanation lies in the
gradient notions of perceptual salience and semantic transparency (as herein defined).
Referring to studies of native and secondary language acquisition (e.g., Slobin, 1985), it is
postulated that any language learner’s task is to map forms in the speech stream onto
associated concepts or meanings. Clearly, an individual grammar provides a particular
form (through the interface of syntax and phonology) with relative prominence based on its
28
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
significance in an utterance (Caplan, 1992, p. 338). Perceptual salience, therefore, refers
to the characteristics of a form and the degree to which it can be recognized and isolated,
corresponding to the implications contained in clines of grammaticalization which suppose
that morpheme types generally consist of significant degrees of phonological form.
Semantic transparency, defined broadly as the successful linkage of form to meaning,
includes degrees of perceptual salience, but it also introduces types (hence, degrees) of
meaning or semantic content—therefore, the association of the concept of semantic
bleaching with semantic types. Because individual languages link form and meaning in a
multitude of ways—viz., they distribute meanings along indices of synthesis and fusion,
and among the various form classes and semantic types in accordance with the
morphological structuring of their lexicons, the transparency of meaning of a particular
form from Y (to speakers of Y as well as to speakers of X, given the assumption of
bilinguality) is seen as crucial in its relative borrowability into X.
Chapter 5 applies the insights and principles gained from Chapters 2— 4 to an
analysis of data from Modem Mexicano (N£huatl), a native American (Amerindian)
language of Mexico heavily interlarded with borrowings from Spanish—said to be in the
range of 60% of its lexical material. Mexicano is one of the most thoroughly documented
of the Amerindian languages (as a result of careful record keeping from the very beginning
of the Spanish Conquest), which provides this study with two positive rewards. First,
temporal claims of borrowability can be examined in light of historical evidence. While we
are once again limited by the fact that only one contact situation is under study, there is,
nonetheless, the potential to disprove certain claims (e.g., that certain grammatical affixes
are not borrowable). Second, the effects of length and intensity of contact can be
investigated with regard to the amount and types of borrowing one expects to occur. It is
29
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
obvious that after nearly five centuries of close contact, there will be a great deal of
convergence, borrowing, and so on.
The concluding chapter (Chapter 6) is a discussion of the proposed lexical-
typological model (including principles of system compatibility and incompatibility and
notions of hierarchies of borrowability) against a backdrop of language contact phenomena.
It centers around the programmatic nature of the model and its extension to other contact
situations, including code-switching and the emergence of mixed languages (including
creoles). Hypothetical contact situations of various possible recipient (X) and donor (Y)
languages are discussed along with specific examples from known contact situations.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2.0 . C h a p t e r 2— Mo r ph o lo g ic a l St r u c t u r in g a n d Sy ste m Co m p a t ib il it y
This chapter begins with a brief discussion of the analysis of morphological
structure and the ways it has been applied in various branches of linguistics, from
comparative and descriptive studies to those concerned with language typologies and the
establishment of language universals. The kinds of distinctions that have been formulated
over the years have provided researchers involved in a variety of disciplines with the
terminological wherewithal to observe and discuss patterns that emerge in the various
processes and phenomena they investigate. Obviously, accurate descriptions of processes
and the elements affected by them are a necessary prerequisite towards explanation.
Relevant to the present work, the terms derived from descriptive and typological studies
have made significant contributions to the many recent advances in the diachronic study of
grammaticalization and to work in bilingual/contact phenomena, including lexical
borrowing, code-switching, and the emergence of entirely new language varieties.
The analysis of words into types of morphemes (free-standing versus bound, as
well as word, root, stem, and affix) has remained fairly constant in recent years, although
there still may be some controversy concerning the general criteria for distinguishing
among certain types of bound morphemes, e.g. between derivational and inflectional
affixes, and so on. There is also some discussion regarding which types of elements figure
most significantly in language contact phenomena: should we look at open- versus closed-
class or lexical versus function elements? While there is an obvious high correlation
between the kinds of entities to which these terms are meant to refer (e.g., lexical items
usually belong to the open classes and functional elements are normally members of closed
classes), some researchers prefer a third set of terms, for instance, content versus system
morphemes (Myers-Scotton, 1993a and 1993b), which takes note of the fact that some so-
31
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
called grammatical words act like lexical items. This usage is at least partly due to
observations that some independent function words are selected for semantic reasons and
participate in thematic role assignment, a function usually reserved for lexical items.
Moreover, in languages such as English, pronouns may exhibit NP-like behavior by filling
slots as grammatical subjects, objects, and so on, and modal auxiliaries can take the place
of entire VPs.
From a Sapirian perspective, it may be difficult to imagine a complete and
intelligible sentence composed of only so-called function words (“I do!”) and no content
(concrete or radical elements) items, though he did recognize the possibility (1921, p. 93).
Nevertheless, with these caveats in mind and to avoid creating additional terminological
confusion that might possibly obscure subsequent discussion, the present work will stick
to the use of terms more typical of studies in borrowing and grammaticalization, even
though their definitions may appear a little squishy to those accustomed to clear,
unequivocal categorical distinctions; but, given the gradient nature of the phenomena being
treated, such terms may be hard to come by and difficult to sustain. Attempting to preserve
more traditional terminology, we will use the terms lexical and content (item) more or less
synonymously, as well as the terms grammatical and function (word or element). Of
course, examples will be given in the case of ambiguous reference.
One of the express aims of this work is to account for hierarchies of
borrowability—why some things are borrowed more frequently than others. An additional
goal is to demonstrate that, while many of the hierarchies discussed here may, indeed, be
cognitively based (owing their characteristics to such recognition and production factors as
perceptual salience, semantic transparency, and so on), they cannot reconcile two
apparently contradictory observations: anything is or should be, in principle, borrowable,
32
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
but not everything is borrowed, especially to the same extent Haugen (1950, p. 224)
states, “All linguistic features can be borrowed, but they are distributed along a SCALE OF
ADOPTABILITY... [emphasis his]”. It is true that all sorts of elements have been borrowed
at one time or another (see, for example, Thomason & Kaufman, 1988, p. 83ff.). It seems
that whatever is leamable should be borrowable (see, also, Hudson, 1980, p. 60 and
Bynon, 1977, p. 255). At least a portion of the total number of bilingual speakers of the
two (or more) languages participating in contact situations will be proficient enough in both
(or all) to be able to draw freely upon the resources of either (all), viz., to use their
elements, structures, and processes alternatively or simultaneously (e.g., in inter-sentential
or intra-sentential code-switching, or for short- or long-term borrowing), for whatever
sociological and/or psychological reasons that might be present
2 .1 . SCALES, INDICES, HIERARCHIES, AND CLINES: CONTINUA OF FORMS AND
MEANINGS
Formal distinctions such as those established in comparative and typological studies
allow us to observe the ways language systems are organized and to identify what aspects
of language are involved when it is said to change. Regarding the nature of linguistic
systems, Sapir (1921) helped turn attention away from languages as “wholes” (to be typed
variously according to clusters of morphological properties) and onto the significance of the
word in generalizations of the structure of language as a mental faculty; he noted that
language “stmggles towards two poles of linguistic expression—material content and
relation—and that these poles tend to be connected by a long series of transitional concepts”
(p. 109). Sapir’s work became the seed for many conceptualizations of language that have
33
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
followed. Many of his basic distinctions have endured into studies of grammaticalization,
language acquisition, and language processing.
The conception of a continuum or index of forms assumed a fundamental
correspondence of independent words (or radical elements) and basic (concrete) concepts.
It is at the other pole, with regard to secondary or relational concepts, that we find that they
are “sparsely developed” in some languages and “elaborated with a bewildering
exuberance” in others (Sapir, 1921, p. 95). It is precisely in the matching of meanings and
forms that we find that languages possess their own unique logic, or “genius”. This is true
when looking at them synchronically as coherent systems and diachronically, as always in
the grip of change, for both form and meaning are “ceaselessly changing” (p. 98); what is
evidenced now is the result of previous processes. The different scales, indices,
hierarchies, and clines are fitting conclusions to these insightful observations, concerning
both synchronic states and diachronic processes.
2.1.1. The indices of synthesis and fusion
In the synchronic analysis of language, we see that meanings are distributed among
the forms available in a language in ways which are particular to that language. Some
elements may consist of direct (one to one) mappings of specific concrete meanings
(concepts) onto discrete phonological forms, while others appear to have no actual, primary
meaning (or definition) at all, except to indicate purely grammatical concepts such as tense,
number, case, and so on. The description of how languages distribute kinds of meaning
among forms constitutes one basis by which they have traditionally been classified. For
example, the fundamental character of some varieties can be distinguished along an index
34
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
of synthesis, i.e., of isolating-analytical versus polysynthetic character. An isolating
language (e.g., Vietnamese, Mandarin, etc.) is one that exhibits the maximal correlation of
morpheme and word. That is, each word consists of one, discrete morpheme representing
a single semantic concept or distinction with no affixal morphology whatsoever. At the
other end of this index, a continuum of language types, are polysynthetic languages
(including a subset that are incorporating) characterized by relatively long words composed
of a number of morphemes, from one to many (e.g., Inuit varieties such as West
Greenlandic and such Algonquian languages as Cree).
A second index, that of fusion, types varieties according to agglutinating versus
fusional character, that is, according to how much lexical or grammatical information can
be represented in one morpheme. On the one hand, words in an agglutinating language
(e.g., Turkish) typically contain a sequence of discrete morphemes. However, distinct
boundaries exist between individual morphemes, and each one expresses only one lexical
or grammatical concept On the other hand, in a fusional language (e.g., Spanish), such
clear boundaries may not exist A particular morpheme may represent a number of
categories that have fused into one, unsegmentable (morphophonological) unit To
characterize the diversity of the world’s languages, most (if not all) are best typed
according to each of these indices, especially in view of the fact that an individual language
will occupy a position relative to other languages between the extremes of both indices
(Sapir, 1921, p. 120ff.; Comrie, 1989, p. 47).
Within an individual language, lexical and grammatical distinctions can be
represented in diverse ways. For example, in English, which is more isolating and less
fusional than, say, Spanish, some grammatical forms are expressed by individual words
while others are expressed through inflectional affixes (e.g., comparatives expressed by the
35
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
affix -er versus those by the independent word more). Crosslinguistically, the lack of
precise correlations of form and meaning creates potential mismatches in such areas as the
translatability of individual lexical or grammatical items from one language to another (see,
e.g. Croft, 1990, pp. 11-18). To illustrate the often subtle effects of possible mismatches,
some researchers consider the ways words are structured in the lexicon of a first or native
language (morphological structuring) to be significant factors in bilingual comprehension
and production and the relative ease with which a second (or subsequent) language is
acquired. It is apparent that the more a second language resembles the first in the ways that
words are structured, the easier it is to learn (Ard & Homberg, 1993, pp. 62-3; Green,
1993, p. 250; cf. Schreuder & Weltens, 1993 and Frauenfelder & Schreuder, 1992).
In general, morphological structuring may have a number of consequences on the
lexicon of a specific language. For example, a more isolating language may have a
relatively large number of discrete content items—i.e., content-bearing roots and words
(e.g., the Romance languages), while a (poly)synthetic language may have a relatively
small number of these (e.g., Inuit varieties). In languages of the former type, referential
capacity may be realized through sheer numbers of content words and morphemes, while in
the latter types, it may be accomplished morphologically through the systematic application
of larger numbers of derivational and inflectional affixes to a more limited number of roots
or stems. What constitutes synonymous or antonymous expressions in one language may
be accomplished through a diversity of radical elements—in Sapir’s terminology, and in
another by means of morphological complexity. This will obviously be reflected in the
paradigmatic relationships among items within particular semantic domains in their
respective lexicons. In an isolating language, such a paradigm will most likely consist of a
variety of distinct roots. In a highly synthetic language, corresponding paradigms may
consist of a single root attached to a variety of affixes; borrowing a separate root from an
36
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
outside source may produce competition among forms and eventually act to break up a
paradigm (one of many possible structural effects of borrowing). It seems clear, then, that
patterns of borrowability will depend on the ways specific languages structure their
lexicons flexical structuring).
2.1.2. The cline of grammadcality
A number of other scales or continua of forms and/or meanings have been proposed
in diachronic studies, as well. Hence, one finds in much work on grammaticalization a
continuum of meanings, from purely lexical (with clear sense and reference to unique
objects) to purely grammatical (which represent language-particular grammatical
categories—from one to a fusion of several). This continuum of meanings, in turn, is
represented by a corresponding continuum of forms and structures that extends from free
standing forms (morphemes) to bound forms (including zero forms) and structures or
positions within structures. Midpoints between purely lexical (concrete or primary) and
purely grammatical (relational or secondary) include elements expressing a range of less
concrete semantic distinctions, e.g., certain types of modifiers (e.g., quantifiers) and
expressions of mood or modality whose meanings are secondary (or relational) to the
forms they usually accompany—in the case of the former, nominals, and in the case of the
latter, verbals.
Among researchers specializing in grammaticalization theory, Heine etal (1991a)
apply this reasoning as a basis for their descriptions of the gradual “thinning out” of
meaning (referred to as “semantic bleaching” in earlier studies) in historical processes of
grammaticalization (pp. 9-10). In a similar vein, Hopper & Traugott (1993) focus on the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
inherently gradient character of these continua and the dichotomy between meaning and
form implied in Sapir’s statement in their notion of the cline (or scale) of grammaticality (p.
7):
(1) content item > grammatical word > clitic > inflectional affix
This cline represents the process of grammaticalization, the gradual historical development
of a grammatical form from content word to inflectional affix, as exemplified in (2), below:
(2) Vulgar Latin *amare habeo > Spanish amare
Moving from left to right, there is a gradual diminution of form, from free-standing,
autonomous word to grammatical affix. There is also a concomitant lessening of semantic
content (“semantic bleaching”), or, rather, change from narrow lexical meaning to broad
grammatical meaning. As a consequence, the meanings associated with linguistic
expressions are distinguished in ways similar to the forms of language, as oppositions of
lexical (or content) versus grammatical (or functional) meanings. The cline of
grammaticality serves to illustrate that the various form and meaning types evinced in a
particular language constitute a continuum whose individual members may be quite
indistinguishable except on the basis of an abstract set of formal and semantic
characteristics.
Terms like “continuum” or “cline” refer to some kind of abstraction, a theoretical
line containing certain focal points at which clusters of formal properties may occur
(Hopper & Traugott, 1993, p. 7). Cross-linguistically, boundaries between such focal
points may seem quite arbitrary—one language makes a distinction in one place while
38
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
another makes it elsewhere. However, in an individual language, relevant characteristics
marking the relative positions of various form classes on a scale or index are, nonetheless,
somewhat less subject to dispute (for example, the distinction between adposidon and
inflectional affix). A cline of grammaticality also captures the general insight that there is
an opposition of specific, concrete lexical meaning often expressed by words or other
“radical elements” and an increasingly generalized and abstract set of meanings expressible
by grammatical elements. Even though, in a particular language, there is a high
coincidence of discrete (content, open-class) word and transparent lexical meaning on one
end of the scale and inflectional affix and grammatical information on the other, there is a
significant lack of exact correspondence across languages at potentially matching points
along the length of each continuum.
This conceptualization of a cline brings to the discussion the natural separation of
form and meaning1 . What may appear to be an idiosyncratic, perhaps random allocation of
(a) a range of meanings and meaning types on to (b) a variety of forms and form types
turns out to be quite systematic, nevertheless. In addition, the mapping of form onto
meaning (and vice versa) surely indicates that they are, indeed, distinct and separable.
Consequently, meaning and form are more properly investigated individually, bearing in
mind that they are inherently and inextricably linked in a particular language. In fact,
current psycholinguistic research is uncovering the many ways that forms and concepts can
be accessed independently by the individual language user (Kroll & de Groot, 1997, p.
171). When investigating a previously unknown language variety, especially during its
initial stages, a precise linkage would be unpredictable and would have to be apprehended
through principled suppositions and on the basis of actual evidence.
1 Much recent work towards the establishment of monolingual and bilingual models of lexical
representation stress the need for each of these to be treated in separate, yet connected ways (see, for
example. Smith, 1997 and Kroll & de Groot, 1997).
39
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
When attempting to establish cross-linguistic bases for equivalence of any kind, it
may be helpful to consider that the exercise of translation, which often involves renditions
ranging anywhere from word-for-word interpretations to broad paraphrases, rarely
involves exact formal equivalence; it is typically based on a number of ad hoc factors
employed for the sake of expediency (cf. Gutknecht & Rolle, 1996, p. 2). Potential
mismatches and incongruence of types of form and types of meaning are especially
conspicuous in the comparison of two separate language systems that differ widely with
respect to such lexical structuring (e.g., that are typologically situated at differing points
along the indices of synthesis or fusion). If, within an individual language (X), there are
formal criteria that provide distinctions among elements, and these criteria prevent us from
equating function words and inflectional morphology, then we have even less of a basis to
equate, say, adpositions in one language (Y) with affixes in another (X), except by positing
some very general and perhaps abstract semantic correspondence (e.g., observations that
various forms share particular functions); such attempts may be very difficult to defend on
any formal basis. Moreover, the formal operations applying to affixes as opposed to
independent words are obviously distinct in individual languages; this fact alone may be
even more telling when attempting to compare or contrast forms from distinct systems.
2.1.3. The cline of lexicality
The cline of grammaticality, stated in (1) above, portrays formal aspects of the
process of grammaticalization in the gradual progression of a form from content item (word
or root) to inflectional affix and captures the observation that a continuum of forms exists
synchronically (at any one point in time) within a single language. In the same way,
Hopper & Traugott (1993, p. 7) illustrate the gradient quality of semantic content and the
40
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
gradual evolution of a concept from one that is strictly lexical (its occurrence restricted by
discourse factors such as choice of topic) to one that is increasingly grammatical (its
generalized, abstract meaning enabling it to occur in a greater number of contexts) with
their cline of Iexicality in (3), below:
(3) a basket full (of eggs...) > a cupful (of water) > hopeful
A number of researchers working in grammaticalization studies mention an increase in
abstractness (though this does not necessarily lead to grammaticalization), but also describe
the semantic processes as a gradual emptying or loss of semantic complexity, pragmatic
meaning, syntactic freedom, and phonological substance (Heine & Reh, 1984, p. 15;
Traugott & Heine, 1991, pp. 3-5; Traugott & Konig, 1991, p. 189ff.; Greenberg, 1991,
p. 301ff; Hopper & Traugott, 1993, pp. 68, 87-93; Heine etal, 1991a, pp. 39-45). It is
not simply the case that the concept represented by a particular word or morpheme is
merely fading to nothing; once again changes are gradual and quite likely invisible to
strictly synchronic analyses.
Regarding the most likely elements to undergo grammaticalization, recent cross-
linguistic studies of change have uncovered considerable evidence that only certain sub
classes of lexical items within fairly restricted semantic domains are potential candidates.
Often cited examples are adpositions from body parts or verbs of motion, tense and aspect
markers from spatial expressions, modals from possession terms, and so on (Traugott &
Heine, 1991, p. 8). It is also evident that the progression from one type of form to another
is step-by-step rather than in leaps and bounds across larger formal or semantic domains
(Heine etal, 1991a, pp. 112-113; Heine etal, 1991b, p. 161ff). Two additional
examples, cited below in (4) and (5), characterize the path of grammaticalization of noun-
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
to-affix and verb-to-affix, and serve to illustrate gradual changes in meaning and a
concomitant diminution of form. The noun-to-affix cline (C. Lehmann, 1986, p. 3),
however, shows that there is not only the loss of specific, concrete meaning, there may
also be an increase in the amount of abstract, grammatical information contained in a single
form. That is, an individual form may become the repository of a number of concepts (that
may have previously been represented individually) that have coalesced (or fused together)
onto that form (perhaps a phonological remnant of fusional processes).
(4) relational noun > secondary adposition > primary adposition > agglutinative case
affix > fusional case affix
This gradual decrease in lexical content is also illustrated in the verb-to-affix cline (Hopper
& Traugott, 1993, p. 108):
(5) full verb > (vector verb >) auxiliary > clitic > affix
An often cited example of the latter is the development of an inflected future tense (in
various Romance languages) from a verb + have construction (Hopper & Traugott, 1993,
pp. 42— 5; Bynon, 1977, p. 249) as represented in (2) above, repeated as (6):
(6) Vulgar Latin *amare habeo > Spanish amard
A scale, or cline, based solely on lexical meaning generalizable from (4) and (5)
above can now be expressed as (7):
(7) primary lexical meaning > secondary semantic distinction > single grammatical
category > fusion of categories
42
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The point at which the processes of grammaticalization eventually stop may be
determined by the morphological character of the individual language (see, also, Bybee,
1995, pp. 227-229). For example, in the case of postposition to case affix, pictured in
(4), above, such developments have yet to be documented in a language of the isolating-
analytic type, prompting claims that language internal processes of grammaticalization are
not likely to motivate the development of inflectional morphology (of any kind) (Traugott &
Heine, 1991, p. 9). Presumably, there is an intermediate stage in which clitic elements
gradually evolve (perhaps as a result of phonological processes). However, this would
entail the (more or less) simultaneous emergence of obligatorily expressed inflectional
categories, as well.
2.1.4. A scale (or continuum) of morpheme types
To sum up the discussion of clines of grammaticality and lexicality to this point, it
is evident that the units (morphemes or words) that link form and meaning, that are
involved at all points in the construction of language, nevertheless, consist of a fairly
restricted set of general forms and form types which combine to form utterances in highly
idiosyncratic, language-particular ways, for instance. Moreover, it is possible to generalize
further from the clines in (1) - (7) above and propose an additional hierarchical ordering of
forms (i.e., at any point in time) to represent the relevant morpheme types available to any
language. Therefore, the Scale (or Continuum) of Morpheme Types (stated in (8), below)
is proposed. This hierarchy, consistent with both the grammaticalization clines of C.
Lehmann (1986), Heine et al (1991a), and Hopper & Traugott (1993) and with observed
language particular patterns of borrowing (discussed in Section 2.2 below), is set forth as
follows:
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(8) independent word, bound root > agglutinating affix > fusional affix
Though the scale of morpheme types presented in this way is similar to the cline
represented in (1), above, the emphasis has shifted to reflect a somewhat wider view of
language types and the types of morphemes they consist of. Specifically, on the far left,
the category “independent word, bound root” is inclusive of all so-called content items
(independent words and contentive roots2) and function words (likewise, words and roots)
that can be realized as independent words in their respective languages. These are the
particular morpheme types permitted in isolating-analytical languages and consequently
realized as discrete words. Agglutinating languages have these (independent words and
roots) and types one position to the right, and fusional languages have the full complement
of morpheme types.
2 .2 . HIERARCHIES OF BORROW ABILITY
Turning now to issues of borrowability, the striking parallels between (diachronic)
clines of grammaticality and lexicality, on the one hand, and (synchronic) hierarchies of
borrowability, on the other, should become immediately evident Taken together, there is a
rather obvious inference that degrees of grammaticalization and degrees of borrowability
are somehow intimately linked. This, in turn, suggests that the parallels are much more
than coincidental and points to underlying causes. There appears to be substance to
2 In the Sapirian sense, these are “radical elements” or content-bearing morphemes of a language that can
be distinguished by its speakers, whose linguistic knowledge includes the identification of “...words,
significant parts of words, and word groupings” (Sapir, 1921, p. 25ff). Whether particular radical elements
are actually realized as independent words or not (the vast majority may not be in synthetic languages), they
are, nonetheless understood and acquired as identifiable, non-derived stems (see Fortescue & Lennert Olsen.
1992, especially pp. 136-7 for a discussion of the acquisition of polysynthetic morphology in West
Greenlandic).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
observations that the more structural (or grammaticalized) an element is, the less likely it
will be borrowed from one language to another.
It has long been noted that some linguistic elements are borrowed more freely than
others. The linguist most often cited as the first to make this observation was the
Sanskritist William Dwight Whitney, who in 1881 noted that nouns are the most frequently
borrowed elements of language, followed by other independent words (“other parts of
speech”), then suffixes, inflections, and individual sounds (in that order). A hierarchy
based on these orderings would appear as (9) below (van Hout & Muysken, 1994, p. 41):
(9) nouns > other parts of speech > suffixes > inflections > sounds
Haugen (1950) suggested a similar ordering in his scale of adoptability, based on a
synthesis of data from American Norwegian and American Swedish (p. 224):
(10) nouns > verbs > adjectives > adverbs, prepositions, interjections
In this scaling, nouns are borrowed more frequently than verbs, and verbs more frequently
than adjectives, the latter an order not reflected in a report on English borrowings in Hindi
(Singh, 1981, cited in van Hout & Muysken, 1994, p. 41), illustrated in (11):
(11) nouns > adjectives > verbs > prepositions
In general, such hierarchies illustrate borrowing patterns that are specific to certain
contact situations, and they are consistent with the statement that speakers of subordinate
varieties borrow from a dominant variety content items more frequently than grammatical
45
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
items and grammatical words more frequently than inflectional affixes (Comrie, 1989, pp.
209-210). In (9) above, there is a much broader generalization that includes among form
types nouns at one extreme end of the scale and inflections at the other. In (9), (10), and
(11), nouns occupy the same positions (as the most often borrowed), but the discrepancy
that appears in (10) and (11) among content items concerns which form class should follow
nouns with respect to frequency (and so on) of borrowing (discussed below).
2.2.1. Borrowing hierarchies: lexical items
It is also possible to make finer distinctions among each of the three categories
(lexical items, grammatical words, and inflectional affixes). For instance, all hierarchies
appear to agree that the most likely content items to be borrowed are nouns, followed by
either adjectives or verbs. Similar observations can be made concerning the relative
borrowability of diverse types of function words3, as well. It is very likely that
explanations for many potential discrepancies (e.g., whether verbs are intrinsically more
borrowable than adjectives) will be found by closely comparing lexical structuring and the
allowable form classes in each of the participant languages. Hence, while nouns (labels for
people, places, things, and so on) are reportedly the most frequently borrowed in all cases,
what comes next in a proposed subhierarchy of content items may vary. Consequently, the
following subhierarchy is proposed:
(12) nouns > adjectives, verbs
3 Van Hout & Muysken, 1994 investigates borrowing preferences of Spanish content and function words
into Bolivian Quechua and make a number of distinctions among both classes of words.
46
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Some have suggested that verbs follow nouns in frequency of occurrence as a result
of their relative semantic and syntactic complexity. Whether verbs precede or follow
adjectives in a particular hierarchy “may just reflect the distribution of grammatical
categories in native-language materials rather than the propensity of specific items to be
borrowed” (Romaine, 1995, p. 65). For example, some languages (e.g., Cree and other
Algonkian languages) have no adjectives; attributes of nouns are expressed through verbs,
relative clauses, and so on. This may make borrowing adjectives from a donor language
that may be rich in adjectives (or, “adjective rich”) problematic, though not necessarily
impossible4. In addition, it may be possible that the complexity of the processes that lead
to the highly synthetic (and incorporating) verbal morphology of some languages inhibit the
incorporation of borrowed verbal roots (Bakker, 1994, p. 21). Irrespective of the status of
adjective versus verb, however, nouns (or nominal roots/stems) are the more frequently
borrowed of any other class of content item.
2.2.2. Borrowing hierarchies: grammatical items
The hierarchy in (9) also points to the fact that languages may also borrow
grammatical elements; more specifically, languages are known to borrow independent
function words and different types of inflectional affixes though not as frequently or
extensively as content items. Characteristically, borrowing hierarchies (to date) identify a
general category of inflectional affix as occupying one end point. However, they stop
4 There are two possible ways these foreign adjectives may be handled. First, if they are reinterpreted
according to matrix form-classes, of course, they can not be adjectives. They would have to be treated as
belonging to an already existing form class (perhaps as verbs in this case). The second is the creation of a
new form class (identifiable as having a foreign source). In either case, the borrowed morpheme will be
analyzed according to the morphological possibilities of the borrowing language (content item, function
word, and so on).
47
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
short of differentiating between agglutinating-type affixes, with one-to-one
correspondences of form and meaning, and fusionai-type affixes, which represent a
coalescence of a number of grammatical categories onto a single, often phonetically
minimal, form. With respect to kinds of inflectional affixes and degrees of borrowability,
morphological typology is obviously key: it is . .more likely that clearly segmentable
[i.e., agglutinating] affixes will be borrowed than fusional morphology...” (Comrie, 1989,
p. 210). Therefore, the following subhierachy is proposed:
(13) function word > agglutinating affix > fusional affix
The hierarchical relationship of grammatical forms thus stated is consistent with a number
of observations of borrowability and various grammaticalization hierarchies5. It also serves
to illustrate implicationally that the more grammaticalized a form is, the less likely it is to be
borrowed. Being located to the far right of the hierarchy, fusional affixes are clearly the
least likely of all forms to be borrowed.
2.2.3. Summary of hierarchies
As anticipated, borrowing hierarchies reduce in much the same ways as
grammaticalization clines and appear to be identical with the Hierarchy of Morpheme Types
in (8) above. Thus, it is now possible to propose a preliminary scale of borrowability,
stated as (14) below:
5 See. for example, Croft, 1990 p. 191) and his tentative hierarchy of grammatical concepts; the noun-to-
affix cline of C. Lehmann. 1986 (pp. 3-4) (already cited in this text); and Hopper & Traugott, 1993 (p.
I08ff) (also cited above). The link between grammaticalization and borrowability is made more obvious
when considering both reduction of form (salience) and concomitant semantic bleaching Goss of specific,
concrete meaning), which can be linked also to degrees of semantic transparency.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(14) independent word, bound root > agglutinating affix > fusional affix
Items furthest to the left are the (a) content items and (b) independent function
words, i.e., those free-standing and/or bound morphemes (acting as roots or stems) that
are involved in the formation of classes of Noun, Verb, Adjective, and Adverb, and
independent function words (free grammatical morphemes and/or bound roots capable of
receiving markers of inflectional categories), including sub-classes of Determiner,
Pronoun, Auxiliary, Adposition, and Connective (inclusive of Coordinators,
Subordinators, Complementizers). Based on degrees of grammaticalization, we can divide
the morpheme type “independent word, root” into the following:
(15) content item > function word
Once again linking grammaticalization and borrowability, we can characterize the
borrowability of function words versus affixes as (16) below (repeated from (13), above):
(16) function word > agglutinating affix > fusional affix
Combining (15) and (16) into a single hierarchy, we have the more specific Hierarchy of
Borrowability, as represented in (17):
(17) content item > function word > agglutinating affix > fusional affix
The implicational nature of the hierarchy in (17) is two-fold. First, there is a
quantitative claim which states that X will borrow (from Y) a greater number of content
items than grammatical words, more grammatical words than agglutinating affixes, and so
49
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
on. Secondly, there is a temporal claim which states that if language X has borrowed
fusional affixes from Y, then it has already borrowed agglutinating ones; if it has borrowed
agglutinating affixes, it has already borrowed grammatical words; and, if it has borrowed
grammatical words, it has also (previously) borrowed content items.
The investigation of borrowing in a specific language will always depend on the
particular languages involved, the word and morpheme types that exist in each language,
and the ways meaning is distributed across the forms available in each. On the one hand,
there are simple, one-to-one correspondences of salient phonetic forms that possess readily
identifiable (transparent) meanings. These specific form-meaning sets require little or no
language particular knowledge and are among the form types most frequently borrowed
from one language into another. On the other hand, there are general markers of
grammatical concepts that are typically less salient and less semantically transparent, as
well. These markers may take a variety of forms, from free-standing function words to
bound inflectional affixes, and represent more generalized, abstract meanings.
Consequently, function words are less likely to be borrowed than content items.
Inflectional affixes are the least likely forms to be borrowed. Stated implicationally, if a
language has borrowed inflectional affixes, it will have also borrowed some grammatical
words; and if it has borrowed grammatical words, it will have borrowed lexical items
(Comrie, 1989, p. 210). Such a scaling is consistent with Thomason & Kaufman's
“borrowing scale” (1988, p. 74ff.) and statements concerning form types from
typologically distinct languages and possible mismatches of form that can affect patterns of
borrowing (hence, borrowability).
The strongest argument for a strictly cognitively based explanation seems to be with
respect to fusional-type affixes. In the donor system to which they belong, fusional affixes
50
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
typically occupy positions within tight paradigms, and they are not generally interpretable
outside of their paradigms; their identification, therefore, requires a more intimate
knowledge of the donor language, including knowledge of entire paradigms and the
oppositions of form and meaning they serve to indicate. They have no one-to-one mapping
of form and meaning, which undoubtedly affects the efficiency and speed with which their
(relational) meanings are retrieved. In addition, a single borrowed fusional affix must
either become part of an existing recipient paradigm (replacing a native form) or augment
the paradigm by creating a new position and a new distinction (i.e., barring a complete
reanalysis of the borrowed form). In either case, the make up of original paradigms of the
recipient language will be altered and the processes which distribute individual paradigm
members will be necessarily affected. This involves a much greater degree of change than
the simple addition of a borrowed lexical item, especially factoring in the frequency in
which inflectional forms are likely to occur. We can also assume that such change takes a
longer period of time to diffuse within a community—the larger the community, the longer
it would take.
It has been implicit in studies of mixed languages that matrix inflectional affixes of
any type are not included in processes of relexification, that is, according to a dichotomous
view of language forms as either lexical or grammatical, and that the reason for this is
cognitively based (in some yet to be defined way). The search for a universal cut-off point
to borrowing based on such a vague notion of cognition, however, may be a red herring—
after all, everything should be borrowable that is, in fact, leamable. And, assuming that in
a bilingual/contact situation, at least a subset of fluent bilinguals have the ability to know
and use both languages (in principle, not restricted with respect to type), such a claim may
be somewhat difficult to defend. The solution proposed here is that the limits to borrowing
51
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
and borrowability are established by the language systems themselves, and are, as a
consequence, linguistic in nature.
2 .3 . THE PRINCIPLE OF SYSTEM INCOMPATIBILITY
It is possible to take into account both the implicational nature of proposed
borrowing hierarchies and the individual characteristics of the participating languages and
formulate a principle that will be able to identify precisely the forms that are borrowable in
the broadest sense. Due to its bilateral nature, its negative formulation will also expose
morpheme types that are not borrowable. The identification of compatible form classes is
done by superimposing the morphological typology of language X (a borrowing variety)
over that of Y (a lexical donor) (see Figure 2.1). Consequently, if X is isolating-analytical
(i.e., contains only independent words), all independent words in any Y are, in principle,
borrowable. If X is agglutinating (contains independent words and roots and agglutinating
affixes), all independent words or roots and agglutinating affixes are borrowable from any
Y. Finally, if X is fusional (contains independent words and roots and both agglutinating
and fusional affixes), all morphemes are borrowable from Y (see Figure 2.2).
F ig u r e 2.1. Language types and allowable morphemes.
FUSIONAL,
SYNTHETIC
independent w ords, roots ag g lutinating a ffix e s
AGGLUTINATING,
SYNTHETIC
independent w ords, roots ag g tafm aan g afiB x es
ISOLATING,
ANALYTIC
independent w ords
52
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
F ig u r e 2.2. Compatibility and incompatiblity.
T y p o lo g y o f X Y f o r m s c o m p a tib le w tth X Y f o r m s in c o m p a tib le w ith X
FUSIONAL,
SYNTHETIC
AGGLUTINATING,
SYNTHETIC
ISOLATING,
ANALYTIC
Y FORMS COMPATIBLE WTTH X
independent words, roots
agglutinating affixes
fusional affixes
independent w ords, roots
agglutinating affixes
independent words, roots (analyzed as
discrete words in an isolating X)
Based on the morphological structuring of the languages involved, we can state the
Principle of System Compatibility (PSQ as follows:
(18) The Principle of System Compatibility (PSQ:
Any form or form-meaning set is borrowable from a donor language if it conforms
to die morphological possibilities of the recipient language with regard to
morphological structure.
All content items are borrowable from one language to another as are grammatical
morphemes (free or bound) that can fill slots typically occupied by elements of the
borrowing (recipient) language (irrespective of their meanings). As a consequence,
compatibility may include any item from content word to inflectional affix. While
hierarchies of borrowability assign grammatical affixes to positions that reflect the degree
of difficulty with which they are borrowed, there is no principled basis for their exclusion
other than to say that, in a specific language X, forms from Y are systematically blocked
that cannot be recognized and processed according to the formal characteristics of forms in
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
X. In other words, as possible forms in X, they do not exist (at least in the ways they exist
inY).
Following from the PSC as formulated in (18) above, the Principle of System
Incompatibility (PSI) can be stated as follows:
(19) The Principle of System Incompatibility (PSI):
No form or form-meaning set is borrowable from a donor language if it does not
conform to the morphological possibilities of the recipient language with regard to
morpheme types.
Using the typology of X (the recipient language) as the basis for comparison, forms
in Y that lie to the right on the hierarchy of morpheme types are rendered unborrowable.
This formulation captures the fact that some agglutinating inflectional affixes are, indeed,
borrowed by other agglutinating varieties (Heath, 1981) (or from fusional ones, for that
matter) despite claims that inflectional affixes are not borrowed (e.g., in mixed languages).
In essence, the PSI states that borrowability, in a broad sense, is constrained by the
morphological structuring of the languages in contact To be more specific, strictly
analytical languages (with no affixal morphology) are blocked from borrowing affixes of
any kind (without perhaps reanalyzing affixes as independent words), and agglutinating
languages are blocked from borrowing fusional affixes (without reanalyzing the fusional
affixes as possessing one-to-one correspondences of form and meaning). Actual cut-off
points will undoubtedly vary in language-specific ways relative to the typology (of
morphological structuring) of the participating varieties and the ability of the recipient
language (viz., its speakers) to identify donor forms and meanings.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2.3.1. Predictions within general classes
According to the PSI, language X (a recipient or borrower) can borrow from Y (a
donor) any form that is consistent with its own morphology (that of X), providing that Y
exhibits such forms. In other words, though change may be inevitable (especially as a
result of contact), a borrowing language will act within its own typological parameters to
preserve its morphological integrity: it will not borrow items that are morphologically
incompatible (to the right on the indices of synthesis and fusion). The PSI simply points to
formal typological constraints that define an individual item’s borrowability—violations
would result in typological anomalies in the borrowing language that would need to be
reconciled.
To illustrate, if X is predominantly isolating, it can, in theory, borrow independent
words and roots6 freely from Y. If Y is primarily synthetic (some of its forms lie to the
right in the hierarchy of morpheme types), all inflectional (and derivational) affixes will be
constrained from the mix. In addition, the hierarchy expressed in (13), above, (which
expresses relationships among classes implicationally) predicts that X will not borrow
function words from Y unless it has borrowed content items first7. The reverse, however,
is excluded. Even if X has borrowed content items from Y, it may not borrow grammatical
6 Borrowed words or roots are typically treated as steins in the recipient language and take the usual affixes
for the appropriate form class (Thomason & Kaufman, 1988, p. 37). These stems may contain derivational
affixes or other elements (e.g., clitic articles in the case of borrowings from French) that are analyzed as part
of the root, for example in Michif zafer “business** from French les affaires (Bakker, 1997. p. 103).
7 The implicadonal nature of the hierarchy does admit the theoretical possibility that a particular function
word may be borrowed at the same time as a particular content item. The likelihood of elements from
different general classes (e.g., content items and function words) being borrowed simultaneously is much
reduced compared to elements from neighboring subclasses of borrowable types (e.g., nouns and adjectives).
55
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
words. The frequency of borrowed items will always be gradiendy skewed to the left of
the hierarchy—viz., X will borrow content items more ffequendy than function words8.
Synthetic languages, depending on their position along the index of fusion, may
borrow content items, function words, and some types of inflectional affixes. If X is
primarily agglutinating, permitting content items, function words, and agglutinating
affixes, all independent words, roots, and agglutinating affixes occurring in Y are
borrowable; the only items formally constrained from borrowing are fusional affixes (if
they are permitted in Y). It is also implied that, if X borrows agglutinating, inflectional
affixes from Y, it has already borrowed function words and content items. If X is
fusional, all forms in any Y are theoretically borrowable. In each case, items gradiendy
positioned to the left along the hierarchy of morpheme types will be borrowed more
ffequendy.
In theory, all content items and function words in any Y are potentially borrowable
by speakers of any X irrespective of morphological typology because all languages have
content items and function words9. In practice, X will not borrow all the forms of Y
(whether they are borrowable or not). Perceptual salience and semantic transparency, in
themselves relative notions, will conspire together to promote individual forms from among
individual form classes. Other factors, for example frequency and intensity of exposure
and relevance, will further restrict the list of possible candidates. Obviously, the actual list
of borrowed forms may, in fact, vary from speaker to speaker depending on such factors
8 There may also be a cumulative effect of the frequent borrowing morphemes of the analytical types; for
instance. Modem Mexicano and Michif (with a Cree matrix), highly synthetic languages, have become
more analytical as a consequence of borrowed content items from Spanish and French, respectively.
9 Recall that the discussion concerns languages in contact in which degrees of proficiency in Y by speakers
of X are presumed. Content items (nouns, verbs, adjectives) are acquired first in both NL and SL
acquisition. Thus, the correlation of leamability and borrowability may hold with respect to compatible
items.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
as degree of education (and, therefore, familiarity with and exposure to Y), occupation
(restricting exposure to certain semantic domains), and so on.
2.3.2. The occurrence of anomalies: reanalysis
In the naturally occurring speech of bilingual (X, Y) speakers, one might expect to
find the occurrence of a number of apparent anomalies, for example in the case of so-called
nonce1 0 borrowings or during code-switching when the switch is constituted by a single
lexical item (with or without inflection) from Y embedded into an utterance in X. The
PSI’s basic predictions refer specifically to processes that necessarily occur subsequent to
such isolated occurrences that may lead to the integration of heretofore foreign elements
into the recipient language. Just as phonological anomalies are resolved phonologically
(perhaps by some process of imitation), morphological anomalies must be resolved
morphologically. We, therefore, propose that the PSI be taken with the following caveat,
as stated in (20), below, the Principle of Reanalysis:
(20) The Principle of Reanalysis (PR):
For a foreign element to be borrowed that is incompatible with the recipient system
by virtue of its position outside or to the right on a scale of allowable morpheme
types, it must be assigned to a position to die left that is within the typological
parameters set by the recipient system.
10 The term nonce borrowing refers to the spontaneous, one-time borrowing of a form from Y that has yet
to be fully integrated into X. It may be morphologically and syntactically adapted, but show minimal
phonological integration. For opposing views of the usefulness of this term, see Myers-Scotton, 1993a
(e.g., pp. 20-23) and Poplack, Sankoff, & Miller, 1988 (e.g.. pp. 47-50) or Sankoff, Poplack. &
Vanniarajan (1990).
57
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The process so construed is analogous in a sense to the process of substitution in the
phonological reproduction of foreign items, providing the mechanism by which a
morpheme can be reanalyzed (so that it can “imitate” the functions of the morphemes
permitted in the recipient). It amounts to the reinterpretation of an element’s morpheme
status to be in conformity with the recipient system regarding morphological structure. The
borrowing language must (a) locate the nearest possible point along the Hierarchy of
Morpheme Types that is within its own morphological parameters and (b) assign the
heretofore incompatible item to that position. Assuming that morphological character varies
along the axes of synthesis and fusion, reanalysis requires the assignment of the
incompatible element to a position far enough to the left along one (and only one) of these
axes to be within the morphological parameters set by the recipient, hence, matrix system.
Accordingly, strictly isolating languages cannot borrow affixes, per se, because the
occurrence of even one affix (i.e., via processes identifiable as affixation) would constitute
a systematic anomaly. In order for this incompatible element to be fully integrated into an
isolating-analytical morphosyntactic system, it must become compatible: it must be
reanalyzed to the left of the index of synthesis, as an independent word, thus preserving the
morphological integrity of the recipient language. Similarly, synthetic languages that are
primarily agglutinating cannot borrow fusional affixes, per se. In the event that a fusional-
type affix occurs in an otherwise agglutinating language (e.g., as part of a nonce borrowing
or single-item code-switeh), it cannot be fully integrated into the recipient language unless
or until it is reanalyzed as agglutinating—it must be assigned to a position to the left on the
index of fusion. Semantically and morphologically, this amounts to its reinterpretation as
having a one-to-one correspondence of meaning and form, which is likely to appear as a
case of simplification. One can imagine that reanalysis of this sort is quite rare; however, it
58
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
has been said to occur, for instance, in Mednyj Aleut, which “reduces Russian sets of
endings, unifies their range of meanings, etc...” (Golovko, 1994, p. 116).
2.3.3. Predictions within subclasses
It is yet again possible to be more specific within smaller subclasses; however,
predictions cannot be made solely according to general morphological structure; it is much
more likely that individual semantic characteristics will play the greater role. For example,
within subclasses of noun, concrete nouns are borrowed more often than abstract nouns
(see Chapter 4). While the frequency of occurrence of an individual item in the language
may be a factor with respect to content items/roots (i.e., as measured in various corpora),
frequency of exposure and relevance1 1 undoubtedly play greater roles in whether or not a
form is learned by an individual speaker. That is, for a particular content item to be
learned, a speaker needs to be exposed to it, and it will need to be relevant (there is a need
to leam it). Consequently, those who work in an environment in which nautical terms are
in frequent usage leam nautical terms; those who are involved in agriculture (e.g., on
farms, co-operatives, plantations, and so on) leam agricultural terms. As a semantic
domain, the nomenclature associated with a particular occupation, social institution (e.g.,
religious terminology), and so forth will be familiar and relevant to those involved in such
areas. The same can be said of other content words (verbs, adjectives) as a whole, though
nouns will be more frequently borrowed than verbs or adjectives.
1 1 Salience may be associated with and reinforced by frequency; the more times a person hears a particular
form, the easier it will be to recognize and recall. Likewise, the desire to attach a meaning onto a form will
most likely depend on its relevance (how important knowing its meaning is to perform a particular job and
so forth).
59
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
With respect to function words, individual inflectional affixes, and grammatical
categories in general, their diversity and language specificity present numerous problems in
borrowing and acquisition. Nevertheless, some forms may be easier to pick out of the
speech stream and associated onto meanings because they are more salient than others (as a
result of certain content-word-like behaviors) and/or because they are more semantically
transparent For example, pronouns in English can behave in NP-like ways. They have
reference, occupy grammatical subject position and are assigned semantic role, can receive
phrasal and clausal stress, participate directly in syntactic processes, and so on. In
comparison, the articles of English carry out numerous and sometimes conflicting
discourse and semantic functions, e.g., indicating such distinctions as old versus new
information, definite versus indefinite, specific versus nonspecific, count versus non-count
noun, etc. As a subclass of pronominals, personal pronouns in languages such as English
appear quite early in second language acquisition, while mastery of the articles (i.e., a/n
and the) is notoriously problematic leading to later acquisition1 2 .
2.3.4. The relative timing of borrowed elements
The hierarchies and subhierarchies of borrowability predict that when contact is
initiated, the first elements to be borrowed will be content items. The current work also
points out that they are borrowed first as a general class because they are the most salient
and transparent of all potentially borrowable elements. Forms (labels) with visible,
tangible referents are the easiest to leam; so, nouns (as a subclass within the class of
content items) will be among the first content items to be borrowed. Nouns will then be
12 See Chapter 4 for a comprehensive discussion of the formal and semantic characteristics that pertain to
degrees of identifiability and. hence, borrowability of items within various subclasses.
60
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
followed by other content items, for instance adjectives or verbs, depending on the
morphological structuring of the languages in contact. It is very unlikely that any X will
borrow verbs (and only verbs) from Y without having already borrowed nouns;
consequently, languages whose contact is fairly limited may borrow only nouns.
There is a certain amount of logical necessity involved in this ordering, as well,
especially with respect to the order of acquisition in a particular language. In order to
attribute qualities to a thing, it is well to know what that thing is and what it is called,
although it is not a logical necessity—the label for a general attribute may in many cases be
acquired before the name of all particular persons, things, and so on that may possesses
that attribute. Because actions involve complex relationships among participants (human,
animate, and inanimate agents, themes, and instruments) and things acted upon (patients,
goals, and so on), the prior knowledge and possession of names for these participants will
be required, even though it should be obvious that when two languages are involved (i.e.,
in bilingual/contact settings), the actual labels may originate from either X or Y.
The hierarchy formulated in (2) essentially rules out the possibility that any X will
borrow adjectives (or verbs) more frequently than nouns from any Y. However, it may be
possible (though not likely) for X to borrow, say, a particular adjective or verb from Y
before a particular noun (assuming that any element is, in principle, borrowable), but it will
have to do so without violating the general tendency to borrow nouns more frequently.
Another possibility naturally follows that is consistent with the implicational nature of such
hierarchies: particular adjective-noun or verb-noun pairs may be borrowed simultaneously.
That is, a particular adjective may be borrowed at the same time as a particular noun, or a
particular verb may be borrowed at the same time as a particular noun. This may be
especially likely when their co-occurrence is the norm, as in cases where frequent
61
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
collocations occur within relatively restricted semantic domains, for instance, involving
occupational nomenclature (agricultural, nautical, religious, and so on).
Borrowing, especially that which may lead to relexification, is necessarily a gradual
process. Consistent with observations of doubling phenomena in language change, in
general (Hawkins, 1990, p. 98), it is quite likely that X will pass through stages in which
single referents have two labels existing side by side. Each label can become more
specialized and acquire more exact, differentiated meanings; the labels may split the
semantic load, each inheriting only one of the possible meanings. In other cases, “older”
forms may be remembered only by older (or more conservative) speakers, while the “new”
forms may increase in currency for a variety of reasons. Original (X) forms may be
completely forgotten (especially in cases of language shift) and lost Differences between
garden variety borrowing and relexification are seen here as essentially quantitative.
Intense contact along with the acute nature of shift in some former colonial communities
has had the obvious capacity to accelerate change of all sorts, especially within particular
domains, and borrowing is no exception. What is remarkable about relexification is that
vocabulary replacement is so complete that—where both continue to exist—relexified X
may no longer be understandable to speakers of conservative varieties of X (that are not
relexified to the same extent), even to those speakers who may be familiar with Y.
62
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.0 . C h a p t e r 3— Fo r m C l a s s e s a n d Se m a n t ic T y p e s
To briefly sum up what has been covered to this point, we can say that lexical
borrowing appears to involve the importation of form-meaning sets from one language into
another. How these form-meaning sets are assembled into words to be used in everyday
speech will, of course, depend on the morphological structure of the recipient language.
Some sorts of reanalysis may be necessary as borrowed forms are integrated into a new
linguistic system, which may be accomplished in a number of ways and at different levels
of grammar (phonologically, semantically, morphologically, and syntactically), perhaps in
parallel fashion. It is also clear that some donor forms are borrowed more easily than
others (e.g., content versus function items) and that not all words or morphemes from a
donor will be borrowed in practice (there appears to be a saturation point, most likely
determined by social factors), despite the fact that any word or morpheme (content or
grammatical) that is leamable may be borrowable.
Observations such as these have led to the postulation of the Principle of System
Incompatibility (PS I), based on morpheme type, which identifies a single cutoff point in
any borrowing situation past which forms cannot be borrowed. Adherence to the PSI
preserves the morphological integrity of the recipient language for the sake of continuity
and stability, though the cumulative effects of borrowing may eventually lead to
morphological and syntactic change. As stated in previous sections, boundaries between
morpheme types and, hence, among word classes (e.g., on clines of grammaticality) are
not always distinct This situation becomes more complex when looking at different
systems in contact because exact form class correspondences at the morpheme or word
level may be relatively difficult to establish. An important question to be posed at this
juncture, and which should be answerable by any framework purporting to offer a model
63
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
for borrowing, may be the following: what exactly is borrowed when a form, morpheme
or word, goes (is copied) from one system to another? The answer to this must include
instances in which an item appears to undergo a change in class membership, as in the case
of a Spanish adjective being reinterpreted as a Chiricahua verb (see Section 1.2.1., above).
The overriding theme of this chapter is the centrality of meaning in processes of
borrowing. It is argued that borrowing involves form-meaning sets irrespective of word-
class membership. Obviously, borrowed elements originate as either words or parts of
words (morphemes) in the donor language. Individual speakers of relative proficiencies in
either or both donor and/or recipient must identify form-meaning sets from among an
extensive inventory of linguistic elements for incorporation into the recipient, a list of
potential candidates that is essentially the donor lexicon minus incompatible forms. This
necessarily involves word and morpheme recognition and a degree of morphological
analysis of candidate forms in the donor. Nevertheless, this is quite distinct from issues of
morphological reanalysis that are addressed by the PSC and its correlate, the PSI.
3.0.1. The relationship of word class and semantic type
Despite the readily apparent diversity in particular languages and potential for cross-
linguistic discrepancies, any bi- or multi-lingual contact situation that involves the
interaction of distinct linguistic systems1 brings with it the tacit understanding that there are
equivalences or correspondences of some kind in such areas as the translatability of
individual words and phrases from one language into another, even if this can only be done
1 Recall that two (or more) languages are said to be in contact when they are spoken by members of a
single community (Chapter 1); the internal interaction of linguistic systems—in the heads of individuals—
manifests itself as performance phenomena (transfer, borrowing, code-switching, and so on).
64
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
through broad paraphrase. For translation to be possible at all, bilingual speakers must
assume that they can find corresponding utterances in each of their languages— irrespective
of typological distance. It would be naive to think that any two languages could exist
having exact morph-for-morph equivalents (beyond perhaps particular subclasses of noun
or verb). This is true between languages that are close genetic relatives or dialects of the
same language, as well. Subtle differences will occur despite any close typological fit A
rough translation based on functional equivalence (for expedience in communication) is not
formal linguistic equivalence. Conversely, it would be unreasonable to speak of formal
equivalence as a rough translation, at least in the vast majority of cases, partly because lack
of direct correspondence of individual forms cannot block effective translation (see Croft,
1990, p. 13). The two issues are clearly separate despite the fact that they may
occasionally coincide. Based on these observations, we may assume that bilinguals find at
least some intuitive grounds for their determinations of correspondence.
The present work proposes that the primary basis is, in fact, semantic by positing
that what is actually borrowed is a core meaning or concept and the label (a phonetic string)
for that concept, that is, a form-meaning set. Previous word class membership (being
assigned by the donor morphosyntax) is rendered moot by the very act of borrowing.
Such a view is compatible with a number of current models of morphological processing
that consider morphemes (and perhaps words) as epiphenomenal and lacking independent
status (Bybee, 1995, p. 233)2. It seems much more likely that assignment to particular
word-classes comes later, as a result of borrowing. Granted, in most cases this means
assignment will be to a similar class (e.g., nouns in a donor language usually function as
nouns in a recipient). If donor items were to be imported strictly on the basis of syntactic
2 For additional discussion of these proposed models, see, for example, Anderson, 1992 (p. 56ff.),
Matthews, 1991 (p. 21f.), and Spencer, 1991 (p. 52,434).
65
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
affiliation, a language could only borrow those items for which it already possessed
equivalent (sub)classes. For example, a language without a class of adjectives (identified
by its unique distributional and morphological characteristics in the donor) would not be
able to borrow an adjective without, perhaps, bringing the class designation along with it.
This hardly seems plausible for a single item. Even in cases where a word class is
apparently borrowed (as in Michif), there may be a number of factors in play, including
obvious structural borrowing.
However, if the borrowed form is first associated with a particular semantic type or
subtype, its assignment into an appropriate word class for appropriate distribution and/or
morphological marking will be automatic and analogous to other members of that semantic
(sub)type in the recipient system. This can occur whether the recipient word class is
identifiable as equivalent to that from which it originated or not. Lexical borrowing, then,
may begin when an individual speaker (or group of speakers) of the recipient language
associates a desired form-meaning set with an equivalent semantic type or subtype on the
basis of semantic congruence3. For instance, a “name for a concrete object” is interpreted
as belonging to the same semantic subtype as other “names of concrete objects” and
emerges as a morpheme of the appropriate type in the recipient Because it is identified
with other morphemes of a particular subtype, it can be assumed to have certain basic
morphological properties (free or bound, root or affix) in accordance with its type in the
recipient language system. On-line determination of exact cross-linguistic word-class
equivalence becomes irrelevant In addition, morphological (re)analysis is relatively
unencumbered and natural—according to the recipient system.
3 Semantic congruence is probably best viewed as gradient The degree of congruence may be significant
when, for instance, the meanings of two forms overlap in certain situations, as in cases of hyponymy and
near synonymy. Minimal semantic congruence may be construed as (merely) belonging to the same
semantic subtype.
66
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.0.2. Form-meaning sets and semantic types
While the number of form-meaning sets (“concepts” in the Sapirian sense)
represented by individual words and the morphemes they are composed of is in principle
limitless due to their sheer numbers and potential for specificity, the number of semantic
types constitutes a range of possibilities that is significantly reduced as a consequence of its
more general character. For instance, names of people and objects can be classed as
CONCRETE nouns, concepts of time as ABSTRACT nouns, and so forth. We can represent
these relationships according to a somewhat simplified version of set theory, avoiding, for
the present, matters of cross-classification, overlaps, multiple sets, and so on. This is
portrayed in (1), below:
(1) form-meaning sets —> semantic type (morphemes or words)
Semantic types and subtypes cluster (as subsets of a sort) into word classes (so-called parts
of speech) which compose a list of classifications even more restricted in number resulting
from its considerably greater general character. For example, names of people, objects,
and places are commonly represented by only one general form-class, Noun. In fact, every
major form class can be divided into collections of semantic types (cf. Dixon 1991, p. 6ff).
We see these relationships illustrated in (2), below:
(2) semantic types (morphemes, words) —> word class (N, V, Adj...)
The path that a borrowed form-meaning set takes as it enters a recipient lexicon and is
consequently integrated (morphologically and syntactically) into the recipient system can be
represented according to these same basic organizing principles, as pictured in (3), below:
67
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(3) form-meaning set —> semantic type —> word class
Across languages, groupings may be quite similar, though we have learned to
anticipate that there will be differences as a consequence of language-particular grammatical
characteristics. What may be a proper subset of one word class in X may be a subset of
another in Y (Sapir, 1921, p. 117; Dixon, 1991, pp. 9,77). The ways concepts (form-
meaning sets) are assigned to syntactic (word) classes may be subject to some variation.
The starkest examples of this are grammatical concepts (e.g., function words), but that
does not necessarily preclude the possibility, or even likelihood, that lexical concepts
(content items, whether independent words or bound roots) will be subject to similar
classification problems, as well.
3.0.3. Organization of this chapter
The discussion begins with an overview of notions of word and word class and on
descriptions of and comparisons among morpheme types and their roles in the construction
of words. Consequently, formal contrasts are presented between items occupying extreme
ends of the Scale of Morpheme Types, i.e., content items versus inflectional affixes,
moving on to the comparison of the remaining types, content items versus grammatical
words, and grammatical words versus inflectional affixes (and clitics). From there, we
proceed to more specific discussion of various form classes (noun, verb, adjective, etc).
Subsequently, general semantic contrasts are made, first between content items and
grammatical affixes—focusing on the Sapirian distinction between primary (“material
content”) and secondary (“relational”) meaning. Function words, at midpoints between
content items (words and bound roots) and inflectional affixes are discussed at length,
68
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
especially in light of the fact they constitute obvious points of conflict (or mismatch of
forms) between two languages in contact. Finally, discussion shifts to semantic types,
representing the kinds of meanings that are allocated into various, language-specific form
classes. This semantic approach allows us to characterize linguistic borrowing as the
copying of semantic entities with accompanying labels from one lexicon into another,
which affords the recipient language maximum freedom of reanalysis in the preservation of
its morphological integrity.
3.1. NOTIONS OF WORD AND WORD CLASS
The most significant units of language (written and spoken) are words, and these
are composed of different types of morphemes. Discussion of one necessarily involves the
other. Sapir (1921) portrayed the meaningful elements of language according to
fundamental “notions” of subject or content matter (so-called radical elements) and
additional concepts of a more abstract nature (grammatical elements), “of person, number,
time, condition, function, or of several of these combined” (Sapir, 1921, p. 25). In
general, words can be independent entities (consisting of individual, free-standing
morphemes), combinations of bound roots representing primary, concrete meanings joined
with language-particular subsidiary or relational concepts (represented by types of affixes),
or other language-specific aggregates of morpheme types (e.g., inflected function words
composed of a bound root plus various grammatical markers, etc.). The possibilities
intrinsic to a particular language are determined by its morphological typology (Chapter 2).
In any case, words form the essential building blocks of speech, representing the topics
(content), attitudes (intent), and concepts that must be expressed if language is to fulfill its
referential and expressive functions.
69
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Sapir (1921) discusses the integrity of the word in terms of a native speaker’s own
intuition. For instance, whether it consists of one or several morphemes, a word is not
divisible without rendering the parts “meaningless” in some sense: “[A word] cannot be
cut into without a disturbance of meaning, one or the other or both of the severed parts
remaining as a helpless waif on our hands” (Sapir, 1921, p. 34). He also refers to
phonological properties, specifically “accent” (stress), as marking the internal cohesiveness
of the word, even though the boundaries between words may become rather blurred in
actual speech (as a result of language-particular prosodic patterns)4.
Bloomfield (1933) describes form-classes as determined by function. For example,
substantive expressions (nouns) share many functions, e.g., the positions of actor or goal
with a verb, point of reference with respect to adpositions, underlying the identification of
pronouns and possessives, and so on (p. 265). Thus, he distinguishes form classes, for
example, according to the English actor-action construction in which the form class of
nominative expressions (in languages such as English) precedes the form class of finite
verb expressions. Bloomfield equates position with function in that the positions that
forms may occur in indicate their functions, and, collectively, their function. Forms that
can fill given positions, therefore, constitute form classes: any English word or phrase that
can fill the actor position in the actor-action construction is, therefore, a member of a “great
form-class” called nominative expressions; any word or phrase occupying the action
position is a member of another great form-class, finite verb expressions (Bloomfield,
1933, p. 185). Thus, word class can be defined along the lines of function or role within a
sentence and position with respect to other expressions (words and phrases). In other
4 In this respect, a number of current approaches offer definitions of phonological word that appear to be at
odds with those that are primarily syntactic or morphological (e.g., Di Sciullo & Williams, 1987; Levelt,
1992; Bierwisch & Schreuder, 1992). Phonological factors affecting word boundaries may have additional
consequences in the on-line recognition of forms.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
words, a word is primarily classified by its characteristic relationships with other words
and their distributional properties. That is, nouns or verbs appear in certain contexts (e.g.,
as heads of NPs and VPs, respectively).
There may be a degree of circularity in this reasoning (nouns appear where nouns
appear), though it is clearly true that context restricts the occurrence of nearly every
linguistic unit or class (except sentence) that may characterize its distribution. However, in
many cases there is a certain degree of ambiguity involved5, as when members of different
form classes can occupy identical slots (e.g., following a copula in languages such as
English or Spanish). Distributional criteria also reflect the fact that members of a particular
class may be distributionally equivalent, where each member occurs in similar contexts, or
in complementary distribution, where particular elements have no common contexts, which
can apply to certain phonemes, affixes, and words of particular semantic subtypes (Lyons,
1968, p. 70-72). Nevertheless, functional and distributional characteristics apply to any
morpheme type in a given language; this applies to affixes, as well. For example, one
distributional restriction on inflectional affixes is that they can only appear on members of
particular word classes: nominal affixes denoting such categories as number, case, and
gender/class usually appear only on noun stems (or are spread by processes of concord),
and verbal affixes denoting tense or aspect appear only on verb stems.
Patterns of occurrence lead to the paradigmatic relationship of one form with all
other forms that may appear in the same context and syntagmatic relationships with various
forms that constitute its context. Certain subsets of nouns, for instance, form paradigms
(e.g., units of measurement, ounce, pound, and ton, that may occur in such expressions as
5 This is typical of the difficulty in distinguishing between members of noun and adjective classes in
languages such as Spanish. See Chapter 6 for discussion.
71
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
an ounce of coffee and a pound of coffee...). Syntagmatic relations lead to the precise
patterning of these terms within the expression “a/n of coffee”. As a direct
consequence of such grammatical patterning, these forms or words (ounce, pound, and
ton) are said to belong to the same grammatical class (in this case, a sub-set of noun).
Obviously, notions of word and word class can be looked at from a number of
perspectives, which may be in large part due to the fact that there are usually clear
connections between distribution (i.e., position and word order) and grammatical function.
Moreover, function is typically linked to meaning (semantic agents are often grammatical
subjects). Theoretically, the two (semantic role and grammatical function) remain distinct
(Lyons, 1968, p. 73), a fact that is clearly demonstrated by the diverse strategies with
which languages encode grammatical and semantic relations, e.g., through word order,
morphological markers, and so on6.
Because the surface appearance of forms does not change from class to class in
more isolating/analytical languages like Vietnamese, it might be difficult to distinguish to
which word class a particular form belongs, especially in the absence of a grammatical
and/or semantic context. Unless form-class membership is specifically and uniquely
marked (i.e., morphologically), there is always the potential for a certain amount of
ambiguity. The form-meaning set qua word love in English seems to belong to both great
classes of noun and finite verb expressions, as in the following: (a) Love is kind and (b)
John and Mary love each other. In (a), according to positional, functional, and
distributional criteria, “love” belongs to the form-class of noun, while in (b), it is a finite
verb. It is apparent that the former inflects for the form class Noun, while the latter, the
6 We would be remiss if we failed to mention ergative languages as an example of this potential for
diversity.
72
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
form class Verb. What is the same about them is a core meaning and its associated
phonetic string /Ia v /.
Whether each individual occurrence of a form in different syntactic contexts must be
treated as a manifestation of a separate lexical entry or not may be somewhat controversial.
It most likely depends more on one’s viewpoint of the nature of the mental lexicon and so
forth than anything else. Whatever the final outcome of this debate might be, the vast
majority of the morpheme types in any language are, nonetheless, classifiable as either free
standing words (or bound roots, stems, or bases) or various types of affixes according to
fairly unambiguous morphological criteria.
3 .2 . CONTRASTING POINTS ON A CONTINUUM
Establishing and comparing types of morphemes (as we have seen in Chapter 2) is,
in fact, an attempt to contrast points on a continuum. Distinctions between adjacent
members may seem somewhat arbitrary at times. It is considerably easier (and clearer) to
look first at points at opposite extremes of a spectrum, in essence, contrasting black and
white, rather than shades of grey. Points toward the middle can be said to more greatly
resemble one extreme or the other. The criteria will be formal, based, for example, on
distributional and morphological characteristics, and semantic, perhaps, a more specific and
explicit measure regarding major word classes.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.2.1. Formal characteristics: grammatical affixes versus content items
The number of inflectional affixes that a language possesses depends in some
measure on the number of inflectional categories it obligatorily expresses (cf. Bybee, 1995,
p. 228), mediated by the number of categories expressible in a single morpheme in that
language. As with function words, they constitute closed classes, whose membership is
generally fixed. In fusional languages, grammatical affixes may represent a coalescence of
categories whose multiple (simultaneous) meanings can only be determined in opposition to
other members in their specific paradigm; this stands in static contrast to content items
which can be associated with individual visible, tangible referents (and so on)7.
With respect to their internal characteristics, affixes are bound, by definition, and
are, therefore, subordinate in form while content items can be either separate words or
constitute bases or stems to which affixes attach. They lie within word boundaries and are
systematically removed from stems and replaced by equivalent morpheme types (affixes) or
other members of their paradigms8. Inflectional affixes typically do not draw word stress
except as a result of normal, language-specific patterns on words; the capacity to receive
primary word stress is generally reserved for the radical (or content) elements to which they
are affixed (Hopper & Traugott, 1993, pp. 145ff.). As a consequence, they quite often
lack perceptual salience9.
7 This will be discussed at length in Chapter 4.
8 In this respect, the Semitic languages behave similarly to Indo-European ones even though, generally
speaking, inflectional morphology is primarily expressed by changing vowel (and some consonant) patterns
of a word and not direcdy by application of discrete affixal morphemes (Bybee, 1995, p. 233).
9 The inflectional affixes of English are a case in point There are, however, a numerous exceptions:
nearly all verbal markers of tense/aspect in Spanish receive primary word stress with the exception of the
basic present tense (but not most present forms of estar, to be).
74
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Contributing to a relative lack of on-line phonological prominence, their realizations
are often characterized as brief sequences of sounds typically ranging from a single syllable
to a single phoneme, in contrast with content items, which can be of varied length and
phonological complexity. There are a number of possible explanations for this: one is that
the relatively simple phonological form of an individual affix results from the fact that it
belongs to a limited set of single, bound morphemes whose occurrence is relatively
frequent in speech, and, as a consequence, that identification requires less phonetic
information (i.e., subsequent to the acquisition of the entire paradigm). Another lies in
their possible historical origins (as descended from independent words) and the processes
of grammaticalization (Hawkins & Cutler, 1988, p. 310; Hall, 1988, p. 335-44; Heine et
al, 1991a, pp. 19-20; Hopper & Traugott, 1993, pp. 145-6).
Positionally, they appear before, after, or within appropriate radical elements
(bases); viz. they are prefixes, suffixes, or infixes. Their placement within a phrase,
clause, or other larger unit of discourse is derived exclusively from the positional character
of the word class members to which they customarily attach. In general, they apply
“outside” all derivational processes; i.e., they are affixed to stems or bases subsequent to
the application of derivational morphemes (in the formation of a stem) and may even be
applied to larger syntactic units (such as phrases) acting as formal, semantic units, even
though they may adhere to only one of the collection of elements. For example, the
genitive marker's in English phrases like “the Mayor o f Lancaster’s limousine, where,
although the mayor is the possessor of the limousine, the - s inflection is attached to
Lancaster” (Katamba, 1993, p. 209). Finally, regarding position, they may be
configurational: “These are so called because the choice of a particular inflection is
determined by the place occupied by a word in a syntactic configuration, i.e. its position
75
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
and function as a constituent of a phrase, or some other syntactic structure” (Katamba,
1993, p. 209).
With respect to their functional roles, inflectional affixes may serve a variety of
language-specific grammatical and semantic functions (while content items cany the basic
semantic content of an utterance):
(a) they mark grammatical or semantic role, enabling content forms to fit into particular
syntactic slots;
(b) they provide markers of language-specific tense, person, or number distinctions
without changing the referential or cognitive meaning or word class of the stems
they attach to;
(c) they indicate inherent properties such as the gender, animacy, or other class of a
noun, that must be accessed by agreement rules;
(d) and, their application is obligatory when required by syntax, e.g., when marking
agreement of grammatical categories across or within phrasal boundaries.
Distributionally, selection may be made according to the inherent properties of the
particular content items to which they attach, e.g., according to specific conjugation classes
in such languages as Spanish (whose so-called theme vowels serve to designate such
classes). They may apply to members of broad paradigms. For instance, some affixes
apply only on certain kinds of nouns while others apply only on particular verbs or verbal
auxiliaries. The content items (or radical element) to which particular inflections are joined
are allowed, as a consequence, to become members of a relatively broad, grammatical
paradigm: sea and seas are opposed within a paradigm constituting the opposition of
singular and plural, which (only) applies to countable nouns (Matthews, 1991, p. 38).
While the word class that is affected by this distinction is large (in fact, as an open class, its
membership is theoretically unlimited), the number of inflections necessary to mark this
very general distinction is two (one typically being a zero form). Generally, then, the
76
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
smaller the number of grammatical affixes belonging to a specific paradigm, the larger the
potential number of items (or class of items) that can be morphologically marked. The
alternate use of a singular or plural form (especially in subject position) will trigger
agreement patterns that are strictly grammatical and language specific (e.g., in the case of
English, upon certain demonstratives within the noun phrase—this/that with singular count
or mass nouns and these/those on plural—and on the tensed verb within finite verb
phrases).
3.2.2. Semantic characteristics: grammatical affixes versus content items
On the one hand, content items are generally understood to carry the primary
semantic content of an utterance, identifying the topic(s) under discussion (the persons,
objects, activities, etc.), and so on. On the other, inflectional affixes function to specify
additional information of a strictly grammatical nature. For example, some grammatical
affixes may mark certain temporal characteristics such as aspect or tense that are highly
relevant to the central action represented by a verb (Bybee, 1985, p. 15). Other
grammatical affixes may mark such categories as number or gender that are directly relevant
to classes of noun (i.e., to identify how many of a particular semantic type are involved in
an action). In either case, neither the action nor the actors change identity or reference.
Constituting the vast majority of words in any language, content items are
customarily defined as having stateable lexical meaning. Forms that designate people,
places, things, and so on, traditionally called nouns, make reference to concepts that exist
(at least in the imaginations of speakers) in some objective or subjective way (i.e„ having a
physical, visible, and tangible referent or referring to psychological states such as
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
sentiments, emotions, etc.)- Items that typically indicate activities or actions, states of
being, and are the main elements of VPs (Lyons, 1968, p. 423ff) are traditionally called
verbs. A third class, Adjective, includes items that specify certain attributes of nouns. It is
assumed that all languages have classes of Noun (nominals) and Verb (verbals), less
frequently classes of Adjective (the attributes of nouns being expressed by alternative, e.g.,
verbal, means)1 0 .
Content items are often characterized metaphorically as vocabulary items or entries
in a dictionary (to indicate clear links to people, objects, visible qualities, etc.). Such
characterizations lead to the awareness that they do, indeed, have identity outside of any
syntactic context though actual usage may limit the possible meanings an item can have in a
specific utterance. This property is particularly significant when investigating word or
morpheme meaning in synthetic-fusional language types such as Spanish or Italian in
which nouns and verbs typically receive various inflections. In each, nouns obligatorily
express gender and number and finite verbs tense-aspect and agreement of person and
number with grammatical subject via inflectional morphemes which attach to word roots.
The associative meaning of a particular content item must be acknowledged apart from its
inflection, as “existing outside of any particular syntactic context” (Aronoff, 1994, p.
II)1 1 . Content items are also organizable into semantically based groupings (or fields)
such as color terms, verbs of saying, relational terms (e.g., kinship terms), taxonomies
(hierarchies such as animal, mammal, dog, spaniel including such part-whole hierarchies as
1 0 Many traditional grammarians identify a fourth major class. Adverb, which in a language like English is
said to modify the meanings of verbs, adjectives, and other adverbs; others prefer to lump adverbs and
adjectives together into a single classification. To enter into this particular controversy is, needless to say,
beyond the scope of this work, even at the risk of creating a deafening silence. Linguists, however, often
refer to certain classes of words indicating temporal distinctions in various languages as adverb(i)al elements
(e.g., derivational affixes, particles, words, phrases, etc.) whether or not they wish to distinguish a special
(open) word class in a specific language.
1 1 The fact that content items have identity that is separable from language particular morphology and
syntax undoubtedly contributes to the increased likelihood of borrowability.
78
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
finger-arm-hand-body), complementaries (e.g., right-wrong), antonyms (gradable pairs
such as tall-short), directional oppositions (e.g., bring-take), synonyms, polysemies, and
so on (Hopper & Traugott, 1993, p. 97).
Semantic content is formally associated with open-classes. Human experience
informs us that the range of possible referents is open-ended (perhaps inexhaustible), and
our linguistic devices need to reflect that reality. In some approaches, adpositions
(members of a more grammaticalized closed-class) are included as lexical (or content) items
(Katamba, 1993, p. 41). However, the concepts they express are always relational and
secondary (in the Sapirian sense), even in instances when they are semantically selected—
marking location in space or time, for example—and not obligatorily expressed as a result
of a syntactic rale or selectional restriction1 2 . As likely products of grammaticalization
processes, they may show a range of semantic independence; the increased abstractness
and generality characteristic of their meanings may straddle some sort of lexical-
grammatical frontier (between function word and grammatical affix), perhaps being
represented in one morpheme type or another depending on the degree of
grammaticalization1 3 .
Grammatical affixes, as noted above, are associated with inflectional categories,
although it may be difficult at times to distinguish among inflectional and derivational
processes. Fusional affixes, the most grammaticalized of the morpheme types, are always
(by definition) members of tight-knit grammatical paradigms, and, as such, cannot be
12 The term “lexical” is sometimes used with a slightly different sense, also in opposition to
“grammatical”, but referring more to independent word status. For example, a lexical (or analytical)
strategy refers to the practice of employing a word (unit of vocabulary) to convey a specific meaning which
may also be (in some equivalent sense) represented by a grammatical affix. An example of this would be “a
picture of my mother” versus “my mother’s picture”. The use of the genitive marker -’s does, however,
create an ambiguity that is not present in the lexical version. (See Crystal, 1991, p. 200f.)
13 See Chapter 4 for a more comprehensive discussion of grammaticalized meanings.
79
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
understood except in context (i.e., when attached to an appropriate root) and in relation to
other members of their paradigm. Obviously, the kinds of meanings that are expressed by
inflectional affixes are quite different from the meanings associated with the words they
help to form (cf. Beard, 1981). While content items maintain the relationship of signifier
and signified (in the Saussurean sense) to the fullest possible extent, inflectional affixes,
especially those that are fiisional, have no tangible referent possible—they mark language-
particular inflectional categories. Even agglutinating affixes, which may appear to to be
semantically simple in a sense by virtue of their one-to-one linkage of meaning and form,
deal with the “bound realization of syntactic categories” (Aronoff, 1994, p. 15) and remain
outside possible derivations of their lexical hosts.
3.2.3. Function words: formal and semantic characteristics
Standing at midpoints between content items and inflectional affixes are function
words. The behaviors they exhibit (i.e., their functional and distributional characteristics)
and the meanings they express run the full gamut of those shown by nearly every other
type of element, from content (lexical) to inflectional. Upon careful examination, they
clearly occupy points stretching nearly the entire breadth of a continuum of morpheme
types. However, in all cases, they are members of closed classes. Positionally,
functionally, and distributionally, they represent perhaps the most diverse group in that
they constitute numerous language-specific subclasses with a wide variety of functions and
distributional characteristics. Perhaps their only unifying characteristic is their status as
independent words, even though some of them may receive markers (or assume suppletive
forms to complete paradigms) for tense, person, and so on (e.g., the various forms of the
English auxiliary “to be”).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In any comparative study of languages, function words pose special problems,
especially in view of the fact that what may be expressed by an affix in one language (e.g.,
an agglutinating one) may be similarly expressed by an independent word in another,
especially in varieties that are relatively high on the index of synthesis. Even within an
individual language, certain redundancies may appear: Spanish yo as nominative, first
person, singular pronoun and verbal first person, singular, present tense suffix-o.
Languages such as English which mark number on count nouns show a similar redundancy
in such expressions as “Four-score and seven years ago...”.
Function words are distributed into either nominal or verbal structures or occupy
positions along phrasal or clausal boundaries. For example, determiners such as articles,
demonstratives, quantifiers, numerals, and possessives are relegated to positions within
noun phrases while pronouns can be functional, positional, and distributional equivalents
to entire noun phrases. Auxiliary verbs are considered to be subsidiary to lexical (or main)
verbs, and are often included in descriptions of VP (or as occupying separate or adjacent
syntactic nodes). Adpositions and connectives (various types of conjunctions) serve to link
elements and to indicate logical relations.
With respect to general semantic characteristics which can affect translatability and
borrowability, the articles, in languages such as English, are the “main subset of
determiners” (Crystal, 1991, p. 100) and mark a number of discourse functions such as old
versus new information, definite versus indefinite expression, general versus specific
reference, and so on (Hawkins, 1991, p. 405ff). Others, e.g. demonstratives, quantifiers,
possessives, and numerals, serve a variety of deictic and expressive functions, i.e.,
indicating a broad range of semantic contrasts such as quantity, number, and so on.
Pronouns (some of whose membership overlaps with determiners, above) are traditionally
81
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
included in the set of nominal expressions. Some approaches distinguish among personal,
possessive, demonstrative, interrogative, reflexive, indefinite, and relative pronouns
(Crystal, 1991, p. 281). As a group, their semantic or discourse function may be to
substitute for an individual noun or entire noun phrase, but syntactically, their behaviors
are typically diverse. In addition, they constitute paradigms of limited membership and
may include expression of such grammatical categories as case, gender/class, person, and
number corresponding to those of the noun or phrase they replace. Terminological
conflicts often surface regarding demonstrative, possessive, and relative pronouns, for
instance, because they characteristically have multiple functions.
Auxiliary verbs are considered a special subset of Verb, and, therefore, may be
marked for such grammatical distinctions as tense and agreement (e.g., of person-number
and gender). In some approaches, they are included in descriptions of VP. In languages
such as English and Spanish, auxiliaries may express semantic distinctions such as mood
or modality, aspect, and voice1 4 . However, the semantic distinctions made by auxiliaries in
one language are often found expressed in some form of verbal morphology in another (or
in the same language), especially in varieties that are high on the index of synthesis (e.g.,
polysynthetic languages). Auxiliary verbs do not function to designate the primary action
of a phrase or clause, but may substitute for full VPs in languages that allow for tag
questions or ellipsis, e.g., abbreviated responses to yes-no type questions, e.g., Spanish
“jSf, puedo!” (Yes, I can!) in response to, “^Puede Vd. hablar espanol?” or in such
English sentences as “I love reggae, and my wife does, too.”
14 In traditional descriptions of auxiliaries in English, including be, have, do, and the modals, it is
commonly noted that they (a) have special negative forms (e.g., isn’t, haven't, don’t, couldn’t, and so on);
(b) participate in subject-aux inversion to form questions (and occupy V2 position after words such as
hardly or seldom); (c) replace or refer back to full verbs; have cliticized forms (e.g.. I’m and he’s), and so
on. Modals also ate distinguished by the fact that they do not receive the third-person, present-tense -s.
82
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Adpositions (pre- and postpositions) are often considered to be heads of their
associated phrases, which renders their identification somewhat simpler on formal (viz.,
functional, positional, and distributional) grounds, especially vis-^-vis the other elements
with which they normally co-occur. Distributionally, they mark boundaries of phrases that
can function adverbially or adjectivally, especially those performing temporal and spatial
locative functions. However, as members of a closed class, they perform grammatical as
well as semantic functions: they may be assigners of inherent case (as the adpositions of
German), express logical relations, and indicate abstract grammatical (syntactic)
relationships when subcategorized for by specific nouns, verbs, or adjectives. Some
adpositions (e.g., after, before, and until in English) appear to take entire clauses as their
objects, which causes some to class them as subordinating conjunctions—perhaps another
instance of functional overlap. Consequently, forms recognizable as adpositions may be
classed as adverbs (when they have no overt objects) or particles, adpositions (when they
do), or types of conjunctions.
Various kinds of connectives constitute a diverse class and perform a range of
grammatical functions. They include coordinating conjunctions (or coordinators) that
conjoin syntactically equivalent words, phrases, and sentences (e.g. English and, or, but)
and subordinating conjunctions (e.g., because), relative pronouns (e.g.,who, what, and
which) and (so-called) relative adverbs (e.g., where and when), and other complementizers
(e.g., whether and that), all of which may be markers of embedded sentences (clauses).
As a general rule, such connectives are also positioned at the edges of the respective
elements they function to connect, that is, at phrasal (e.g., NPs, VPs, PPs, etc.) or clausal
boundaries, and, as a consequence, occupy salient positions. There have been proposals to
classify some subordinating conjunctions in English as prepositions with sentential
complements (Crystal, 1991, p. 334— 335; Radford, 1988, p. 133ff). This suggests that
83
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
distinctions between certain subordinators and adpositions may be somewhat blurry in a
particular language, and perhaps even more so cross-linguistically.
While it is beyond the scope of the present work to treat each type of function word
individually and comprehensively, a few remarks can be made based on observations of
borrowability. When the role of the function word is strictly grammatical, for instance
reflexive pronouns of Spanish that are subcategorized for by specific lexical verbs (e.g. -se
of caerse, “to fall”, often translated into English as “to fall down”), it is much less likely to
be borrowed. However, there are a number of instances of borrowed personal pronouns
that can behave like nouns or full NPs—even in English (W. Lehmann, 1992, p. 136):
they often express case distinctions and so on usually reserved for nominal expressions;
can appear in isolation as responses to questions (as in “Who, me?"); and can refer to the
principal people, objects, and qualities mentioned in discourse in ways similar to common
and proper nouns. NP-like behavior ensures perceptual salience as elements capable of
receiving stress, and so on (this may apply to some determiners, as well). Modal
auxiliaries capable of VP-like behavior (as noted above) and adpositions that are heads of
phrases expressing isolable semantic distinctions (e.g. locative under, after, between, etc.)
are likewise capable of representing reasonably transparent and independent meaning and,
therefore, can constitute potentially salient stretches of speech.
3.2.4. Function words versus content words
With respect to syntactic behaviors, the line between function and content words
may seem blurred, which is especially true across languages regarding pronouns and
adpositions (cf. Myers-Scotton, 1993a, p. 99). As the preceding discussion of function
84
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
words points out, content Gexical) and grammatical words are in principle distinguished on
a number of formal bases, most often on that of open versus closed class. Both are
internally cohesive; that is, as independent words, they are identifiable according to similar
internal criteria, e.g, as potential stress-bearing units and other phonological cues such as
pauses or juncture points. In general, however, function words associated with nominal
classes (e.g., determiners) cannot assign markers of such grammatical categories as case or
number or receive agreement markers of person-number or gender independently of the
specific content item with which they co-occur (or replace), signifying their structural
dependence. Those associated with verbal classes (e.g., auxiliaries) are subject to similar
restrictions, with the notable exception of adpositions (which, in some languages, can
assign grammatical case). While main verbs automatically assign semantic roles (and
associated case markers) according to their semantic type, auxiliaries (as a result of their
subsidiary nature) cannot (except in cases of ellipsis and so on, noted above).
3.2.5. Function words versus inflectional affixes and elements in between
As outlined above, inflectional affixes are distinguishable from content items on
numerous formal grounds, and grammatical words are distinct from content items. In any
discussion of the grammatical elements/forms of a language (excluding for the time being
gradations of meaning and the like), distinguishing among grammatical words, inflectional
affixes, and clitics may frequently be one of shifting boundaries. The contrast in
morpheme status (i.e., free versus bound) supports the conceptualization that function
words are formally situated between content items and inflectional affixes (in languages that
allow inflectional affixes). As independent words/roots, syntactic procedures that result in
word order changes may apply to certain kinds of function words (e.g., personal pronouns
85
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
and auxiliary verbs in English). However, regarding affixes in general, language-particular
word formation rules will apply (e.g., inflectional affixes are applied “outside” derivational
affixes). In addition, the functions of both function words and inflectional affixes are in
some sense subsidiary. Function words are generally subordinate to the heads of their
respective phrases (with the exception of adpositions, for example, which are heads of
phrases themselves, and various pronominals which stand in the place of entire NPs). The
functions of inflections are clearly secondary to the stems to which they adhere.
The remaining class in the cline of grammaticality, that of clitics, is traditionally
positioned between grammatical words and inflectional affixes, and are often viewed as
occupying midpoints of form between free and bound. Their characteristic degree of
bonding relative to either derivational or inflectional affixes is, however, slight because
they are neither inflectional nor affixes (Bybee, 1985, p. 121 5 ). With respect to their
functions, they consist of phonologically reduced forms of grammatical words which
coalesce with immediately adjacent forms. In some cases they attach to content items (e.g.,
clitic pronouns of Spanish and articles of French); in other cases they involve two closed-
class function words (e.g., contracted forms of pronoun plus “be” in English; the clitic is
realized as a reduced form and attaches to the preceding pronoun (as in I ’m and we’re).
Cliticized elements (both proclitics and enclitics) are outside the word, including its
possible inflected forms, while derivational and inflectional affixes occur within word
boundaries. Their peculiar mix of behaviors is more a result of their ambiguous
morphological status. For instance, they may be considered separate (but contracted)
words by native speakers (or separable in their full, phonological forms). They constitute
15 Bybee includes clitics with particles and auxiliaries as free morphemes because none are obligatorily
bound to any lexical item. In her approach, clitics merely resemble inflections as a result of positional
restrictions and because they are obligatory in certain contexts.
86
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
sets whose membership is very restricted, and whose distribution is correspondingly small
(e.g., articles in French, certain pronouns occurring at the end of a verb in Spanish, or
negative elements following auxiliary verbs in English). Therefore, they have a very
limited range of functional, positional, and distributional characteristics, limited by the
short list of grammatical forms they reduce from.
Apparently, their dual behavior is an indication of change towards a more general
meaning (Heine et al, 1991a, p. 213). That is, they appear in an increasing number of
contexts, and, as a result, can become phonologically reduced (recognition is assured
through frequency). However, semantic change is not necessarily in the direction of a
more grammaticalized meaning; that is, a new inflectional category is not being represented
that might require a new set (or paradigm) of obligatory markers (cf. Bybee, 1985, pp. 42-
3; Hopper & Traugott, 1993, p. 150f.). As a consequence (when viewed synchronically)
of their position along a scale (or cline) of grammaticalization, clitics, as a general rule, are
not borrowed, i.e., as clitics (cf. Muysken, 1988, 414 and 1981, p. 61; Bakker, 1997, p.
226).
In situations of intense language contact, grammatical elements which have
cliticized to content items are often reanalyzed as an integral part of the (content) word, i.e.,
inside word boundaries, for example Michif lamur (love) from French I’amour (Bakker,
1997, p. 103)1 6 . This is more likely the result of language-specific, donor phonological
and morphological processes that occur coincidentally with various other semantic factors.
For example, clitics usually come from special classes of pronouns, copular or auxiliary
verbs, discourse particles, or, in specific languages, other classes of function word (e.g.,
1 6 Phonological reanalysis of French articles plus nouns is common in French based creoles, as well.
Much of this reanalysis undoubtedly is a result of French liaison and elision rules.
87
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
certain adpositions) from which cross-linguistic generalizations are difficult to make
(Hopper & Traugott, 1993, p. 5). Any reduction of phonological form will make these
form-meaning sets less salient as well (see Chapter 4 for discussion of the role of salience
in borrowing).
3.2.6. Derivational affixes: between content and inflection
There is little unanimity among linguists regarding the exact differences among
types of derivational and inflectional affixes (Katamba, 1993, p. 47; Bybee, 1985, p. 81).
Nevertheless, most agree that, in the more straightforward cases, the application of
derivational morphemes leads to the creation of new, discrete, words or stems to which
inflectional affixes (markers of inflectional categories) may be subsequently added.
(Inflectional affixes, in contrast, create so-called different versions of the same word and
never change the class to which a particular word belongs.) Derivational affixes may be of
two basic sorts, so-called category preserving, those that only change the meanings of the
roots to which they attach in some specified way (e.g., un-, re- and dis- in English) and
category changing, those that change the grammatical class of the root (e.g., -ly, -non, -ize)
(Katamba, 1993, pp. 47, 51; Bybee, 1985, pp. 82-83). Thus, we can easily derive
adverbs from adjectives, nouns from verbs, and verbs from nouns. We can also derive
both nouns and verbs from certain subtypes of adjectives, and so on (e.g. strong —>
strength, strengthen). Both sorts of derivational process result in the creation of a new
word, usually of a similar semantic type (Bybee, 1985, p. 81).
In many cases, derivational morphemes have clearly stateable meanings (e.g., bi-,
supra-) similar to content items; hence, they may be referred to as bound lexical items. For
88
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
example, the English derivational prefix un- may signify negative with adjectives (as in
unable and unhappy) or some sort of reversal of meaning in verbs (for example, in
oppositions of tie, and untie), but it never changes the syntactic class of a root word.
Derivational affixes often make semantic distinctions that can be expressed lexically (i.e.,
with a separate lexical item), for example inoperable versus not operable, or redo that,
please versus please do that again. While derived forms may have complex meanings
(consisting of radical plus derivational concept, in Sapirian terms), derivational morphemes
themselves have either one associated meaning or one function (e.g., nominalizing or
verbalizing). In any case, they have one-to-one correspondences of form and meaning and
resemble agglutinating affixes in that respect—one might say they are agglutinating without
being clearly inflectional.
Semantically, derivational morphemes lie at points between content items (words or
bound roots) and inflectional affixes, between lexical and inflectional expression (Bybee,
1985, p. 12), and may share characteristics of both. On the one hand, derivational form-
meaning sets can be quite idiosyncratic (e.g., retro-, mini-, -dom, -ship, -ful, and so on),
and, on the other, they are bound (i.e., they require bonding to a contentive root) and can
never appear in isolation (as independent words). Among the usual criteria employed to
distinguish between derivational and inflectional affixes is obligatoriness (Bybee, 1985, p.
27; cf. Carstairs-McCarthy, 1993, p. 174ff), which in essence reflects the grammatical
status of the respective types. That is, the expression of inflectional categories of person,
number, and (occasionally) class agreement and verbal categories of tense, mood, aspect,
voice, and so on is required in languages that have those inflectional categories (according
to some approaches, as a result of a syntactic rule or procedure). The application of
derivational affixes is applied according to semantic criteria linked to their ability to provide
89
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
specific semantic modification or a change in word class, and, as such, are options for
which there may be a number of alternatives.
However, in the cross-linguistic comparison of derivational and inflectional
processes, lines may not be easily drawn. For instance, in one language, a particular affix
may be considered derivational, while in another, a semantically similar affix may be
considered inflectional, thus indicating a greater degree of grammaticalization. For
example, a number of languages (e.g., Diegueiio and Kwakiutl) express the normally
inflectional category of number derivationally (Bybee, 1985, p. 103). Even within a
particular language, a particular suffix may appear to be inflectional in one instance and
derivational in another (e.g., English -ing in “He is reading”, which marks progressive
aspect, or in the derivation of a noun from a verb form, as in “Reading is fun”). As a
consequence, distinctions among derivational and inflectional affixes may be somewhat
less clear and subject to controversy than between affixes, in general, and stems or roots.
To illustrate the gradient nature of many such distinctions, Greenberg (1966)
proposed a number of implicational universals with respect to morphological types. For
example, he states, in essence, if a verb (in a specific language) is marked for person-
number (agreeing with the grammatical subject), then it has categories of tense and
modality (Universal 30). If the verb agrees with grammatical subject with respect to
gender, then it also agrees in number (Universal 32); if a language has the category gender,
it has the category of number (Universal 36). Thus, we can arrange a hierarchy of
inflectional concepts according to (4), below (items to the left representing more basic, i.e.,
frequently encountered, categories in his sample of languages):
(4) tense-aspect-modality > number > gender
90
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Greenberg also states implicationally (Universal 29) that any language which has
inflectional morphology (i.e., inflectional affixes) has derivational morphology (i.e.,
derivational morphemes). As a result, the following hierarchy (adapted from Croft, 1990,
p. 191) illustrates the gradient and hierarchical relationships that exist among derivational
and inflectional concepts:
(5) derivational concepts > tense-aspect-modality > number > gender
Despite the frizziness of discussions of derivational morphemes, it is still
sufficiently clear that the form-meaning sets used in the application of word-formation
processes that are unequivocally derivational show (a) unifunctionality (i.e., one stateable
meaning or a single function) and (b) clear segmentability (distinct morpheme boundaries)
whether the particular affixes involved might be considered expressions of inflectional
categories or not. These are two significant factors that will figure prominently in the
subsequent discussion of transparency and salience (Chapter 4).
3 .3 . Sem a n tic t y p e s: g r o u pin g s o f m o rph em es a c c o r d in g t o m e a n in g s
Linking borrowability and grammaticality assumes a twofold interaction of form
and meaning. In the previous section (3.2.), discussion centered on various sorts of forms
(morphemes and words), morpheme types, and word classes, and their various
characteristics, assuming that the principal task in a borrowing situation is to (a) identify
form-meaning sets that are compatible with the recipient system and (b) integrate those
form-meanings sets into the recipient language. To accomplish (a), individual proficient
bilingual speakers of the recipient and donor (a subset of the total number of members of a
91
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
bilingual community) must be able to recognize and isolate forms in the donor (see Section
3.1, above). We can conclude that they do, indeed, possess the requisite morphological
knowledge in each of their languages, despite that fact that in most cases of borrowing, the
recipient is their dominant (first or native) language. They know, however unconsciously,
the roles that different morpheme types play in the formation of words and word classes
and in the formation of larger discourse units (e.g., phrases), being sufficiently able to
distinguish among such forms as content-bearing items (free or bound), function words,
and types of inflectional affixes17.
However, a more semantic focus is most likely required to accomplish (b), above,
that is, to successfully integrate borrowed form-meaning sets into semantic types in a
recipient lexicon. It is proposed here that, in the case of loanwords, concepts and their
accompanying labels are borrowed, not form-class members (see 3.0.1, above). Partly
due to the questionable nature of such terms as “semantic bleaching” and so on, a great deal
of attention has been paid in recent grammaticalization studies to various kinds of
meanings, especially as they may lend themselves to placement along scales. Meaning
types, representing clusters of semantic characteristics that can be situated along a cline
(hence, continuum), are especially relevant to the present work because they have the
greatest potential to shed light on the semantic aspects of borrowability.
One way that has been advanced to categorize meanings is to posit a scale
representing the evolution of a concept as it is abstracted away from its original concrete
meaning (i.e., direct reference to person or object) towards an increasingly general,
relational, and, hence, abstract concept (cf. Lyons, 1968, p. 406). To this end, for
17 As mentioned elsewhere, it is taken for granted that the individual bilingual's knowledge of each
language in his/her linguistic repenoire will be situated along a continuum of proficiency.
92
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
example, Heine et al (1991a) construct a scale of meaning types (in (6), below) based on
ontological categories that depict general domains of conceptualization according to which
we are able to view and organize experience. Individual categories represent prototypical
entities that encompass a wide range of perceptually based and, therefore, linguistically
expressable concepts (p. 49). The particular ordering is based on degrees of metaphorical
abstraction; that is, one category may be used to conceptualize another immediately to its
right (so-called categorial metaphors), for example OBJECT-to-SPACE or SPACE-to-TIME. It
is also consistent with the ways languages typically represent entities according to kinds or
degrees of animacy (cf. Croft, 1990, p. 113; Comrie, 1989, p. 42f. and p. 185ff.)18.
(6) PERSON > OBJECT > ACTIVITY > SPACE > TIME > QUALITY
An example of this in English would be the word “back”, which, as a source concept
originally signifying a body part, may refer to position in space or time, and then (in
compound form) to an attribute or quality, “backwards” (Heine et al, 1991b, p. 151).
These authors also posit a correlation between metaphorical (ontological) categories and
both word class and constituent type (construction) (Heine et al, 1991a, p. 53), portrayed
in (7), below:
18 For further discussions of the role of metaphor in grammaticalization processes, see also Traugott &
Hopper, 1993, p. 86-87, Heine et al, 1991a, p. 48ff. and 1991b, p. 157ff; Traugott & Kdnig, 1991, pp.
190,207-212; and Keesing, 1991, pp. 325,334-6 (cf. Bybee & Pagliuca. 1985 and Sweetser. 1988).
93
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(7)
Category Word Type Constituent Type
PERSON
OBJECT
ACTIVITY
SPACE
TIME
QUALITY
Human noun
Concrete noun
Dynamic verb
Adverb, adposition
Adverb, adposition
Adjective, state
verb, adverb
Noun phrase
Noun phrase
Verb phrase
Adverbial phrase
Adverbial phrase
Modifier
Numerous linguistic phenomena are used to support this generalization (and those
of its type); one example is the range forms and meaning types elicited in response to the
so-called Wh-words (interrogative pronouns) who, what, where, when, and how—a class
of forms held to be fairly constant across the world’s languages. While other categorial
lists of this type have been proposed (e.g., Jackendoff, 1983; Clark, 1993, p. 43ff.; cf.
Landau, 1993, p.l91ff), the categories proposed by Heine etal (1991a) should be
sufficient for the present purposes. One may wish to add a category or two for specificity
(e.g., AM OUNT), but this list is specific enough to allow the organization of semantic types
into ordered groups (that, incidentally, would qualify as responses to basic who, what,
where, when, and how question types).
Any process of abstraction will involve the extraction of a basic or core meaning
and its extension (broadening) in some fashion along a scale of this type. However, it is
also necessary to view form-meaning sets (words) according to the amount of detailed
information they entail. For example, people usually know more about the persons and
objects that are near to them; children show this tendency by learning words for the people,
things, and activities in their immediate environment first (e.g., Clark, 1993, p. 30; cf.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Markman, 1989, p. 5ff.). These items represent their most specific and concrete referents,
and descriptions of them will be based on direct (specific) and tangible (concrete)
experience. Close objects have clear edges and are more clearly structured; distant ones
have blurred edges and are much less structured (Heine etal, 1991a, p. 44). Metaphorical
processes allow us to apply our direct experience with tangible objects and the concepts
they help organize (e.g., into taxonomies of similar objects) to more distant, therefore
abstract, concepts. This tendency to go from close to far, concrete to abstract, and specific
to general provides one basis for claims of unidirectionality in processes of
grammaticalization (cf. Traugott & Heine, 1991, p. 4ff.; Traugott & Konig, 1991, p.
192ff.; Heine etal, 1991a, p. 50; Hopper & Traugott, 1993, p. 94ff.), though it is quite
clear that not every term that becomes abstract and general will become grammaticalized1 9 .
In the remaining portions of this chapter, the kinds of meanings that fall under
headings provided in the above scale of metaphorical/ontological categories are combined
with a comprehensive list of semantic types to portray the layered ways meanings and their
corresponding forms may be organized in a language, vertically according to ontological
category and horizontally according to semantic type-morpheme type correspondence. The
purpose is to construct a set of criteria by which distinctions can be made among the types
of words and morphemes that are starting and end points of borrowing processes,
distinctions that can be used to identify more specific characteristics of borrowed forms. In
so doing, it may be possible to predict more precisely which forms are the most or least
likely candidates for borrowing.
19 See Section 4.2.1. for a more thorough discussion of abstraction.
95
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.3.1. Semantic types and subtypes of N, V, and Adj
There have been numerous suggestions concerning the ways languages (and their
users) actually organize words into semantically based sets and types, for example in
studies of child language acquisition (see, e.g., Clark, 1993 and Markman, 1989, 1994).
One rather extensive study of English (upon which the ensuing discussion is primarily
based) was done by Dixon (1991), who set out to identify possible groupings with an eye
towards the ways languages, in general, encode concepts. One assumption (shared here)
was that syntactic (word) classes may vary considerably across languages (influenced by
morphological structuring), but the basic kinds of concepts to which languages must make
reference and, perhaps, need to express may be quite similar based on human experience.
Drawing from the 2000 most frequently used words in English (from West, 1953), Dixon
proposes a listing of semantic types and subtypes of content items which are linked to
classes of Noun, Verb, and Adjective. About 900 of these were verbs that were organized
into types based on their syntactic and semantic characteristics (i.e„ similarities of
meanings, number and type of semantic roles, and other characteristic syntactic behavior).
In all, there were five major Noun types, ten Adjective types, and six Verb types.
Among types of Nouns, there were CONCRETE, ABSTRACT, STATES (or PROPERTIES),
ACnvmES, and SPEECH A CTS, each divided into numerous subtypes. Among Adjective
types, there were those expressing DIMENSION, PHYSICAL PROPERTY, Speed, AGE,
COLOR, VALUE, DIFFICULTY, QUALIFICATION (e.g., definite, possible, etc.) and HUM AN
PROPENSITY (some organized into subtypes). Major Verb types include PRIMARY-A A N D
B, AND SECONDARY-A through SECONDARY-D (all divided into various subtypes). In
general, PRIMARY verbs make direct reference to an activity or state, while SECONDARY
types provide some sort of semantic modification to another (main) verb (e.g., modals,
96
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
semi-modals, or other auxiliary verbs). PRIM ARY-A verbs are distinguished from
PRIMARY-B on the basis of complement types; the former take only NPs as grammatical
subjects and objects, while the latter may take either NPs or complement clauses in those
grammatical functions.
Assuming that Dixon’s groupings are reasonably comprehensive (at least for
English), we divide again the semantic types and subtypes according to the six
metaphorical categories delineated above in the following sub-sections (3.3.2-3.2.7).
Nouns are separated according to kinds of referents, for example, whether direct reference
is made to a person, object, activity, and so on. Most verbs represent activities, but can be
sub-divided according to semantic roles, for example whether the semantic role that
obligatorily maps onto the usual grammatical subject (external NP) must be filled by a
human (person), an animal (animate object), or other. Adjectives normally represent
qualities of some sort but (as with verbs) may be restricted with respect to the types of
referents to which they may be applied (e.g., person or animate object). In each section,
we also include more grammaticalized forms traditionally associated with a particular
meaning type. For example, included in the category of PERSON are pronouns and various
inflectional affixes that are markers of person: in the category of SPACE are adpositions and
an assortment of inflectional affixes. These additional forms are included because they
provide a category-by-category listing of forms linked to specific meanings indicating
degrees of grammaticalization and reflecting the semantic tendencies we seek to
characterize.
At this point, a note about animacy seems in order. In languages such as English,
there appears to be a fairly clear delineation among general semantic categories or domains:
there are human, animate, and inanimate referents. It should be clear that among the
97
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
world’s languages, not only will these distinctions not necessarily correspond to those
made in English20, a particular language may possess a relatively long list of semantic
categories or word classes based on such distinctions (among others) as animacy.
3.3.2. Person
Forms which refer to the category of PERSON may belong to classes of Noun and
Pronoun, and various inflectional person and agreement markers. Nominals referring to
PERSON are all included in the semantic class of CONCRETE nouns, for example kinship
terms (father, mother, sister, and so on), social group (e.g., nation, tribe, army, crowd...),
and rank (chief, captain, etc.). Most words referring to social groups cover a range of
terms that are, in essence, more general than those referring to individuals and may, in
many cases, require more language (and culturally specific) knowledge than basic kin
terms. Terms denoting rank also require knowledge of specific patterns of social
organization.
These and other similar terms can be viewed from near to far, as well, for instance
from the standpoint of an individual language learner (e.g., a child): body parts, family,
rank, and social group. In addition, individual subgroups like body parts (human or non
human animate) can be applied metaphorically to inanimate objects (the mouth of a river).
Source concepts such as foot may occur (with metaphorical meaning) in any of the
subtypes. Body parts also offer a convenient pool for terms that can be utilized in spatial
orientation (e.g. points of reference), and as such, constitute possible source concepts for
20 In Manam (New Guinea), for example, has a category of animacy that includes higher order animals
(pigs, dogs, birds, and so on) along with humans. Generally, animals are considered higher order when
domesticated, optionally so when wild (Croft, 1990,113).
98
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
grammaticalized (relational) forms. As a result, this semantic subtype (within the type of
PERSON) has great potential for grammaticalization (Heine etal, 1991a, p. 34).
Pronouns can refer to both persons and objects; certain kinds (i.e., personal
pronouns) may also occupy positions of grammatical subject, object, and so on. They can
receive any semantic role that can be associated with nouns, noun phrases, or nominal
complements and are also subject to the same syntactic processes or patterns of occurrence
as nouns, for example in Subject-AUX inversion in English direct questions. In contrast,
inflectional affixes indicating PERSON distinctions are applied according to language-
specific word formation rules. In many languages, verbal markers co-exist alongside full
pronouns, which may render pronouns superfluous in many (non-emphatic) contexts, for
example, in so-called Pro-drop languages, though there may be other pragmatic
considerations that contribute to this tendency.
In specific languages, full pronouns and affixes marking person may be contrasted
in other ways, as well. For example, Spanish fusional affixes marking tense and person-
number agreement do not mark gender while the full pronouns (£1, ella) do, agreeing with
their referents in person, number, and gender. Hence, in addition to being more
grammaticalized than pronouns, these affixes have more generalized meanings that are
capable of referring to either gender: while they often consist of only a single vowel
(stressed or unstressed), they also indicate conjugation class as well as tense-aspect,
resulting in a 1-to-many mapping of form to meaning(s). A different sort of verbal
inflection exists in Modem Israeli Hebrew; while it preserves gender in the third person,
singular, present tense (as a separate suffix marking the feminine), tense and aspect are
usually indicated by certain vowel patterns that occur between root consonants (cf. Bybee,
1995, p. 233 in reference to Arabic). With different aspects of meaning distributed at
99
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
different points in the word, the contrast between full pronoun and affix in Hebrew is
certainly complex, clearly showing long term results of grammaticalization.
3.3.3. Object
Content items that have visible and tangible referents are normally classed as
CONCRETE nouns. Most CONCRETE nouns are underived roots, but some may be derived
from verbs, for example pig, flower, moon, valley, or window, but building from the verb
“to build”. Depending on their position on an animacy scale in the particular language,
CONCRETE referents can be AN IM A TE (from domestic to wild animals) or INANIM ATE,
which can cover a wide range, including flora (flowers, fruits and other edible plants or
parts of plants, trees, and the like), celestial and weather (sun, moon, star, wind, storm,
rain, snow...), environment (air, water, stone, metals, forest), and artifacts (including
names for types of lodging, articles of clothing, tools and utensils, transportation vehicles,
and perhaps foods).
ABSTRACT nouns, which have referents that are less visible and tangible, may refer
to results, products (of processes) or abstract notions relating to various types of activities.
For example, LANGUAGE terms refer to Oinguistic) sounds, words, phrases, or sentences
and GENERAL ABSTRACT TERM S refer to such concepts as ideas, problems, methods,
results, truth, and so on. Such referents, though not visible in the sense of stone or
coyote, are, nonetheless, tangible, at least in the imagination of the speaker.
In similar fashion, the category termed STATES and PROPERTIES also makes
reference to entities that exist in a more psychological sense and typically refer to aspects of
100
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
human existence. For example, this subtype of noun typically refers to M ENTAL (e.g.,
emotions) or CORPOREAL (hunger, pain, strength) categories; many are bound roots, but
others are derived from adjectives (e.g., jealousy from jealous). The direction of derivation
may also be the reverse, for instance envious (the adjective) from envy (the noun). If, in a
language there is some categorial variation, then, cross-linguistically, one would suspect
that the syntactic category employed to represent a particular semantic type may be subject
to considerable variation.
3.3.4. Activity
Words referring to an ACTIVITY generally belong to two separate syntactic classes,
Noun and Verb. Words referring to (i.e., naming) activities, acts, or states as entities are
nouns, and, as such, they can occur as grammatical subjects, objects, and so on and be
morphologically marked as other members of the Noun class. While most nouns
(independent word or roots) are derived from the word class Verb (e.g., division, thought,
and seat) some are basic, non-derived forms (e.g., war and game) , . For nearly every
ACTIVITY noun, there is a corresponding verb, whether that verb is a cognate (i.e.,
represented by the same, basic phonetic label) or not (e.g., game the noun and play the
verb).
Included here are nominal expressions referring to SPEECH ACTS and STATES (and
PROPERTIES) because many of the concepts they represent can be referred to as (a) entities
and function as nouns or as (b) acts, activities, events, or processes (terms used
traditionally in descriptions of classes of Verb) and function as verbs. With respect to
SPEECH ACTS, in every instance, there is a corresponding, often cognate, verb, e.g.,
101
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
response-respond and command-command, but question!request-ask and story!the truth!a
lie-tell. Regarding basic sorts of STATES (and PROPERTIES), there are two main types:
M ENTAL (e.g., pleasure, joy, honor...) and CORPOREAL (e.g., hunger, thirst, etc) with
subtypes of ACHE and STRENGTH. Some terms for STATES (and PROPERTIES) are derived
from adjectives (e.g .Jealousy—> jealous) while others may be derived from verbs (e.g.,
delight).
In most instances, however, ACTIVITY and its closely related types belong to the
general class Verb, and occur as the central elements of verb phrases (or predicates). Verbs
are clearly the most diverse of the major word classes in terms of syntactic behavior (most
likely resulting from the ways semantic roles map onto grammatical relations).
Semantically (and, consequently, syntactically), PR IM A RY VERB types usually require other
elements—the bare minimum perhaps being some sort of grammatical subject (which, if
understood by context and so on, may be left unexpressed). They may also require
grammatical objects, complements, and so on, which function according to specific
semantic roles (e.g., agent, instrument, etc.). SECONDARY verb types (e.g., modals or
other auxiliaries) most often co-occur with PRIM A RY types and may be morphologically
marked for (verbal) inflectional categories such as tense or aspect, thus resembling primary
verbs. However, their meanings will be subsidiary in some fashion to the main (or
PRIMARY) verb.
Among the PRIM ARY-A verbs, M OTION verbs may be the most relevant for the
present purposes, being identified as a semantic type with potential for grammaticalization
(Hopper, 1991, p. 20; cf. Hopper & Traugott, 1993, p. 109). Many of the members of
this particular subclass are everyday activities (i.e., core vocabulary) and often have
accompanying semantic roles that must be filled by CONCRETE, ANIM ATE subtypes (e.g.,
102
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
as grammatical subjects), and are, presumably, among the most frequently used forms
(Hopper & Traugott, 1993, p. 114)
3.3.5. Space
Forms representing spatial concepts can belong to a number of grammatical classes
and morpheme types. Cross-linguistically, they can cover a variety of meaning types from
concrete and abstract reference (e.g., place names and directions, respectively) to relative
position (adverbials such as here, there, and somewhere) to more grammaticalized locative
meanings linked to adpositions and certain case markers. It is safe to say that words and
morphemes relating to spatial concepts may cover a full range of form-meaning sets
contained on a cline of grammaticalization or hierarchy of borrowable types.
Concrete place names can serve as points of reference; they include reference to
environmental objects such as forest, river, and hill or artifacts such as house, market, city
(particularly when marked for definiteness, as in the river or the market, respectively).
Abstract nouns represent such concepts as position or direction (e.g., front, edge, north) or
units of measurement (e.g., meter, mile, foot). Function words such as adpositions can
indicate location (on, beneath, and at) or direction (to and from), but only relationally, i.e.,
one object in relation to another.
103
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.3.6. Tune
Similar to spatial concepts, forms representing temporal concepts belong to a
number of grammatical classes from noun to affix. They span a range of meanings from
ABSTRACT reference to units of time (e.g., month, day, night, season) and more general
terms indicating time (as an abstract notion) and position in time (e.g., time, present, past,
future, yesterday) to abstract notions of tense and aspect. At midpoints between these two
extremes are adverbials (e.g., now and then); forms that can be adverbs, subordinating
conjunctions, or adpositions in languages like English, depending on grammatical function
(e.g., before and after); and such function words as conjunctions (e.g., while) and
adpositions (e.g., during) and those adpositions that occur with temporal complements
whose meanings have been metaphorically applied from spatial expressions (e.g., in, on,
and at used with such NPs as “the morning”, “the third day”, “five o’clock”, respectively).
3.3.7. Quality
Form-meaning sets that refer to Q U A LITY may refer to a broad range of loosely
related concepts (cf. Heine et al, 1991a, p. 49). At one end of the spectrum are ABSTRACT
entities (organized into classes of Noun) such as VARIETY (e.g., type, kind, character,
shape, and types of shapes such as circle, line, and so on) and concepts referring to a wide
variety of qualities conceptualized as A BSTRA CT entities. In languages like English, some
QUALITY nouns are derived from adjective classes (for example, redness, happiness, and
narrowness from semantic subtypes COLOR, HUM AN PROPENSITY, and DIMENSION,
respectively). One particular type, QUALIFICATION, is composed of several subtypes that
are obvious semantic relatives to M O D A LS (SECONDARY-A verbs). These are the subtypes
104
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
of DEFINITE (e.g., probable, true), POSSIBLE (e.g., impossible), USUAL (e.g., normal,
common), LIKELY (e.g., likely, certain), SURE (one member), and CORRECT (e.g., right,
wrong, appropriate, and sensible).
Various qualities (or types of QUALITY) may be associated in semantically restricted
ways to only particular sorts of entities. For example, the HUM AN PROPENSITY type can
only be used to add particular semantic information to (modify) CONCRETE (HUM AN and
ANIM ATE) referents, to refer to states of mind, emotions, and so on. This type consists of
a number of subtypes, for example FOND (fond of...), ANGRY (e.g., angry, jealous of,
sad), HAPPY (e.g., anxious, keen, happy, thankful, careful, sorry, proud, ashamed, afraid
of), UNSURE (e.g., certain, sure, unsure, curious about), EAGER (e.g., eager, ready,
prepared, willing), CLEVER (e.g., stupid, luck, kind, cruel, generous).
Generally speaking, the following types also refer to qualities attributable to
CONCRETE (ANIMATE or INANIM ATE) entities: DIM ENSION (e.g., big, small, narrow,
short, tali), PHYSICAL PROPERTY (e.g., hard, strong, sweet), SPEED (e.g., quick, slow)
, COLOR (e.g., red, yellow, blue). A subtype of PHYSICAL PROPERTY, CORPOREAL,
most often refers to ANIM A TE types (e.g., well, ill, absent). However, a number of types
may modify a wider range of CONCRETE or ABSTRACT referents, for example AGE (e.g.,
new, old, young, modern), DIFFICULTY (e.g., tough, simple, complex), and SIMILARITY
(e.g., unlike, different from), VALUE (e.g., good, bad, odd, necessary, lucky).
As a general semantic type, meanings associated with classes of Adjective appear to
be more general as a whole than those of Noun; viz., an adjective may appear in a greater
number of contexts. The concepts (attributes) they represent can be applied to more than
one noun or grouping (subtype) of nouns (e.g., COLOR terms with almost any CONCRETE
105
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
noun). Even though one would never wish to say that adjecdves are grammadcalized from
nouns (in the sense referred to in the current work), there is, nevertheless a certain
asymmetry that underlies their relationship as modifier and modified that surfaces in cross-
linguistic patterns of gender/class, number, and other sorts of agreement (Hawkins, 1988,
pp. 7-9)—in languages that have such agreement, adjecdves will agree with the nouns they
modify and not vice versa. As mentioned above, one can find frequent nominalized forms
of the concepts contained in subtypes of Adjective in English; a number of verbs are
derived from adjectives, as well (e.g. redden, strengthen, and widen from COLOR,
PHYSICAL PROPERTY, and D IM ENSION subtypes, respectively). If a set of meanings can
freely move from word class to word class (derivationally) with relative ease and frequency
in English, then we can look for similar processes to operate cross-linguistically, as well.
As a category of concepts that are attributable to a wide variety of referents, types
referring to QUANTITY may, indeed, need to be considered a separate category21; however,
because of their importance in the description of persons and objects (of various types),
QUANTITY is included here as a subcategory of QUALITY. Form-meaning sets also range
from making reference to concepts of QUALITY as abstract entities (e.g., number, amount,
age, width, depth) to the expression of various semantic contrasts associated with the
classes of Determiners (function words), to grammatical categories of number (e.g.,
singular, dual, plural, paucal, uncountable) that map onto inflectional nominal, and (by
agreement) verbal affixes. In languages that have a grammatical category of number,
QUANTITY terms may be distributed (in complementary fashion) among objects that are
countable or uncountable (for instance, in languages like English).
21 As does Jackendoff (1983), whose ontological categories include THING, ACTION, EVENT, PLACE.
DIRECTION, MANNER, and AMOUNT. Without arguing for or against this listing as opposed to any other
(or a synthesis of any sort), we include QUANTITY as a type within QUALITY, for the stated reasons.
106
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.4. S u m m a r y a n d c o m m en ts
In this chapter, it has been proposed that lexical borrowing involves the association
of borrowed form-meaning sets into recipient semantic types and their subsequent
assignment into syntactic word classes. In steps, this includes (a) the recognition of
system compatible forms, (b) the linkage of appropriate meanings onto those forms, (c) the
assignment of form-meaning sets into semantic types, (d) subsequent allocation into word
(form) classes, (e) and the full integration of the borrowed form into recipient
morphosyntax. It is likely that every borrowed form-meaning set undergoes degrees of
reanalysis (phonological, morphological, etc.) when being integrated into the new (host)
system. With respect to word class, they will probably occur in recipient classes that are
similar or equivalent to those from which they originated, but this may be only
coincidental. The degree of similarity of donor and recipient word class may merely be a
function of the type of meaning expressed by the borrowed form.
The synchronic investigation of relationships between word classes and semantic
types has given indirect but significant support for this proposal by pointing out that certain
concepts can be represented by members of a number of syntactic classes whose forms
may function in various ways, even in languages like English. Types of ACTIVITY, for
example, can be members of the two great form classes, Noun or Verb, depending on their
grammatical function in a sentence; some QUALITIES may be members of any major word
class (e.g., strength, strengthen, strong, strongly). Certain types of meanings (e.g., terms
for body parts and so on) appear to weave their ways through a wide range of form types
from content items with CO N CRETE reference to inflectional affixes with highly
grammaticalized meanings. Other concept types may become more abstract and general,
but never actually become fully grammaticalized (as in processes of derivation). It can be
107
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
inferred, then, that the ways concepts organize into syntactic classes may depend in large
part upon language-specific morphological structuring—and there is no reason to assume
that similar investigations of other languages would yield significantly different results.
Patterns have emerged in many studies such as the one done by Dixon suggesting
that the creativity of human language has certain specific characteristics and limitations.
For instance, form-meanings sets that undergo grammaticalization pass through a gradual
evolution of form and meaning: forms become more frequent and phonetically less
complex (in general); their associated meanings become more general (allowing them to
occur in a greater number of contexts), more abstract (moving away from reference to
visible, tangible referents), and eventually associated with obligatorily expressed
grammatical (inflectional) categories. Borrowability mirrors grammaticalization: the more
grammaticalized an element is in the donor language, the less likely it is to be borrowed into
a recipient. Hence, the links between processes of grammaticalization and those of
borrowing are most likely the result of the systematicity of language and the language
faculty—correspondences are too close to be random.
We now examine the characteristics and properties that are linked to
grammaticalization, especially those that may help account for the striking parallelism.
With respect to form, phonetic substance has the ability to influence the recognizability of
specific forms and form classes; diminution of form, therefore, is a factor. Regarding
meaning, generality and abstractness of meaning and linkage to inflectional categories are
semantic concomitants. Consequently, the next chapter goes into greater detail regarding
these properties and their roles in borrowing.
108
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4 .0 . CHAFFER 4— THE IDENTIFICATION OF FORM-MEANING SETS
Apart from the social forces inherent to all bilingual/contact situations, any
comprehensive model of linguistic borrowing and borrowability needs to offer some sort of
plausible explanation for observations that the more grammatical(ized) an item is, the less
likely it is to be borrowed (all borrowed forms being necessarily compatible with the
recipient morphosyntax). Even after defining which donor forms are potentially compatible
with the recipient/matrix system and those that are not, questions remain regarding why,
among all borrowable (i.e., system compatible) forms, certain types of form-meaning sets
are preferred over others. Moreover, the evidence suggests that individual speakers do not
borrow members of abstract, language particular classes. Initially, they borrow form-
meaning sets (i.e., new concepts with associated labels) and then, perhaps, forms only
(new labels for concepts already known and for which forms already exist in the recipient);
they may also borrow new and unfamiliar concepts that can be represented with native
forms Goan translations and semantic extensions). In any case, for borrowing processes to
take place at all, forms must be picked out of the linguistic environment and linked to
appropriate concepts. In other words, form-meaning sets must be learned; then they can be
put to use.
It certainly does not seem to be coincidental that intense bilingual/contact situations
also show a variety of contact phenomena involving different kinds of language mixing
(Grosjean, 1982,1995, 1997, in press). The more intense the bilingualism, the more one
expects bilingual phenomena. With respect to how bilingualism is attained, there are only
two basic routes: (a) the sequential acquisition of a first (native LI) and then subsequent
109
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
language (second or L2) and (b) the simultaneous acquisition of both languages1 . Hence,
in any contact situation (on individual and societal levels), the first stage in bilingual
acquisition necessarily includes the identification of forms in each language and the
successful matching of meanings onto those forms. This, of course, applies to all kinds of
learning/acquisition situations, primary, secondary, or other. In simultaneous acquisition,
certain form-meaning sets in both languages are learned more or less simultaneously, while
in sequential acquisition, forms (i.e., from the L2) will be learned for concepts that already
possess labels (in the LI). In the latter case, significant effects of transfer are to be
expected; in the former, various kinds of mixing behaviors are likely to occur. In addition,
the more proficient speakers become in their languages, the more their individual resources
(forms and structures) become available (cf. Kroll & de Groot, 1997, p. 170).
One can say intuitively that content items are easier to identify and incorporate into a
borrowing system than, say, inflectional affixes. Among other things, they are generally
more semantically transparent and salient and, therefore, easier to pick out of the speech
stream. Because this undertaking appears to be essentially the same irrespective of the
source language, it can be inferred that candidates for borrowing (donor forms) and
original, native forms may be learned in many of the same ways2. Consequently, the
current chapter follows a basic learner’s strategy (cf. Slobin, 1985) regarding spoken
language and attempts to identify particular properties that may coincide in varying types
and degrees to render one form-meaning set more readily identifiable than another.
Isolating the phonological properties of items that facilitate identification appears to be a
1 Distinctions between the two acquisition scenarios obviously depend on the age of onset of L2 learning,
which can vary. Consequently, phenomena associated with one or the other may differ in individual cases.
2 Degrees of attainment will ultimately depend on a number of factors such as acquisition history (i.e..
according to sequential or simultaneous acquisition), function (i.e., with whom and for which purposes the
languages are acquired), modality (e.g., through spoken and written media), and so on, but most likely on
an interplay of all the above (see Grosjean, 1995; Kroll & de Groot, 1997; Smith, 1997; Green, 1993; de
Groot, 1993).
110
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
logical place to start Analyzing the types of meanings constituting the semantic arsenal of
a particular language naturally follows—especially in view of the fact that, in practice, the
two are inextricably linked.
4.0.1. The spread and integration of borrowed form-meaning sets
There are a number of issues of a practical nature to consider with respect to the
integration of borrowed items into a recipient/matrix language. For instance, content items
become members of open classes where overlaps of meaning (degrees of synonymy),
hyponymy, and so on are quite usual. In fact, the hierarchical organization of many
semantic fields allows the addition of new terms on a number of levels, for example, in
names for individual, characteristic parts, or groups of entities (including persons, animals,
objects, places, and so on). In contrast, borrowable function words and inflectional affixes
constitute significantly shorter lists as members of closed classes. In addition, they may
face stiff competition from native forms, which are also relatively few in number, already
solidly entrenched, and very frequent in normal speech patterns as a result of their roles in
the structural organization of the language. The emergence of new grammatical distinctions
(i.e., obligatory inflectional categories) on the basis of borrowed forms seems, on its face,
to be a much more difficult and complex process, one that would take a relatively long
period of time, if it were to occur at all (cf. section 2.1.3, above)3.
Another factor affecting the spread of donor forms and form-meaning sets
throughout an entire community is the borrowing population’s gradient levels of
3 Clearly, inflectional affixes have been borrowed. Thomason & Kaufman (1988) point out the borrowing
of an aspectual marker from Balochi into Brahui (p. 93), and Heath (1981) mentions numerous instances of
borrowed case markers among the Australian languages of Amhemland (p. 336).
I ll
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
proficiency in the donor language. So-called “balanced bilinguals” (a classification
generally accepted as ideal rather than actual) should be able to borrow freely from either
language, lifting the upper limit on borrowability to those forms that are leamable to them.
However, not every member of the bilingual community, a large portion of whom are
assumed to be primary speakers of the subordinate variety, will have the proficiency
required to command and borrow any and all forms of the donor language. In fact, it is
rare that an individual (let alone a community) is equally proficient in both (or all)
languages, especially with respect to a broad range of genres and registers (Romaine,
1995, p. 19; Hoffmann, 1991, p. 21; Grosjean, 1982, p. 232). Even supposing that
relatively proficient bilinguals are the initial conduits of borrowing (a relatively small sub
set of the total set of bilinguals), individual borrowed forms must diffuse throughout the
community among speakers of a wide range of proficiencies.
While it may not be completely accurate to say that borrowing is, therefore, reduced
to the lowest common denominator (e.g., to the borrowing of concrete nouns by
monolingual speakers of the recipient), a lack of proficiency in the donor may inhibit some
members of the bilingual community from having complete access to the more
grammaticalized form-meaning sets. That is to say, the list of potential candidates for
borrowing appears to be constrained by the collective abilities of another perhaps greater
sub-set of the community of bilingual speakers to (a) successfully identify donor forms and
(b) associate appropriate meanings onto those forms. Exposure to these forms may come
from three sources: monolingual speakers of the donor; bilingual speakers of donor and
recipient (speaking either language or in code-switehes); and monolingual speakers of the
recipient who have already borrowed the forms. If this reflects a sequence of any kind (as
forms from Y gradually pass into the speech of monolingual X speakers), one would
expect a kind of filtering or sifting effect: only the strongest forms will survive.
112
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
There may be a number of reasons why a donor form wins out over a more
traditional, native one in such situations. Frequency in the donor language, the relative
prestige and social dominance associated with donor forms, and the waning influence of a
recessive, perhaps dying, language (among other things) all undoubtedly play important
roles4. However, distinguishing between donor and native forms is not necessarily simple.
At times, forms considered to be native by the most proficient speakers (i.e., according to
folk etymologies) were themselves borrowed at more remote points in the history of the
language, their origins obfuscated through the cumulative effects of phonological and
morphological integration (i.e., reanalysis).5
4.0.2. Identifying clusters of properties
To begin, a general definition of semantic transparency is proposed as the
successful matching of an acoustic signal (a phonetic form or amalgam of forms) with a
semantic interpretation, “with the least possible machinery and with the least possible
requirements regarding language learning” (Seuren & Wekker, 1986, p. 64). The optimal
or ideal case is associated with a one-to-one correspondence of meaning and surface
representation. A number of researchers refer to a direct one-to-one linkage as primary in
acquisition and in processing. For example, Roger Andersen’s One-to-One Principle
(1989, pp. 385-393 and elsewhere) asserts that children assume that there is only one clear
4 It may well be that the donor, because of its cultural and economic dominance, is used more often and in
a greater number of contexts by bilinguals of varying competencies than the original (ethnic) language of
the community. In such situations (characteristic of advanced shift), old native forms, victims of neglect,
may be forgotten by younger speakers (Reid, 1994a), providing a practical incentive to borrow seemingly
ubiquitous donor labels.
5 Which is certainly the case with Spanish borrowings in Modem Mexicano that have been around for
centuries (e.g.. Hill & Hill, 1986, pp. 124-141, p. 157).
113
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
and invariant form or construction for every intended meaning6. Consequendy, the
meaning of a novel form must contrast in some fashion with one already known. If a new
form does, in fact, refer to a familiar object or other referent (perhaps to a specific attribute
or characteristic of that object), ascertaining its precise meaning may involve taking on a
different conceptual perspective, an ability that apparendy is present in very young learners
(E. Clark, 1997). At any rate, the one-to-one mapping of form and visible, tangible
referent is maintained even when that visibility is primarily conceptual (mental).
Levelt (1989) states unequivocally that a one-to-one mapping guarantees speed and
accuracy in connecting a word-meaning with an appropriate form (p. 200). Semantic
transparency, therefore, must include the ability to map a form onto a meaning and the
reverse, the ability to find a form for a particular concept (cf. Heine et al, 1991a, p. 119).
This sort of conceptualization of a one-to-one mapping, however, may need to be modified
or elaborated in some sense to account for the possibilities inherent in different kinds of
form-meaning sets. Implicit in the various scales and continua discussed so far is that
mappings of any sort will be somehow dependent on the nature of the form (e.g., how
easy it is to separate out of the speech stream). They will also depend to a degree on the
nature of the referent (e.g., from concrete to abstract and so on).
It is assumed that the content of human communication is relatively constant across
cultures (i.e., the types of meanings that are encoded into language) even though the formal
characteristics of individual languages may vary considerably. Content items in all
languages make reference to entities that exist in some fashion from concrete to abstract or
6 Because there may be more than one toy, brother, friend, and so on, children apparently assume that
words can refer to types of objects and not necessarily individual ones (E. Clark. 1993, p. 50). There is no
reason to think that adult learners of a second language would assume anything to the contrary, though it is
likely that they will be influenced by the ways their primary language organizes items into semantic types.
114
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
from specific to general. On the one hand, they express meanings that are concrete (in the
Sapirian sense); they link to visible and tangible referents that represent the primary
concepts of all human discourse that can be classed into semantic types and subtypes (e.g.,
topics, participants in activities, etc.). On the other hand, inflectional (i.e„ fusional)
affixes have highly abstract and generalized meanings that link to language-specific
inflectional (grammatical) categories. Being language-specific, these categories are by
definition not found in all languages and are not common to all human discourse. By
identifying the relevant characteristics of forms and meanings at both extremes, the stage is
set for the analysis of the formal and semantic properties that may cluster in varying
strengths or combinations at different points along the scales.
4.1. FORM IDENTIFICATION CHARACTERISTICS: THE ROLE OF SALIENCE
The term salience is often used to refer to the phonological prominence afforded to
certain words or bound roots in the speech stream relative to various kinds of function
words and/or affixes. That is, some forms receive greater phonological stress or certain
other prosodic features (e.g., intonational contour) that make them more conspicuous than
others as a result of their intended significance or functional role within a word, phrase,
clause, or sentence (cf. Lyons, 1968, pp. 273,435-8). Discussions of the syntax-
phonology interface (e.g., Pullum & Zwicky, 1988, p. 255; Carr, 1993, p. 228) point to
the ability of phonology to respond to syntax to indicate word groupings and relationships
among words (Hawkins, 1994, pp. 3-4,116), distinguishing such forms as content items
(e.g., identifying specific topics, objects, and actions) from function words (expressing
more general semantic distinctions) and so on. Moreover, grammars afford salient
115
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
positions to members of particular syntactic (word) classes relative to their modifiers, for
example, to those that function as heads of phrases (i.e., heads extreme left or right).
Words may receive stress according to normal word-level prosodic patterns and
additional phrasal and/or clausal stress. For example, the status of a noun as a stress-
bearing content item coincides with its salient position within the English NP7. Words in
so-called focus positions may also receive extra phonological emphasis to increase their
prominence in relation to other elements in a clause. Word, phrase, and clausal stress can
work together or in conflicting fashion to contrast members of particular word classes
and/or morpheme types in an utterance for a number of reasons. The converse is also true:
salience can be accomplished in diverse ways. For instance, in some languages, the
relative importance of a specific form can be indicated by other grammatical or
morphological means, through the use of particles or the reduplication of forms or syllables
to mark emphasis, intensity, and so forth. Nevertheless, it is clear that these strategies also
add to the perceptual prominence of the marked form.
As a result of language particular lexical and metrical prosodic patterns (Cutler.
1989, p. 352), certain words are easier to pick out. However, whether a particular form is
salient or not, the ability to correctly identify it, that is, to associate it with an appropriate
meaning, involves more than general prominence. It does not necessarily follow that a
form will be any easier to comprehend just because it is made louder or more conspicuous
by a combination of phonological foregrounding strategies. It still needs to possess
properties that make it unique and recognizable. Ultimately, form identification is
7 While this may seem a bit circular, it is not. The fact that syntactic position and phonological stress
may, indeed, coincide is a built-in, systematic redundancy that undoubtedly increases intelligibility.
Moreover, in languages such as Japanese, where phonological stress is played down to an extent, one would
expect word order and other morphosyntactic devices to pick up the slack.
116
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
dependent on the makeup of the Individual form. All other factors being equal, however,
salient forms stand a greater chance of being noticed, thus providing a kind of incentive to
hearers. Salience, therefore, appears to be the result of a number of aspects of grammar
that conspire together to provide clues to the hearer concerning the relative significance of
selected forms in the speech stream (that s/he desires to understand).
Salience typically favors content items. Some function words can also become
salient, for example, personal pronouns in various argument positions; less frequently will
affixes of any kind be perceptually salient in systematic ways (e.g., Spanish tense markers
with the exception of those indicating present). In the event that an affix is allocated
primary word stress, that stress may be the result of normal word stress patterns or
triggered by inherent characteristics of the affix itself (and other members of its paradigm).
In either case, the affix is an integral part of the full content item to which it attaches and
occupies a position within phonological and syntactic word boundaries (cf. Levelt, 1992,
10ff.). It provides one basis for the identification of word class membership and, perhaps,
among other things, the grammatical and semantic function of the content item. The fact
that various affixes substitute for one another also provides an important perceptual clue by
which a speaker/hearer can distinguish between the bound root and the particular form-type
that marks any and all inflectional categories and their respective category values that are
obligatorily expressed on words of that class.
4.1.1. Transparency versus opacity
Forms at either end of the scales of borrowability and grammaticalization can be
contrasted in a number of respects regarding identifiability. True transparency
117
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
encompasses the ability to distinguish one form from all others together with the
assumption that an individual form has one (and only one) intended meaning (one form —
> one meaning; one meaning —> one form). Consequendy, the bidirectional nature of
semantic transparency entails not only ascertaining a unique meaning for an individual form
(in perception), but also the ability to recover an appropriate form to indicate a specific
meaning (in production). Speakers, therefore, are able to retrieve appropriate meanings or
concepts when presented with particular forms; they can also select from their lexicons the
individual forms representing the entities to which they wish to refer and the ideas they
intend to express. Opacity, at the other end of a spectrum of relative transparency, denotes
the inability to adequately connect an individual form to a particular concept or meaning
and/or the reverse, to link an individual meaning with a single, appropriate form.
One of the reasons given for the relatively detailed phonetic shape of content items
is that specific phonological information (a collection of distinguishing characteristics) is
required to make accurate identification. In order to retrieve the correct meaning of a
particular content form, a hearer will likely attend to specific portions of the phonetic string
or label. For example, in languages such as English, words are often recognized at or
before the point at which they are different from all other words in that language beginning
with the same sequence of phonemes—their so-called uniqueness point8. The fact that
forms are often correctly recognized before this point is a remarkable feat undoubtedly
facilitated by semantic context (Marlsen-Wilson & Tyler, 1980; Marlsen-Wilson, 1989, pp.
4-6; Hawkins, 1994, p. 4). Considering the speed with which an item is encountered in
the incoming speech stream and the sheer number of possible meanings that could link to a
8 Even assuming that lexical access is ultimately a complex process involving "narrowing-in and
monitoring stages, correcting strategies, post-access decision stages, and even look-ahead and look-back
operations...” (Grosjean, 1995. pp. 264-265; Grosjean & Gee, 1987), form recognition seems hardly
possible without at least some sort of one-to-one mapping.
118
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
specific content item, there must be an equally rapid and efficient mapping procedure that is
capable of identifying forms based on initial sound sequences (from left-to-right), at least in
many cases. Other relevant portions (middles and endings) may be important for full and
accurate identification (Hawkins & Cutler, 1988, p. 295ff.; cf. Cutler, 1989, pp. 343-
344).
The amount of phonological information necessary for identification is undoubtedly
dependent on the number of possible meanings to be retrieved. The larger the number of
possibilities (in a sense, limitless with respect to content items), the greater the amount of
detailed information required by individual identifying labels. In other words, the label
needs to contain enough phonetic information to achieve the uniqueness that is required to
pull up the correct (equally unique), corresponding meaning. This is similar to the
requirements of telephone numbers or social security numbers for individuals or license
plates for cars. The mathematical possibilities represented by the symbols themselves stand
in direct proportion to the number of items to be identified. This also suggests that an
efficient system employs labels that possess optimal amounts of phonetic information for
the purpose of economy. On the one hand, a system composed of forms with too little
phonetic substance would yield a high degree of homonymy, making the retrieval of
appropriate meanings problematic, at best On the other hand, too much information can
make retrieval inefficient, as well. Returning to the analogy of telephone numbers, it
certainly seems that too many unwarranted digits would make the recognition and
association of numbers to appropriate persons unnecessarily tedious. Sequences of
phonemes, as phonological addresses (see, for example, Butterworth, 1992, p. 264),
possess similar characteristics, for instance in the form of sequential constraints, which can
119
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
be viewed as optimizing form-meaning associations and maintaining the efficiency of
retrieval processes9.
Concerning the amount of detailed phonetic information that may be available for
the construction of morphemes, English, for example, has between 6,500 to 7,000
syllables (Levelt, 1992, p. 17)1 0 . Taking the lower figure and excluding form-meaning
sets that are not syllabic (for the purpose of discussion), as many as 6,500 unique
meanings can be linked to individual, unique forms—with no resultant synonymy,
homonymy, or polysemy. The number of possible two-syllable form-meaning sets soars
well into the millions, which can be added to the original number of one-syllable
possibilities. Needless to say, a language that possesses such potential in the formation of
morphemes or words has the capacity to refer to an enormous number of concepts. With
affixation (derivational and/or inflectional), compounding, and other word-formation
strategies (all restricted by language-particular word-formation processes or rules), the
potential is clearly inexhaustible in principle—the exact requirement that needs to be m et
Uniqueness of form also implies that the form itself has definition; it possesses
limits. That is, it must contain beginning and end points which can serve as morpheme
boundaries of some sort, irrespective of language-specific metrical patterns that can shift
syllable and even word boundaries (see Levelt, 1992, pp. 10-16). Consequently, some
sort of segmentability (by speaker/hearers, not just linguists) must be part of any
9 Phonological processes that streamline labels by reducing superfluous or redundant segments or features
certainly may also be seen as improving the efficiency of the system by reducing the amount of information
that must be processed. This is particularly consistent with views of individual languages diachronically as
“expression-compacting machines” (Langacker, 1977, p. 106; cf.. Hopper & Traugott, 1993, pp. 64-65).
10 According to Levelt (1992), English and Dutch (which resembles English in the number of possible
syllables) can be contrasted with (Mandarin) Chinese, which “has no more than about 400 syllables” (p.
17). Obviously, not all possibilities and combinations of phonemes in languages like English or Dutch
play equal roles. Certain combinations occur with much greater frequency; hence. Levelt proposes
language-specific frequency effects in various on-line tasks.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
conceptualization of identifiability. Accordingly, one property that is clearly applicable to
transparent forms is sufficiently detailed and segmentable phonetic information that can
function as a unique phonetic fingerprint that is, consequently, traceable to an appropriate,
individual meaning.
The discussion of uniqueness of form also directly leads to its necessary co-factor
in transparency, uniqueness of meaning (e.g., there is one, unambiguous referent).
Uniqueness of form serves no logical purpose unless that uniqueness also indicates a
clearly identifiable meaning. Therefore, the second property that is specified has to do with
the mapping possibilities provided by the intrinsic properties of the form: one unique form
maps onto one unique meaning. Perhaps the recognition of other form types (e.g.,
inflectional affixes) that normally consist of minimal phonetic information (a consequence
of the so-called diminution of form in processes of grammaticalization) is based on
something other than detailed phonetic shape, which may not be as critical when the form is
obligatory and drawn from a very limited pool of candidates (e.g., members of such tight-
knit paradigms as specific conjugation classes in Spanish)1 1 .
We now come to the postulation of a third property pertaining to selectional
possibilities: as a consequence of a clear one-to-one mapping of form and meaning,
transparent items can be selected on the sole basis of the particular meanings they convey
(identifying topics, participants, and so on). Such forms stand in opposition to those
whose selection is obligatory, that are assigned to positions within an utterance to satisfy
some grammatical requirement, which applies to inflectional affixes and certain function
1 1 Given the systematic complexity of language in all its aspects, it is difficult to accept that this clear
contrast is not a result of something and serves no logical purpose. This analysis also provides implicit
support for processing models of language that suggest content items are processed differently from
inflectional affixes.
121
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
words that fulfill the subcategorization requirements of certain nouns, verbs, and so on.
Instances of homonymy and polysemy, in which there are multiple but not simultaneous
meanings for a single form, and synonymy, in which there may be more than one form for
a single meaning, may make form recognition and the interpretation of particular utterances
somewhat tricky at times. However, any resultant ambiguity is generally temporary and
quickly resolved by the eventual effects of context (see, for example, Frazier & Clifton,
1996, pp. 4).
Nevertheless, just as there are unequivocal cases of semantic transparency (e.g.,
regarding CONCRETE nouns with clearly visible and tangible referents), there are cases of
complete opacity where there is no possible bidirectional linkage of form and meaning. As
shown in the following sub-section (4.1.3), there are forms that do not make direct
reference to unique entities of any kind (they refer to abstract inflectional categories), and
their multiple, simultaneous meanings cannot be linked back to a single form (i.e„ many
categories are fused onto one form).
Table 4.1 portrays the three properties that may cluster in various numbers and
degrees to render specific forms relatively transparent or opaque:
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 4.1. Form-meaning interpretation characteristics (FMICs): forms
Transparency: Opacity
(a) shape: unique (detailed, segmentable)
(b) mapping: 1-to-l link to meaning
and back to a single form
(a1 ) minimal, no (or zero) form12
(b1 ) 1-to-many simultaneous links to
multiple meanings; no return link to a
single form
(c) selection: optionally selected on the sole
basis of particular, individual meaning
(o’ ) obligatory; occur in response to the
requirements of language-specific grammar;
give appropriate syntactic form
Form-meaning sets sharing all characteristics represented in the left-hand column,
(a) - (c), are content items (for example, members of the major classes Noun, Verb, and
Adjective). Those sharing all the characteristics in (a1 ) - (c') are fusional affixes.
However, function words and affixes that are derivational and/or agglutinating may have
characteristics of either column. For example, agglutinating-type inflectional affixes
maintain 1-to-l linkage of form to concept even though they typically have minimal
phonetic shape and their application may be obligatory. In a similar vein, various kinds of
function words (as a diverse group) generally have 1-to-l form-meaning correspondences
resembling those of content items, but are occasionally required by the grammar (e.g.,
particular prepositions in languages such as English, German, Dutch, etc.).
Fusional affixes have a 1-to-many mapping of form to meaning, but this should not
be confused with synonymy. For example, the Spanish verbal affix -o represents a
12 The phonological requirement also refers to the fact that zero (or unexpressed) form and no form are not
the same. For example, while the regular plural is marked (with -s) in English, an unmarked form can
indicate two separate grammatical conditions, either singular (when it is countable) or uncountable noun.
In the former, the absence of a particular marker is. nonetheless, an indication of singular number, in the
latter, there is no form for either plural or singular. There are also instances of unmarked plurals, as in
sheep, fish, reindeer.
123
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
coalescence of present tense and person-gender agreement (with the intended grammatical
subject)1 3 . These three meaning types are mapped onto the single form simultaneously.
The ambiguity resulting from the kinds of synonymy and polysemy that occur with content
items are normally disambiguated by semantic and/or grammatical context, and meanings
can be found one after the other that are appropriate only in a given context. They are not
simultaneously expressed on a single form in the same sense that the discrete inflectional
categories are on a Spanish affix (see Figure 4.3). They are integrally fused together and
cannot be teased apart or segmented (clearly the result of grammaticalization processes).
The phonological segmentability characteristic of bound roots also acts to isolate affixes (of
any type), even though affixes, as phonologically and semantically subordinate (bound)
forms, bond in different ways to their roots. In some cases, there is a high degree of
phonological bonding (sometimes referred to as fusion) resulting in a shift of word stress
and so on. However, with respect to the internal characteristics of fusional affixes as
discrete forms, there has been both complete phonological and semantic fusion so that
neither phonological nor semantic segmentability is possible.
4.1.2. The opacity of fusional affixes
As a consequence of the intricate web of mapping possibilities created by their
fusional character, fusional-type affixes are completely opaque and represent the extreme
opposite form type from CONCRETE nouns. There are no phonetic fingerprints that can be
used to identify any one of their individual meanings; all meanings associated with a
particular form have to be retrieved simultaneously (not one after another depending on any
13 There is a homophonous form, -o, which marks masculine gender on nouns. However, their separate
grammatical contexts prevent any ambiguity that could conceivably arise.
124
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
kind of semantic or grammatical context). Therefore, hearers cannot segment an affix of
this type to locate a single meaning; no portions of the Spanish -o referred to above can
mean first person, or singular, or present tense sequentially (the way agglutinating affixes
can) because all of them are expressed at the same time. Neither can speakers select an
individual meaning and locate a single form (first person maps onto eight separate forms
that also express singular or plural in four separate tenses, in the first conjugation class
alone). Multiple mapping possibilities exist in each direction that cannot be disambiguated
by a formal (Le., strictly grammatical) context, at least in the same sense that “love” the
noun can be distinguished from “love” the verb in English by position or function within a
sentence. Fusional affixes can substitute freely for members of their respective paradigms.
There are other cases of one-to-many and many-to-one mappings on grammatical
affixes that are, nonetheless, relatively transparent. English has three grammatical affixes
that are homophonous in identical phonological environments (realized as [s], [z], or [tz]
depending on the preceding phoneme): -s, marking plural, noun; -s, a possessive marker,
and -j, marking third person, singular, present tense on verbs. All three appear in
distinctly different grammatical contexts, and, therefore, can be easily disambiguated. In
German, plurality on nouns can be expressed by a number of complementary forms, e.g.,
by the pluralizing affixes -e, -(e)n, or -s, or by means of a vowel change in the root (i.e.,
umlaut), combination of umlaut plus suffix (e.g., umlaut +er), and so on: Pferd —>
Pferde; Frau —> Frauen; Auto —> Autos; Haus —> Hauser representing horse(s),
woman/women, car(s), and house(s), respectively. However, in each instance, one, and
only one, concept is expressed: plurality. Both English -s and the German plural can be
contrasted with the Turkish plural -lar, portrayed in Figure 4.1, below:
125
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 4.1. Mapping patterns of grammatical affixes.
T u rkish G erm a n E n g lish
-lar -e, -(e)n, -s, 3+er
plural 3pspt. poss.. plural plural
The Turkish plural marker -lar shows a clear 1-to-l mapping; there is no ambiguity
possible. English -s has a l-to-3 form-meaning linkage that can be disambiguated by
context (e.g., in comprehension tasks)1 4 . German plural has (at least) 4-to-l form-meaning
mapping, but in each instance, the meaning of the form is clear and unambiguous (also) in
context The correct forms marking plural must be individually learned (which may make
second language learners wince in production), but they still make up a relatively small
number of possibilities1 5 . The Spanish fusional affix -o, however, maps l-to-3 (form to
meaning), but those 3 meanings simultaneously map back to a minimum of 25 possible
forms (i.e., excluding suffixes representing subjunctive mood and/or conditional). Suffice
it to say that this is not a simple case of semantic transparency. The complex mapping
scheme of the Spanish fusional affix -o is summed up in Figure 4.2, below:
14 Note that, in English, there are so-called strong forms (e.g.,foot-feet, tooth-teeth) and zero forms
(sheep-sheep, deer-deer), but they are not numerous enough to say that they are in direct competition with
one another as are the various plural forms of German.
15 German may not be the best example here. Clearly, there is a homophonous form -en which occurs on
verbs; however, the two forms are easily disambiguated by syntactic context And, genitive-s is also
homophonous with the plural marker, but again, is disambiguated in context. In addition, some nouns do
not change form when expressing plurality (e.g.. Tier, “animal”)—plurality being indicated by plural forms
of definite and indefinite articles, through the expression of agreement in other phrasal constituents, and so
on. Nevertheless, each of these individual markers have direct links to the category of number and mark
plural on nouns.
126
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 4.2. The mappings of Spanish verbal -o .
PLURAL SINGULAR
2 inf. PERSON 3 & 2 2 formal & 3
[-an] [-amos] [-as] PRESENT
[-aron]
[ - < 5 1
PRETERIT
[-aban] [-at a] [-abas] [-aba] IMPERFECT
[-aran] [-aremos] [-arl] FUTURE
Among all fusional affixes indicating four tenses, a total of 8 forms express first
person; 12 mark singular number; and 5 indicate present tense (only)1 6 . Summing the
number of possible mappings for the members of a single tense paradigm expressing three
persons, singular and plural (plus second person, singular, informal), there are 138
mapping possibilities involving five forms, as depicted in Figure 4.3.
16 For the purpose of this calculation, the present-tense suffix -amos is taken as homophonous with that
marking preterite because it is a member of a separate paradigm and can be counted separately as a
consequence. However, the affix -a marking the singular version of both second-person, formal and third-
person belongs to this one paradigm only, and is, therefore, not divided into two homophonous forms, one
marking second-person and the other third.
127
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 4.3. Calculating mapping possibilities (present tense, first conjugation class)*.
SINGULAR PLURAL
form meaning no. of forms for form meaning no. of forms for
given meaning given meaning
-o first person 8 -amos first person 8
singular 12 plural 8
present tense
_ i
present tense _ i
3+25 3+21
-as second person, informal 4
singular 12
present tense
-1
3+21
-a second person, formal 8 -an second person, formal 8
third person 8 third person 8
singular 12 plural 8
present tense
_s
present tense
_£
5+22 5+22
totals: 80 58
* The leftmost column lists the forms; the middle column represents the types of meanings expressed by
that form; and the rightmost represents the number of forms that express that particular meaning (mapping
backwards from meaning to form).
4.1.3. Borrowing continua of forms
The first basic set of predictions can now be made regarding borrowing preferences
based on oppositions of the general properties most relevant to the identifiability (and
transparency) of particular form types. These contrasts can be stated implicationally in (1)-
(3), below. First, form types whose individual phonetic shape is relatively detailed (i.e.,
consisting of a sufficient number of phonemes and/or syllables for efficient retrieval of
meaning) will be preferred over those characterized by minimal or no phonetic information
(e.g., a single phoneme or zero expression):
128
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(1) sufficiently detailed shape > minimal or no shape
Second, it is predicted that forms which map onto individual referents or concepts
and back to the same individual forms will be preferred over those that map onto more than
one concept simultaneously, as in the case of individual Spanish fusional affixes on verbs
that express tense-aspect plus person and number agreement In addition, they will be
preferred over forms whose numerous simultaneous (or coalesced) meanings are also
separately and equally expressed by a number of other forms, as in the various Spanish
affixes expressing first-person or present tense. These preferences are presented in (2a)
and (2b), respectively:
(2a) 1-to-l form-meaning mapping > 1-to-many simultaneous form-to-meaning
mappings
(2b) 1-to-l form-meaning mapping > 1-to-many simultaneous meaning-to-form
mappings
The complex mapping procedures inherent in fusional-type affixes are a consequence of
both conditions; one isolable form maps simultaneously and equally onto many meanings
and each one of those meanings maps simultaneously back to a number of distinct forms
(which also express another set of meanings simultaneously).
Lastly, it is predicted that individual forms that may be selected by speakers as a
direct consequence of the meaning or concept to which they refer (that is, they link to
entities or referents of some sort, and their selection is, therefore, semantically based) will
be preferred over those that are obligatorily selected or applied as a consequence of word
129
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
order (co-occurrence) restrictions1 7 or morphosyntactic rule (i.e., the concepts they
represent are obligatorily expressed, and selection is, therefore, syntactically required):
(3) selection optional > application/selection obligatory
These three sets of oppositions of form identification characteristics represent
scalings of properties that can occur in varying degrees and combinations and, therefore,
may cluster in different complementary (or potentially conflicting) ways on individual
forms and form types. It is now possible to identify which form types are likely to be the
most to least preferred in any borrowing situation.
Content items (words, bound roots) typically have each of the properties to the left
of the arrows indicating maximal degrees of identifiability or transparency (and potential for
salience) and are, therefore, always the most highly preferred form types. Fusional-type
affixes, possessing every characteristic to the right of the arrows, will be the least preferred
(assuming that they are, in fact, compatible with the recipient morphosyntactic system).
Other bound form types (e.g., agglutinating-type inflectional affixes) will be preferred over
fusional types, but less than either function items (words, bound roots) and content items.
Function items (words, bound roots), which occupy medial positions on scales of
borrowability and grammaticalization, will be borrowed more easily and more frequently
than inflectional affixes of any type1 8 , but not as easily or frequently as content items.
17 For instance, in languages such as Spanish, English, or German, particular verbs (and their derived
nouns or adjectives) are subcategorized for specific accompanying (obligatory) prepositions as a result of
syntactic processes of phrasal organization (see, e.g., Friederici, 1985, pp. 136-139).
18 Regarding segmentability, an implicit component of complex phonetic shape, function items are
typically analyzable as having word boundaries; they also take part in syntactic processes affecting the
ordering of phrasal or clausal constituents. Bound grammatical forms, that are subsidiary in form even
relative to function items (words and bound roots), are subject only to word formation rules. We can
assume that the additional syntactic property (i.e., word as well as morpheme boundaries) acts to increase
identifiability.
130
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
As a general form type, function items may possess the properties represented in
(1) - (3), above, in different numbers and strengths. For example, locative (spatial and
temporal) prepositions (e.g., those of Spanish or English) typically possess detailed
phonetic shape (relative to inflectional affixes), 1-to-l form-meaning mappings, and are
selected as a result of the kinds of concepts (i.e., semantic relations) they express. This is
in spite of the fact that they belong to a closed class and that their meanings may be
considered relational in that they function to indicate relationships among other phrasal and
clausal constituents. Other sorts of adpositions that are obligatorily selected for syntactic
reasons are less transparent and will be preferred less in borrowing situations despite their
relatively detailed phonetic form.
Other function items may be described in similar terms. For example, particular
subclasses of Determiner and Pronoun may be semantically selected and have 1-to-l
mapping possibilities (others not); but, in all cases, they are typically less detailed in form
than members of the open classes as a result of grammaticalization processes, which
include reduction of phonological form as a result of relative frequency of occurrence.
They may also be subject to processes of cliticization and so forth. All classes and
subclasses of connectives share similar characteristics with respect to mapping possibilities
and selection. However, with respect to auxiliary verbs, in general, some may be
semantically selected (e.g., modal verbs) while others that function as markers of tense or
aspect may be obligatory because of the categories they express.
In all classes of function items, various members of particular subclasses may have
noticeably different configurations of the stated form identification properties (in degree or
number) from members of other subclasses, and this can result in degrees of transparency
even within an individual class. As a consequence, those items exhibiting the greater
131
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
number and degree of properties enhancing their identifiability will be among those that are
preferred in borrowing. In addition, when a particular form type has multiple functions
(e.g., subclasses of Pronoun such as possessives or those with special clitic forms),
salience may be affected. In one context, a form may be salient, and in another, maybe
not In one case, salience may conspire with other phonological factors to enhance overall
identifiability; in the other, the lack of salience may decrease identifiability.
4.2. SEMANTIC CHARACTERISTICS
The Sapirian notions of concrete versus relational concepts have been foundational
in much work dealing with grammaticality, borrowing, and so on, and have provided a
basis for a number of recurring ideas about meaning. Many of the characteristics that Sapir
proposed can be analyzed according to two notions typically associated with descriptions of
semantic bleaching. One refers to levels of generality and the other to degrees of
abstractness. In each there is a set of oppositions: generality can be contrasted with
specificity, and abstractness with concreteness. General terms identify types or classes of
entities and so on, but specific terms provide detailed semantic information, enough to
distinguish among members of particular groups. Within individual taxonomies, the
meaning of a superordinate form-meaning set, a hyperonym, is always more general and
inclusive of the various members than a subordinate one, or hyponym. As a consequence,
the more general a term is, the greater number of contexts there will be in which it is likely
to occur (e.g., dog versus chihuahua)1 9 .
19 For discussion the relationships of hyponymy and hyperonymy, see, for example, Levelt, 1989, pp.
201,212-214,1992 p. 6ff.; and Bierwisch & Schreuder, 1992, p36ff.
132
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The quality of concreteness is descriptive of entities that exist in the real world, can
be actual referents, or, like unicorn and Klingon, have real (i.e., imageable) attributes but
exist only in the imagination or in the realm of fiction. Being visible and tangible, actual
referents possess physical properties (cues) by which mental images are obtained (see
Rosch e ta l, 1976; Rosch & Mervis, 1975; Rosch, 1975). Hence, the term “imageable”
refers to processes by a which a speaker assesses and analyzes features of physical objects
either to secure an appropriate label or to construct a mental image (i.e., concept) that can
function as the meaning in a particular form-meaning set ABSTRACT nouns do not point to
referents that are imageable in this same sense; their referents are not physical entities that
can be touched, watched, or used (de Groot, 1993, p. 41). Therefore, abstractness is
characteristic of concepts that have no direct concrete existence and are not objects or
entities that can be signified in a Saussurean sense20.
In cases of polysemic extension, in which forms representing objects are extended
to refer to processes associated with those objects (e.g., to wallpaper, to carpet, to wax) or
forms for actions extended to cover their results (e.g., a hit, a run, a fall) some sort of core
meaning is pulled out, abstracted, and applied to a different kind of referent (OBJECT — >
ACTIVITY, and so on). In this respect, processes of grammaticalization and derivation can
both result in the emergence of concepts that are increasingly abstract and, perhaps, general
(cf. Bybee, 1995, p. 226), yet relatable to their sources. Consequently, a third property is
required, based solely on degrees of grammaticalization viewed synchronically, that can
screen out both (a) derived meanings that have merely become more general or abstract,
and (b) meaning types that lie at different levels of generality within individual taxonomies.
In each case, direct reference (i.e., a mapping of form to meaning) is maintained, albeit to
different types of referents that are not necessarily more grammaticalized.
20 See Section 3.3, above, for discussion of abstractness and ABSTRACT nouns.
133
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
An opposition of meaning types based on grammaticalization is also implicit in
Sapir’s distinctions. It is based on characteristics that are language-specific and depends on
whether the kind of meaning expressed by a particular form belongs to a broad lexical
system (semantic type or subtype) or whether it belongs to a language-specific inflectional
category. On the one hand, the meanings associated with some forms are members of
general semantic types (e.g., CONCRETE nouns or ACTIVITY verbs) or so-called semantic
fields (i.e., subtypes like nouns for BODY PARTS and MOTION verbs, discussed in Chapter
3) and can, as a result, be listed in a dictionary. That is, representations of clear referents
and/or explanations of characteristic properties serve as definitions and can be listed
alongside appropriate forms. On the other hand, the meanings expressed by inflectional
affixes link to particular inflectional categories (e.g., number, tense, or gender) whose
“definitions” may elude even the most proficient speakers (linguists, as well). This third
opposition also reflects the fact that the topics speakers choose to discuss are not limited by
a particular language grammar (i.e., morphosyntactic infrastructure) although the form
types and grammatical categories they have at their disposal to express those topics may
vary considerably.
4.2.1. Inflectional meanings
Much recent work has focused on the types of meanings that inflectional categories
are known to express. For example, Bybee (1985) catalogs grammatical categories
associated with verbs based on their occurrence in a relatively large and heterogeneous
sample of languages. A cross-section of 50 languages was chosen judged to be relatively
free of genetic and areal bias (cf. Carstairs-McCarthy, 1992, p. 173) concerning the relative
frequency of certain types of inflectional categories, the order in which they appear in
134
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
relation to a verbal stem, and so on. They were identified according to three criteria:
boundedness (the morpheme cannot be separated from a stem); obligatoriness (specific
forms must appear whenever the category is applied); and predictability of meaning (the
meaning is the same when applied to any verb). The list represents the kinds of categories
one might expect to uncover in a particular language, though it is not assumed to be
exhaustive of all the sorts of abstract, grammaticalized concepts that may occur in all of the
world’s languages. It does represent, however, a clear basis for a general contrast with
semantic types (lexical categories).
Table 4.2. Inflectional categories associated with verbs (according to Bybee, 1985).
CATEGORY
Valence
Voice
Aspect
Tense
Mood
DEFINITION
the number and types of arguments a verb can take (e.g.,
causative, which affects the arguments and semantic roles
taken by a specific verb)
the perspective from which the event or act described by the
verb stem is taken (e.g., active and passive voice, which also
affects the assignment of semantic roles)
the way the internal temporal constituency of a situation is
viewed (e.g„ as an act in progressive, perfective, and so on)
the location in time of the event depicted by the verb (i.e., in
relation to the moment of speech or some other point in time)
the way the speaker represents the prepositional value of a
statement (probable, possible, certain, etc.)—this includes
evidential and expressions of modality
Number agreement agreement with an argument (most frequently grammatical
subject) in number (e.g., singular, plural, dual, paucal...)
Person agreement agreement with respect to first, second, third person, and so on
of an argument (grammatical subject, direct object, and so on)
Gender agreement agreement with argument with respect to gender or class (e.g..
masculine or feminine gender)
135
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In addition to the above, nominal inflections of number, person, gender-class, and case
may also occur as a result of language-specific concord (obligatory patterns of agreement)
on particular constituents of NPs in addition to classes of Noun, e.g., various determiners
and adjectives.
As with form, the kinds of meanings that are typically expressed by function words
are positioned somewhere between extremes, between those associated with content items
and those of fusional affixes. For example, adpositions have no direct referents themselves
but typically indicate physical relationships such as location of one entity with respect to
another in space or time (cf. Sapir, 1921, p. 89; Hopper & Traugott, 1991, p. 107). Such
relational concepts are apparently restricted to the possibilities that exist in the world; in
some approaches, they constitute an ontological category of their own termed RELATIONS
(cf. E. Clark, 1993, pp. 43-^19). However, this type of meaning is integral to human
communication irrespective of the ways it may be expressed in an individual language.
Such highly isolating-analytical languages as Vietnamese or Mandarin are obviously
equipped to express temporal and spatial relationships through the use of adpositions,
particles, etc. However, neither has obligatory inflectional categories indicating tense-
aspect or case that may express similar, though more abstract, general, and
grammaticalized, concepts (e.g., which occur in more synthetic types of languages such as
German, Russian, Italian, and so on).
Other classes of function words can be contrasted cross-linguistically with differing
sorts of inflectional affixes (sometimes within individual languages) in similar ways, for
example, modals (which often receive tense markers and so on) versus affixes expressing
mood; pronouns (in many languages indicating grammatical categories of gender and
number) versus tense markers also indicating person and number agreement; and
136
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
subclasses of determiner such as articles (at times marked for such grammatical categories
as case and number and gender agreement) in contrast with affixes indicating definiteness.
Regarding inflectional categories in a specific utterance, however, it is not very likely that a
speaker/hearer is concerned with recognizing (or accessing) an abstract category of TENSE,
PERSON, or NUM BER. The task is to correctly identify the particular instance of any and all
distinctions expressed by a discrete form (e.g., present or past tense; first, second or third
person; and singular or plural). Logically, there can be no general occurrence of TENSE
that could be significant in any specific situation; an inflectional category is neither an entity
nor an imageable referent. In the case of fusional affixes, the identification of meaning
requires a strategy that recognizes all such meaning possibilities simultaneously.
Accordingly, in the presentation of form-meaning recognition characteristics, the
factors by which kinds of meanings can be compared or contrasted (in the context of
transparency versus opacity) are listed as follows in Table 4.3, below:
Table 4.3. Form-meaning interpretation characteristics (FMICs): meaning
Transparency Opacity
(a) concreteness: reference to distinct entities
or concepts that exist
(a) abstract: apart from concrete or physical
existence; no distinct reference
(b) specificity: explicit: distinguishes among
members of a species or type; typically
occurs in a limited number of contexts
(b) general: applies to many (all members of
a group); consequently, may occur in a larger
number of contexts
(c) grammaticalization: belongs to a broad
lexical subsystem (i.e.. semantic type
or subtype)
(O
belongs to a language-specific inflectional
(i.e.. grammatical) category
137
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Meaning types sharing the characteristics listed in the left-hand column, (a)-(c), are
those typically represented by content items which indicate what Sapir termed primary
concepts; they designate the principal topics of discourse, e.g., persons, objects, activities,
and so forth. Those types possessing all of the characteristics represented in the right-hand
column, (a X c 1 ), are the meanings that belong to language particular inflectional affixes.
Moreover, the various properties may cluster to identify types of meanings that occur at
midpoints on scales of grammaticalization and borrowability. For instance, pronouns may
represent meanings that are relatively concrete and specific based on their antecedents, but
reference is made through the expression of such language particular categories as PERSON,
NUMBER, GENDER, and/or CASE. The meanings associated with auxiliary verbs (e.g.,
modals expressing obligation, permission, and so on), in contrast, are more abstract and
general than the primary verbs with which they may occur but, nonetheless, belong to a
semantic subtype of Verb21. Other meaning types, such as those linked to certain
subclasses of connectives and determiners, appear to occupy midpoints in each respect.
Even though they are relatively abstract Gack direct reference) and general (can appear in a
large number of contexts), they often express semantic distinctions that are easily definable
in terms of logical relations, quantities, and so on. In addition, the concepts associated
with classes of connectives and determiners appear to be much more universal (i.e., typical
of human discourse) than the more grammaticalized meanings of, say, TENSE or GENDER.
4.2.2. Borrowing continua of meanings
The second set of predictions, expressed implicationally in (4)-(6), below, is based
on the three oppositions of meaning types outlined in the preceding portion of this section.
21 See Section 3.2.3, above, regarding the potential salience of modals, pronouns, and so on.
138
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
First, it is predicted that form-meaning sets whose meanings or referents are concrete will
be preferred in borrowing over those that are characterized as abstract This should be
especially evident within particular semantic domains (e.g., within occupational
nomenclatures) with respect to additions to the recipient lexicon (see subsection 1.1,
above). On the one hand, concrete meanings are associated with individual or identifiable
groups of persons, things, and so forth that have physical existence in reality and
experience and that constitute observable entities (e.g., wind, radio waves, etc.). They
include reference to close objects that have physically recognizable attributes (e.g., they are
green or may become furious)', consequently, they have a clear definition and can be
contrasted with other objects or entities on the basis of their physical properties. Objects
that have physical existence can be pointed to, seen, picked up, and put to use, and,
therefore, have the ability to provide the perceptually-based semantic information (e.g.,
shape, size, texture, material composition, etc.) necessary to establish a mental
image/concept capable of being a concrete referent22. Other speakers will have the near
identical concept (they see the same image) and receive similar amounts of exposure to its
label (phonetic form).
On the other hand, abstract terms generally refer to concepts that do not have
physical existence, and as a consequence, can have no physical properties of shape,
texture, and so on). They may include processes, methods, ideas (truth or lie), or the
results (products) of processes. They may be remote or theoretical in some sense (existing
in psychological or social space) and are typically more difficult to define (or translate from
LI to L2) as a consequence (Heine et al, 1991a, p. 41ff). Concepts associated with
22 It is also noted studies of bilingual lexical access, that concrete nouns are translated more quickly than
abstract nouns, and animate nouns more quickly than inanimate (see, for example, de Groot, 1993. p. 40).
One explanation is there is a greater density of conceptual features on a semantic (i.e., conceptual) level
with respect to concrete versus abstract form-meaning sets. Some researchers have, therefore, proposed a
“distinctly perceptual component of the semantic representation” (Kroll & de Groot, 1997, p. 189).
139
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
abstract nouns may require demonstration and illustration (e.g., how to use a particular
implement) and/or conscious instruction (e.g., regarding the meanings and significance of
terms commonly used within various governmental, religious, educational, or medical
settings). As a consequence, the first prediction is as follows:
(4) concrete > abstract
Meanings may become more abstract either through processes of grammaticalization
or through those associated with derivation. The term “abstraction” generally indicates a
loss and/or change of particular semantic specifications; for example, the meaning
associated with a particular form-meaning set may be reduced to its semantic core (a
generalizing abstraction) or to a certain aspect of its meaning (an isolating abstraction).
Metaphorical abstraction, most often linked to grammaticalization, occurs as a meaning is
abstracted from one form-meaning set and applied in a more diffuse or fuzzy way to
another, more grammaticalized form in discrete steps (Heine et al, 1991a, pp. 44ff.; Heine
et al, 1991b, pp. 160-161). It may be either structure-preserving, which does not change
the word class of the affected form-meaning set, or structure-changing, which does change
class membership. In the former, there is an extension of meaning and, in the latter an
extension of function. Heine and his collaborators offer as examples of both the Ewe noun
ta' “head”: as structure preserving, its meaning is extended to refer to such concepts as
“intellectual ability”, “main issue”, “group, party”, and so on. As structure changing, ta' is
used as a postposition and subordinating conjunction meaning “over”, “on”, etc (Heine et
al, 1991a pp. 44— 45; cf. Heine eta l, 1991b pp. 160ff.).
Concerning the second characteristic, specificity, the initial expectation may be to
predict that form-meaning sets whose meanings (referents) are specific, individual entities
140
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
will be preferred over those that are more general because general terms refer to kinds of
entities (as opposed to individual objects) that require some sort of classificatory scheme.
In preliminary fashion, therefore, the second basic prediction is set forth as follows:
(5) specific > general
However, this assumes that the most easily recognizable concepts are those that are
linked to the most explicit semantic information (i.e., maximally specified concepts)
because it is in some way necessary for the identification of particular referents. Perhaps,
the expectation should follow by analogy from the prediction concerning phonetic detail
(see subsection 4.1.1, above), which states that form recognition is facilitated by
sufficiently detailed shape in its function as a precise phonological address, allowing the
speaker/hearer to access an exact and appropriate meaning (i.e., figuratively speaking, the
occupant of the address). What is required for identification is an optimal, not maximal,
amount of information. Hence, specificity more properly refers to the number and kind of
physical features (qua semantic information) required to recognize kinds of objects and to
distinguish among or between individual members of a kind or larger, more general
grouping. For instance, many different types of animals have the ability to fly; however,
we can distinguish among them on the basis of several other physical properties, e.g.,
whether or not they have feathers, wings, and so forth. Among types of birds, additional
properties (i.e., more information) will be required in order to make further “category cuts”
(Rosch et al, 1976)23. As a consequence, the prediction needs to be amended somewhat in
23 This may evoke discussion of the naturalness of semantic types (categories). Items included or excluded
from specific categories (so-called category cuts) may be open to some cross-linguistic variation based on
cultural factors and/or levels of expertise. For example, porpoises are not fish, but share certain attributes
with them; having expertise in the area of biology provides individuals with sufficient numbers and relevant
kinds of attributes to enable sharper category cuts (inclusive/exclusive) (see, for example, Rosch et al,
1976. p. 430ff).
141
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
order to clarify the circumstances under which gradient levels of specificity and generality
are most relevant in particular borrowing situations.
Regarding degrees of generality, Rosch et al (1976) argue that there are semantic
(conceptual) levels at which the mental representations of entities possess systematic kinds
and amounts of information by which they can be differentiated and grouped. This
includes a basic (entry) level which represent the most useful terms for kinds of objects,
superordinate level(s) which organize objects into larger families, and additional
subordinate levels which provide increased specificity of reference. Within taxonomies of
common concrete nouns, they posit that basic objects are identified and named according to
aggregates of physical characteristics; they have similar shapes, are identifiable from
averaged shapes of other class members, and so on “in terms of cognitive economy...” (p.
384). Within individual taxonomies, superordinate terms (hyperonyms) have a relatively
small number of characteristics by which distinctions can be made, and are, as a
consequence, more inclusive. Subordinate terms (hyponyms) have more attributes in
common, and are more exclusive. For example, nouns such as dog, apple, chair are basic-
level terms in English, while corresponding superordinate terms are mammal, fruit, and
furniture, respectively. At a more subordinate level, one also finds such words as
chihuahua, Mackintosh, and kitchen chair, respectively. Hence, while dog is a hyponym
of mammal, it is also a hyperonym of chihuahua. However, dog remains the basic/entry
point term24.
24 Regarding the ways object names are learned and organized during child native language acquisition,
which appear to be consistent with this often cited approach, see, for example, Waxman, 1994; Markman,
1994, 1993, 1989; Huttenlocher & Smiley, 1987; Waxman, 1994; Landau, 1994, 1993; and E. Clark,
1993.
142
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Rosch and her collaborators link abstractness, the identification of a somewhat
reduced collection of characteristics that form the semantic core of a particular term (an
intrinsic or non-derived kind of generalizing abstraction), with gradient levels of generality.
But, for the present purposes, it is important to note that abstractness and generality are,
nonetheless, distinct. For instance, the concept represented by one concrete noun may
certainly be more general than another without any loss of concreteness, e.g. bird versus
sparrow, and the concept associated with an abstract noun may be more specific than
another, e.g., psycholinguistics versus linguistics. It is safe to say that processes of
abstraction can lead to increased generality as a co-effect. Nevertheless, basic-level objects
are those that are first sorted and named by children (cf. E. Clark, 1993, p. 50ff) and are
the most codable (representable), most coded (represented), and most useful in an
individual language (Rosch et al, 1976, p. 382 and 428). In addition, they are also the
most likely to be found in cross-linguistic analyses of individual language lexicons. As a
result, it is hardly surprising to find the striking resemblances among word lists in
introductory foreign language texts and core vocabulary among the world’s languages.
The ways levels of generality may affect borrowing patterns may vary somewhat.
In the event that the recipient community is exposed to a novel entity (e.g., a specific
animal, agricultural implement, and so on), there may be a number of possibilities (see
Subsection 1.1., above). One is to use an already existing native form or collocation of
forms and extend their range of possible referents via semantic loans or caiques; another is
to borrow the donor form-meaning set outright and integrate it into the recipient system (or
perhaps a combination of both, so-called loan blends). When there is already an
appropriate family (i.e., a semantic subtype) in the recipient lexicon to which a borrowed
form-meaning set may be conceptually linked, that form can easily be assigned to that
subtype. It is not in direct competition with any recipient forms—i.e., other hyponyms.
143
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
However, if there is no existing semantic subtype, an entire taxonomy may be borrowed,
including hyperonyms and hyponyms. This often appears to be the case regarding families
of entities that are entirely new to the recipient language and culture, for instance, concepts
belonging to technological fields pertinent to particular professions or industries, religious
hierarchies and institutions, and so on. In this case, the most useful and basic term may be
borrowed first—to provide a family name that can function as a hyperonym under which a
new taxonomy will develop. Knowledge of individual form-meaning sets included within
the new taxonomy will likely depend on exposure (frequency), relevance, levels of
expertise, and so on.
The third and final prediction pertains to the effects of grammaticalization, that is,
the synchronic status of meaning types aiong scales of grammaticalization. As stated in the
previous sections of the present work, semantic types and subtypes (at one extreme end of
a scale of meaning types) represent concepts common to all languages. However, these
concepts may be gradiently concrete and/or specific. Inflectional categories (at the other
end of the scale) are by definition language-specific and, as an apparent consequence,
idiosyncratic with respect to the concepts they express, their functions, and restrictions on
appropriate usage. Such categories are typically abstract (implying the loss or lack of
concreteness) with respect to the types of meanings they express and typically general
regarding their functions (implying the loss or lack of reference to specific entities).
Concerning specificity, however, individual instances are more specific than the general
category to which they belong—the greater the number of members in a paradigm, the
more specific the meaning (or complex of meanings) expressed by each member. In
addition, there is a direct correlation between specificity (as applied to occurrences of
inflectional concepts) and the amount of grammatical knowledge that is required for
identification within the speech stream—more so in production than in comprehension.
144
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
It is predicted, therefore, that form-meaning sets whose meanings are members of
broad lexical systems (semantic types or subtypes) will be preferred in borrowing over
those that are expressions, viz., specific instances, of inflectional categories. This is
represented in (6), below:
(6) semantic (sub)type > inflectional category
The three sets of oppositions regarding types of meaning discussed above also
represent scalings of properties that can occur in varying degrees and combinations and,
consequently, may cluster in complementary (or conflicting) ways. It is now possible to
identify meaning types that are most to least preferred in any borrowing situation.
Form-meaning sets whose meanings typically possess each of the properties to the
left of the arrows (indicating maximal degrees of transparency and identifiability), i.e.,
those whose meanings belong to broad lexical categories (semantic types and subtypes) and
whose referents are concrete and specific25, will always be the most preferred (and most
often borrowed) in any contact situation. The least preferred will be those that possess all
of the characteristics to the right of the arrows, that is, those that are are abstract and
general and whose associated meanings are expressions of inflectional categories. In-
between types of meaning, e.g., concepts representing gradiently concrete relations
(locative, directional, and so on), modality (e.g., ability, permission, etc.), and so on (cf.
25 This includes the caveat that basic level categories may be preferred over either superordinate or
subordinate donor categories under certain specified circumstances, i.e.. depending on the prior existence of
equivalent taxonomies in the recipient lexicon. It is also possible that basic level terms may be more
frequently borrowed as a result of their general frequency, although those belonging to recipient core
vocabulary will likely win out as a result of various social factors—in that case, recipient labels (forms) are
retained as ethnic markers and so on (Held, 1997b).
145
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Heine et al, 1991a, pp. 42-43)26, will be preferred less than members of semantic types
and subtypes, but more than expressions of inflectional categories.
4.2.3. Linking form and meaning
All form-meaning sets possess properties according to the six interpretation
characteristics discussed above (expressed as oppositions of form and meaning). That is,
every form will display (i) comparable amounts of phonetic information, (ii) different
sequential and/or simultaneous mapping possibilities (1-to-l, 1-to-many, and so on), and
(iii) selectional restrictions governing their occurrences; every meaning type will exhibit (iv)
degrees of abstractness (e.g., from totally concrete to completely abstract), (v) specificity,
and (vi) grammaticalization. In addition, these characteristics indicate properties that can
cluster in various ways. For instance, forms with detailed phonetic shape often map onto
individual referents (1-to-l); specificity and concreteness (and their oppositions generality
and abstractness) often coincide even though they are distinguishable. Moreover, concrete
concepts are usually represented by unique and segmentable forms, and inflectional
meanings are typically represented by minimal phonetic shapes that are obligatorily
expressed. Whenever the expression of a particular category value is required (e.g., in
English, tense on finite verbs or number on count nouns), a form indicating that category is
obligatorily selected and applied to the stem (or root) of a member of the syntactic (word)
class to which the category applies. Even though selection of a specific instance of a
26 Perhaps, the form-meaning sets expressing such relations and attitudes belong to closed classes because
there is a limit to the number of possible concepts they can signify. Their meanings are relatively specific,
yet abstract; they are frequent as a result of their useful (and versatile) functions. Even though they do not
belong to inflectional categories, per se, they appear to exhibit the effects of grammaticalization. We may
speculate that the processes by which they emerge may lie somewhere between derivation and
grammaticalization.
146
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
category may be for stylistic reasons (e.g., the conscious choice of singular versus plural
forms of generic nouns), it remains true, nonetheless, that one (and only one) member of a
relevant paradigm must appear on the appropriate stem or root, even if that form is zero
(see Bybee, 1985, p. 27 and p. 202).
While form-meaning sets that exhibit the six interpretation characteristics to the
highest degree (i.e„ on either end of the oppositions) may provide the most obvious cues
regarding identifiability, grammaticalization, and borrowability, those occupying midpoints
between extremes may still be distinguished according to particular clusters of
characteristics. For instance, adpositions whose semantic functions are most clearly
locative (e.g., the so-called “lexical” prepositions of German and English) will be preferred
over those that are subcategorized for by individual content items, so-called “grammatical
prepositions” (cf. Friederici, 1985) on the basis of form (e.g., reduced phonological form,
mapping possibilities, and selectional restrictions—in the case of grammatical prepositions,
they are obligatory) and on that of meaning (e.g., gradient levels of abstractness and
generality and degrees of grammaticalization). Similar predictions can be made of
pronominals (e.g., the full pronouns of Spanish versus reflexives that obligatorily co-occur
with certain verbs), conjunctions expressing certain types of logical relations (e.g.,
coordinators versus subordinators), determiners expressing quantification versus
definiteness, and auxiliary verbs that are semantically selected and belong to semantic
subtypes of Verb (e.g., modals expressing ability or obligation) versus those expressing
obligatory categories (e.g. those used in the so-called compound tenses of English), i.e., in
those particular highly grammaticalized functions.
Many correlations such as these can be linked to grammaticalization and/or
derivation, suggesting that each process operates concurrently on both form and meaning.
147
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
However, actual outcomes cross-linguistically are mitigated by the fact that not every
language will show the same degrees of grammaticalization (Bybee, 1995, p. 228f.).
Moreover, some gradable characteristics of meaning and form will be present irrespective
of the extent of grammaticalization and/or derivation; they may be inherent to the individual
forms and concepts themselves. Overall, form-meaning sets can, nevertheless, be
compared and contrasted according to clusters of these properties. One may conclude,
then, that the similarities found between grammaticalization clines (or chains) and
borrowing hierarchies are each intimately linked to morphological character, that is, the
ways individual languages structure their particular form-meaning sets into morphemes and
words. In the case of borrowing, the morphological structuring of each language plays the
critical role.
Because the establishment of property clusters may be obfuscated somewhat by the
gradient character of the properties themselves, there may be a number of judgment calls
with respect to the investigation of particular language data that may influence a particular
prediction (e.g., A is more concrete and less specific than B, when I am the arbiter). This
appears to be true when attempting to distinguish between derivational and inflectional
categories in particular languages (Bybee, 1985, p. 81; Croft, 1990, p. 191; Heine eta l,
1991a, p. 17). Regarding the multiplicity of gradient factors and the complexity of the
issues involved in cross-linguistic comparisons of continua of forms and meanings, Sapir
opined, “There are too many possibilities” (Sapir, 1921, p. 107). Nevertheless, the
number of properties that are assumed to be factors in the case of borrowability (gradient or
not) can contribute to the accuracy of the predictions, especially when properties cluster in
their most clearly identifiable ways.
148
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4.3. ISSUES OF SEMANTIC COMPLEXITY
Before summing up the various predictions based on form and meaning types, one
remaining issue requires discussion which concerns the relative borrowability of various
major word class members (N, V, etc.). It is commonly noted that nouns are always
among the first elements of a language to be borrowed, followed by other content items. It
is also axiomatic in studies of language acquisition that nouns are learned before verbs.
These observations have prompted a number of possible explanations suggesting that the
obvious asymmetry goes beyond the mere recognition of form-meaning sets and that it is a
result of the syntactic and semantic complexity of verbs relative to nouns. Numerous
studies or early acquisition indicate that the ease with which verb meanings are associated
onto their appropriate forms is mediated by the learner’s attention to the syntactic
environment (C. Fisher et al, 1994, p. 333; Choi & Gopnik p. 96; Pinker, 1996, p. 39ff.;
cf. Gleitman, 1993 and Landau & Gleitman, 1985). On the one hand, there is considerable
evidence that nouns can be successfully identified on the basis of word-to-world mappings
alone. On the other, verb meanings are apparently determined in large part by their
syntactic and semantic context. This, of course, includes relationships among one to many
argument and/or complement structures (and the nature of their referents) that are requisite
in the phrases (VPs) in which they appear as heads. This has led to the conceptualization
of verb learning as context-sensitive and involving sentence-to-world mapping procedures
(Gleitman, 1993, p. 19 Iff.).
Implicit in this is that a rather comprehensive, prior knowledge of nouns (often
representing participants in the action expressed by the verb), specifically, of those situated
in the immediate context where a target verb form occurs, are prerequisites to the accurate
extraction of verb meanings. The gradually accumulated and complex knowledge of verbs
149
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
and their behaviors (e.g., in the creation of subcategorization frames) is drawn from
inferences obtained by observing the patterns—syntactic and semantic contexts—in which
they usually appear, specifically those that may restrict the kinds (semantic subtypes) and
number of nouns that can participate in the expressed action, and the modifiers co
occurring within relevant phrasal and clausal boundaries. Therefore, a greater awareness
of the structural (syntactic) and semantic characteristics of the language being learned or
acquired is necessary for the proper identification of verbs (C. Fisher et al, 1994, p. 336).
The more knowledge required to master a form-meaning set or structure, the more time it
will take to acquire.
As discussed previously (subsection 3.3.4), verbs exhibit a much wider range of
syntactic behaviors than nouns, based on their semantic requirements. Primary verbs are
syntactically the main elements of the VP and, as a consequence, are crucial to the
organization of the sentence (van Hout & Muysken, 1994, p. 55). They also assign
semantic roles appropriate to their core meanings, and, therefore, entail a great deal more
grammatical knowledge in performance than the simple naming of PERSON, OBJECT, or
ACTIVITY. The complexity implicit in the selection process is not merely related to the
number of complements but is also intrinsically related to the number and type of semantic
roles. For example, a PERCEIVER must be a CONCRETE noun that is ANIMATE, probably
HUMAN. Metaphorical extension implies some knowledge of original semantic parameters:
if one says something along the lines of “That stone was just sitting there, staring at me...”,
it is assumed that the metaphor is accomplished by conscious choice, and presumably for
literate effect (a figure of speech). The overall concreteness of the term stone does not
change.
150
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Regarding individual members of the word-class Verb, the number of complements
may also vary depending on the intended specificity of a particular utterance. If the number
of arguments were the sole determinant of syntactic or semantic complexity and, as a
consequence, were decisive in respect to leamability or borrowability, then any term
representing the semantic subtype GIVE would be among the most complex (requiring three
semantic roles: a recipient and a gift in addition to a giver) and the subtype MODAL would
be among the least complex (requiring no independent roies). Based on a simple count of
required arguments, modals should be learned or borrowed before giving-type verbs, and
that has never been observed to be the case. Given the inherent complexity of verbs in
general, the leamability or borrowability of particular verb types or subtypes undoubtedly
depends on a number of other factors. General issues of complexity in acquisition or
borrowing regarding classes of Noun, Verb, and Adjective are listed (Table 4.4):
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 4.4. Content items according to semantic and syntactic complexity.
W ORD CLASS
Noun
Syntactic:
Semantic:
Adjective
Syntactic:
Semantic:
Verb
Syntactic:
Semantic
CHARACTERISTICS
Occupies syntactic slots as head of NP; may receive markers of
situational number, inherent gender-class, or grammatical case; relative
syntactic importance outside its own domain (NP) depends on
grammatical role (subject object, etc.); position (with respect to
grammatical role) is relative to V of VP.
Identifies the topics of discourse (participants of processes, events, or
activities specified by V of VP); generally more specific than verbs,
especially those they accompany (i.e.. there are more names for
possible participants than there are for activities, which is reflected by
sheer numbers in an individual lexicon).
Position, function, and distribution is relative to head of NP; any
syntactic process that applies to head of NP (by virtue of co-occurrence
patterns) applies to Adj-modifier, dependent on head noun if marked for
categories of agreement; less likely to receive nominal inflections (via
agreement or concord) than head of NP; syntactic behavior more
restricted relative to verbs (e.g.. to the types of complements they take).
Meaning modifies head of NP; selection is determined by characteristics
of modified noun; usually more abstract (expresses QUALITY, not
OBJECT) and general (may appear in a greater number of contexts) than
noun it modifies.
As the nucleus of a clause, establishes syntactic links among clausal
constituents; focal point of the greatest number of syntactic procedures;
contains markers of the greatest number of inflectional and derivational
categories; in many languages, must agree in number, gender-class with
grammatical subject; assigns specific grammatical case in languages
that mark this category.
Determines semantic role of its arguments, one to three places in
English, with secondary roles marked by prepositions—markers
resulting from semantic (and. therefore, syntactic) requirements
constitute most of the significant differences among the worid’s
languages regarding morphological structuring; relative to adpositions,
assigns the greatest number of semantic roles (and case markers).
In sum, there is an overall asymmetry of relationships among these three word
classes. For example, there is a dependence implicit in the relationship of modifier-
152
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
modified; adjectives are subsidiary in a number of aspects and characteristically respond to
the syntactic and semantic categories of the nouns with which they co-occur (Hawkins,
1988, p. 8-9). Nouns, as labels associated with the most concrete and specific of
(ontological) referents, generally involve the most consistendy direct, one-to-one mappings
possible in a language; however, they are dependent (positionally, functionally, and
distributionally) upon the head of VP and are inert with respect to the syntactic makeup of
the clause. This general pattern of syntactic and semantic dependency is portrayed in (7),
below (in increasing levels of semantic dependency):
(7) V (head of VP) > N (head of NP, argument of clause) > Adj (constituent of NP)
This suggests, regarding the leamability and borrowability of content items, that
(primary and secondary) verbs require the most extensive knowledge of a particular
language because the interpretation of every other element is directly or indirecdy dependent
on them (for example, recognizing the word class noun can often be accomplished only
with respect to its position relative to the main verb). One would expect that nouns and
adjectives would be learned or borrowed in roughly the same proportions. However, that
conclusion must be tempered by the fact that the number of form-meaning sets of particular
semantic sub-types assigned by an individual language to the word class Adjective may be
from zero (in languages like Cree) or very few (e.g., in Igbo or the Bantu languages) to
many (e.g., Dyirbal and the European languages) (Dixon, 1991; cf. Waxman, 1994, pp.
249— 250). In addition, adjectives are almost always subsidiary in meaning and generally
less transparent than the nouns they modify. These basic semantic and syntactic contrasts
are essential to the identification of content items. The numerous factors discussed above
paint a picture of a dynamic process affecting leamability and/or borrowability that can only
be applied to one language (or, perhaps, pair of languages) at a time.
153
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4 .4 . Su m m a r y a n d G e n e r a l pr e d ic t io n s
It is now possible to assemble the various aspects of a semantically based model of
borrowing and state the predictions that naturally follow. It is apparent that languages
borrow forms and form-meaning sets for a variety of reasons. Despite the inherent
uniqueness of every human language relative to its particular genealogical (genetic) history
and typological position on indices of synthesis and fusion, it has been suggested that all
forms are borrowable, in principle27, even though not every form is borrowed in practice
(cf. Hudson, 1980, p. 60; Bynon, 1977, p. 255; Haugen, 1950, p. 224). Nevertheless,
there is a consensus in studies of bilingual/language contact phenomena that certain forms
are borrowed more often and in greater numbers than others; for instance, classes of nouns
are always borrowed more easily and more frequently than classes of verbs or adjectives
(see, for example van Hout & Muysken, 1994. Therefore, the following Hierarchy of
Borrowability was developed (see subsection 2.2.3) as a generalization of the patterns
represented in the various individual borrowing hierarchies (rankings) found in the
literature to date (repeated in (8), below):
(8) content item > function word > agglutinating affix > fusional affix
Considering the systematic variability of linguistic forms and the inevitable
possibility of cross-linguistic mismatches, the present work has come to the conclusion that
the morphological structuring of each language in a specific contact situation is a significant
27 This is based on two assumptions. First, individual bilinguals may achieve high levels of proficiency
in both (or all) languages in contact to the extent that they will have access to any form in either language.
Second, form-meaning sets representative of every type portrayed in the Scale (or continuum) of Morpheme
Types have, at one time or another, been borrowed by one of the world’s languages.
154
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
factor in the ultimate determination of borrowability. Moreover, the recipient system,
which acts as a morphosyntactic matrix into which borrowed items are placed, plays the
determining role in the assignment of donor form-meaning sets to appropriate form types
and, subsequently, the distribution of such form types as content items into appropriate
syntactic word classes based on semantic type or subtype. Consequently, two principles
have been proposed to represent systematic limitations on borrowing inherently imposed by
the matrix: (a) the Principle of System Compatibility (PSQ, which defines the
morphological compatibility of donor forms with a recipient system, and its correlate, (b)
the Principle of System Incompatibility (PSI), which identifies a cutoff point along a scale
of morpheme types beyond which forms cannot be borrowed (Section 2.3). In essence,
the PSI states that no form or form-meaning set is borrowable that does not conform to the
morphological possibilities of the recipient system with respect to form (i.e., morpheme)
type. This is based in part on the assumption that a recipient language will always act to
preserve its own morphological integrity (for the sake of continuity and intelligibility within
the community of its speakers). Table 4.5, below, presents a synthesis of the PSC and
PSI.
Table 4.5. Compatibility versus incompatibility.
DFXIS COMPATIBLE Y-FORMS INCOM PATIBLE Y-FORMS
Isolating-analytical words/roots (analyzed
as independent words)
affixes of any kind
Agglutinating independent words/roots
+ agglutinating-type affixes
only fusional-type affixes
Fusional any Y-form borrowable, i.e.,
word/roots and affixes of
any type
155
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The first general set of predictions, therefore, pertains to types of form-meaning
sets. First, incompatible forms cannot and, therefore, will not be borrowed. Second, all
others (i.e., compatible form-meaning sets) will be borrowed according to the rankings
indicated in the Hierarchy of Borrowability. Consequently, analytical-isolating languages
will not borrow affixes of any kind from any type of donor though they can freely borrow
content items and function words (in each case, bound words or roots analyzed as
independent words). In addition, agglutinating languages will not borrow fusional affixes,
all other form types being borrowable. Fusional-type languages have the broadest range of
borrowing possibilities: they can borrow any type of form-meaning set from any type of
language—in principle2 8 .
Even though the PSI limits the bare possibilities by identifying specific form types
that must be ruled out, it does not account for the fact that, among those that are system-
compatible, some are borrowed more frequently than others, and some types of
morphemes are rarely, if ever, borrowed. By concluding that borrowable items are actually
form-meaning sets that become words or morphemes in the recipient, the focus shifts to
individual characteristics of types of form and those of types of meaning. In addition,
acquisition of form-meaning sets in bilingual contexts clearly involves processes linked to
simultaneous and/or sequential bilingual acquisition. Thus, it is assumed that form-
meaning sets are learned in much the same way irrespective of language of origin; some
sets are more difficult and take longer to learn than others because they require greater
knowledge of the source language grammar. The task is, therefore, to isolate the properties
that may cluster in various combinations and degrees to make form-meaning sets gradiently
28 This, however, needs to be modified to more accurately reflect actual borrowing scenarios; in all
likelihood, numerous restrictions apply to form and meaning types that would restrict borrowing to areal
neighbors that are genetically related as well as typologically similar (for discussion, see Section 6.1.
below).
156
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
identifiable and leamable, and, consequently, borrowable, and that can account for
hierarchies (scales or continua) of borrowability. Following the scenario that
learners/acquirers assume initially that utterances within the linguistic environment are
meaningful, and that their primary task is to isolate individual forms (and structures) out of
the speech stream and map them onto appropriate meanings (cf. Slobin, 1985), sets of
interpretation characteristics (oppositions linked to degrees of identifiability) based on form
have been proposed (subsection 4.1) and, subsequendy, on meaning (4.2). Form-meaning
interpretation characteristics (FMICs) identified and discussed in this work are presented in
Table 4.6, below, with those pertaining to form listed first:
Table 4.6. FMICs: form-meaning sets
Transparency
SHAPE: unique (detailed, segmentable)
MAPPING: 1-to-1 link to meaning and back to
a single form
SELECTION: optionally selected on the sole basis
of particular, individual meaning
CONCRETENESS: reference to distinct entities or concepts
that exist
SPECIFICITY: explicit: distinguishes among members
of a species or type; typically occurs
in a limited number of contexts
GRAMMATICALIZATION: belongs to a broad lexical subsystem
(i.e., semantic type or subtype)
157
Opacity
minimal, no (or zero) form
l-to-many simultaneous links to
multiple meanings; return linkage
of each separate meaning to many
forms
obligatory; occurs in response to
the requirements of language-
specific grammar; gives syntactic
form
abstract; apart from concrete or
physical existence
general; applies to many (all
members of a group); consequently
may occur in a larger number of
contexts
belongs to a language-specific
grammatical (i.e.. inflectional)
category
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
A second set of general predictions begins with the most obvious cases. First,
form-meaning sets possessing all of the characteristics in the left-hand column, typically
content items, will be the most highly preferred in any borrowing situations. According to
the PSC, any such donor form, irrespective of the morphological possibilities of the donor
language, is borrowable into any recipient, irrespective of its morphological character, as
well. In principle, content items are borrowable from any language into any (in fact,
EVERY) language. Second, form-meaning sets exhibiting all of the characteristics in the
right-hand column, typically inflectional affixes of the fusional type, will be the least
preferred. More specifically, the fusional affixes of one language can only be borrowed by
other fusional languages—probably only by those that also have close social, genetic, and
areal relationships. Implicit in the PSI is that the borrowability of any grammatical form
will correspond to its relative position along the scale of morpheme types. A third general
prediction follows concerning form-meaning sets with properties clustering at midpoints
between the extremes (certain subclasses of function words): they will be intermediate with
respect to identifiability, degrees of grammaticalization and, therefore, borrowability. More
specific predictions concerning individual oppositions of form-meaning interpretation
characteristics that have been proposed thus far are summarized in Table 4.7, below:
Table 4.7. Summary of predictions regarding FMICs
FORM:
shape
mapping
selection
sufficiently detailed, segmentable shape > minimal or no shape
1-to-l form-meaning > 1-to-many simultaneous form-to-meaning;
1-to-l form-meaning > 1-to-many simultaneous meaning-to-form
selection optional > application/selection obligatory
MEANING:
concreteness concrete > abstract
specificity specific > general
grammaticalization semantic (sub)type > inflectional category
158
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Regarding individual subclasses of content item (independent word, root),
particular syntactic and semantic factors may also be powerful predictors. For example,
regarding the borrowability of nouns, verbs, and adjectives, the relative semantic and
syntactic complexity of verbs compared to nouns offers a tentative explanation for ratios of
borrowing frequency that are reflected in various borrowing hierarchies. If these properties
are combined with form-meaning characteristics and applied in individual cases, much
more specific measures are obtained by which precise predictions of borrowability can be
made, even within particular semantic subtypes. Regarding the relative borrowability of
types of inflectional affixes, morphological criteria are available (according to the PSI), in
addition to form-meaning interpretation characteristics.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5.0. C h a p t e r 5— b o r r o w i n g P a t t e r n s in M o d e r n M e x ic a n o
Mexicano (Ntihuatl) may be the most studied of the languages indigenous to the
Americas. Before the advent of Spanish, it was the dominant language of Mesoamerica,
the language of the ancient Toltecs, and spoken also by the Aztecs, Tlaxcalans, and other
peoples of Mexico and Central America. Numerous works concerning its grammatical
structure (morphology, phonology, and so on) and the nature of its various dialects have
been published since the latter parts of the 16th century. In addition, many teaching
grammars have appeared in more recent times in such languages as Spanish, English,
French, German, and so on1 . This extensive documentation provides a rather unique
backdrop against which the specific claims presented here can be examined, especially
concerning the long-term effects of borrowing and certain other temporal aspects (e.g.,
nouns are likely to be borrowed before inflectional affixes).
Largely as a result of the clear, uninterrupted historical links with the so-called
classical language (i.e., the language prior to the conquest and subsequent colonization of
its speakers by the Spaniards) to its modern-day descendants, there has been little
controversy regarding its genetic origins. It is traditionally classified as a member of the
Uto-Aztecan family of languages, even in the face of its greatly evolved character—the
rather obvious product of its intense and prolonged contact with Spanish. The tendency to
stick religiously to a genealogical perspective has been referred to by some as the Adamic
Model (Aarsleff, 1982) and seems to be fairly typical of linguists whose work tends toward
the establishment of proto-languages, perhaps in attempts to discover evidence of an
original human language (Hill & Hill, 1986, pp. 55-56). Thomason & Kaufman (1988,
pp. 1 — 12) devote much space to a discussion of the predispositions of mainstream
1 See Ledn-Portilla (1972) for an extensive bibliography of Ndhuatl studies.
160
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
historical linguists to ignore all sorts of language mixing (and their cumulative effects)
which can result in the emergence of varieties not easily classed according to family
lineage, that is, whose origins are not traceable to a single progenitor (which evidently
includes pidgin and creole varieties).
Maintenance of the traditional classification scheme, therefore, implies that the
development of Modem Mexicano has proceeded according to normal processes of change.
Surely, this betrays the biases of individual researchers more than anything else, especially
in view of the fact that modem versions of the language have changed so much as a result
of contact with Spanish that they are no longer even the same morphological type. That is,
on the basis of morphological typology, Modem Mexicano has drifted substantially away
from its original polysynthetic and incorporating character towards analysis to its present
point along the index of synthesis (Hill & Hill, 1986, p. 249ff.). According to
grammaticalization scenarios— seen as one of a number of normal processes leading to
language change that are internally motivated—the expected direction of change is away
from analysis towards synthesis. However, borrowing is also a normal process and an
obvious factor in change that is externally motivated; it is not merely some kind of linguistic
oddity. The effects that it produces demonstrate its ability to interact with and, perhaps,
significantly alter other normal processes of change in dynamic ways.
Judging from the results, however, it seems rather clear that the transmission of
Classical Mexicano has been something other than according to a normal, genetic model.
As a consequence of different kinds of borrowing, it is no longer possible to attribute the
entire lexicon and structure of present-day varieties of Mexicano to one and only one
source. This includes (among other things) the proliferation of Spanish-like prepositional
phrases and co-occurrence patterns within the noun phrase (see subsection 5.3.2, below).
161
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
According to the present definition, it is a mixed language (see section 1.2)2. Regarding
the bilingual nature of such mixtures, Thomason (1997a) states that “...linguistic material
from each source language is adopted wholesale, without the kind of distortion that would
occur in the absence of bilingualism” ( p. 6). This is evident in Modem Mexicano, as well,
where there is little phonological reanalysis of borrowed Spanish elements according to
native (original) Mexicano phonological patterns3. Spanish-origin loans tend to be
pronounced as they are in Spanish (Hill & Hill, 1986, p. 198), which is a marked change
from the extensive sorts of phonological reanalysis that took place in the earliest years of
contact (see Karttunen & Lockhart, 1976)4. Native Mexicano elements are pronounced as
they were in its so-called classical progenitor, with phonemic contrasts among long and
short vowels (e.g., 5 versus o in toca “to bury” and toca “to follow, pursue”) (Hill & Hill,
1986, p. 62). Consequently, the phonemic inventory of the modem language has
expanded to include a number of Spanish consonants (which appear only in Spanish
borrowings)—reminiscent of the phonological characterizations of Michif (see Bakker,
1997, p. 80ff).
Mixed languages (bilingual mixtures) such as Michif, Media Lengua, Ma’a (Inner
Mbugu), Pecu' (Petjo or Petjoek), Mednyj Aleut, and so on exhibit many of the same
structural effects of contact as Mexicano, but they have not been treated in the same
2 It is true that the definition of mixed language in the first chapter differs in a number of respects from the
one presented in Bakker, 1997, which is stated as follows: "A mixed language is a language that shows
positive genetic similarities, in significant numbers, with two different languages” (p. 195). However,
from the point of view taken here, the two are not necessarily mutually exclusive or contradictory. In the
long run, Bakker tends to restrict the number of languages he classes as mixed languages. In the present
work, language mixing/intertwining is looked at more as a process that may occur in a broader range of
contexts. The differences seem to be definitional, more or less. The unique characteristics of Michif are
remarkable from either standpoint. (See. also, Bakker, 1988,1994, and Bakker & Pappen, 1997.)
3 See Chapter 1 for discussion of the phonological adaptation of loanwords. Apparently, the kind of
phonological reanalysis that occurs in Media Lengua (discussed in Muysken, 1981,1988, 1994, and 1997)
is not an absolute criterion for relexification and/or language mixing.
4 Older loans have become relatively infrequent in the modem, spoken language, often being replaced by
newer, nonnativized forms. At times, such nativized forms as compalehtzin "co-father” and compalehGnn
“co-mother” (ritual kinship terms) occur alongside nonnativized forms compadrito and comadrita.
162
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
conservative manner regarding genetic classification. The special treatment afforded them
is very likely due to the circumstances surrounding their investigation (the when and where
of the research). Many of these mixtures are said to have developed within a relatively
brief span of time; and, inasmuch as the emergence of many of them has been somewhat
recent, each has a relatively short history. In addition, the effects of what appears to be
greatly accelerated change have been systematic and extensive. As a consequence, attention
has been paid in a much more contemporary context (as opposed to, say, more traditional,
historical approaches) to such linguistic and social issues as the roles of primary and
secondary language acquisition (e.g., native versus non-native acquisition) and code
switching in processes of change.
Irrespective of the special status mixed languages such as Michif and Media Lengua
receive in current work in bilingualism/language contact phenomena, it seems reasonable to
say that the linguistic consequences of powerful, dynamic social processes can certainly be
cumulative as well as acute5. The presence of a literature tracing the historical development
of an individual language certainly should not prohibit the unbiased (synchronic)
assessment of its character at particular points in time. Neither should the amount of time
it has taken to develop (the diachronic), for instance, whether it has emerged rather
suddenly or over longer periods of time as a consequence of the social, economic, and
consequent linguistic subjugation that typically accompanies colonization. Nevertheless,
the knowledge garnered from studies of language mixing of various kinds, sudden or
gradual, presents a challenge of some import to more traditional approaches. It offers an
alternative perspective with respect to the many linguistic phenomena frequently occurring
in our increasingly multilingual world (cf. Milroy and Muysken, 1995, p. 1). Such
5 According to Thomason (1997a and 1997b), Ma'a (Inner Mbugu) arose as a result of “long-term
linguistic persistence in the face of intense cultural pressure from Bantu...” (1997a, p. 6). See 6.1.4,
below, for discussion of Ma’a.
163
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
information has the potential to shed significant light on language change in general and,
viewed retrospectively, on the origins of many known languages.
5.0.1. Bilingual phenomena
A number of seemingly contradictory observations have been made about the nature
of Mexicano-Spanish bilingualism. On the one hand, nearly all Mexicano speakers of the
Malinche region are fluent in Spanish, although the reverse is not necessarily true. In most
areas where the use of Mexicano still flourishes, there remains considerable diglossia
despite the massive borrowings that would indicate advancing shift towards Spanish. This
suggests a fundamental asymmetry in the relationships between the two languages. On the
other hand, the intense and relatively stable bilingualism characteristic of the region has led
to the description of modem forms of Mexicano as “syncretic” by such researchers as Hill
& Hill (1986, p. 55-58). In addition, inter- and intra-sentential code-switching are normal,
everyday occurrences. This indicates that both autonomous language systems are integral
and active linguistic components in the construction of the social identities of individuals,
and, indeed, of the community in general.
The fact that Mexicano still occupies its own social domains underscores its
function as an ethnic language and marker of social identity, though this may be gradually
changing as shift progresses. The purist attitudes of some older speakers place a positive
value on the knowledge of legitimo mexicano (i.e., legitimate Mexicano) and hold
perceived Mexicano forms in high esteem. Such behaviors as borrowing and types of
code-switching are often considered (i.e., by purists) to be corruptions of the ancestral
language (Hill & Hill, 1986, p. 99ff). Nevertheless, it may very well be that the opposing
164
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
forces of shift and maintenance work against each other to establish a kind of equilibrium
that enhances its ultimate potential for survival (cf. Woolard, 1989, 355— 363). It is also
important to note that change may not be as easily recognized and militated against as the
purists might wish. The ability to distinguish true ancestral forms from borrowed ones can
be obfuscated by a number of factors, not the least of which is the basic adaptability of the
language and its speakers (i.e., the ability to draw effectively from familiar outside sources
for linguistic support). In the case of the Malinche and other regions where Mexicano is
still spoken, code-mixing (borrowing) and code-switching have been part of the social
fabric for nearly five centuries (Hill & Hill, 1986, p. 2).
Speakers of Mexicano appear to borrow words and structures quite freely from
Spanish. Therefore, the identification of Spanish form-meaning sets is a necessary part of
the proficient bilingual’s linguistic knowledge. It can also be assumed that highly
grammaticalized elements (e.g., fusional affixes) are recognizable and interpretable by
proficient bilinguals—which may not depend on any conscious ability or knowledge.
However, the linguistic capabilities of proficient bilinguals (e.g., in simultaneous
translation tasks) cannot be equated with the compatibility of competing languages in regard
to their respective morphosyntactic matrices. Even if one assumes that extensive
knowledge of Spanish is characteristic of the bilingual Mexicano-Spanish speaker/hearer
(who uses either language separately or during various kinds of code-switching and
borrowing behaviors), it is not adequate to state simply that all form-meaning sets are
borrowable from Spanish into Mexicano because, in fact, morphological typology is likely
to erect absolute barriers between languages that are neither genetically related (however
that may be defined) nor typologically similar.
165
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5.0.2. The corpus
The data base of this chapter consists of a 23,272-word portion of the oral
interviews contained in the corpus (of approximately 80,000 words) gathered by Kenneth
and Jane Hill of the University of Arizona. The data were collected over a ten-year period
beginning in December of 1974 in the ethnographically distinct Malinche region, which
includes communities on the western and southwestern slopes of the Malinche Volcano in
the adjacent Mexican states of Tlaxcala and Puebla6. At the time, the population estimates
of the towns and villages in which the interviews were conducted ranged from 500
(Santiago Ayometitla) to 20,000 (San Pablo del Monte). Hill & Hill describe the region as
a “particularly indigenous cultural island within the Spanish-speaking communities of
Tlaxcala and Puebla” (1986, p. 7), basing their description on such cultural factors as ritual
kinship relationships and other patterns of social organization. They paint an intimate
portrait of a community in the midst of enormous socio-economic pressures, whose
members are being drawn by irresistible forces of modernization. The Spanish language,
in opposition to the indigenous Mexicano, is the medium through which much social
change finds linguistic expression. Its use has come to symbolize technological progress
and economic advancement in many socially relevant areas.
Overall, sociolinguistic surveys (which included interview materials) were given to
ninety-six subjects in the Tlaxcalan towns of San Pablo del Monte, San Antonio
Acuamanala, Santiago Ayometida, Santa Marfa Acxoda del Monte, San Luis Teolocholco,
San Rafael Tepadaxco, San Felipe Cuauhtenco, and Santa Ana Chiautempan and in the
6 The text of this survey can be found in Hill & Hill, 1986, Appendix A.
166
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Pueblan towns of San Miguel Canoa, La Resurreccion, and San Lorenzo Almecatla7. Each
was administered by a native Mexicano speaker (a native of San Miguel Canoa)—partly to
keep the conversation from drifting into Spanish, the language preferred in speaking with
outsiders. Because of the interviewer’s relatively young age (16 at the beginning of their
research), the Hills felt that they were allowed to minimize the feeling of formality that
usually accompanies such interviews. The survey itself was composed of five sections:
Part A—Base Data on Speaker, in which background information of the respondent was
gathered; Part B—General Conversation, containing prompts designed to elicit reports of
important events, near-death experiences, and local legends; Part C—Morphology, which
involved the translation of words, phrases and short sentences from Spanish to Mexicano;
Part D—Language Attitudes and Self-Reports of Usage, which drew out the circumstances
under which Mexicano is used by the speaker and so on; and Part E—One-Hundred Word
Lexicostatistic List, used to measure knowledge of “pure” Mexicano forms. One of the
purposes of the survey was to create a large enough sample to investigate correlations
among attrition, the accompanying loss of functions that typically result, and the expected
narrowing range of structural possibilities.
One of the many things Hill & Hill encountered was a continuum of usage which
served to demonstrate a speaker’s pragmatic competence and which could be linked to
varying numbers of Spanish or Mexicano forms (p. 71). In this respect, Hill & Hill
disclose that the nature of the interviews appeared to bring out a relatively formal posture
by the participants despite the fact that they took place in the form of visits or visitas (p.
74)8. As a result, many of the samples that they obtained (in what the participants may
7 The portion of the Hill & Hill corpus used here is from the towns of San Antonio Acuamanala
(population 3.18S), Santiago Ayometitla (500), and Santa Maria Acxotla del Monte (800), all located on or
near the western slope of the volcano.
8 These visits were of four types. A casual home visit involved a chat in a kitchen area where women were
generally quite comfortable. A second, more formal kind of home visit entailed sitting in the main room of
167
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
have perceived as somewhat official settings) may have been skewed more toward
Spanish—the so-called power code—than might be expected under more normal
circumstances. In more domestic settings, an increase in Mexicano forms (the purist code)
was a sign of speech among intimates. The registers evident in the interviews, however,
did not contain the amount of Spanish material as those typically used when the topic
drifted towards politics, for example. In other words, topic and register clearly affected the
ratio of Spanish or Mexicano lexical items that would occur in a particular speech event. In
discourse settings at large, there was also the rather frequent occurrence of code-switching
of various kinds (and, presumably, nonce borrowings). In spite of these and other
potential drawbacks9, the corpus contains many linguistically revealing encounters. It
remains true, nonetheless, that the accumulated effects of contact and the considerable skill
with which these speakers manipulate their linguistic systems are remarkable.
5.0.3. Organization of this chapter
The remaining portions of this chapter deal with actual borrowing patterns found in
the data. Section 5.1. begins with a brief overview of the two participant morphosyntactic
systems (the donor, Spanish, and the matrix language, Mexicano) and describes the
various kinds of Spanish form-meaning sets that have been borrowed into Modem
Mexicano. Subsection 5.1.1. goes into some detail in describing the morphological
the house (near a family altar), where the visitors were served food and drink by women scurrying back and
forth from the kitchen. Two other contexts included informal gatherings of men (drinking the traditional
pulque) on their days off or in the evenings, and during work groups—while people are mending farm
equipment, shelling com, and so on.
9 Hill & Hill devote considerable space to the explanation of their interview techniques and the content of
the survey (pp. 67-89). Those who may be critical of any and all such techniques may benefit from reading
the pages referred to. The many examples offered by the authors clearly leave one with the impression that
much good and valid is obtained despite the inherent problems, especially in light of the fact that
respondents often reacted to the prompts in very creative and unpredictable ways.
168
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
structuring of each language. In 5.1.2., borrowing patterns are compared to various
hierarchies discussed in Chapter 2, and a specific hierarchy is constructed for Mexicano. It
is shown that Mexicano has, indeed, borrowed form-meaning sets from every compatible
type, from content items to agglutinating-type affixes. This subsection concludes with a
brief discussion of the Principle of System Compatibility (PSQ and its corollary, the
Principle of System Incompatibility (PSI) and the ways the two principles delimit
borrowable and unborrowable form-meaning types in this particular contact situation. The
patterns found validate both: only compatible forms are borrowed, and preferences clearly
follow from those depicted in the Hierarchy of Borrowability (see subsection 2.2.3.).
Section 5.2 breaks down borrowing patterns according to form-meaning
interpretation characteristics (FMICs). Subsection 5.2.1. deals with the three
characteristics pertaining to form and each prediction that has been made with respect to
borrowing preferences (see subsection 4.1.3). Each prediction is upheld across the board,
which is reflected quite clearly in the data. Recall that the PSI rules out any occurrence of
Spanish fusional-type inflectional morphology. Form-meaning interpretation
characteristics merely predict that certain types of form-meaning sets will be more or less
favored in borrowing situations. 5.2.2 takes up issues of meaning and focuses on
subclasses of Noun in making comparisons between concrete and abstract nouns; it then
treats oppositions based on specific versus general (i.e., hyperonyms contrasted with
hyponyms). Abstractness links to the lateral (horizontal) organization of the lexicon in
terms of semantic types and subtypes, while specificity correlates with the hierarchical
(vertical) relationships of items within those types. Borrowing patterns clearly reflect these
characteristics in systematic fashion. The final prediction of this particular subsection, that
form-meaning sets linked to semantic (sub)types will be preferred over those linked to
inflectional meanings, is fulfilled, as well. There is a strong (though not exhaustive)
169
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
correlation with characteristics of form (e.g., whether or not selection is optional or
obligatory). Once again, similarities to synchronic grammaticalization dines is brought to
the fore.
5.3 discusses the long-term effects of borrowing on the Mexicano matrix and, in
5.3.1, some of what can be gleaned from past studies concerning the temporal aspects of
the predictions. As indicated in the work of Lockhart & Karttunen (1976), borrowing
patterns have also substantially followed the predictions implied in the Hierarchy of
Borrowability with respect to the order in which elements have been borrowed. In
addition, there are important indications that, for nearly 300 years, Mexicano resisted
structural borrowings to a great extent Only towards the very end of the 18th and
beginning parts of the 19th centuries did large numbers of borrowed function items begin
to occur, at the very time a new national identity was emerging (that was not strictly
associated with allegiance to the Mexicano language) and Mexico began to pull away from
the colonial yoke of Spain (Wright, 1992). This may be especially significant when one
considers the relatively remote status of the Malinche region. Subsection 5.3.2 focuses on
some of the effects various borrowings have had on Mexicano syntax, that is, on phrasal
and clausal organization.
The discussion in 5.4 turns on the centrality of meaning to borrowing. Apparently,
borrowing hierarchies to date have focused on form, while semantic characteristics have
largely been ignored (with the possible exception of formal links to inflectional categories).
Subsection 5.4.1 covers the role of form in the determination of borrowable form-meaning
sets and links between borrowing hierarchies and clines of grammaticalization; it reaffirms
intuitions that the more structural or grammatical an item is, the more difficult (i.e., the
more time) it takes to leam/acquire and the less likely it is to be borrowed. In addition,
170
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
rankings among classes of N, V, and Adj coincide with (a) predictions based on semantic
and syntactic complexity and (b) the distribution of form-meaning types in Mexicano.
5.4.2 suggests that borrowing patterns reflect different kinds of meaning, and that meaning
may be more significant than characteristics of form in accounting for the distribution of
borrowed forms into the recipient lexicon.
5.1. Overview of the Participants: Mexicano and Spanish
In general, the morphosyntactic matrix of Modem (Malinche) Mexicano retains
much of the agglutinating nature of its progenitor1 0 . Consequently, it is positioned higher
on the index of synthesis than Spanish. It is still considered to be polysynthetic and
incorporating despite losing much of the incorporating character of its classical ancestor as
a result of contact (Hill & Hill, 1986, p. 249-266). An extensive repertoire of compounds
(including those incorporated into verbs) has diminished considerably as a result of the
tendency to borrow Spanish forms to label new concepts (le x ic a l strategies having replaced
morphological ones). This extends to the reduction of other sorts of “adverbial” material
previously incorporated into the verb, as well. As a consequence, its morphosyntactic
matrix has drifted away from synthesis towards analysis. Typical of the agglutinating
nature of the Uto-Aztec an family of languages, Mexicano verbal affixes generally express
one, and only one individual inflectional or derivational concept1 1 . In contrast, Spanish is
positioned much higher on the index of fusion. Typical of the Romance languages, its
verbal morphology is strictly fusional. Multiple inflectional concepts are semantically and
10 For more comprehensive studies see Andrews (1975) regarding Classical Ndhuatl, Karttunen & Lockhart
(1976) concerning Ndhuatl during the colonial period, and Hill & Hill (1986) for the particular variety
spoken in the Malinche region.
11 See example (14) and its accompanying explanatory note (fn. 6), below.
171
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
phonologically fused together onto single, unsegmentable forms, which results in the
formation of verb paradigms (in three conjugation classes). Nevertheless, Spanish is much
more analytical than Mexicano— as a general rule, it does not tend to construct extremely
long, multimorphemic words typical of polysynthetic languages.
While a comprehensive treatment of Mexicano grammar goes well beyond the scope
of this work, the following passage illustrates some of the characteristics of Mexicano
relevant to this study. For the sake of consistency, the analysis is based on that of Hill &
Hill (1986), including terminology and spelling conventions, which, in turn, is drawn to a
large extent from the terminology of Andrews (1975)1 2 . (P signifies the interviewer, and R
indicates the respondent. The top line is Mexicano; the second consists of English glosses,
which is then followed by Spanish and English translations, in that order.)
P. ...itech nil puebloh aquin ocachi cualli yec-laht-oa demexicanoh.
...in this town who else well WELL-speak-TRNS of Mexicano
^En este pueblo, quien mas habla bien de mexicano?
“Who else speaks well of Mexicano in this town?”
R. poz run puebloh de mexicanoh poz pocos, aun miec genteh
well this town of Mexicano well some even many people
Pues, en este pueblo de mexicano, pues, poca, aun mucha gente
“Well, in this Mexicano town, well, a few, even many
tlaht-oa-h mexicanoh huan miec genteh tlaht-oa-h en castellanoh...
speak-TRNS-PL Mexicano and many people speak-TRNS-PL en castellano...
habla mexicano y mucha gente habla en castellano...
speak Mexicano and many people speak in Spanish...”
huan nin puebloh porque cada puebloh mo-patla-tih run
and this town because every town REFL-change-T este
y este pueblo porque cada pueblo est£ cambiando este
“.. .and this town, because every town is changing this
12 Abbreviations used for the Mexicano data are essentially those of Hill & Hill. 1986: ANT-anterion IMP-
imperative; APPL-applicative; T-outbound purposive; REFL-reflexive; TRNS-transitive; INTRN-intransitive:
F-ftiture; PL-plural; HON-honorific; 3P-third person; and DEF-definite.
172
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
para saludar para platicas...
in order to greet for conversations...
para saludar, para platicar...
to greet (each other), to talk (have conversations)... (S51)
S. 1.1. Morphological structuring: the words of each language
In Spanish, nouns are inflected for gender (masculine or feminine) and number (the
marked form being plural). There is also gender and number concord which spreads to
other members of the individual NP. The Spanish plural marker -s is an agglutinating-type
affix; when present on classes of Noun, it indicates only one concept, plural (more than
one). For instance, in such form-meaning sets as los ninos “the boys” (from el nifio “the
boy”) and las ninas “the girls” (la niha “the girl”), the plural marker is added to the stem in
each case. This marker (and some characteristics of Spanish number) has been borrowed
into Mexicano. With respect to gender marking, both masculine and feminine are marked,
usually by the suffixes -o and -a, respectively1 3 . Words not overtly marked must,
nonetheless, be assigned lexically to one gender or the other (e.g. la luz “light” is feminine,
and el pastel “pastry” is masculine). An occasional neutral form occurs (e.g., lo curioso
“the curious thing”), a probable remnant from Latin neuter (Corbett, p. 215). Hill & Hill
(1986, p. 266) note that Spanish gender concord, while it operates in regional forms of
Spanish, does not systematically occur in Mexicano borrowings—Mexicano has no
grammatical category of gender. Adjectives generally retain a masculine form, which is
apparently a result of the relative frequency of such forms.
13 See Aronoff, 1994, pp. 67-74 for a revealing account of the complexities of grammatical gender in
Spanish. One point that is made there concerns the inconsistency of different markers; e.g.. -o typically
marks the masculine and -a the feminine, but this is not always the case. Gender, then must often be
determined by looking at other members of the NP, i.e.. forms of determiner and adjective.
173
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In clear contrast, nouns in Mexicano receive numerous inflectional and derivational
suffixes, for instance, absolutive (i.e., non-possessed state) -tl(i), plural number -meh or
its variant -tin1 4 , possession (i.e., object possessed), reverential (whose forms also
function as honorifics and diminutives), and pejorative. For example,
(1) cone-tl “child”
child-ABS
(2) no-cone-h “my child”
my-child-POSS
(3) justicia-tzin “justice”
justice-HON
(4) ranchoh-zol “old run-down ranch’
ranch-PEJ
Some non-possessed nouns may be pluralized by reduplication of the initial syllable or by
the suffixation of -h,-meh or -tin, or by the Spanish plural -s, as in the following examples:
(5) tlaca-tl
tt__ »»
man
man-ABS
(6)
tlaca-h
tt T t
men
man-PL
(7)
tlatlacah men
(8) tlatlaca-meh men
men-PL
14 The classical language (Ndhuatl) marked plural only on animate nouns; as a result of contact with
Spanish, plurality has spread to include classes of inanimate nouns (Karttunen & Lockhart, 1976, p. 24).
174
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(9) yolca-meh “animals”
animal-PL
(10)
mulah-tin “mules”
mule-PL
(ID
chiquihuite-s “baskets”
basket-PL
Mexicano nouns also receive affixed pronominal elements (bound pronouns) to indicate
possessor. When nouns are preceded by adjectives, these elements are prefixed instead to
the adjective, for example
(12) no-ahuelitah “my grandmother’1
my-grandmother
(13) to-mero mexicano “our real Mexicano’
o ur-real mexicano
These bound pronominal elements may co-occur with either Mexicano or Spanish plural
forms suffixed to the nominal root:
(14) no-familidh-huan1 5
my-family-PS PL
or “my spouse and children”
no-famflia-s
my-family-PL
One final item which was affixed to the noun stem or root in the classical language
and is increasingly rare in Malinche Mexicano usage (Hill & Hill, 1986, p. 247) was a
spatial or temporal locative suffix, for instance:
15 This particular form may show a degree of phonological and semantic fusion, that is, of -h (PS) and -(i)n
(PL) and of possessed-state and plural, depending on one's analysis. Both Hill & Hill (1986) and Karttunen
& Lockhart (1976) merely indicate that it is “possessed form plural” (Karttunen & Lockhart, 1976, p. 20).
175
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(15) lunes-tica
Monday-LOC
“on Monday”
In the works of Karttunen & Lockhart (1976), Hill & Hill (1986), and various
other authors, these locative forms are often referred to as postpositions, irrespective of
their status as bound or free morphemes16. For example, Sullivan (1988) writes: “The
Nahuatl postposition is equivalent to the English preposition, with the difference that the
postposition is a suffix [emphasis mine] placed after the noun or pronoun instead of before
it” (p. 107). However, Andrews (1975), in direct contradiction, states that “.. .relational
suffixes have often been called postpositions [emphasis his], a name that suggests that they
are prepositionlike elements that merely occur after a substantive instead of before it.. ..
The name is incorrect” (p. 304). From the perspective taken here, they are assuredly not
function words in the generally accepted sense and are rightly considered suffixes. They
do not occur as free-standing elements; when they do occur, they are always affixed to a
nominal stem1 7 . Semantically, they indicate such relational concepts as location, direction,
and instrument (cf. Sullivan, 1988, p. 107-137). Indeed, they appear to be quite similar to
form-meaning sets that are considered case markers in other Amerindian languages such as
Quechua (cf. Lastra, 1968, p. 29).
The possible occurrence and ordering of morphemes relative to the noun (root) in
Mexicano can, therefore, be represented as (16), below:
(16) possessor — nominal root — absolutive/object possessed, singular or plural —
(or) honorific (reverentials or diminutives), pejorative — locative, etc. (case)
16 See Crystal, 1991, p. 269) for a definition of the term postposition which includes word status.
17 For some corroboration of this point, see, also Andrews (1975), which states, “.. .relational suffixes
have often been called postpositions [emphasis his], a name that suggests that they are prepositionlike
elements that merely occur after a substantive instead of before it.... The name is incorrect” (p. 304).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Spanish verbs receive individual markers (morphemes) that contain a fusion of
specific category values of tense-aspect, person, and number that obligatorily agree with
the grammatical subject. Hence, every tensed verb in Spanish receives a fusional-type
morpheme that simultaneously encodes individual category values of (at least) three
inflectional categories: tense (or aspect), person, and number (see subsection 4.1.2,
above). Mood or modality distinctions can be expressed through morphological means
(i.e., subjunctive, imperative, and indicative), which involve separate paradigms which
also indicate person-number values (and tense, in subjunctive), and/or through a relatively
small number of modal auxiliary verbs (i.e., compared to relatively long list of modals in
Germanic languages such as German, Dutch, or English). Neither Spanish verbal
categories nor their individual values are borrowed into Mexicano, with two apparent
exceptions: (a) a few frozen forms such as es que “it is that...” and (b) a small number of
verbs that are treated as Mexicano modals (e.g., debe “must, ought to”, depende “it
depends”, and conforme “it conforms” in the sense of “accordingly”)1 8 .
Morphology relating to the Mexicano verb is considerably less straightforward than
noun morphology (see, for example, Karttunen & Lockhart, 1976, p. 29). Here, the focus
is merely on the ordering and kinds of elements that are part of the verb complex. For
instance, (according to Hill & Hill, 1986) Mexicano verbs receive agglutinating-type
suffixes for future tense (-z) but are prefixed for anterior aspect (o-). Individual verbal
suffixes may include the following: so-called applicative objects (-lia)—“an object in
whose favor (or against which) the action is performed” (Hill & Hill, 1986, p. 159),
18 Hill & Hill (1986, p. 160) state that these forms occur with third-person, singular present tense
inflection; however, in the present analysis, it appears that these are better viewed as individual form-
meaning sets that cannot actually express those categories. In addition, the morphological connection
between es and ser is one of drasdc allomorphy (see Section 1.1, above). And. despite their relative
frequency, modals seldom occur in the infinitive forms. (In Spanish, all verbs receive an infinitive marker
(-ar. -er, or -ir) depending on conjugation class; their are no so-called base forms as in English, to read.)
177
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
outbound (-tih) or inbound purposive, and transitive (-oa) or intransitive (-(i)hui). Prefixes
include imperative mood (xi-) and reflexive (pronominal elements such a s , mo-
“you/yourself”), honorifics (on-, literally “away”), along with a number of derivational
prefixes (such as morphemes yec and nen meaning “well, completely” and “badly,
unfortunately”, respectively). The following are but a few examples from the corpus1 9 :
(17) quin-costar-oa trabajo “It costs them work.”
them-cost-TRNS work
(18) costar-ihui in neca trabajo “That work is costly.
cost-INTRNS DEF that work
(19) ti-nech-prepar-huili-z ce pantzln “You will fix me a piece of bread.”
you-me-prepare-APPL-FUT one bread
(20) xi-c-om-preparar-o in centavos “Get the money ready.”
IMP-it-o/i-prepare-TRNS DEF money
(21) o-t-c-arreglar-oh-queh “We have already arranged
ANT-we-it-arrange-TRANS-PLUR the business.”
in as unto
the business
(22) nen-cuatrear-oa-h “They make terrible mistakes.’
badly-make mistakes-TRNS-PLURAL
(23) t-qui-mo-on-yec-tender-hu-ilia “Do you understand it well?’
you-it-REFL-on-well-understand-TRNS-APPL
(24) ye ni-c-tehtender-oh-tih “Now I go along regularly
now I-it-understand-TRNS-T understanding it.”
19 Each of these can also be found in Hill & Hill, 1986, pp. 158-159.
178
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The ordering of affixal material (morphemes) relative to the verbal root is,
therefore, portrayed in (25), below.
(25) anterior aspect/ imperative — subject marker— object marker — honorific (on-) —
reflexive — derivations (e.g., “well”, “badly”) — verbal root — thematizing
suffixes (transitive, intransitive) — applicative (/plural) — outbound/inbound
purposive — future tense
The distribution of classes of Adjective in each language undoubtedly plays a role in
borrowing patterns (i.e., with respect to their frequency relative to that of, say, verbs). On
the one hand, Spanish has large numbers of adjectives (attributive and predicative) typical
of the Romance languages. On the other, Mexicano has relatively few, as do other
members of the Uto-Aztecan language family. In the classical language, modifiers were
not typically free-standing. Adjectival elements were incorporated into nominal
compounds, and adverbs were incorporated with verbs. Even today, Mexicano speakers
use adjectives sparingly, but do, nevertheless use Spanish borrowings such as bueno
“good”, malo “bad”, chico “small”, and grande “big”. There is a certain ambiguity
concerning Spanish adjectives from a Mexicano perspective in that they can appear alone
(e.g., as predicate nominatives) in such constructions as es viejo “he is old” and in nominal
form as in el viejo (“the old one/man”). Nevertheless, independent Mexicano adjective
forms may be derived by application of the suffix -tic; for instance, mulatic is glossed as
“stupid” and taken from Spanish mula “mule”. There are numerous other Spanish loans in
particular subclasses of Adjective, for instance, many that can also function pronominally
and/or as determiners: algo “some, a little”, alguno “some”, and so on.
A number of Spanish adverbs (e.g„ exacto, pronto, etc.) have been borrowed that
are used with relative frequency. However, in many traditional Ndhuatl studies, a separate
class of adverbs is generally not acknowledged as such, at least not in the sense that
179
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Spanish and other Indo-European languages appear to have a separate word class Adverb
(cf. Andrews, 1975, pp. 27— 34). In general, adverbials and various other free-standing
forms are referred to in the literature as particles (e.g., connectives of various sorts,
exclamations, and hesitation forms). Mexicano also possesses other sorts of function
words, for instance, independent determiners, pronouns, and a small number of
uninflected modal verbs. In contrast, Spanish has a full complement of Indo-European
sorts of independent function words/roots (pronouns, determiners, auxiliary verbs,
prepositions, and conjunctions). In spite of these apparent differences, numerous Spanish
loans that have fallen into the generic class of particles can be divided into either
(subclasses of content items (adverbials) or independent function words/roots based on
their semantic and syntactic characteristics.
In the following tables, adverbials include those form-meaning sets which are
borrowed from Spanish that function precisely as they do in their source language and that
maintain the same range of conceptual meanings as their glosses and/or translations. The
same criteria will apply to kinds of function words/roots. Form-meaning sets that are
clearly members of more than one class (e.g., antes “before”) are treated separately
according to their functions and/or positions within phrases or clauses20. The number of
borrowed Spanish function words is quite striking (see Table 5.2). For instance, there are
numerous borrowed prepositions (e.g., de “of \p a ra “for, in order to”, sin “without”,
etc.), coordinating conjunctions (e.g., o “or”, pero “but”), and subordinating conjunctions
20 See Section 32, above, especially subsections 32.3-3.2.5, for discussion of the formal and semantic
characteristics of function words forming the basis for the distinctions made here. In addition, very frequent
conversational particles such as puesfpoz “well, then” are classified here as function words (see Hill & Hill,
1986, pp. 190-194). which differs from the classification scheme of Myers-Scotton (1998 and elsewhere).
This inclusion is based on the following: conversational particles (a) are semantically empty at the level of
the individual form-meaning set or morpheme (i.e.. their discourse-level meanings are much more abstract
and general than a literal interpretion of the word, itself), (b) are members of a very small closed class, (c)
and neither assign nor receive grammatical or semantic role. Consequently, they are less controversially
classed as function words—not full content items (cf. Lyons, 1995. pp. 65-71).
180
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(e.g., mas in the sense of “even though”, hasta “until”, porque “because”, como “as, since,
like, how”, etc.), including the complementizer que “that”.
The examples in (26) - (30), below, illustrate the role of Spanish function words in
the construction of Mexicano phrasal and clausal frames.
(26) tlacpac hasta tlatzlntlan
above towards below
“.. .(from) high to low”
(27) de huei puebloh
from big town
“from the big town”
(28) pero yenon ocacni hueli in molli
but that (is) more well DEF mole (kind of sauce or gravy)
“.. .but it is even more delicious (or tasty) in mole”
(29) porque in tehhuan nican to-tlahtol cahcorrecto
because DEF we here our-language is correct
“because around here, our language is correct...”
(30) poz neci que acmo nech-pactia
well (it) seems that not yet me-pleases
“Well, it seems that I don’t like it yet”
5.1.2. The borrowing hierarchy of Mexicano
To establish a borrowing hierarchy for this particular bilingual/contact situation,
Spanish borrowings in the portion of the corpus treated here were analyzed according to
types (actual borrowed form-meaning sets) and tokens (the number of times these
1 8 1
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
individual items occurred in the text)21. In Table 5.1, below, borrowed Spanish content
items (words/roots) are displayed (with the number of tokens in parenthesis). Similarly,
Table 5.2 portrays types and tokens of independent function words (or roots) and different
sorts of inflectional affixes.
Table 5.1. The occurrence of Spanish content items.
NOUNS: 570(2,420) VERBS: 81(268) ADJECTIVES: 74 (332) ADVERBIALS 44 (411)
Table 5.2. The occurrence of Spanish function items.
FUNCTION WORDS: 46 (3,221) AGGLUTINATING AFFIXES: I (164) FUSIONAL AFFIXES: 0(0)
A synthesis of the two is presented in slightly different terms in Table 5.3, below.
In view of the fact that function items as a whole (a) are members of closed classes (i.e., in
the donor, and presumably in the recipient, as well), (b) consist of various heterogeneous
subclasses from independent pronoun to fusional-type affix (see, for example, subsection
3.2.3), and (c) generally occur in discourse more frequently than individual content items,
the hierarchy reflects borrowing preferences according to the number of form-meaning sets
that are actually borrowed, that is, types (again, with tokens in parenthesis).
21 The ambiguity resulting from the two separate usages of the term type is, unfortunately, somewhat
unavoidable. In an attempt to minimize the terminological confusion, semantic types (i.e., the
classification of words according to kinds of meanings) will be referred to as semantic (sub)types.
Regarding the following tables, numbers in parenthesis represent tokens to coincide with types in the more
numerical sense. See Appendix II for a complete listing of borrowed form-meaning sets.
182
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 5.3. The occurrence of Spanish form-meaning sets in Mexicano.
CONTENT ITEMS (N, V, ADJ. ADV):
FUNCTION WORDS (PARTICLES):
AGGLUTINATING-TYPE INFLECTIONAL AFFIXES:
FUSIONAL-TYPE INFLECTIONAL AFFIXES:
767 (3,431)
46 (3,221)
1(164)
0 (0)
Clearly, borrowing patterns (in terms of types) accord with the Hierarchy of
Borrowability (see subsection 2.2.3, above), which, in turn, corresponds to the Hierarchy
of Morpheme Types (2.1.3, above). The information presented in Tables 5.1-5.3 gives
substance to observations such as those suggesting that the more structural (or
grammaticalized) an element is, the less likely it is to be borrowed (i.e., independent of its
relative frequency in either language). The striking parallels with grammaticalization clines
suggest that characteristics shared by content items (whatever they might be) allow them to
be borrowed quite easily, with variability among major classes of content items in all
likelihood a reflection of the (a) language-particular distribution of class members in regard
to form and (b) relative syntactic and semantic complexity of members of individual major
classes (see Section 4.3). All things being equal (and apparently they are not), one would
expect types and tokens of each individual form class to pattern out in similar (if not
identical) ways—which has not been the case in any contact situation. The preferences also
suggest that those characteristics which tie function items to language-specific matrices
(i.e., grammatical infrastructures) make them less than ideal candidates for borrowing.
The number of tokens with respect to function words, however, is a sure indication
that Mexicano borrowing patterns are not “normal” in a quantitative sense. Out of 23,272
words in the text, 3,221 are Spanish function words—for instance, prepositions,
conjunctions of some kind, and so on. This means that 1 out of about 7.2 words (or
183
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
13.82% of the total)22 is an independent Spanish function word. This extraordinarily high
rate of occurrence undoubtedly reflects their complete adoption into the Mexicano matrix
and, consequently, their roles as integral parts of the language proper. One must,
nonetheless, bear in mind that the list of potential content items is open-ended and can be
expected to grow. Additional loans (types) can be expected to occur depending on such
factors as topic (e.g., when the discussion shifts to politics or occupational concerns).
Irrespective of these particular functional characteristics (and relative frequency), a general
hierarchy representing borrowing preferences in Modem Mexicano can, nevertheless, be
portrayed according to the following:
(31) content item (independent word, root) > function item (independent word, root) >
inflectional affix (agglutinating-type only)
The information presented in each of the tables, above, also indicates that the
morphological structuring of each participant language is certainly relevant. Mexicano has
borrowed form-meaning sets from all compatible types according to the PSC. The
converse is true, as well: it has not borrowed an incompatible form-meaning set. This
cannot be accounted for by (a) frequency (fusional-type affixes occur on each and every
tensed verb in Spanish), (b) the linguistic abilities of individual bilingual speakers (at least a
subset of bilinguals will be proficient in Spanish as a result of native/first language
acquisition), or (c) lack of inherent borrowability (i.e., fusional affixes have been
borrowed from one fusional language to another, though this is admittedly rare). In fact,
most form-meaning sets of Spanish are compatible with the morphosyntactic system of
Mexicano. According to the PSI, the fusional affixes of Spanish are the only form-
meaning sets that are incompatible, and, therefore, not borrowable.
22 All percentages are rounded off to hundredths.
184
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reminiscent of the situation in Media Lengua (Field, 1997b; Muysken, 1997),
Modem Mexicano has borrowed from Spanish an inflectional affix, the plural -s, and a
small number of derivational affixes—all of the agglutinating type: the diminutive -itolita
(which function in both Mexicano and Malinche Spanish as reverentiais as well as
diminutives23), and the agentive suffix -terolero. A few others such as -mente, which is
used in the derivation of adverbs from adjectives in Spanish, -ado, which creates the past
participle which, in turn, can be used adjectivally, and so on, appear in the corpus on a
small number of words that are probably borrowed whole, as single, unanalyzed units.
This is also consistent with the occurrence of borrowed derivational affixes in Media
Lengua and/or Quechua (Muysken, 1997). In addition, it is possible to create double
plurals and double diminutives by using both the Spanish and Mexicano forms (Hill &
Hill, 1986, pp. 165 and 196).
It needs to be emphasized, however, that Spanish, in fact, possesses only a very
small number of agglutinating-type affixes—most pertaining to classes of Noun and other
constituents of the NP. Almost every category and category value expressable in classes of
Verb are expressed via fusional-type affixes, with the only exceptions being the participial
forms representing specific values of Aspect (which combine in constructions with
auxiliary verbs that are marked with fusional-type affixes of tense, person, and number2 4).
Therefore, Mexicano has, indeed, borrowed a significant number of compatible affixes
relative to the available pool. In keeping with the PSC, one can anticipate the possible
23 They appear to be analyzed along the lines of of native Mexicano forms. According to Myers-Scotton's
distinctions between content and system morphemes (e.g., 1993, p. 99ff.; 1995, p. 238ff.), reverentiais are
considered system morphemes—in present terms, highly grammaticalized. while diminutives are very often
considered to be derivational. Gray areas between derivational and inflectional categories are problematic for
many analyses of morpheme types, and this is no exception (cf.. Bybee, 1985, p. 12; Croft. 1991. p.
190f.). In the present work, the diminutive itolita will be considered derivational, more as a precaution than
anything else, and excluded from the tabulations of borrowed inflectional affixes (types and tokens).
24 These individual Spanish verbal suffixes are borrowed into Media Lengua, which has. in essence, the
morphosyntactic matrix of Quechua (see, for example, Muysken, 1997, pp. 385-386).
185
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
occurrence of Spanish derivational affixes, though the number of loans will most likely be
relatively small compared with the number of possibilities. In spite of the depth of
borrowing (the length and intensity of contact), there are no occurrences of Spanish
fusional affixes, except in frozen expressions (e.g., creo que “I believe that”) and code
switches from Spanish. They are systematically barred from inclusion and incorporation
into the Mexicano morphosyntactic matrix.
5 .2 . THE ROLE OF FORM-MEANING INTERPRETATION CHARACTERISTICS (FM ICS)
In Sections 4.1 and 4.2, above, characteristics associated with the ease of
interpretation of various form-meaning sets were discussed in relation to their likely effects
on borrowing. Within particular language contact situations involving degrees of bilingual
proficiency, it is assumed that those form-meaning sets that are the most easily recognized
within the speech stream will be preferred in borrowing processes over those that are the
least easily recognized. This is in part due to the fact that individual members of a bilingual
community will exhibit different degrees of fluency in each of their languages—forms that
are more easily identified in the speech stream will be easier to borrow. That is, bilinguals
of varying proficiency will be more familiar with those forms. For this reason, predictions
pertaining to characteristics of form are argued to fall into three general areas. First, form-
meaning sets that possess optimal amounts of phonetic information will be preferred over
those with minimal (or zero) phonetic shape. In order to account for the rapid on-line
identification that is necessary for individual words and morphemes (as form types),
correlations have been made between the amount of phonetic information necessary and the
number of possible associations (concepts) that must be made; motivating factors cited
include the economy and efficiency of the retrieval process. Second, uniqueness of form
186
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
must correspond with uniqueness of meaning; the point being that a unique phonological
address must locate an equally unique concept (the inhabitant of the address). Third,
because form-meaning sets are customarily classified according to their selectional
possibilities, form-meaning sets that are selected optionally will be preferred in borrowing
situations over those that are required by language-particular syntactic requirements of the
donor variety.
It is also argued that characteristics of individual form-meaning sets can be
contrasted on the basis of meaning—the other component of the form-meaning set These
characteristics also fall into three areas. Two of these, concreteness and specificity, are
often mentioned in descriptions of the evolution of kinds of meaning in processes of both
grammaticalization and derivation. The first, concreteness, involves the opposition of
concrete versus abstract concepts and figures prominendy in studies concerning the
organization of the lexicon (monolingual or bilingual) and the accessibility of individual
concepts. As a result of their demonstrated link to ease of access (and speed in translation),
it is predicted that those meaning types that can be classed as concrete will be preferred over
those considered abstract. The second, specificity is often mentioned in opposition to
generality. It was initially predicted that the more semantic information associated with a
particular concept (the more specific the term), the easier it would be to distinguish it from
other concepts25. However, in ways analogous to amounts of phonological information
required for individual form recognition, this prediction has been amended somewhat to
suggest that optimal amounts of semantic information will be preferred over minimal
amounts. Based on the hierarchical relationships that exist within families of terms
25 That is, the number of perceptual features (specific physical characteristics) required to define a specific
term is greater than that required for a general term. For instance, the difference between chihuahua and dog-.
the definition of the word chihuahua includes all general characteristics associated with the class of
mammals known as dog, plus additional physical characteristics qua semantic specifications that would
distinguish it from other kinds of dogs.
187
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(semantic types and subtypes), the modified prediction states that entry-level hyponyms
(more specific terms), as individual members of families of terms, will be preferred over
hyperonyms (general family names containing fewer semantic specifications).
The third characteristic pertaining to meaning involves associations of particular
meanings with concept types, for instance, those that are linked to semantic types and
subtypes (see Section 3.3) versus those associated with inflectional categories and their
respective category values. Consequently, it is predicted that the kinds of meaning
associated with individual semantic (sub)types which require little or no language-specific
grammatical knowledge will be preferred in borrowing over those that require a much
greater degree of language-particular knowledge.
5.2.1. FMICs pertaining to form
Predictions with respect to borrowing preferences according to form hold across the
board, as demonstrated in the hierarchy constructed in 5.1.2 above. The comprehensive
correlations that exist between the borrowing hierarchy and predictions based on FMICs
pertaining to form occur irrespective of individual form-class (e.g., whether borrowed
form-meaning sets are assigned to classes of Noun versus classes of Verb) or specific
semantic (sub)type. This is especially significant in view of the fact that nouns, verbs, and
adjectives should be borrowed in equal numbers if phonetic form were the only
consideration. All classes of content items exhibit (a) sufficiently detailed shape, (b) 1-to-l
associations of form and meaning, and (c) optional selection possibilities. The tallies used
to establish the hierarchy also reveal that the form-meaning sets of Spanish that typically
lack these characteristics are clearly preferred less in borrowing processes. Of the many
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
items available for borrowing, those that possess (a) minimal or no phonetic form (e.g.,
inflectional affixes in general) and (b) 1-to-many simultaneous mappings of both form to
meaning and meaning to form (i.e., fusional-type affixes) are obviously not preferred.
In regard to selectional possibilities, the Spanish preposition de occurs numerous
times in its locative function meaning “from”. This is the meaning cited when the form first
appeared in 1738 (Karttunen & Lockhart, 1976, p. 79). However, de, which does not
occur with a meaning other than locative until 1795 (i.e., de “of”), does appear, though
very infrequently, as an obligatorily selected item in such constructions as antes de “before”
and despuds de “after”, although such an occurrence was not attested at all during the
period covered by the Karttunen & Lockhart work (see subsection 5.3.1, below). In each
of these so-called complex prepositions, de is required; it has no independent meaning.
Table 5.4, below, shows a further breakdown of borrowed form-meaning sets to include
such occurrences.
Table 5.4. The occurrence of Spanish form types based on selection.
Simple tallies that merely record the occurrences of particular form types cannot
make this important distinction. On the one hand, content items (i.e., open classes of N,
V, Adj) are never selected on the basis of language-particular grammatical rule; they are
selected for the meanings they represent Their syntactic class may be a direct reflection of
semantic characteristics (Dixon, 1991) (see, also subsection 3.3. above), but selection of
OPTIONAL (N, V, ADJ, etc.):
OBLIGATORY PREPOSITIONS:
AGGLUTINATING AFFIXES:
815 (6.644)
1 (8)
1(164)
189
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
individual forms is optional. Morphological marking (application of affixes appropriate to
word class) and slots in which they may appear (their distribution) are, nonetheless,
matters of a language-specific grammatical nature. On the other hand, inflectional affixes
are, by definition, obligatorily selected on the basis of language-specific grammatical
requirements. That is, while the selection of a specific form (which represents a specific
category value) may be a consequence of the meaning it expresses, the expression of the
general inflectional category is required. In contrast to both content items and inflectional
affixes, selection of particular function items (word, roots) may be either optional (in the
case of locatives, for example) or obligatory because they are subcategorized for by specific
content items (in the case of so-called grammatical prepositions). For instance, the forms
that obligatorily accompany de are clearly adverbial in nature, expressing imageable
temporal and/or locative relational concepts while de has no independent meaning at all.
The relatively small number of obligatory function words (prepositions) and
inflectional affixes that occur do illustrate two matters of importance: they are borrowable
(because they are borrowed), and they are numerically far from preferred in terms of types
(as opposed to tokens). Recall that fusional affixes are the only form-meaning sets that are
categorically ruled out as a result of morphological incompatibility.
5.2.2. FMICs pertaining to meaning
A number of interesting issues emerge as the analysis of borrowing preferences
turns to semantic characteristics. For the sake of continuity, discussion begins here with
the third FMIC pertaining to meaning, that individual meanings can be associated with a
range of concept types that range from those associated with semantic types and subtypes
190
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
to those associated with inflectional categories. The prediction that follows from this
opposition of meaning types states that meanings which belong to semantic (sub)types will
be preferred in borrowing situations over those that are linked to inflectional categories. In
this respect, the preceding discussion serves to make an important distinction: selectional
requirements are formal requirements, and do not necessarily form comprehensive
correlations with either semantic (sub)types or inflectional categories. This becomes even
more obvious when investigation extends to individual languages that are isolating-
analytical, that possess no obligatorily expressed inflectional categories (and, obviously, no
individual category values). Functional elements (such as the noun classifiers of
Vietnamese) can certainly be obligatory and not inflectional.
The semantic (sub)types discussed in Section 3.3 are direct expressions of
ontological categories and represent the kinds of meanings that every language and every
speaker needs to express. They identify (make reference to) topics of conversation,
activities and states in which persons and objects participate, attributes (e.g., qualities and
quantities) of the people and things in the environment, locations in space and time, and so
on. These concept types are not dependent on any language-particular grammar, although
their formal representation may vary to an extent The kinds of independent meanings
expressed by optionally selected function words are traditionally placed somewhere
between extremes, according to a Sapirian perspective. Here, they are listed as
RELATIONAL types of meaning according to Clark, 199326 (and can, as a consequence, be
26 See Clark (1993. p. 47) for her listing of ontological categories. She states the following: “Children
must also make use of their ontological categories when they create meanings for adjectives that pick out
properties like shape, size, or color (e.g., round, small, red) and for prepositions that mark relations in space
or time (e.g., in, on, near, or above). In Classical Mexicano, locative functions were expressed only via
suffixation; hence, it may be that some inflectional affixes may be viewed as possessing independent
meanings associated with semantic types so characterized. However, due to the fact that such affixes also
express additional, maximally grammaticalized meanings via their maximally grammaticalized forms, exact
correspondences vis-it-vis individual adpositions are rare, if they occur at all. Here, one is dealing with
obviously blurred and possibly shifting boundaries.
191
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
properly included under the heading of semantic type or subtype). In clear contrast with
semantic (sub)types, inflectional concepts involve abstract conceptualizations that have
become obligatorily expressed in a particular language. As noted previously (see Section
4.2), the presence of general inflectional categories and the particular values they express
are language-specific and, as a consequence, highly idiosyncratic (Bybee, 1985).
While the data displayed in Table 5.4 (above) reflects borrowing preferences based
on form, the following, Table 5.5, illustrates types of meaning:
Table 5.5. The occurrence of Spanish concept types.
FMICs regarding concreteness and specificity produce a variety of patterns within
particular noun subclasses. For instance, degrees of concreteness are instrinsically
associated with particular semantic types and subtypes (see Section 3.3 for discussion of
groupings based on semantic characteristics). It is likely that borrowing patterns also have
much to do with the identification of concepts newly introduced by Spanish speakers into
the cultural and linguistic environment and that were either laterally assigned to already
existing semantic subtypes or placed into newly created semantic subclasses.
For the purpose of the present study, borrowed nouns are divided into a number of
different classes and subclasses of CONCRETE or ABSTRACT which represent a synthesis of
the kinds of meanings developed in Dixon (1991) (discussed in subsection 3.3.1 and listed
192
SEMANTIC TYPES:
RELATIONAL MEANINGS:
NO INDEPENDENT MEANINGS:
INFLECTIONAL CONCEPTS
769(3,431)
46 (3,213)
1(8)
K164)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
in Appendix II) and those used by Karttunen & Lockhart (1976)27. They include (a)
CONCRETE ANIMATE (e.g., human2 8 and animal), CONCRETE INANIMATE (including
artifacts, materials, etc.), QUASI-CONCRETE29 (e.g., names of organizations, jurisdictions,
etc.). ABSTRACT terms consist of those types of nouns traditionally considered abstract
(e.g., ideas, actions, and procedures, measurement terms, and so forth) and include
a c t iv it ie s (e.g., results of particular physical acts), SPEECH ACTS (e.g., question,
promise, and so on), and STATES (and PROPERTIES) (e.g, embarrassment, hope, etc.).
Borrowed Spanish nouns are displayed below (in Table 5.6) accordingly.
The counts in the first column represent Spanish form-meaning sets found in the
portion of the Hill & Hill corpus treated here and, therefore, borrowings found in Modem
Mexicano. The second and third columns are presented here for the purpose of
comparison. They contain lists of borrowed items that can be presumed to be available to
Malinche speakers in addition to those recorded in the spoken language of the Hill & Hill
study. The counts in these columns come from two sources. The one on the far right
comes from the word lists of Fray Alonso de Molina in an early Spanish-language grammar
of Mexicano and an accompanying vocabulary (Arte de la lengua mexicana y castellana and
Vocabulario en lengua castellana y mexicana), both published initially in 1571. These
works contain the first wave of Spanish borrowings into the classical language immediately
following the conquest. The middle column, designated Karttunen & Lockhart, includes
27 Their terminology (traditional for Ndhuatl studies) is somewhat more limited in that the kinds of nouns
that needed classification appeared to refer primarily to concepts imported to Mexicano culture (cf. Karttunen
& Lockhart, 1976, p. I6ff). The more specific terms of Dixon (1991) are necessary to encompass the
expanded numbers of borrowed noun types evidenced in the Modem Mexicano text. However, preserving
the spirit of the earlier terminology makes these preliminary comparisons possible.
28 According to Dixon (1991), HUMAN includes subclasses of kinship terms, rank, social group, and so
on. Hence, in the subclassed designated individual under QUASI-CONCRETE, characteristics pertain to such
things as trades and governmental or religious function.
29 This term is taken from Karttunen & Lockhart (1976) to refer to concrete referents that are distinguished
by an abstract quality, for instance names for and titles of religious and governmental officials, professions,
trades, and so on (e.g., terms regarding marital status). These are also considered concrete by Dixon (1991,
p. 76).
193
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
word lists these authors developed from a number of texts dating as early as 1540 and
continuing up to the end of the colonial period, i.e., through the end of the 18th century
(Karttunen & Lockhart, 1976, pp. 52-84) (see, also subsection 5.3.1, below).
Table 5.6. Borrowed Spanish nouns.
Hill & Bill Karttunen & Lockhart Molina (1571)
CONCRETE
Animate: 126 54 62
Inanimate: 78 245 60
Quasi-concrete: 85 228 34
Other _122______________________________
Total Concrete: 418 527 156
ABSTRACT:
General: 77 193 45
States (and properties) 15
Activities: 30
Speech acts: 2Q__________________________________
Total Abstract: 152 193 45
Table 5.6 shows that concreteness is, indeed, a likely factor. In the portion of the
corpus analyzed here, 73.33% of borrowed nouns refer to concrete concepts of one sort or
another; put differendy, the ratio of concrete to abstract nouns is 2.75 to one. With the
caveat that the tallies indicated in the other two columns and the ratios that can be inferred
are drawn from very different sources, their similarities are, nevertheless noteworthy. The
middle column shows that 73.19% of borrowed nouns occurring in the middle period also
had concrete referents, as did 77.61% of the loans appearing in the Molina text. These
percentages appear to have remained fairly constant.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 5.7, below, displays borrowed Spanish noun types according to the semantic
domains in which they occur. Table 5.8 shows preferences based on specificity (entry-
level hyponyms versus hyperonyms) within particular word families30.
Table 5.7. Borrowed Nouns: Semantic Types.
Hill & Hill Karttunen & Lockhart Molina (1571)
CONCRETE ANIMATE
human 30 -0- -0-
rank 14 -0- -0-
kinship terms 33 -0- -0-
body parts 1 -0- -0-
plants & fruits 11 17 47
animals 16 17 8
derived products 15 17 7
diseases 5 3 - 0 -
other 1 -0- -0-
126 54 62
CONCRETE INANIMATE
materials (cloth, dyes, etc.) 5 25 3
artifacts (novel types) 47 162 55
physical complexes/buildings 25 58 2
other 1 - 0 - - 0 -
78 245 60
QUASI-CONCRETE
individuals 67 165 24
organizations/institutions 8 31 2
places 10 8 2
other 122 24 £
215 228 34
ABSTRACT
general 13 -0- -0-
religious 9 44 18
legal 7 55 2
other cultural I 38 1
measurements 47 56 24
states (& properties) 15 -0- -0-
activities 30 -0- -0-
speech acts 30 -0- -0-
152 193 45
30 Tables 5.7-5.9, below, display only counts according to types and not tokens; zeros may indicate either
that specific information was unavailable or that there were zero instances. With respect to the Karttunen &
Lockhart text, information concerning tokens was simply not available. At any length, relative frequency
of individual nouns is not at issue here.
195
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 5.8. Borrowed Nouns according to Specificity: Ratios of hyponyms to
hyperonyms.
CONCRETE ANIMATE
hyperonym hyponyms
human gente(h) "people" 28
persona "person"
rank 14
kinship terms familia(h) "family" 31
pariente “relative”
body parts 1
plants & fruits “frutah" 10
animals 16
derived products 15
diseases 5
other _L
5 121
CONCRETE INANIMATE
materials (cloth, dyes, etc.) 5
artifacts (novel types) clothing "ropa” 45
instrumento “instrument”
physical complexes/buildings 25
other cosa(h) “thing” Q
3 75
QUASI-CONCRETE
individuals 67
organizations/institutions/trades trabajo "work” 6
escuela “school”
places 10
other 129
2 221
ABSTRACT
general idea “idea” 12
religious regelion [sic] "religion” 8
legal 7
other cultural costumbre “custom” 0
measurements lugar “place” 44
dempo(h) "time”
dinero “money”
states (& properties) 15
activities 30
speech acts idioma “language” 28
palahra "worrt" _
8 142
Totals hyperonyms hyponyms
18 552
196
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The FMIC of specificity appears to be quite relevant, as well, manifesting itself in
the hierarchical relationships among loans within particular word families (semantic
subtypes). Of the 570 borrowed noun types occurring in the text, only 18 could be
construed as possible hyperonyms (names for families of terms). That is, 96.84% of the
total number of noun types were hyponyms (nearly 32 hyponyms to one hyperonym).
Clearly, the patterns indicate that specificity is a factor. It may very well be that names for
basic concepts (i.e., members of word families) are necessary in daily conversation for
accuracy and clarity and are, therefore, apprehended and learned more frequendy then
general terms (family names). In this respect, it is also very likely that concreteness and
specificity (as properties that may cluster) can and do work in concert to greatly facilitate
and/or motivate borrowing.
5.3. THE EFFECTS OF BORROWING
The extensive borrowing illustrated above has had significant effects on the nature
of the Mexicano matrix system (see subsection 6.2.4, below). First, the numbers of
Spanish content items that are potentially available to bilingual Mexicano-Spanish speakers
offer significant competition for the compounding and incorporating word formation
processes of the classical language (see, for example, Hill & Hill, 1986, pp. 249-266). As
a probable result of the relatively isolating-analytical character of Spanish and its form-
meaning sets, Mexicano speaker/hearers (as members of a bilingual community) may select
from an almost inexhaustible store of semantically transparent content items. Second, the
numerous borrowed function words create additional lexical alternatives to other,
previously exclusive morphological strategies. Such a drift away from (poiy)synthesis
197
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
towards analysis is certainly not uncommon, especially in situations of shift and
subsequent attrition (see, e.g., Mithun, 1984, 1989; Romaine, 1989, p. 376).
5.3.1. A chronology of borrowing
It is generally accepted that the structural changes resulting from such borrowings
accumulate gradually. In this respect, the hierarchy of (31), above, makes both quantitative
and temporal claims of an implicational nature. Quantitatively, it has proved to be very
revealing concerning the depth of borrowing. The texts used by Karttunen & Lockhart
(1976), whose book attempts to bridge the gap between the classical language (Nahuatl) of
pre-conquest times and the present (represented by the various published articles and book
by such authors as the Hills), provides interesting, though at times admittedly thin, support
for temporal aspects of the hierarchy, as well.
These texts came from the national archive (Archivo General de la Nacion) and
National Museum of Anthropology and include such documents as testaments (wills), land
documentation (grants, investigations, etc.), municipal documentation (minutes from
various meetings and so on), litigation, petitions, and various kinds of correspondence
(including personal letters). As a language sample, the sources obviously differ from the
type considered here (i.e., the spoken language of the visitas), but the tendencies are well
worth noting for a variety of reasons. Karttunen & Lockhart report that nouns were
practically the only borrowings during the 16th century and began appearing in various
texts between 1540 and 1560. They state, “Since loans in almost every imaginable
category put in their appearance by 1550, trends must be expressed as proportions” (p.
16). Generally speaking, larger numbers of Spanish verbs did not begin to occur until the
198
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
late 17th century, at which time a particular morphological strategy for the integration of
Spanish verbs (i.e., infinitive + oa)began to appear (p. 29). Borrowed adjectives are
relatively sparse throughout the entire colonial period.
Appendix I in the Karttunen & Lockhart text lists all of the borrowed items that
were found according to the year of their first occurrence, from the first entry dated c. 1500
to the last in 1795. From the years 1500 to 1550, only nouns were recorded with the
exceptions of several tides (e.g., don, dona), the number fourteen (catorce), and the phrase
word etcetera. In 1550, two words that originated as Spanish adjectives appear (castano
“chestnut” and alazdn “sorrel”), both describing kinds of horses; the word cristiano
“Christian” does not appear until 1560. One verb (apelar “to appeal”—in the Spanish
infinitive form) appears in 1553, followed by another (agostar “to pasture catde on stubble
in the summer”) in 1562. Clusters of verbs do not begin appearing until the beginning of
the 18th century (c. 1700), and multiple occurrences of adjectives (more than two) do not
occur at all during this span of nearly 300 years. The first so-called particles, the
prepositions de “from” and a “to”, occur in 1738—more than two hundred years after the
conquest (1519). Note, too, that the only entries recorded in the final year of the sampling
were function words {de “o f’, mientras “while”, and pero “but”). (The authors are careful
to note that words cited were in all likelihood in use prior to their first occurrences.)
At first glance, nouns are obviously the first and most frequently borrowed.
Significant numbers of verbs came much later, with adjectives accumulating gradually over
the entire period. While independent function words come relatively late, they apparently
come into the language in numbers. Proportions of borrowed items are represented in
Table 5.9, below, in terms of the dates of individual works and collections of documents in
which they are first recorded, beginning with the work of Molina and followed by that of
199
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Pedro de Arenas (Vocabulario manual de las lenguas castellana y mexicana), a teaching text
designed for Spaniards living in or visiting Mexico who wanted to become acquainted with
Ndhuad, dated 1611. (Lists from the Molina volumes and that of Arenas are included in
the Karttunen & Lockhart (1976) volume.) The two rows labelled Texts represent the
additional collection of documents investigated by Karttunen & Lockhart (1976, p. 16f)
and described above (in the first paragraph of this subsection); Texts (pre-1650) include
those documents dated prior to that cutoff, and Texts (post-1650) were those dated 1650
and later
Table 5.9. Spanish content items according to Karttunen & Lockhart (1976).
N OUNS V ERBS ADJECTIVES PARTICLES
Molina (1571) 201 I 0 0
Arenas (1611) 57 0 0 0
Texts (pre-1650) 496 9 7 0
Texts (post-1650) 224 24
6 IQ
Totals 720 33 13 10
As a consequence, the temporal nature of the proposed hierarchy appears to be substantially
confirmed, as well—with one noteworthy exception.
The Spanish nominal plural marker -s seems to have been borrowed very early on,
appearing in the earliest documents. Until quite recently, it only occurred on Spanish
loans. If this were a general tendency (that affixes are borrowed at any time in the
process), it would present a fairly serious challenge to the predictions made in the present
work. However, recall that the hierarchies merely indicate preferences and not absolute
barriers; the PSI has been completely accurate in that regard. It could also be true (yet
200
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
unattested in texts investigated by Karttunen & Lockhart) that types of content items and
function words were either borrowed prior to or simultaneously with the borrowed affix,
therefore preserving the validity of the temporal claims implicit in the hierarchy (see
subsection 2.3.4 in reference to the timing of borrowed elements). Nevertheless, the
occasional exception does require explanation. Moreover, the temptation to resort to
glibness or to recklessly respond in ad hoc fashion needs to be avoided.
Consequently, the following is offered as a reasonable supposition (with the
obvious caveats). First, the Spanish -s is, after all, a borrowable morpheme type and the
numerical predictions hold, nonetheless (though this is less than explanatory). Second, in
view of the fact that Mexicano speakers already had distinctions of number, applying the
Spanish plural marker to borrowed Spanish nouns merely matches a relatively transparent,
agglutinating-type affix (with one-to-one correspondence of meaning and form) of a type
already known, which expresses a category value also already known, to a form
recognizable as coming from the same, foreign lexical source (Spanish). An example of
this very same phenomenon comes from English and illustrates the apparent ease with
which a plural marker can be borrowed along with a noun form from a typologically and
genetically distinct language31: Some speakers of English have borrowed both the Modem
Israeli Hebrew word kibbutz and its appropriate plural marker, -im. Hence, both kibbutz,
as the singular form, and kibbutzim as the proper plural form, are now considered (at least
by some) to be English words. English already possesses the inflectional category of
number and the particular category value of plural (masculine gender indicated in the
particular affixal form is totally irrelevant to the speaker of English). In addition, English
does not apply the borrowed affix to words of non-Hebrew origin, which is also true
31 A frequently cited example is phenomenon/-a from Greek. For the Modem Israeli Hebrew example, see
Webster’s New World Dictionary of American English (circa 1978), where it is the only plural listed.
201
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
regarding the treatment (i.e., non-application or complete absence) of Spanish -s on native
Mexicano forms—with extremely rare exceptions. Ultimately, explanation most likely
resides in the particular bilingual/language contact situation, itself.
According to the chronology discussed above, the increased number of borrowed
function items coincides with (perhaps, narrowly predating) a number of significant social
changes. In considerable evidence during the first three centuries of colonization was the
obvious social stratification according to ethnic and language status (i.e., European
colonists and their descendants, or criollos, versus colonized “Indians”). The likely
linguistic correlate, that language maintenance was largely responsible for the resistance of
Mexicano to much structural borrowing during that time, is indicated by the careful
documentation provided by the Karttunen & Lockhart text. However, many authors note
such cultural changes as (a) the spread of Spanish throughout Mexico and increased
bilingualism during longer periods after the conquest (e.g., Karttunen & Lockhart, 1976,
p. 50), (b) the rise of a national identity that was not necessarily linked to ethnicity or
language during the 19th century (e.g., Wright, 1992, Berdan, 1982, Vigil, 1980), and (c)
the rapidly accelerating emergence of a people of mixed race (i.e., mestizos) and the mixing
of Spanish and indigenous cultures, also in the 19th century (see, for example, Meyer &
Sherman, 1987, p. 218; Vigil, 1980, p. 11).
In addition, Hill & Hill (1986, p. 104ff.) discuss the gradual decline of literate
forms (and formal registers) of Mexicano and the encroachment of Spanish into an
increasing number of social domains beginning in the early parts of the 19th century, as
well. They discuss these changes in terms of the social differentiation of “inside” (or in
group) versus “outside” language—terms parallel to Gumperz’s (1982) distinctions of
“we” and “they” codes—as Spanish was adopted by growing numbers of indigenous,
202
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Mexicano-speaking people. As a result, regional forms of Spanish have become
increasingly capable of expressing social and/or ethnic status, as well. These types of
socio-cultural changes surely suggest that significant language mixing has become
increasingly prevalent in more recent times, bringing into question the widespread
assumption that the accumulated changes in Modem Mexicano were, in fact, gradual—at
least, as gradual as believed. It can be easily inferred that the many changes wrought by
various kinds of borrowing have, indeed, accelerated at different times in its history.
5.3.2. Phrasal and clausal organization
The numbers (tokens) and types of borrowed Spanish function words suggest that
borrowing has had far-reaching effects and that these effects provide ample evidence of
significant structural changes. For example, borrowed prepositions have led to the
emergence of adpositional phrases patterned after Spanish constructions that compete with
applicative, purposive, and locative markers (morphological strategies). Hill & Hill (1986,
p. 247f) state that Mexicano does have a small set of native particles which functioned like
prepositions, but that received possessive prefixes (e.g., Ipan, Itech, Itzintlari), as in the
following examples in (32):
(32)
I-pan in cama on the bed
it-on D EF bed
i-tech in no-chan in my house
it-in DEF my-house
i-tzintlan in cama under the bed
it-under D E F bed
203
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The authors also note: “Hawkins’s (1980) proposals for word-order universals
specify that a language such as Malinche Mexicano should exhibit prepositions” (p. 147).
That is, a language with the co-occurrence patterns of the classical language should actually
have prepositions (i.e., according to other word-order characteristics associated with VSO
languages) and not postpositions as often suggested in the literature. However, the bound
suffixes on nouns that express locative distinctions cannot be called postpositions (see
5.1.1., above). Moreover, the fact that the so-called particles above are prefixed for
possession does not necessarily rule out their status as function items (independent word,
root). It seems much more reasonable to say that the locative suffixes that are apparently
quite rare in the modem language (which were, however, productive in Classical Nahuatl)
have essentially been replaced, and that their replacement is consistent with the
implicational nature of Hawkins (1980) and his later work (e.g., Hawkins, 1983, 1991)32.
In this case, it can be said that Spanish prepositions joined an already existing subclass of
function items (with independent, relational meaning) and that their borrowing and
subsequent effects may have been facilitated by the existence of these particles.
Additional word order changes are also reported: from Classical Mexicano Adj N
to N Adj according to the Spanish co-occurrence patterns and from Gen N to N Gen with
the addition of de to indicated possessor (Hill & Hill, 1986, pp. 237-241). These also
accord with Hawkins’ predictions. Change, whether gradual or acute, proceeds in an
orderly, therefore, predictable fashion.
As noted in subsection 5.1.1, above, other form-meaning sets that serve to mark
phrasal and/or clausal boundaries have been integrated into the Mexicano matrix, for
instance, various coordinating conjunctions (e.g., o “or” and the very frequent pero “but”),
32 See also Field, 1994b for a discussion of implicational universals and the origins of mixed languages.
204
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
subordinating conjunctions (e.g. como “how”, cuando “when”, and porque “because”),
and the complementizer que “that”. These are further indications that lexical alternatives
have replaced strategies that were once strictly morphological prior to the advent of those
borrowed free-standing form-meaning sets. Taken along with the massive borrowings of
individual form-meaning sets into classes of Noun, Verb, and Adjective, it is certainly not
surprising that Mexicano has drifted towards analysis to its present position on the index of
synthesis.
5.4. DISCUSSION: THE R O LES OF FORM AND M E A N IN G IN BORROWING
The patterns of borrowing that develop within any bilingual/contact situation will
reflect the social conditions of the communities in contact and the resultant motivations of
bilingual speaker/hearers of varying proficiency levels for linguistic borrowing. On the
community level, particular form-meaning sets from a culturally dominant language may be
borrowed into a recessive one to expand its referential capacity within specific semantic
domains (e.g., those involving entities/concepts newly introduced to the community), to
compensate for the loss of native forms in situations of advancing shift (as a result of
attrition), and/or for social affect (Grosjean, 1982). Individual, form-meaning sets from a
culturally, hence, socially, economically, and institutionally dominant language variety are
most often borrowed because the kinds of referents or concepts they express are an integral
part of the linguistic and social milieu, and knowledge of their associated labels are required
for successful reference to be made. It is also true that meanings can be borrowed without
donor labels in the case of caiques and semantic extensions. Conversely, donor labels
(without their associated meanings) may be borrowed as a result of the perceived
attractiveness of the forms, for example, for social affect when the prestige associated with
205
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the donor variety is relatively high. For example, in processes of relexification, a donor
form replaces one from the recipient and presumably acquires the meaning of the
corresponding recipient form-meaning set. In that event, form may at times be more
important than a particular meaning, though both are intrinsically linked. Nevertheless, the
role of meaning in borrowing is essential, particularly in view of the fact that the occurrence
of loanwords pattern in such predictable ways.
In the case of Mexicano, initial patterns of borrowing reflected the conquered and
colonized status of the community itself. Terms were borrowed en masse to represent
concepts that were brought into their world having to do with the imposition of social
practices (e.g. governmental, military, social, and legal organization along
Spanish/European lines), religious customs (according to the spread of the Roman Catholic
form of Christianity), occupational (including such aspects of business as
Spanish/European monetary principles, implementation and methods of farming, and so
on), and many other cultural accoutrements (e.g., clothing standards in various official and
unofficial areas of life). From this perspective, there certainly seems to be some substance
to the intuitions and observations of Karttunen & Lockhart (1976), who write: “We believe
that there is distinct, significant diachronic and topical patterning in the Nahuatl
incorporation of noun loans” (p. 16). Most studies of similar bilingual/contact situations
corroborate the premise that meaning is central, that the principal motivating factor for
borrowing of such depth is the desire to express linguistically the concepts that are present
physically in the social environment It is also evident that initial limitations of a
particularly linguistic nature relate to the matching of form and meaning in the respective
participant languages.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5.4.1 Issues of form
The present work has proceeded along two distinct lines: according to form and
according to meaning. The influence of form on borrowing in the case of Mexicano has
served to illustrate that form restricts what is or is not borrowable in an absolute sense.
With respect to form-meaning sets that are not borrowed, form is, perhaps, the only
criterion necessary. Fusional-type affixes of Spanish have not been borrowed irrespective
of the meanings they might express, as correctly stipulated in the PSI. Issues of form also
provide links to formal aspects of grammaticalization clines in support of the long-reported
intuitions of scholars that the more “structural” an element is (i.e., grammaticalized on
scales of grammaticalization viewed synchronically), (a) the longer it will take to learn, (b)
the less likely it is to be borrowed, and (c) the more difficult it will be to find a place in a
language that already possesses its own full store of grammaticalized elements. In
addition, form apparently interacts with function in significant, patterned ways. Among
content items—which are, in principle, compatible with and borrowable by any
recipient/matrix system—nouns are favored over all other types. Hence, grammatical
aspects regarding the semantic and syntactic complexity of individual semantic classes and
subclasses are pertinent. It has been concluded that the particular characteristics associated
with classes of nouns and verbs (and adjectives, when applicable) appear to hold across
language boundaries, irrespective of morphological type (i.e., with respect to position
along indices of synthesis and fusion).
Mexicano has borrowed content items of all types and classes from Spanish despite
the fact that inflectional morphology that applies obligatorily to each general class is not
borrowed. Numerous independent function words are borrowed, as well, demonstrating
that all forms are borrowable that conform to the recipient/matrix system’s morphological
207
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
structure. However, as strongly indicated in Mexicano, the borrowing of structural
elements such as adpositions, conjunctions, auxiliary verbs, and so on will have a
structural effect. Their adoption, integration, and usage indicate that the organizational
patterns of Spanish (which is considerably more analytic-isolating than Mexicano) is being
interwoven with those of Mexicano and its predecessor(s), a fact that reveals that the
morphosyntactic matrix is undergoing changes far beyond the effects of garden-variety
borrowing (e.g., the kind that merely fills lexical gaps). If the transmission of Mexicano
had proceeded along normal genetic lines, one would expect that traditional syntactic
strategies would suffice, and, therefore, would be maintained. Native syntax is subject to
normal processes of gradual language change; the structural changes that are evident as a
result of contact with Spanish are obviously distinct. Evidence of a rapidly evolving
morphosyntactic matrix system immediately following the colonial period lends significant
support to the claim that the resultant linguistic system constitutes a mixed language (i.e.,
defined on a priori grounds), and that grammatical aspects of two languages are being/have
been intertwined.
Clusters of properties associated with form identifiability are also evidenced in
Mexicano borrowing patterns (i.e., rankings according to borrowing preferences). Those
forms that clearly show optimal amounts of phonetic information, one-to-one form-
meaning mappings, and optional selection (content words/roots and certain independent
function words/roots) are clearly preferred, while those that have minimal to zero form,
one-to-many (and many-to-one) mapping possibilities, and obligatory selection
requirements (specifically fusional affixes) are not borrowed at all. Various other
clustering possibilities conform to the preference predictions, as well. For instance, items
that have one-to-one mapping possibilities are borrowed despite the fact that they (a) have
minimal phonetic form and (b) are selected obligatorily (e.g., obligatory/grammatical
208
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
preposition de and the aggludnating-type plural noun affix -5); however, they are borrowed
to a much lesser degree (in terms of types) than those possessing all three characteristics
associated with ease of identification (e.g., content items). Typically, minimal-to-zero
shape clusters with obligatory selection, and, as a consequence, can be associated with
increased grammaticalization. It appears that specific aspects of form play a role with
respect to borrowability as well as identifiability; it is likely that these correlations also
reflect the general tendency of all languages to display certain form-meaning sets in very
predictable fashion, for the same reasons, and with the same results in acquisition (some
are easier to learn) and borrowing (some are easier to borrow).
5.4.2. Issues of meaning
While characteristics of form account for many of the borrowing patterns in the
Mexicano data, the role that meaning plays fills out the picture in much greater detail and
indicates reasons for a marbling effect in the lexicon. Veins of Spanish content items
project into specific semantic subclasses representing new concepts. In fact, entire word
families composed of borrowed contentive roots form coherent strands that weave their
way through the entire Mexicano lexicon. It is very apparent that these form-meaning sets
have diffused throughout the entire Mexicano speaking community. Hence, oppositions
based on meaning may illustrate even more than general borrowability and/or overall
borrowing preferences.
The strong reference for concrete over abstract nouns suggests concrete (visible and
imageable) concepts are the most easily borrowed, which is supported by work indicating
that they are also the fastest in translation. It can be inferred, therefore, that the semantic
209
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
transparency associated with such items has its base in physical, perceptual reality. This is
especially evident in the earliest borrowings noted by Molina and others, extending into
current forms of the language (see Table 5.6, above). It is a very small leap to the simple
conclusion that concepts formed on the basis of physical, perceptual attributes are the
easiest to identify, and, as a consequence, the easiest to label—in all likelihood, this applies
to any type of acquisition. Regarding the borrowability of classes of Noun, it seems that,
once a concrete object is labelled, that label can be used by anyone wishing to make
reference to that object33.
Concepts based on specificity deal with the hierarchical organization of particular
semantic classes and subclasses in terms of hyperonyms and hyponyms. While it cannot
be stated with absolute certainty that the relatively great amounts of semantic information
provided by maximally specific hyponyms make them the most easily identifiable, it
remains valid to say that entry-level sorts of terms are among the most useful, probably the
most frequently used, and, hence, first learned (e.g., in native or second language
acquisition) in any particular semantic domain. In addition, by virtue of their specificity
and possible links to concreteness (two properties that typically cluster together), concepts
that are both concrete and specific ensure very high degrees of (a) imageability and (b)
uniqueness of meaning (a unique conceptual/perceptual address to correspond with a
unique phonetic label). The converse, form-meaning sets associated with abstract and
general concepts invoke perceptual requirements beyond the visual and tactile. They often
require demonstration, explanation, and/or relative amounts of cultural knowledge in
translation, which seem somewhat beyond the scope of particular entry-level terms in any
language.
33 In this regard, it would be interesting to find out how many languages have borrowed the form pizza
from Italian as a label for that uniquely Italian food.
210
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In the Mexicano context, abstract and general terms were necessary in dealing with
Spanish colonial persons and institutions; nevertheless, they have been borrowed with less
frequency because they require more effort on the part of speakers of both donor and
recipient In religious instruction, for example, entirely abstract principles were
consciously taught through sermons and catechism books (e.g., concepts of spiritual
(religious), cultural, and family organization). Mexicano speakers were taught labels
representing varying degrees of importance for governmental (king, viceroy, etc.),
religious (pope, cardinal, bishop, priest), and other kinds of social hierarchies (boss/chief
and so on). Anecdotally, this is not merely viewing a horse and learning the label caballo.
It is certainly no wonder, given this type of cultural context, that borrowing patterns are
“topical” in the sense suggested by Karttunen & Lockhart.
Preferences for meanings based on semantic types versus those linked to language-
particular inflectional categories show the clearest semantic correspondence with scales
according to grammaticalization. At first glance, it seems very plausible to say that the
more language-specific a concept is, the less likely it is to be borrowed. However, this
points, once again, to what appears to be the inescapable conclusion that individual
concepts may be language-general—that is, completely free of dependency on specific
language requirements, and, therefore, universal in some sense (i.e., as types). For
instance, the kinds of meanings associated with content items are (by virtue of form
characteristics) borrowable by any other language irrespective of morphological typology.
Certain inflectional concepts (and aggregates thereof) may be gradiently borrowable based
on similar semantic criteria. In general, form-meaning identification characteristics of
meaning may prove to be even more significant than those pertaining to form.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6 .0 . C h a p t e r 6— d is c u s s io n
The basic nature of the proposed model is two-fold. First, the claim that anything
leamable should be borrowable (i.e., from an individual speaker/hearer’s perspective) has
provided an intriguing starting point Intuitively, it seems that if a speaker of X can
leam/acquire relatively full proficiency in Y, then he/she should be able to enlist its
resources and/or mix those resources with those of X. However, in every bilingual/contact
situation studied so far, there are strong borrowing preferences for certain types of form-
meaning sets, and these preferences show systematic tendencies. It has become apparent
that the similarities found in various borrowing hierarchies and grammaticalization clines
(viewed synchro nically) are grounded in the ways individual languages construct words
and other types of form-meaning sets that are relevant to both processes. Moreover, there
appears to be an absolute cutoff point in specific contact situations past which certain
specific form-meanings sets cannot be borrowed; that is, there is a basic systematic
constraint on potentially borrowable forms based on the morphological structuring of the
recipient language relative to that of the donor that goes beyond the preferences depicted in
various borrowing hierarchies.
Therefore, the current formulations of the Principle of System Compatibility (PSQ
and its corollary, the Principle of System Incompatibility (PSI), capture the basic intuition
that restrictions on borrowing are not necessarily linked to the linguistic abilities of
individual speakers, but to the inherent characteristics of the linguistic systems themselves.
Certain specifiable form-meaning sets may be blocked from borrowing because they do not
conform to the morphological possibilities of the recipient with regard to morpheme types.
In the specific case of Mexicano, the PSC predicts that all content items (words, roots),
function words, and agglutinating-type affixes can be borrowed. And, many form-
212
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
meaning sets from each of these classes are, indeed borrowed. The only form-meaning
types ruled out by the PSI are fusional-type affixes; this prediction is satisfied, as well.
Second, beyond the cutoff point identified by the PSI, formal characteristics
apparendy cease to be relevant in an absolute sense, in that many borrowable forms are not
borrowed, and those that are borrowed pattern in ways not necessarily predicted by form
alone. For instance, thousands of perfecdy acceptable (i.e., system compatible) Spanish
nouns are not borrowed into Mexicano. At first glance, one might hazard to guess that this
is for reasons that are primarily social, and this turns out to be quite feasible. Sets of social
circumstances will lead to the exposure of an individual speaker/hearer to certain registers
of speech, occupational nomenclatures, and so on. Those involved in maritime industries
will be exposed to nautical terms, and those involved in agriculture will leam terms
associated with farming implements, cultivation methods, business practices, names for
agricultural products, and so forth. Thus, as a direct result of the physical characteristics
and perceptual accompaniments of a socially and/or economically constructed environment,
individuals, who compose social and linguistic networks, are exposed to terms associated
with certain semantic domains.
6.0.1. The relevance of form-meaning interpretation characteristics (FMICs)
Regarding the three form-meaning interpretation characteristics (FMICs) pertaining
to form (see subsection 4.1), the first, sufficiently detailed phonetic shape (see subsection
4.1.1), expresses the requirement that there is sufficient phonological substance present to
make correct associations with appropriate concepts, and that borrowing preferences will
favor forms with optimal (and segmentable) phonetic shape over those with minimal or no
213
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
form. This is realized in the clear preference for content items (words, roots) over other
more grammaticalized forms. In addition, the Spanish preposition a (“to”), though
occurring in the text relatively infrequently, and the nominal plural suffix -s are borrowed;
they each apparently possess sufficient phonetic form for identification and to indicate that a
concept is expressed1 . The second characteristic, which predicts a strong preference for
forms with direct, one-to-one mappings of meaning and form is also fulfilled as evidenced
by the total exclusion of Spanish fusional-type affixes. The third, based on selectional
restrictions, is likewise satisfied with the anticipated result that very few obligatory forms
(i.e., donor inflectional affixes or obligatory function words) are borrowed relative to the
total number of borrowed content items and independent function words (e.g., adpositions,
conjunctions, etc.).
FMICs of meaning are also highly relevant to the patterning of borrowed items (see
section 4.2, above). Overall, the data reflect clear borrowing preferences for concepts that
are concrete versus abstract, which can be interpreted as a preference for form-meaning sets
that represent concepts easily identified on a perceptual (hence, visible, tangible) basis.
However, the actual patterning corresponds also with particular semantic domains. That is,
whether or not abstract form-meaning sets are borrowed at all may depend more directly on
such factors as the occupational, institutional, and, perhaps, ontological domain in which
they occur across semantic domains from a lateral, rather than hierarchical, perspective.
For instance, with respect to religious terminology (which was consciously and deliberately
taught as opposed to observed or handled in a physical environment), form-meaning sets
referring to people and offices seem to be freely borrowed, along with numerous terms
referring to religious (hence, abstract) concepts and processes. Within semantic groupings
1 While this is certainly consistent with observations concerning segmentability, it may call into some
question claims of syllabicity (see, for example, Thomason & Kaufman, 1988, p. 56). However, each is
surely pronounceable in isolation.
214
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
referring to animal and/or plant life, native forms have been preserved (and even integrated
into regional varieties of Spanish), with the exception of a number of terms indicating
species newly introduced or foreign to the culture (additions) and subsequendy learned in a
variety of settings, in the work place, school, and so on. One conclusion that can be drawn
from this is that concreteness undoubtedly facilitates recognition and, therefore, enhances
borrowability, but the actual motivations for borrowing are more likely to be found in areas
that are primarily social, for instance, in human responses to changing social, cultural, and
consequent linguistic circumstances.
This leads to preferences according to specificity. Predictions assumed to be based
on processing ease or leamability associated with specific versus general types of meaning
appear to require cognizance of semantic domains, as well, but this time hierarchically
within specific semantic types and subtypes. Perhaps partly as a result of the fact that
many borrowed items are additions, hyponyms significantly outnumber hyperonyms in
terms of types and tokens; the few hyperonyms that do actually occur correlate somewhat
with entire word families brought to the Mexicano culture via the colonial practices of the
Spaniards. One should anticipate widespread cultural and linguistic bias towards Mexicano
core vocabulary items (where relatively few Spanish borrowings are found). This
attitudinal factor, even though it may be difficult to measure, is clearly present and likely
functions to maintain ethnic and linguistic identity.
The third FMIC pertaining to meaning directly relates to the expected borrowing
preference for kinds of meaning that can be linked to semantic (sub)types, as opposed to
those associated with inflectional categories. The data overwhelmingly support this
prediction. However, it should be noted that this particular characteristic, which involves
oppositions of meaning types, is not coterminous with formal selectional possibilities,
215
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
particularly in view of the fact that numerous forms are obligatorily selected that arc not
markers of inflectional categories. Nevertheless, only one Spanish aggludnating-type
inflectional affix is borrowed, and, while numerous Spanish adpositions are borrowed,
those that are obligatory as a consequence of language-particular syntactic requirements
(e.g., as a result of subcategorization frames of particular content items) are quite rare—in
fact, restricted to the infrequent occurrence of obligatory de 2 (see subsection 5.2.1,
above).
6.0.2. The organization of this chapter
The remaining portions of this concluding chapter include a number of
generalizations that can be made based on the various borrowing patterns in Modem
Mexicano with a look at other language contact phenomena that seem to be related to
borrowing. The departure point in Section 6.1 is the extension of the Hierarchy of
Borrowability, the PSI, and predicted preferences based on FMICs to the entire range of
bilingual/language contact situations and the possibilities that present themselves in regard
to morphological typology. Subsection 6.1.1 treats the types of patterns expected to occur
in contact situations involving languages that are typologically similar; 6.1.2 and 6.1.3 deal
with languages which differ along the indices of fusion and synthesis, respectively. 6.1.4
takes a brief look at the types of analyses that follow from the PSI if applied to apparent
exceptions, that is, to situations that appear to exhibit violations of the PSI.
2 See Jake & Myers-Scotton (1998) for a current and much refined version of the MLF Hypothesis in
which similar distinctions of morpheme types are made on the basis of whether or not they are conceptually
activated, which is similar to distinctions made here in terms of semantic types, and their participation in
Theta-role assignment, which coincides in many respects with formal selectional possibilities. This new
work from a psycholinguistic perspective suggests that comparisons with the approach followed here may
yield many more parallels.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In Section 6.2, discussion proceeds to the role of the most grammaticalized types of
meaning in borrowing and their relevance to the general borrowability of inflectional
categories (i.e., as opposed to individual category values). While some researchers claim
that inflectional categories are never borrowed, most note (in apparent contradiction) that
individual inflectional markers are indeed adopted in specific situations characterized by
intense language contact Some of the conflict over this particular issue may very well be
terminological in nature. However, by unravelling some of the confusion of terms and
briefly discussing the nature of inflectional categories, in general (in 6.2.1 and 6.2.1), a
significant generalization is made: categories as such are not borrowed, but individual
category values (or exponents) may be borrowed under certain specified conditions. This
generalization is tentatively stated in the Principle of Categorial Compatibility (PCC) and its
correlate, the Principle of Categorial Incompatibility (PCI) in subsection 6.2.3, which are
patterned along the lines of the PSI and its correlate, the PSC (see Section 2.3, above).
Next, in 6.3, discussion moves on to some of the connections that can be
established among a number of different bilingual/language contact phenomena, for
instance, among various kinds of code-switching and borrowing (6.3.1). This is done
especially with respect to what these seemingly disparate processes may have in common
and what they may reveal about lexical representation in the bilingual lexicon.
Convergence and the possibility of the emergence of a composite language matrix are
discussed in 6.3.2 in a context of externally motivated language change. Finally, a few
measured conclusions are offered in 6.4 concerning the light that borrowing and other
contact phenomena can shed on the language faculty itself, and what the programmatic
nature of this particular work may suggest as fruitful areas of future research.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6. l . T h e PSI, FMICs, a n d Ot h e r C o n t a c t S it u a t io n s
While the constraints proposed in the PSI and the preferences outlined according to
FMICs successfully account for the borrowing patterns of Modem Mexicano, a number of
other patterns are expected to follow in other contact situations. As in the case of
Mexicano, the first step in the assessment of borrowability involves the morphological
typology of the participating languages relative to each other (see Section 2.3, above). As
an opening move, this typological approach has the ability to deal with the complete range
of possibilities and, as a consequence, to make immediate sense of the apparent chaos in
various contact situations. A sweeping claim that inflectional affixes are never borrowed
and are, therefore, not borrowable is more than an overgeneralization; it is empirically
false. Sometimes inflectional affixes are borrowed, and sometimes they are not. A
comprehensive model of borrowing and borrowability needs to account for such inherent
variability. The second step is to assess the sorts of preferences one expects based on
FMICs, which assumes that items in a language can be classed and placed on a scale of
borrowability (according to gradations of form and meaning), from prototypical content
items (words or roots) to function words/roots, and, in the event that one or both languages
in contact are synthetic, on to inflectional affixes3.
It seems that a system of (universal) constraints is not possible, that is, unless it can
be adapted to (or specified in) particular contexts. The PSI represents precisely that. At
times, clear constraints are needed to account for the absolute unborrowability of specific
form-meaning sets (e.g., particular types of inflectional affixes in certain contexts). At
other times, however, typological limits simply do not hold. In such situations, FMICs
3 Blurred boundaries among classes (e.g., among certain classes of adverbs, prepositions, and
conjunctions), items with multiple or overlapping membership, and so on, are similarly anticipated.
218
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
predict preferences (only)—recall that just because an item is borrowable in principle does
not mean that it will be borrowed in fact. Particular FMICs such as concreteness and
specificity also point to the likelihood that individual and community-wide social patterns
may result in clusters of loanwords within specific semantic domains. (Knowledge of
such patterns may allow us, nonetheless, to predict fairly accurately in which semantic
domains loanwords are most likely to occur.)
The typological barriers indicated by the PSI pertain specifically to situations in
which morphological structuring differs significantly in a donor and recipient along either
index of synthesis or fusion (see sub-section 2.1.1). For instance, when the recipient is (a)
isolating-analytical, it cannot borrow inflectional affixes of any kind (neither agglutinating-
nor fusional-types) from a synthetic language; when the recipient is (b) an agglutinating-
type language (and, hence, is positioned relatively high along the index of synthesis), it
cannot borrow fusional-type inflectional affixes from a language positioned higher on the
(other) index of fusion. These barriers are one-way; typological constraints are irrelevant if
the patterns are the reverse (i.e., when a synthetic language is the recipient and the donor is
isolating-analytical). In the event that participant languages are similar with respect to
morphological structuring, no absolute limits hold. Borrowing patterns, then, will be the
consequence of language-specific characteristics, not absolute (universal) constraints.
6.1.1. Typologically similar languages
When participant varieties are typologically similar (i.e., positioned similarly along
indices of synthesis and/or fusion), one anticipates the relatively unencumbered flow of
form-meaning sets from one to the other, at least without typological constraints. This may
219
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
occur in borderland areas where languages in contact are also genetically related (see 6.2.3,
below). Borrowing patterns will, nevertheless, demonstrate preferences according to
FMICs, which will reflect degrees of grammaticalization (for instance, content items will be
preferred over function words, and so on) and such factors as semantic and/or syntactic
complexity4 (e.g., nouns will be preferred over verbs). Consequently, in terms of types,
nouns will be preferred over verbs and/or adjectives, which will be preferred over function
words and inflectional affixes of any type (when relevant), in that order. In addition,
regarding subclasses of Noun, which are borrowable in any bilingual contact situation
irrespective of morphological typology, concreteness and specificity are always relevant,
accounting for the distribution of borrowed form-meaning sets laterally, into semantic
(sub)classes, and vertically, according to the proven strong preference for hyponyms over
hyperonyms.
To illustrate the ease with which typologically similar languages appear to be able to
borrow from each other, Heath (1981) discusses situations among Australian languages, all
positioned high on the index of synthesis but low on that of fusion. They apparently have
borrowed agglutinating-type affixes quite freely even though some of the varieties involved
are not genetically related. Large numbers of borrowed content items (types) are also
found; one expects that they would pattern along the lines established by FMICs (and
degrees of grammaticalization) and according to semantic and syntactic complexity (nouns
being the most frequently borrowed content items). Another well-known situation
involving fusional-type languages occurs in bordering areas of northern Uruguay and
southern Brazil, where varieties (putative dialects) have arisen (so-called “ffonterizo”) in
which grammatical elements from both Spanish and Portuguese occur in such high
4 See subsection 4.3., above.
220
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
numbers that classification is difficult: Are they dialects of Spanish or dialects of
Portuguese (see Hensey, 1975 and 1993)—perhaps both or neither?
With respect to isolating-analytical languages, barriers imposed by morphological
typology are simply not there. As a consequence, borrowing patterns will primarily reflect
semantic and syntactic complexity and the various FMICs—clearly manifest in the massive
borrowing from Chinese into Thai, which, in past studies, has led to proposals (later
rejected) of possible genetic relationships among these typologically similar languages (cf.
Hock & Joseph, 1996, p. 478; W. Lehmann, 1992, p. 86). In fact, the establishment of
genetic relationships among such typologically similar languages that are also in close areal
contact may be quite problematic. Comparisons among Indo-European languages based on
inflectional morphology apparendy have facilitated the organization of these languages into
(handy) family trees; however, the establishment of genetic relationships among isolating-
analytical languages simply cannot be based on this sort of method. What appears to be left
are comparisons of other grammatical characteristics (e.g., noun classifier systems, tone
systems, and so on) and those based on core vocabulary (cf. Wang & Lien, 1993).
6.1.2. Along the index of fusion: agglutinating versus fusional types
Perhaps, the more intriguing cases of borrowing involve languages that clearly
differ in morphological typology. In the event that the typological asymmetry were the
reverse of the situation in Mexicano-Spanish—and the dominant, hence, donor variety
were an agglutinating-type language and the recipient were fusional—no typological
limitations hold. However, patterns may indeed differ in a number of respects. First,
donor inflectional paradigms (associated with fusional-type inflectional morphology) are
221
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
not at issue; fusional-type affixes are simply absent as candidates and their borrowability is
moot. Consequently, the entire repertoire of the most highly grammaticalized donor form-
meaning sets would be available for borrowing, particularly on the grounds of sufficient
(segmentable) phonetic shape and (one-to-one) mapping possibilities. Second, key
inhibiting factors would, therefore, turn on selectional possibilities (optionally selected sets
being preferred over those that are obligatory) and links to inflectional categories (members
of semantic types clearly preferred over those expressing inflectional category values).
Semantic and/or syntactic complexity will play likely roles with respect to the particular
patterning of content items, as will such characteristics as concreteness and specificity in
subclasses of Noun. Unlike the PSI, which establishes inviolable typological barriers in
particular contexts, all of these potential factors merely predict preferences and not absolute
cutoff points. Apparent exceptions to the preferences predicted, therefore, do not constitute
typological anomalies of any kind. On the contrary, exceptions can be expected to occur,
(but) rarely, and possibly due to extralinguistic factors (see 6.1.4, below).
6.1.3. Along the index of synthesis: isolating-analytical versus synthetic types
Situations in which isolating-analytical languages come in contact with those higher
on the index of synthesis, either agglutinating or fusional-types, will show the same kinds
of asymmetrical patterning as those regarding agglutinating versus fusional language types,
with typological restrictions applying to the participant language that is isolating-analytical
and not to the more synthetic variety. For instance, in western China, Mongolian
languages (e.g., Baonan) have been in long-standing contact with Chinese and have
borrowed extensively with little structural effect at all (Thomason & Kaufman, 1988, p.
90). Japanese, which at various points in its history has been under heavy cultural and
222
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
linguistic pressure from Chinese (Le., via highly literate genres), has borrowed thousands
of Chinese words with relatively minor resultant structural changes (Miller, 1967), none
affecting morphological typology, however. As stated in previous sections, Chinese
languages remain isolating-analytical despite intense areal contact with more synthetic
languages along most of its frontiers.
6.1.4. The analysis of apparent exceptions
Claims have been made about the origins of a small number of languages which
show the obvious effects of contact and bilingual acquisition and which appear to contradict
the principles and preferences discussed here (and the intuitions of linguists, as well).
However, the principles and preferences also point to the kinds of information that would
be most relevant to the discussion of a particular bilingual mixture and its emergence. For
instance, in the presence of clear linguistic evidence of a morphosyntactic matrix system
(e.g., Quechua in Media Lengua), one might search for plausible, corroborating evidence
from a historical viewpoint—the account that “best fits the attested data and which requires
the smallest number of unattested steps is to be preferred” (Thomason & Kaufman, 1988,
p. 181). This assumption certainly includes the great likelihood that portions of the lexicon
are more prone to change and/or replacement than (say) inflectional morphology, and,
hence, morphological typology5.
This approach has been taken with significant results in the case of Ma’a (also
called Inner Mbugu by Mous, 1994), a mixed language/bilingual mixture spoken in parts of
Tanzania. Ma’a is described in various studies of language contact as a Cushitic language
5 See 2.1.3, above, for discussion of the likelihood that an isolating-analytical language will develop
inflectional morphology of any kind as a result of contact, i.e., external, rather than internal forces of
language change.
223
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
that has borrowed complex Bantu nominal and verbal inflectional morphology while
retaining a primarily Cushitic core vocabulary, in apparent violation of all previously
known borrowing hierarchies. For one thing, current work brings into question its status
as an independent language (see, for example, Thomason, 1997a and 1997b for fuller
discussion of its possible origins). It seems to be most likely a lexical register of (Normal)
Mbugu, a Bantu language, which is then used in its relatively relexified version as a marker
of ethnic solidarity. Contrary to the earlier and perhaps better-known analyses reported in
Thomason & Kaufman (1988, pp. 223— 228), Hudson (1980, p. 60), and Bynon (1977,
pp. 253-254), Mous (1994) surmises that it has a clear Bantu (rather than Cushitic)
morphosyntactic matrix that appears to have kept Cushitic lexical items, not a Cushitic
matrix that has borrowed Bantu morphology. To account for this particular mixture,
Myers-Scotton (1998) hypothesizes that there was a complete turnover (changeover) of the
morphosyntactic matrix (pp. 301,306-310) from one language to the other. Each of these
alternative analyses concludes, based on various historical accounts, that members of the
community (most being multilingual in at least one other Bantu variety and Swahili, the
national language) were originally speakers of a Cushitic language who reluctantly adopted
Mbugu (which is also intelligible to and/or spoken by Ma’a speakers), and which
consequendy became the new morphosyntactic matrix framing lexical items associated with
their particular ethnic identity (retained largely en masse)6.
Similar approaches may be quite revealing for other apparent contact varieties, as
well. Take, for example, Wutun, spoken in western China. According to Li (1983),
Wutun is a Chinese language which has borrowed some of the word order patterns and
affixes of Anduo, the local Tibetan language (see Thomason & Kaufman, 1988, p. 92).
6 This phenomenon is also known to take place in various versions or dialects of Romani (see Boretzky &
Igla, 1994).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Even though its lexicon is mostly of Chinese origin, it also has significant numbers of
Anduo words. Its speakers, often bilingual in both, are descendants of Chinese
immigrants who were apparently compelled to assimilate into the surrounding Tibetan
culture and to leam its language; few now know another Chinese language. However,
Wutun, as Anduo, lacks a tone system (atypical of any other known Chinese language),
has non-Chinese consonant clusters, and only one noun classifier (a total reduction from
the kinds of classifier systems typical of Chinese languages).
If it is true that an isolating-analytical language has, in fact, borrowed affixal
morphology (of any kind), there is an apparent typological anomaly. The present work
strongly suggests, however, that additional historical information is needed to prove or
disprove Li’s original analysis. On the linguistic evidence alone, it seems more likely that
there was a community-wide changeover to Anduo as the matrix language in a context of
extensive bilingualism. That is, Wutun speakers adopted Anduo as their community
language (under intense cultural pressure or out of necessity) and have retained large
numbers of Chinese lexical items to index special or separate ethnic status, thereby giving
this particular contact variety its distinctive character. Consistent with views that language
mixture is a gradient phenomenon (cf. Thomason, 1997a, pp. 3— 4), Wutun may, indeed,
show degrees of mixing and the intertwining of two systems, perhaps more than Mexicano
but less than Michif. (Mixed-language status is apparently in the eye of the beholder.)
6.2. Th e bo rro w a bility of inflectional categories
Assuming that various types of borrowing and other contact phenomena reported to
date constitute a sufficient and reasonably broad sample from which certain reasoned
225
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
generalizations and predictions can be made, the following observation is made: entire
inflectional categories such as Tense or Aspect associated with various (sub)classes of
verbs or Gender and Number on nouns are not borrowed in language contact situations.
This is despite the fact that knowledge of particular inflectional categories (and the
distinctions they make) of an encroaching, culturally dominant language cannot be said to
go beyond the linguistic abilities of speakers of a subordinate (recessive) variety. It may
become an integral part of an emerging individual bilingual’s skill and usage in that
language. Obviously, this kind of linguistic knowledge is typically available in normal
bilingual/contact situations. Despite the fact that there are no definitive instances in which
entire inflectional categories have been borrowed, certain distinctions, that is, category
values of a type already present in the recipient language, expressed by morphologically
compatible inflectional markers have, in fact, been taken7. For instance, the borrowed
Spanish plural marker -s (which expresses one and only one category value) into Mexicano
is permitted because the recipient already makes distinctions of number (also singular and
plural), and, hence, possesses the inflectional category of Number.
6.2.1. The emergence of category values
To provide a context for this observation, we turn, once again, to the kinds of
issues that are typically treated in studies of grammaticalization. Particularly relevant are
observations of the semantic nature of general inflectional categories such as Tense or
Aspect8. Individual inflectional markers express category values that provide distinctions
7 See Heath (1981) for numerous examples of borrowed inflectional affixes among typologically (though
not genetically) related Australian languages.
8 See Comrie, 1976 and 198S for the definitive works on the meanings and forms associated with Aspect
and Tense, respectively, across languages.
226
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(e.g., between present and past, or first and second person). Therefore, it does not seem
feasible that a language would have only one instantiation or value of Tense, say past
Every action would obligatorily receive a marker for past (or perhaps an unmarked form
indicating this one-member category). Because the usual function of a past tense marker is
to indicate that the action depicted by the verbal element to which it is affixed has occurred
at some point in time in the past, then all so-called finite or tensed verbs would express
events or occurrences located in the past. Surely, such a language cannot exist. Moreover,
the complete absence of tense distinctions (i.e., a category with no members) in a language
cannot be construed as indicating that the language possesses, say present tense as a sort of
default or category with one and only one value. In that case, every finite verb would
obligatorily depict an action or state that is present and/or located in present time. This
concept of inflectional categories applies to other categories, as well, for instance, in regard
to nominal markers of Gender or Number (Croft, 1990, p. 65). A language cannot have
only one value for grammatical gender (say, masculine); the absence of distinctions is
surely an indication of the absence of the category.
In contrast, languages can quite capably indicate reference to time (past, present,
future) without the inflectional category of Tense by means of particular adverbials and so
on (Comrie, 1985, p. 4). In addition, any language is quite capable of indicating natural
gender and specifying number without the existence of inflectional categories of Gender or
Number. The presence of plural markers despite the absence of singular markers in a
particular language shows clear evidence that at least some items are counted (and/or
countable) and that distinctions between individual and groups of items are grammaticalized
in that language. (The reverse is also a possibility, though relatively rare
crosslinguistically, where singular is the marked form and plural unmarked) (Croft, 1990,
p. 66). The existence of a plural marker, then, implies that there is a way to indicate
227
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
singularity (a category value of the inflectional category of Number), even if that
necessarily presumes an unmarked form (i.e., one that has zero expression). A language
cannot have an inflectional category for Number with only one category value, say
singular. This would imply that every object is countable, and there is only one in every
instance. Of course, a language may indicate semantic number or amount and countability
without an inflectional category for its expression. For instance, Vietnamese employs
various noun classifiers/determiners for this purpose without marking plural on nouns.
Even if one takes a primarily Indo-European language perspective, individual
values within each general inflectional category may differ significantly from language to
language within a family proper. However, the emergence of entirely new categories is
quite rare (e.g., Tense, Aspect, and Mood on classes of V and Number, Gender, and Case
on N), though in some languages, entire categories have, in fact, disappeared (e.g.,
Gender in English). What is known seems to indicate the following:
(a) individual markers and the meanings or values they express gradually emerge, most
likely in the expansion of the referential and expressive capacity of a language (and,
of course, its speakers) (see, for example, Bybee, 1985, 137-9; 1995, pp. 226-
227; Traugott, 1982,1988; Traugott & Konig, 1991; Herring, 1991; Hopper,
1991; Hopper and Traugott, 1993; Heine et al, 1991a and 1991b); Lichtenberk,
1991.
(b) individual category values come about as contrasts, even if one of a set of
oppositions has no expressed form (see, for example Bybee, 1995, p. 228 and
1985, p. 191ff; Matthews, 1991, p. 39; Croft, 1990, p. 64ff; Comrie, 1985, p. 9;
Lyons, 1968, p. 270-273; Sapir, 1921, pp. 105-109).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6.2.2. The inheritance of categories and category values
Another problem area concerning inflectional categories in individual languages is
that their existence in modern-day varieties, in the vast majority of cases, is a consequence
of historical lineage. That is, they are generally inherited from common progenitors. For
instance, the occurrences of Number and Gender distinctions on nouns in the Romance
languages are matters of inheritance from Latin; they have not been created out of nothing
{ex nihilo) in each language in the same sense that individual markers and their associated
values may have developed. To illustrate, regarding the often cited example of the
historical development of Spanish future “tense” markers, it is seldom, if ever, mentioned
that Spanish already possessed the inflectional category of Tense, which it inherited from
Latin. This development is portrayed once again in (1), below:
(1) Vulgar Latin *amare habeo > Spanish amar£
The specific emergence of a future marker/tense distinction in the Romance
languages involved the development of an additional category value, viz., an abstract,
grammatical distinction that contrasted with present and so on. This new value augmented
an already existing store; it was certainly not the development of the category of Tense by
any means. Moreover, particular distinctions of Person and Number and, in this case, their
obligatory expression via agreement, are semantically and phonologically fused with the
emerging future distinction9—a result clearly allowable by the morphological structuring of
the morphosyntactic matrix. These are values of inherited inflectional categories, as well.
9 Note that it is clearly the case that Tense is obligatory in Spanish, and not specifically “future".
Individual category values are selected on the basis of the meaning, for example, as the speaker locates the
event depicted by the verb at a place in time.
229
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In stark contrast, if an individual inflectional marker is borrowed, then it seems to
be implied that both the general category and the specific value it expresses are borrowed.
However, where there is no general category, there can be no specific value. Moreover, it
would not be possible for a language to borrow a tense marker (hence, a particular value) if
the category and the kinds of values it expresses were not already present in that language.
Logically, either the category is already present, or the category is borrowed—which does
not occur. Perhaps the effects of contact will facilitate the emergence of a particular
category value if the category already exists, for instance, the development of future in a
language which previously had only present and past However, the simultaneous
borrowing of an inflectional category and the range of values it expresses is not attested.
One of the cornerstones of grammaticalization theory is that the progression from
one concept type to another is in discrete steps in the direction of increased
grammaticalization (cf. Hopper & Traugott, 1993, p. 207f; Heine et al, 1991a, pp. 112-
113; Heine eta l, 1991b, p. 161ff). However, the transition of postposition directly to case
affix has not been documented in languages of the isolating-analytic type (which do not
have affixes of any kind), suggesting that language internal processes of change are not
likely to motivate the spontaneous development of either inflectional morphology or
inflectional categories. Apparently, an intermediate stage is required in which previously
free-standing elements become cliticized and the concepts that map onto them (a) become
gradiently more abstract and general, (b) develop categorial distinctions (category values),
and (c) eventually become obligatorily marked. It seems unlikely that borrowing (as an
instance of external language change) will lead to the acute development of entirely new
cognitively based semantic domains where internal processes of change are gradual,
generally imperceptible (because of the time depth involved), and based on the
accumulation of favorable conditions (e.g., word to clitic prior to the development of an
230
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
affix). This would imply that borrowing can accomplish overnight what
grammaticalization scenarios cannot do over much longer periods of time, for instance,
skipping over intermediate stages of clidcization. Evidence for such a claim comes from
the fact that such phenomena have not been attested.
6.2.3. Limits on types of borrowable meanings/concepts
The point at which the grammaticalization processes stop in an individual language
also appear to be determined by the kinds of meanings that are expressable in that language
(see Bybee, 1995, pp. 227— 229). That is, typological limits that restrict the development
of particular types of form-meaning sets appear to reflect constraints on grammaticalized
meanings, as well. As indicated in indices of synthesis and fusion, inflectional categories
and the distinctions they express can be analyzed along two distinct lines analogous to
those of their corresponding forms. One is according to the type of meaning (e.g., those
linked to inflectional categories), and the second is according to the number of meanings
expressed by a single form. As a consequence, the inflectional concepts expressed
individually by agglutinating-type affixes can be distinguished from those that are
semantically fused onto individual fusional-type affixes. In the former, form-meaning sets
involve the mapping of a single category value onto one, segmentable form (a single
representational unit or type) and, consequently, maintain a 1-to-l correspondence of
meaning to form. In the latter, multiple distinctions coalesce onto one form that is not
segmentable according to individual category values. Therefore, not only are types of
individual meanings relevant (i.e., categories and their values), the number of coalesced
meanings present on one form are relevant, as well. In each case, there is one form/affix.
231
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
As a consequence, a set of principles can be proposed that are analogous to the PSC
and PSI formulated in Chapter 2. The first, the Principle of Categorial Compatibility (or
PCC), identifies the types of meanings that are compatible with those of a recipient/matrix
system:
(2) The Principle of Categorial Compatibility (PCC):
Any concept or fusion of concepts is borrowable from a donor language if it
conforms to the semantic possibilities of the recipient language with regard to
conceptual types (e.g., semantic types and/or inflectional categories) and inherent
fusional possibilities.
In essence, this states that borrowed concepts will conform to matrix requirements
regarding (a) meaning (or conceptual) types of the recipient and (b) the number of concepts
that can be simultaneously expressed by one, discrete form. Hence, isolating-analytical
languages can borrow any concept codable by content items and independent function
words. Agglutinating-type languages may borrow these and individual grammatical
concepts, as long as they already possess types of concepts (categories and category
values) that are equivalent1 0 in some discemable ways (e.g., for purposes of translation),
and that a 1-to-l correspondence of meaning to form is maintained. Hence, the
morphological and semantic integrity of the matrix language is preserved. Fusional type
languages can borrow any meaning (i.e., category value) represented by any other
synthetic language (from agglutinating to fusional type) with the same kinds of semantic
and formal constraints: for a category value to be borrowed, the category must be already
present.
10 With the caveat that crosslinguisdc equivalence of inflectional categories is especially problematic (see
Croft, 1990. p. llff.)
232
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Its corollary, the Principle of Categorial Incompatibility (or PCI) can be expressed
in (3), below:
(3) The Principle of Categorial Incompatibility (PCI):
No concept or fusion of concepts is borrowable from a donor language if it does
not conform to the semantic possibilities of the recipient language with regard to
conceptual types and inherent fusional possibilities.
As a result, an isolating-analytical language can borrow neither the individual,
isolable inflectional meaning types of agglutinating languages nor the multiple,
simultaneously expressed category values associated with fusional-type languages (and
fusional-type affixes). For instance, such highly isolating-analytical languages as
Mandarin and Vietnamese cannot borrow inflectional markers of any kind because they do
not have inflectional categories. Strictly agglutinating languages (e.g., Mexicano) may,
indeed, borrow inflectional markers, but only in cases where they have at least an
equivalent general category (i.e., a category that already consists of two values). In the
same ways that agglutinating languages are constrained from borrowing fusional affixes as
a result of formal mismatch, they are also blocked from borrowing the multiple
simultaneous distinctions (inflectional concepts) represented by fusional affixes—they
neither fuse forms (through phonological bonding or fusion) nor fuse meanings (as a result
of the coalescence of multiple meanings onto a single form). Support for this comes from
the case of Mexicano, which has not borrowed any of the fusional affixes of Spanish11.
For an agglutinating language to borrow an affix from a fusional-type language, that affix
would have to undergo semantic reanalysis: only one concept will map onto one form.
1 1 In the present context, it is more likely to say that, ultimately, the reasons for the complete lack of
borrowed fusional affixes are based on both (a) morphological structuring and (b) conceptual (i.e.. semantic)
representation. Obviously, either serves to constrain borrowing. However, as proposed in the ensuing
discussion, arguments for semantic barriers may be the most persuasive.
233
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Fusional languages offer special, multifaceted challenges because types of affixes
are generally restricted to the types of category values that they can express (see, for
example, Blake, 1994, p. 106). For instance, classes of nouns may express category
values of Gender/Class, Number, and/or Case, but only certain combinations
(coalescences) occur in the languages of the world- On verbs, only certain categories find
appropriate expression. The kinds of meanings that can coalesce in a single affix appear to
be restricted in similar ways (e.g., on nominals versus verbals). With respect to form, a
fusional-type language can borrow any affix from any other language of similar
morphological structuring—in principle. However, according to the PCC, above, it cannot
borrow a nominal affix which represents a coalescence of particular values of Gender and
Number, for example, when it has no category of Gender. A verbal affix which expresses
a coalescence of Tense, Person, and Number category values (which can be phonetically
minimal irrespective of the number and types of distinctions it makes) cannot be borrowed
into a language that does not have any one of those inflectional categories. Ultimately,
semantic barriers may provide the most significant constraints to the borrowing of
individual category values and entire categories themselves.
While it goes well beyond the scope of this paper to argue which meanings go with
which others, it should suffice to say that the borrowing of inflectional categories and
associated category values would be a considerably complex and problematic task with
respect to types, numbers, and specific combinations of meanings expressed in fusional-
type languages. It will depend both on meaning type (e.g., which category is represented
and the values it expresses) and number of simultaneously expressed concepts. This
provides support for the suggestion that the only time that fusional-type affixes can be
borrowed is in situations of intense social and linguistic contact where the participant
languages are close areal neighbors, are genetically related, and close typological fits. For
234
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
instance, this might occur in borderland areas where each variety may be considered a
dialect of the other (i.e., at one time), though it is likely infrequent for reasons of national,
socio-cultural, linguistic, and/or ethnic identity (which may be under some degree of
conscious control).
One would be remiss not to mention a set of related possibilities. The first involves
the special relationships that exist between putative standard and nonstandard varieties of an
individual language (see fh. 64), for example, with respect to reported influence of so-
called Standard Serbo-Croatian on nonstandard dialects (Thomason & Kaufman, 1988, pp.
30-31). The other is the influence of a superstrate (acrolect or standard) on its respective
mesolects and/or basilects within communities of creole speakers. However, it should be
noted that in both cases, there is some degree of pressure exerted (top-down) on individual
varieties that systematically diverge from their “standards” via the conscious,
institutionalized practices enforced and reinforced through the educational systems of their
respective communities (see, for example, Gillman, 1993; Rickford, 1988; cf. Romaine,
1988, pp. 195— 197 and Alleyne, 1980, p. 15)1 2 . In such cases, one would expect a variety
of linguistic and social consequences.
6.3. Connecting borrowing and various contact phenomena
Wherever there is intense language contact, members of a single community
acquire/learn and use two or more languages on a daily basis. Those who become
12 This is in addition to, perhaps, the less conscious pressures that accompany the desire to speak a
prestige dialect or to have a wider potential communication network, for instance, in the case of speakers of
regional or social dialects of limited or marked membership (see Milroy, 1987; cf. Milroy & Wei, 1995, p.
137ff).
235
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
proficient in both (or all, in the case of multilingualism) may develop skills to translate
utterances spontaneously from one language to the other, to alternate from one to another
depending on the situation (i.e., in inter-sentential code-switching), and borrow from the
resources of both at various levels of grammar. Many may be quite adept at switching
from one language to the other intra-sententially, that is, at points in a single utterance
within clausal and/or phrasal boundaries. Throughout this work, the concept of a
morphosyntactic matrix is held to be at the heart of various bilingual/language contact
phenomena. As a result, there are clear links to language acquisition and such matters as
individual and/or community bilingual acquisition history (e.g., simultaneous or
sequential), levels of proficiency in each language in the bilingual’s linguistic repertoire,
and so on. It should not be surprising, then, to find similarities among various contact
phenomena such as borrowing and code-switching. Since both kinds of phenomena
involve the alternation of elements from two languages in some way, we might expect that
the lines separating the two would appear somewhat blurred at times (see Section 1.1.).
To account for the full range of these linguistic behaviors, it has been proposed that
bilinguals are continually faced with a continuum of situation types that call forth particular
language modes: a bilingual mode, in which both languages are activated (in the
psycholinguistic sense), and a monolingual mode for each language, used when speaking
with monolinguals of the particular language (cf., Grosjean, 1995, p. 259). At one
extreme, they are interacting with speaker/hearers of one language. At the other end, they
are speaking with other bilinguals who share their two languages and with whom they are
free to use either or both (i.e., in various types of code-switching and borrowing). At first
glance, it would seem that speakers in a bilingual mode would be able to (perhaps even be
inclined to) engage in code-switching of various kinds. All of the required on-line
236
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
operations and processing can be accomplished rapidly and efficiently because all aspects
of both languages are at the ready, capable of being accessed simultaneously.
In this view, speakers may also find themselves at points between extremes
depending on such matters as the dynamics of the situation, personal choice, and degree of
proficiency in one of the languages. Some may appear to be in a perpetual bilingual state,
for instance, those living in close-knit bilingual communities where conversational code
switching and language mixing are the norm (such as the one depicted in the Mexicano
data). Others, who are not part of such an environment, who disdain switching and mixing
behaviors, and/or who have not acquired conversational code-switching behaviors
(Penfield & Omstein-Galicia, 1985, p. 14), may rarely leave a monolingual mode
(irrespective of proficiency in each language). As a consequence, it is not difficult to see
how actual data can be interpreted in different ways. It may be quite problematic deciding
from which mode a particular utterance has originated in the absence of comprehensive
background information of all participants in the particular speech event. The occurrence of
a single form-meaning set from a donor in a recipient/matrix language may be the
consequence of either code-switching or borrowing. Perhaps, the only clues as to which
process has occurred are degrees of integration into the recipient system and the extent to
which the specific elements (i.e., morphemes, words, and/or phrases) have been accepted
and conventionalized throughout the community1 3 .
In this regard, one significant issue pertains to possible connections between
separate linguistic systems, for instance, in the ways lexical items in two (or more)
languages are represented and accessed in the mind/brain of the bilingual, who may have
13 As a consequence, a distinction must be made between the language of the individual and that of the
community, to which nearly every approach to linguistics makes reference in some respect.
237
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
degrees of proficiency in each. More-or-less equivalent form-meaning sets in each language
(so-called translation pairs) appear to be closely linked, which is particularly evident in the
simultaneous translation abilities of many proficient bilinguals (see section 1.1, above).
Much recent research suggests that bilingual representation may not merely be a question of
one system or two (see, for example, Kroll & de Groot, 1997; Poulisse, 1997, Paradis,
1997). On the one hand, the bilingual lexicon seems to be a unitary store of forms and
meanings (with parallel levels of activation a possibility); on the other, it appears to contain
two distinct components (one for each language), which can be selectively and/or
accidentally activated and, therefore, put to use (Poulisse, 1997, pp. 206— 208, 219f).
De Groot (1993) proposes that the degree of similarity of the conceptual meanings
associated with translation pairs may ultimately determine their representation1 4 . In the case
of sequential second language learning, form-meaning sets from the individual emerging
bilingual’s mental lexicon are conceptually linked and initially stored compoundly (i.e., one
meaning, two forms): “It [the bilingual lexicon] is parsimonious where it is justified to be
so; representational space is not wasted by storing the same meaning twice, once for the
word in each language ” (p. 46). This is based in part on the assumption that a second
language learner often assumes, based on prior knowledge, that the referent of an L i
concrete noun form is identical to that of an Li form already known, which is often the case
based on experience. They are very likely to share numerous physical or perceptual
properties and possess overlaps of meaning based on physical appearance (in terms of
numbers and kinds of properties). More abstract concepts have fewer or fuzzier perceptual
properties and/or conceptual overlaps. As proficiency in the second language progresses,
there is an increasing ability to represent form-meaning sets in each language in more
independent (optimal) ways (Kroll & de Groot, 1997). This developmental tendency
14 See, also, de Groot, 1992, in regard to cognate status of such translation pairs.
238
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
illustrates one way in which degrees of proficiency figure in various language contact
phenomena, including borrowing, code-switching, and other types of language mixing.
6.3.1. Distinguishing code-switching and borrowing
From numerous observations concerning types of code-switching and types of
borrowing, it seems reasonable to assume that there is not only a continuum of situation
types, but of phenomena, as well (for various social and psychological reasons). These
phenomena may manifest themselves naturally during a proficient bilingual’s on-line
speech production, from clear instances of inter-sentential code-switching at one extreme to
obvious cases of borrowing at the other. In a continuum of this sort, exact distinctions
may be somewhat difficult to make in each and every instance, especially in view of the fact
that similar looking phenomena may be products of very different underlying processes.
Nevertheless, it is one matter when a proficient bilingual switches from X to Y to
spontaneously “borrow” an individual form (in its more literal sense), whether or not it is
consciously or unconsciously adapted on-line to X phonologically and/or morphologically.
It is quite another when a unilingual speaker/hearer of X uses that identical form from Y
because it has diffused permanently into the community of X speakers.
A number of generalizations have been made that point to the more obvious
differences (see, for example, Poplack and Sankoff, 1984; Muysken, 1995). For instance,
when the use of forms from different languages proceeds along phrasal or clausal lines
(i.e., a phrase or clause in language X followed by a phrase or clause in language Y), it
may be a relatively simple task to recognize this as code-switching, although phrasal-like
borrowings such as idioms, ritualized greetings, and so on are also distinct possibilities
239
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(e.g., from French into English: fait accompli “a thing accomplished” and faux pas “false
step”). In cases that are not so clear-cut, in what the literature often refers to as nonce
borrowings, single-item code-switches, or lexical transfer (van Hout & Muysken, 1994, p.
40; Romaine, 1995, p. 229), a precise definition of borrowing such as the one proposed
here may help make sharper distinctions (e.g., by pointing out what cannot be borrowed).
In separating the two processes, one issue to be discussed is the level of
grammatical analysis. Borrowing involves the analysis of donor form-meaning sets in the
process of complete integration into the recipient system (see subsection 3.0.1.)1 5 .
Borrowed elements are (re)analyzed according to the morphological structuring of the
recipient system; word class of a borrowed form-meaning set is in all likelihood assigned
by analogy, based on the semantic characteristics of other similar form-meaning sets in the
recipient. Word or morpheme type is, therefore, epiphenomenal. In addition, lexical
borrowing typically involves the permanent adoption of individual constituents of phrases
or clauses (i.e., words or morphemes) from a foreign system, which are, in turn, used to
build up exclusively recipient frames. Hence, structural units are assembled from material
analyzed as belonging to the recipient and subject to the rules and processes of its
grammatical system. Indeed, morphological structuring is always essential to the
discussion of borrowing while the construction of syntactic frames may not be particularly
relevant, except, perhaps, to illustrate the ways syntax and morphology interface in the
final product. This is true whether we consider long- or short-term borrowing.
15 Muysken, 1995 (p. 190f) notes that the integrative processes of borrowing bear striking resemblances to
those involved in derivation. According to the position presented here, the only actual difference is the
source lexicon of individual form-meaning sets. See also Chapter 3, above, especially subsections 3.0.1.
and 3.2.6.
240
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In contrast, code-switching involves running syntactic analyses. In production, it
involves the establishment of entry and exit points in the linear speech stream, so-called
switch points at which the language not in use is deactivated to an extent These analyses
mark syntactic boundaries between phrasal and/or clausal constituents of two separate
language systems (see, for example, Grosjean, 1995, p. 261; Muysken, 1995, p. 177ff.;
Myers-Scotton, 1995, p. 235). Such boundaries may merely serve to indicate one-word
constituents (e.g., an NP consisting of a proper noun or pronoun or a VP consisting of a
single verb). In other words, code-switching entails the identification and construction of
phrasal and clausal frames whose individual (or collocations of) constituents may originate
in either language (cf. Muysken, 1995, p. 191). Occasionally, these frames may appear to
be mixed, consisting of elements of both an embedded and matrix (or base) language1 6 .
Another factor to consider is recognition and control of a matrix and awareness of
non-matrix (i.e., foreign) material (Hill & Hill, 1986, p. 345). A speaker/hearer must
discern whether there are two (or perhaps more) autonomous language systems being
employed in particular utterances. This may be a matter of degrees, as well, and depend on
such factors as the level of proficiency in each language. For example, when a speaker is
significandy more proficient in only one of the two languages whose elements occur in a
single utterance, it is likely that the dominant (according to proficiency) will play a greater
role in the determination of such linguistic characteristics as phonology (including prosody
and so on), word order, and other grammatical characteristics (e.g., application of
inflectional categories and their associated values). Code-switching necessarily involves
the recognition of separate, autonomous systems irrespective of whether or not a clear
matrix language responsible for setting phrasal or clausal frames can be unambiguously
1 6 This point is consistent with most major views of code-switching viewed as altemational, insertionaJ,
or some combination of the two. See Muysken. 1995 for discussion.
241
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
established. In borrowing, there is no question that only the recipient system is relevant to
the utterance. As a consequence, morphological integration may be the only criterion that
can be used to distinguish borrowed from switched forms in specific instances (cf., Hill &
Hill, 1986, p. 346), though it is not completely foolproof1 7 . Attending to morphosyntactic
characteristics—that is, the degree to which particular form-meaning sets are (re)analyzed
and integrated into the recipient morphosyntactic matrix—is much more likely to yield
evidence of borrowing. However, we have learned by now to anticipate numerous gray
areas where identification is made more on intuitive rather than strictly linguistic grounds.
It may not be possible to establish unequivocal cause-and-effect links between
code-switching and borrowing. In cases of deep structural borrowing (including
processes, co-occurrence patterns, and so on), code-switching may play a role similar to
the one it likely plays in lexical borrowing—it models structures and elements of the
encroaching donor language system that can be subsequently borrowed (i.e., copied) into
the borrowing system. It is true, however, that those who engage in such behaviors as
conversational code-switching are very likely to be borrowers, as well. They may also
belong to the subset of individual bilinguals who have the resources and required
proficiency levels to be conduits of borrowing. Even though it may be difficult to speak of
any long-term effects of code-switching on either language, one can refer to its possible
facilitating role in external forces of change, for example, in increasing the frequency of
donor form-meaning sets and structures in the actual speech patterns of fluent bilinguals
and, perhaps, establishing community norms. Extensive societal bilingualism,
simultaneous bilingual acquisition, and community-wide attitudes towards various mixing
17 See Hill & Hill, 1986, Chapter VII, for a thorough discussion of the kinds of problems that exist in
telling code-switching and borrowing apart in Modem Mexicano. For the most part, we take the position
here that code-switching is much more likely to have a temporary effect, and that borrowing necessarily
involves full morphosyntactic integration of form-meaning sets.
242
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
behaviors may very well constitute a set of underlying conditions, predisposing factors,
that can lead to a variety of bilingual phenomena. Hence, widespread bilingualism and
code-switching are likely precursors of externally motivated change.
6.3.2. Convergence and a composite matrix
Convergence in bilingual/contact situations is typically referred to as an assortment
of processes by which two languages or dialects of the same language become more alike
in specific areas of grammar, or when one language or dialect becomes more like another1 8 .
Both borrowing and transfer play significant roles. Assuming that one language is
dominant within the community in certain social domains (Thomason & Kaufman, 1988),
speakers of another, culturally subordinate variety who have become bilingual in the
dominant are very likely to engage in borrowing, especially in cases of advancing shift As
a consequence, their original language will become more like the dominant (within the
morphological parameters stipulated by the PSC and PSI), illustrated in the long term
effects of borrowing of Mexicano from Spanish. The other side of convergence, transfer,
may affect the dominant language in that it may be acquired/learned as a second(ary)
language by relatively homogeneous community (see subsections 1.2.2 and 1.2.3, above).
In the case of the Malinche region, Spanish was the language of interlopers—a
conquering people. At first, it was learned sequentially as a secondary language, and was
most likely spoken in varying degrees of proficiency. Up until the early parts of the 19th
century, maintenance of Mexicano was the rule; massive lexical borrowings and those of a
more structural nature appear to be a later development As Spanish gained in currency in
18 See, e.g., Gumperz & Wilson, 1971 for their classic study of convergence.
243
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the Malinche region and elsewhere, the effects of transfer have become increasingly evident
(cf. Lipski, 1994, p. 279f). The effects of transfer, therefore, can be expected on
individual and community levels and in all areas of grammar, from phonology to syntax to
pragmatic areas of discourse. For instance, Hill & Hill (1986) discuss the Mexicano
“accent” in Spanish (p. 198ff)1 9 , the numbers of Mexicano loanwords in regional Spanish
(pp. 99,143), and even morphological/semantic adaptation of Spanish diminutives for
honorific usage resembling that of Mexicano (p. 196). Spanish sounds, form-meaning
sets, and even discourse-level characteristics (e.g., particles) have been interpreted
according to native forms. As a direct consequence, the Spanish of the Malinche region
has become more like Mexicano. In essence, each language has moved towards the other
in numerous ways (cf. Silva-Corvaldn, 1994, pp. 4-5, 166).
The concept of a morphosyntactic matrix that is a composite of two separate
language systems is a reasonable conclusion and consistent with views of convergence. As
a gradual product of processes associated with internal language change, the
morphosyntactic matrix of a particular language will gradually evolve from one generation
of speakers to the next20. In bilingual/contact situations, the matrix of each language may
also change as the result of external factors (evidenced in patterns of convergence). In
some instances, there is an interaction among these distinct processes, for example, when
structural borrowing (from outside) affects grammaticalization (on the inside). As a result
of the ways individual aspects of language can be affected by internal and/or external
processes of change, the morphosyntactic system of an individual language may show an
assortment of changes, each proceeding in its own discrete steps. In addition, it can be
19 See, also, Gimate-Welsh, 1980. p. 33-35.
20 See, for example. Labov, 1972, p. 160ff for discussion of sociolinguistic factors of change and W.
Lehmann, 1992, pp. 9-14 and Hock & Joseph, 1996, pp. 3-17 for aspects of language known to be subject
to internal change.
244
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
assumed that the matrix of each individual speaker exists as a part of a community-wide
network of individual systems, with overlaps (certain changes taking place more or less
simultaneously from speaker to speaker) and layering (changes primarily occurring in
particular areas of grammar, for example, phonology) producing a variety of effects in
dynamic ways.
Jake (1996) argues that individual interlanguages (ILs) represent the gradual
development of a speaker’s L2 matrix and evidence of the simultaneous operation of more
than one system (i.e., Lj and L2). Hence, the emerging second language system is seen as
a composite of Li and L2 morphosyntactic characteristics (in ways that appear to be quite
similar to those associated with pidginization and creolization). Numerous researchers
have reported observations of spontaneous child language mixing (i.e., code-switching and
borrowing of various sorts) in the simultaneous acquisition of two (or more) languages2 1
and different kinds of mixing in the speech of adults (e.g., Grosjean, 1995, p. 263) in
which a clear, unitary matrix does not appear to be consistently exhibited in uninterrupted
fashion. Individual structures are traceable to separate systems. In the development of an
individual or community-wide interlanguage (qua dialect), fossilization may result in a
composite matrix becoming the conventionalized, regional norm, while spontaneous
mixing in children or adult speech may have a number of effects within a bilingual speech
community. In any event, the incidence of various language contact phenomena like
spontaneous borrowing and code-switching generally tends to increase the numbers of
potential candidates for borrowing (inclusive of form-meaning sets, structures, and even
processes). Lexical borrowing may be facilitated and may increase to include additions and
21 See, e.g., Grosjean, in press; Hoffmann, 1991, pp. 75-79 (cf. Relinger & Park. 1980 and Volterra &
Taeschner, 1978).
245
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
substitutions based on frequency (from the perspective of the individual speaker/hearer) in
either or both languages, most likely within specific semantic domains.
In summing up the conclusions of her study of Los Angeles Spanish, Silva-
Corvaldn (1994) states that “in language-contact situations bilinguals develop strategies
aimed at lightening the cognitive load of having to remember and use two different
linguistic systems” (p. 206). While it may be inferred from such comments that bilingual
speakers employ their various strategies for the purpose of economy, or in order to make
perception and production, and, hence, learning easier, it is much more accurate to turn the
inference around. The implicit goals (in Silva-Corvaldn’s view) of the strategies bilingual
speakers employ are, in fact, the natural results of normal processes. In fact, there may be
few other options available. By using prior (i.e., Li) linguistic knowledge (see, for
example, Corder, 1993; Zobl, 1993), learning is facilitated. Any reliance on first language
experience to bootstrap into a second will always entail a large amount of first language
influence—that is, transfer (cf. Sharwood-Smith, 1994, p. 13).
6.4. CONCLUSIONS
As illustrated in the case of Mexicano, borrowing plays a significant role in
processes of change. Whenever intense language contact occurs, lexical borrowings are to
be expected, and when contact situations become protracted as a consequence of particular
sets of social conditions, borrowing will likely increase, even to the point that fundamental
changes will become evident in the recipient language system, both in terms of lexicon and
in terms of structure. As a result of intense and prolonged contact with Spanish, lexical
borrowing has resulted in a typological drift from polysynthesis to analysis, a process not
246
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
uncommon in such contact situations. In addition, the wholesale borrowing of Spanish
function words has resulted in significant changes in phrasal and structural organization. It
seems likely that (a) the borrowing of such function items as prepositions and various types
of conjunctions and (b) word order changes (e.g., co-occurrence patterns within the NP)
show the effects of change at much deeper levels of grammar. The very nature of
borrowing appears to be a kind of imitation (Haugen 1950, p. 212); speaker/hearers of a
recipient language use (to the best of their abilities) the forms and strategies of another (the
donor) to augment, and, in more extreme cases, to eventually replace the forms and
strategies of their own. To the degree that this is true, structural borrowing (copying the
ways proficient bilingual speakers string together certain ideas and concepts in the
culturally dominant language) may necessarily entail the copying (i.e., borrowing) of the
forms, as well. In this case, form follows function.
It has also been shown that borrowing processes can interact with language-internal
processes of change, especially grammaticalization. As one obvious motivating factor in
the drift from a (poly)synthetic and incorporating character to one that is more highly
analytical and isolating, borrowing provides evidence that the normal, single direction from
analysis to synthesis assumed to be integral to grammaticalization can be arrested. This
includes borrowing of a more grammaticalized nature, which is evidenced by the thorough
incorporation of Spanish function words into the Mexicano matrix and resultant typological
changes. The resemblances borrowing has with normal, language-internal derivational
processes also demonstrates that borrowing is, in fact, an equally normal process, one that
speaker/hearers of a language can utilize to their social (perhaps, cultural) and linguistic
advantage (e.g., to increase expressive and referential power).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Connections among borrowing and other language contact phenomena also
illustrate a number of interesting things. For instance, the relationships between different
types of code-switching and types of borrowing show a logical progression, one that, in all
likelihood serves as indicator of impending shift and/or language death. Both code
switching and borrowing have the ability to reflect patterns of language acquisition,
including acquisition histories (e.g., simultaneous or sequential). However, borrowing
appears to be a much more reliable indicator of lasting change. While both types of
phenomena involve the interaction of social and psychological conditions (i.e., in the
activation of different bilingual or monolingual modes), borrowing of different sorts is a
more accurate predictor of eventual convergence. Moreover, the spread of form-meaning
sets among members of a bilingual community reflects social structure, from individuals to
networks of individuals and onto the community at large (cf. Milroy & Wei, 1995 and
Milroy & Magrain, 1980). The evidence provided by the Mexicano data shows the extent
to which borrowing can spread and points to a potentially interesting topic for future
research: the roles of social and linguistic networks in borrowing and subsequent
externally motivated language change.
This study has provided considerable evidence that linguistic borrowing is not at all
random. It is systematic because each language participating is systematic, particularly in
the ways that form-meaning sets are structured (morphological structuring). Issues of
compatibility are firmly grounded in this fact. In this regard, the concept of matrix
language (with its obvious links to acquisition) is crucial; the similarities (and differences)
among mixed languages and pidgin and creole languages also serve to illustrate the
systematicity that always accompanies the emergence of new language varieties. Its clear
presence in mixed languages provides significant, observable support for approaches to
creolistics that assume (and, hence, search for) evidence of substrate languages. In this
248
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
sense, both pidgins and creoles clearly link to language contact, even if original (substrate)
varieties can only be inferred.
One issue merely broached in this final chapter deserves additional attention and
requires future research: the roles of types of meaning in diverse bilingual contact
phenomena. For instance, to what extent do various types of meanings determine the
character of borrowing? Can this be extended to pidgin and creole genesis and/or code
switching? Research currently underway concerning selectional possibilities of particular
function words and their relationships to a matrix frame in code-switching is one area
where types of meaning will very likely prove to be significant Myers-Scotton and her
collaborators are currently refining the Matrix Language Frame Model to include different
types of morphemes (beyond the previous binary division between content and system
morphemes). For instance, she now differentiates among (i) content morphemes, (ii)
indirectly-elected system morphemes (e.g., certain prepositions that collocate with nouns
for semantic reasons, e.g., at in “look at”), and (iii) structurally-assigned system
morphemes (e.g., inflectional markers of tense/aspect)— the so-called 3M Model (Jake &
Myers-Scotton, 1998 and Myers-Scotton, personal communication).
In sum, extensive lexical and structural borrowing, even though it is often neglected
as a specific area of study, has the ability to shed significant light on the language faculty,
especially in regard to the ways different language systems can interact in the mind/brain of
the bilingual speaker/hearer in the emergence of entirely new language varieties.
249
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Appendix A
Additional Mexicano Text
One additional passage from the data is presented below. While other portions of
the Mexicano text may show considerably greater numbers of Spanish loans, this particular
exchange is, nevertheless, illustrative of many of the characteristics of the content of the
interviews. It was uttered by an elderly man in Acxotla in response to the translation
section of the questionnaire (Hill & Hill, 1986, pp. 120-121). Those familiar with
Spanish, note the ways borrowed content items and function words are fully integrated into
the sentences. The first version in (1) appears as transcribed in the corpus. The line below
the Mexicano text contains Spanish words (in small caps) that are generally word-by-word
translations. However, they obviously do not represent exact morpheme-by-morpheme
glosses. The same portion of text as it appears in Hill & Hill (1986) immediately follows
in (2) with its English translation.
(1)
R. tnamiquih ce tlacatzintli, huan quihtoa in tla cualcan
ENCONTRAMOS A UN SENOR Y DICE SI ES LA MANANA
“quen otonmixtonaldh, cox timopaquiltihtica” in tla tlahcah,
COMO AMANECIO SI ESTA UD CONTENTO SI ES EN EL DIA
“cualli itlah cahtzin Dios quen omitzmomaquilih, cox timezticah
BUEN DIA DE DIOS COMO LE DIO SI ESTA USTED
contento? quemah, tla canin tmoica nican nioh ica in notrabajo,
CONTENTO SI DONDE VA UD ACA VOY CON MI TRABAJO
niah itech monte, niah itech trabajo decampo, niah itech
VOY EN EL MONTE VOY EN TRABAJO DE CAMPO VOY EN
trabajo de tronco de mezontetl nicuilihua, yenon oficio.
TRABAJO DE TRONCO DE MAGUEY ARRANCO ESE ES EL OFICIO
huan yahui quihtoa poz buenoh ximoicatehua ximoicatehua,
Y VA DICE PUES BUENO VAYA UD VAYA UD
250
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
oncan thualmonamiquih dotlac, thualmanamiquih, yenon panoa,
AHI NOS ENCONTRAMOS EN LA TARDE NOS ENCONTRAMOS YA ESO PASA
can tmoica poz niah nican ninemitih nican niah nicyehualotih
DONDE VA UD PUES VOY ACA A CAMINAR ACA VOY A DAR UNA VUELTA
in noaxca canin xamo cox panoa inmilli, nozo amo panoa
EL MIO DONDE SI NO SE PASA LA MILPA O NO PASA
milli, yenonic decampo, mh, yenon, yenon totlahtol, huan tla ye...
LA MILPA POR ESO DE CAMPO MH, YA ESO ES NUESTRO IDIOMA Y SI ES
poz orale poz in cualcan tiahzqueh ica in yunta tdapehuatihua,
PUES ORALE PUES EN LA MANANA VAMOS A IR CON LA YUNTA VAMOS A ARRAR
xtlalican ce tlaquehual ome tlaquehual yeyi tiaquehual para
PONGAN UN PEON DOS PEONES TRES PEONES PARA
ctzinmelahuatih inmilli, cpedahuatih inmilli campa repente
QUE ENDERECEN LA MILPA VAN A DESTAPAR LA MILPA DONDE DE REPENTE
pachitoc ica tlalli, mixpetlahua para mozcaltia, yecah, poz
ESTA TAP ADO CON TIERRA SE DESTAPA PARA QUE CREZCA YA ESTA PUES
xmaca in tlalli, xdalhui, xictzatzacua campa yahui in ad
DALE LA TIERRA PONLE TIERRA CIERRA DONDE VA EL AGUA
den zanja dead xdali zanja dedalli, apamid, para huitz
DE LA ZANJA DE AGUA PON ZANJA DE TIERRA ZANJA PARA QUE VENGA
in ad ctzacuilia, eh? ye non, ye non dahtol ticpiah in tehhuan, amo de
EL AGUA LO ATAJE, EH? YA ESO YA ESO IDIOMA TENEMOS NOSOTROS NO DE
ocachi mas idah...
MAS MAS ALGO...
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(2)
Tnamiquih ce tlacatzlntli (Mhm), huan
quihtoa, in tla cuaican. “^Quen
otonmixtonaltih? ^Cox
timopaqulihticah? (Mhm.) In tla, tla
tlahcah, “^Cualli Itlahcahtzm Dios.
6 Quen omitzmomaquilih? ^Cox
timetzticah contento?’' “Quemah. Tla
cualtzin in tehhuatzin. ^Canin tmoica?”
“Nican nioh ica in notrabajo. Nioh
itech monte, nioh itech trabajo de
campo, nioh itech trabajo de campo,
nioh itech trabajo de trongo de
metzontetl nicuilehua.”
Ye non oficio. (Mhm) ^Hmm?
Huan yahui, quihtoa, “Pos buenos,
ximoicatehua, ximoicatehua.’ Oncan
thuahnonamiquih tiotlac (Aha),
thualmonamiquih, ye non panoa...
“^Can tmoica?” “Pos nioh nican
ninemitih nican, nioh nican
nicyehualotih in no-, no- noaxca, can in,
xamo cox panoa in milli, nozo amo
panoa in milli” Ye n5n ic, de campo.
(Mhm.) ^Hmm? Ye non totlahtol.
Huan tla ye cah, “Pos, pos 6rale. Pos
in cuaican tiazqueh ica in yunta
ttlapehuatihueh.” (Aha.) ^Hmm?
“Xtlalican 5me tlaquehual, tlaquehual
para quimelatih in milli.” ^Eh? (Mm.)
“Para cmelahuatih in milli, campa de
repente pachihtoc ica tlalli, mixpetlahua
para mozcaltia.” <,Hmm? Ye cah, “Pos,
xmaca in tlalli, xtlahui, xictzatzacua
campa yahui in atl den zanja de atl.
Xtlali zanja de tlalli, apamitl, para huitz
in atl ctzacuilia.” j,Hram? Ye non, ye
non tlahtol ticpiah in tehhuan. Amo de
ocachi mds itlah.
We meet a gentleman (Mhm), and he
says, if it is morning, “How did you
greet the day? Are you happy?” (Mhm.)
If, if it is noon, “God’s noon is good.
How did he reward you? Are you
happy?”
“Yes. Surely you are well. Where are
you going?” “I’m going here for my
work. I am going to the countryside, I
am going to the fieldwork, I am going to
the work of pulling out an old maguey
plan.” That is work. (Mhm.) Hmm?
And when he goes, he says, “Well, good
[day], be going, be going, be going.”
Then we come meeting in the afternoon
we come meeting, now he passes by...
“Where are you going?” “Well, I am
going to walk about here, I am going here
to take a turn round my, my my property,
where, perhaps the com is coming along,
or the com is not coming along.” That’s
it, about the fields. (Mhm.) Hmm?
That’s it, that’s our language. Eh?
And if that’s done, “Well, well let’s go.
Well in the morning we will go with the
ox team to plow.” (Aha.) Hmm? “Put
on two laborers, laborers to straighten the
com plants.” Eh? ‘T o straighten the
com plants, where they have just been
covered with earth [by the cultivator in
the “second plowing”], they get
uncovered in order to grow.” Hmm?
That being done, “Well, give it earth, put
it there, block it up where the water runs
in the water ditch. Put in a ditch of earth,
an irrigation ditch, so that when the water
comes it blocks it off.” Hmm? that’s it,
that’s the language we have. It’s nothing
more than that (S76)
252
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Appendix B:
Spanish borrowings in the data
Spanish borrowings found in the data are presented in the following order: I—
Nouns: concrete animate; II—Nouns: concrete inanimate; HI—Nouns: quasi-concrete;
IV—Proper nouns; V—Abstract nouns; VI—Verbs; VII—Adjectives; VIII—Adverbials;
IX—Independent function words & affixes. The final section, X—Word counts by
village, lists the participants/subjects by number and amount of text attributed to each.
Different semantic (sub)types of nouns are listed according to the classifications used in
Chapter 5 (see subsection 5.2.2). Proper nouns (names of individuals), which generally
have visible and tangible referents, are listed separately here to illustrate the cultural impact
of Spanish and its speakers on Mexicano culture1. In nearly every instance, inhabitants of
the Malinche area have adopted surnames of Spanish origin, and, presumably through the
influence of the Roman Catholic church, Christian forenames (and so on). Place names,
on the other hand, are often referred to by either the original Mexicano toponym or by the
name of the locale’s patron saint (Hill & Hill, 1986, p. 11), and ffequendy both, as in San
Antonio Acuamanala.
Under the heading of FUNCTION WORDS, Spanish numerals appearing in the text
are also listed. In the literature on the classical language (Ndhuatl), native numerals are
referred to variously as a subclass of substantives (e.g., Noun) and/or quantitative
pronominals (see Andrews, 1975, pp. 143, 183ff). In such languages as English and
Spanish, such words are typically included as a subclass of Determiner (i.e., function
1 Proper nouns were tabulated under QUASI-CONCRETE, OTHER in Table 5.7 for the sake of convenience;
they represent names for individuals, organizations, and places. Irrespective of status as proper versus
common noun, proper nouns adopted horn the Spanish language and culture certainly have concrete
referrents and are necessarily included in the word counts.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
words). As a consequence, whether or not borrowed Spanish numerals are rightly classed
as content items or function words could not be determined with complete certainty in every
instance. Hence, they were not counted as members of either class. Obviously, word
counts may be slightly affected, specifically regarding the ratios of content items to function
words.
In fact, the occurrence of native numerals was quite rare in the text (i.e., beyond ce
“one” and ome “two”). The wholesale adoption of Spanish/European naming practices and
contemporary lack of native Mexicano numeracy skills (i.e., knowledge of the ancient
vigesimal system of calculating) prompted Hill & Hill (1986) to comment, “.. .the concern
of many people about the fact that their names are not legitimo mexicano ‘genuine
Mexicano’, and the concern that they cannot count correctly in the language can be seen as
a remarkable continuity of resistance to the resymbolization of their social universe in
Christian, Occidental terms.” The impact of colonization is clearly felt in these two
particular domains.
Next to each entry is an English gloss, with the number of tokens in parenthesis.
Hyphens before or after the form indicate that the item typically occurs with either a
Mexicano prefix or suffix. Thus, FULLY INTEGRATED VERBS depicts fully borrowed
Spanish verbal roots. Verbal loans that occur without accompanying Mexicano
morphology (or ambiguously with apparent inflection) are listed immediately after. Forms
such as the latter seem to occupy the frontier between code-switching and borrowing;
however, their occurrence is always within a totally Mexicano context (no other Spanish
forms co-occur). In these and similar cases, it is assumed that the items are borrowed; this
is also in view of the fact that little additional linguistic evidence exists that two separate,
autonomous systems are being used alternately in code-switching.
254
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
I— N O U N S: CONCRETE ANIM ATE
1. HUMAN
alma “soul” (1)
amiga “female friend” (1)
amigo “male friend” (2)
amiguitos “friends HON” (1)
ancianitas “old women” (3)
ancianito- “old man” (1)
ancianoh- “old person” (4)
antiguanos “ancient ones” (1)
antiguitos “ancient persons” (4)
auxilio “aid, assistance” (2)
bandas “bands” (1)
chiquitos “children, boys” (1)
-conocidoh “acquaintance” (3)
difunto “corpse” (1)
dijundito “dead body HON” (2)
dijunto (var. of difunto) (1)
finadito “deceased HON” (4)
gente(h) “people” (24)
gobiemoh “government” (2)
hombre “man” (2)
joven “youth, young man” (5)
jovencito “youth” (2)
juvenazos “juveniles PEJ” (1)
muchacho “boy” (4)
mulatoh “mulatto” (1)
nina “girl” (1)
ninoh “boy” (5)
pendejitos “pubic hair PEJ” (1)
persona(h) “person” (15)
pobre- “poor” (1)
2. RANK
dama“lady” (1)
don (title of respect, m.) (32)
dona (title of respect, f.) (9)
padre (priest, father) (14)
-padrecito (priest, father) (1)
-pastor “pastor, shepherd” (2)
-patronas “patrons, saints, f.” (3)
patron “patron, saint, m.” (3)
presidentah “president, f.” (3)
presidente “president, m.” (1)
principe “leader” (1)
seiior “mister, lord” (17)
-senora (at Madam) (8)
teniente “lieutenant” (1)
3. KINSHIP TERMS
-a(b)ueIitoh “grandmother” (7)
-ahijada “goddaughter” (4)
-ahijado “godchild” (5)
-comadre “godmother” (1)
-comalehtzin (older var. of
comadre) “godmother” (2)
-compadre “godfather” (3)
-compadrito “godfather” (7)
-compalehtzin (older var. of
compadre) “godfather” (2)
-cunado “brother-in-law (4)
-familiah- “family” (20)
-hermano- “brother” (4)
-jefa-“chief, wife” (1)
-jefe- “chief, husband” (1)
-madreh “mother” (4)
-mamacita “mother” (1)
-mamd “mother, mom” (17)
-papacito “father, dad” (4)
-papd “father, dad” (41)
-primoh “cousin” (8)
-suegrah “mother-in-law” (3)
-tiah “aunt” (2)
-tioh- “uncle” (2)
-viudoh “widower” (1)
abuelos “grandparents” (1)
comadrita(h) “godmother” (35)
confianza “fiancee” (1)
cunadah “sister-in-law” (3)
hermana “sister” (1)
hija “daughter” (1)
hijo “son” (1)
parejah “couple, pair” (1)
pariente(h) “parent, relative” (3)
viudah “widow” (1)
4. BODY PARTS
puiio “fist” (1)
5. FRUITS AND VEGETABLES
ajo “garlic” (2)
airocito “rice” (1)
azucar “sugar” (1)
brijolitoh (from frijolitos) “beans”
(1)
cebada “barley” (2)
cebollah “onion” (1)
chavacano “type of apricot” (1)
255
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
cilantroh “coriander” (2)
frutah “fruit” (2)
mangoh “mango” (1)
pldtanoh “plantain” (2)
6 . ANIMALS
animales “animals” (1)
-axnoh “donkey” (1)
bunito “burro DIM” (6)
burro(h) “burro” (6)
caballo “horse” (1)
cabrdn “goat” (usage PEJ) (4)
camaron “shrimp” (1)
carp ah “carp” (1)
lechdn “suckling pig” (PEJ)(2)
pajarito “bird” (1)
-palomax- “dove” (1)
pavitoh “turkey” (1)
toritoh “bull” (1)
toroh “bull (12)
vacada “cattle” (1)
vacas “cows” (1)
7. DERIVED PRODUCTS
abonoh “fertilizer” (1)
alcohol “alcohol” (1)
cachitos “bull homs” (1)
chitoh “beef’ (2)
cigarros “cigarettes” (2)
comidah “food, meal_(l)
manteca- “butter” (1)“
pan-tzin “bread” (4)
producto “product” (1)
puntal “snack” (1)
refresco “drink” (1)
sopah “soup” (1)
toroh-nacad “bull meat” (2)
tortah “round cake” (1)
-tortillah “tortilla” (2)
8. DISEASES
enfermedad “infirmity” (4)
inyeccidn “injection, shot” (2)
pastillas “pills” (1)
pulmonfa “pneumonia” (1)
remedio “remedy” (2)
10. OTHER
-vida(h) “life” (50)
256
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
II— NOUNS: CONCRETE INANIMATE
1. MATERIALS
hilo “thread, string” (1)
ladrillo “brick (1)
maderas “wood” (1)
teja “tile” (1)
vigah “beam, rafter” (2)
2. ARTIFACTS
alambique “still” (1)
arado “plow (1)
bacinica “chamber pot” (1)
bolsa “bag” (2)
cajas “boxes” (1)
cajetitos “boxes, packets” (1)
calzoncilloh “men’s shorts” (1)
calzdn “trousers” (2)
cama “bed” (1)
camidn “truck” (1)
camixah- “shirt” (3)
coche “car, auto” (1)
copa “cup” (4)
diarioh “daily paper” (2)
escopeta “shotgun” (2)
escuperas “cuspidor” (1)
espuelas “spurs” (1)
garrote “club, stick” (1)
gropecia (from gropos?) “special
dyed cloth?” (1)
-hacha “ax, hatchet” (2
homoh “oven, furnace” (1)
instrumento “instrument” (1)
iscuadrah “carpenter’s square” (2)
lozo “kind of crockery ?” (2)
-machete “machete, knife” (1)
macna (var. of m£quina)
“machine” (1)
mdquina “machine” (1)
mesa “table” (2)
palah “stick” (3)
pantalon “pants” (2)
periodico “newspaper” (1)
pistola “gun, pistol” (4)
popoiio-tzin (from punal)
“dagger, sharp point” (1)
-puerta “door” (4)
retrato “portrait, photo” (1)
ropa “clothing” (1)
silla “chair” (1)
sombreritos “hats” (1)
sombrero(h) “hat” (4)
-tapetes “rag” (1)
tocadisco “record player” (2)
tren “train” (1)
tronco “trank, log” (1)
vestido(-v) “dress, garment” (1)
xaloh (var. of jairo) “jar” (1)
yunta “yoke of oxen” (10)
zapatos “shoes” (1)
3. PHYSICAL COMPLEXES/BUILDINGS
calle(h) “street” (2)
canetera “highway” (2)
c&cel “ jail, prison” (3)
cementerio “cemetery” (2)
cocina “kitchen” (1)
corral “corral” (1)
cuartito “room” (1)
cuarto “room” (1)
establo “stable” (1)
fabricah “factory” (3)
hospital “hospital” (1)
iglesia “church” (3)
juzgado(h) “court” (27)
mercado(h) “market” (14)
oficio “office, position” (9)
pasillo “corridor” (1)
patioh(-p) “patio” (2)
plaza “plaza, town square” (3)
pozo “well, shaft” (1)
seccion “section of town” (14)
temploh “temple” (1)
tienda(h) “shop” (19)
troje “bam, granary” (1)
xahuen (from jagiiey) “reservoir,
pool” (1)
zanja “ditch” (2)
4. OTHER
cosa(h) “thing” (20)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
i n — NOUNS: QUASI-CONCRETE
1. INDIVID U ALS
agente “agente” (2)
albanil “mason” (2)
arrieros “muleteef” (1)
barboncito “bearded person” (1)
-boyeroh “cowsherd” (4)
campesino “peasant, farmer” (1)
campista “herdsman” (2)
candidato “candidate” 1)
cantor “singer” (3)
carcelenos “prisoners” (1)
catequistas “catechists” (4)
comercianteh “merchant” (2)
costurah “seamstress, sewing” (2)
criada“maid” (l)
cuatrero “horse-thief ’ (3)
-cuchilero “cutler” (1)
cura(h) “priest, curate” (4)
Dios “God” (27)
director “director” (1)
doctor “doctor” (2)
doctorah “doctor, f.” (2)
eclesiastico “eclesiastic” (1)
escribano “secretary” (2)
figurah “figure” (2)
-fiscal “church steward” (9)
gobemador “governor” (2)
guardah “guard” (1)
huerah “blond” (1)
inocente “innocent” (2)
jomalero “journalist” (1)
juez “judge” (1)
juntah “council” (1)
-lavandera “laundress” (2)
lepero “leper ?” (1)
limosnero “beggar” (1)
maestra “teacher, f.” (1)
maestroh “teacher, m.” (10)
maldades “wicked persons” (1)
malditos “accursed persons” (1)
marchanteh “merchant” (1)
mayor “mayor” (1)
mayora “mayor, f., (1)
mayordomo “steward, major-
domo” (6)
mexicanera (Mex. speaker) (9)
mexicanistas (pro Mexicano) (6)
musico “musician” (1)
obreroh “worker” (1)
paisanos “countrymen” (1)
pistolero “gunman, gangster” (1)
pixcal (var. of fiscal) (2)
-politica “politcal person” (3)
porteria “work of a porter” (1)
portrero “doorman, porter” (2)
presidenta “president, f.” (1)
presidente(h) “president, m.” (2)
presos “prisoners” (1)
profesora- “professor, f.” (2)
revolucionarioh “revolutionary”
(3)
sancristan “sacristan” (1)
senoritas “young ladies” (1)
servidor “server” (1)
terrenos “lands, terrain” (1)
trabajador “worker” (1)
-trabajo “work, job” (29)
-vecinoh- “neighbor” (22)
vigueroh “one who works with
beams, lumber” (2)
zapatistas “poltical group named
after Zapata” (3)
2. ORGANIZATIONS AND INSTITUTIONS
colegio “high school” (3)
escuela(h) “school” (47)
gradoh “grade level” (1)
hermandad “brotherhood” (4)
mayordomia “estate” (1)
primaria “primary school” (3)
regelion “religion” (1)
secundaria “secondary school” (1)
3. PLACES
barrio “neighborhood” (13)
campo(h) “field” (17)
cerro “hiU” (1)
ciudan/h “city” (5)
estado “state” (1)
loma “small hill” (1)
monte “mount, mountain” (1)
mundo “world” (1)
-pueblo(h) “town” (181)
vecindad “neighborhood” (1)
258
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
IV— PROPER NOUNS
l£L INDIVIDUALS
AbraMn (1)
AbraMn Sanchez (2)
Alberto Zepeda Serrano (3)
Albino Lunah (4)
Antonio (1)
Antonio Corte (1)
Aparicio (1)
Ascensidn Manzana (1)
Carlos (2)
Carmentzinco (1)
Camacidnah (2)
Chucha (1)
Concepcidn (1)
Cruz (7)
Dominguez (2)
Esperanza(l)
Eufemia (1)
Eufemia Rojas (1)
Eugenio Zepeda (1)
Faustino (1)
Felipeh (5)
Fidencia Flores (1)
Fidencio (2)
Florencio (1)
Franciscah (1)
Grabiel (2)
Graciela (1)
Guadalupena (1)
Gutierrez (1)
Hemdn Cortez (1)
Ismael (1)
Jesus (2)
Josefa (1)
Josd (2)
Juan (1)
Juan Luna (1)
Juan Rold£n (1)
Juana (2)
Juanita (2)
Leonardo Rojas (1)
Luis (1)
Lunah (1)
Lupe (1)
Manuel (1)
Marcela (1)
Mariana Meza (1)
Maria (3)
Martinez (1)
Mdximo (2)
Miguel (2)
Milio (1)
Murilleroh (1)
Pedro (4)
Pedro Flores (2)
Pepencita(l)
Perez (1)
Petra Serrano (1)
Polonia (3)
Ramos (4)
Reyes Sanchez (1)
Rosah (3)
Rosalia (1)
Rosalia Sinchez (1)
Rosario (1)
Slnchez(3)
Senor Santiago (1)
Silvia (1)
Silvia de Gutierrez (1)
Teodora (1)
Tiburcia (2)
Torrris (1)
Vicente (1)
Victor (1)
Yolanda (1)
lb. RELIGIOUS FUNCTIONARIES
(in) Senor de Canoa “the Lord of
Canoa” (1)
(Serior de San Isidro “the Lord of
San Isidro” (1)
(in) Senor de San Pablo “the Lord
of San Pablo” (1)
El senor de quinto viemes “the
Lord of the Fifth Friday (of
Lent)” (1)
Jesus Santa Mariahtzin “Jesus and
Holy Mary” (1)
Judas (1)
La Purisma “the Blessed Virgin
(i.e., Mary, mother of Jesus
Christ)” (1)
Marfa de Pilar “Mary of Pilar” (1)
Maria Santisma “Most Holy
Mary” (2)
Sagrado Corazon “Sacred Heart
(ofJesus)” (1)
259
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Santisma Trinidad “Holy Trinity”
(1)
Virgen “the Virgen (Mary)” (1)
Virgen de Carmen “Virgen of
Carmen” (3)
Virgen de Guadalupe “the Virgin
of Guadalupe” (7)
(in) Virgen del Rosario “the
Virgin of the Rosary” (1)
Virguemaria “Virgin Mary” (1)
2. ORGANIZATIONS
Accidn Catdlica “Catholic Action
(association)” (4)
Associacidn “Association” (4)
Associacidn Accidn Catdlica
“Catholic Action Association”
(1)
Federacidn “Federation” (1)
3. PLACES
del Monte “(San Pablo) of the
Mountain” (1)
espana “Spain” (1)
Estados Unidos “the United
States” (3)
Madalenah (4)
Munoztla (1)
Puebla(h) (15)
San Antonito (5)
San Bartoloh (4)
San Bemaldino (2)
San Cosme (4)
San Crisol (1)
San Diego (1)
San Estabdn (2)
San Francisco (14)
San Isidro (13)
San Juan (1)
San Juaneros (1)
San Luis (11)
San Marco (4)
San Martin (2)
San Miguel (9)
San Miguel Canoa (2)
San Pablito (1)
San Pablo del Monte (7)
San Pedreiios (2)
San Pedritoh (1)
San Pedro (5)
San Simdn (2)
Santa Ana (5)
Santa Catarinah (5)
Santa Cruz (5)
Santa Inds (1)
Santa Isabel (3)
Santa Marfa (3)
Santiago(h) (4)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
V— a b s t r a c t n o u n s
1. GENERAL
carga “load” (2)
cargo(h) “charge, duty” (18)
caso “case, instance” (3)
contenidoh(-c) “contents” (1)
contrario “contrary” (2)
eje(m)plo “example” (8)
estdo(h) “style” (1)
idea “idea” (1)
juventud “youth” (1)
limpieza “cleanliness” (2)
-modo “way, style” (12)
punto/a “point” (2)
suerte “luck” (1)
2. RELIGIOUS
bautismo “baptism” (2)
confesidn (w/var. confisidn)
“confession” (2)
doctrina “doctrine, teaching” (2)
i€ “faith” (1)
milagroh “miracle” (5)
misah “mass, church service” (3)
-sacramentado “sacrament” (1)
-sacramento “sacrament” (1)
santo- “saint, holy person” (1)
3. LEGAL
causa “case, cause” (1)
justicia “justice” (5)
ley “law” (4)
pleito(h) “lawsuit, case” (4)
-pleitista “plaintiff, quarelsome
person” (1)
registro “registration” (1)
seguro “insurance” (1)
4. OTHER CULTURAL
costumbre “custom” (1)
5. M EASUREM ENTS
almon (from almud) “grain
measure (5)
anterior- “front, previous” (3)
ano “year” (12)
base “base, foundation” (1)
centavito “cent, part of peso” (1)
centavo(h) “cent” (12)
clase “class, kind” (1)
cobro “charges, price” (1)
dia “day” (4)
dinero “money” (1)
domingo “Sunday” (1)
edad “age” (9)
falta(h) “lack” (11)
fecha“date” (2)
fin “end” (1)
gastoh “expense” (1)
hora(h) “hour” (13)
igual “equal, the same” (1)
julio “July” (1)
junio “June” (1)
kiloh “kilogram” (6)
lado(h) “side” (8)
litroh “liter” (1)
lugar- “place” (10)
lugarcito “place” (1)
lultimo “last one, thing” (4)
lunes “Monday” (1)
mayor(-m) “major part” (3)
mayoria “majority” (5)
mdrtes ‘Tuesday” (1)
medio “half, means” (5)
parte(h) “part” (12)
pasado “past” (1)
pedacito “piece” (1)
pedazo “piece” (1)
pesoh “monetary unit” (8)
precio “price” (1)
presente “present” (1)
principio “beginning, source” (1)
rato “short time, while” (2)
rumboh “direction, route” (1)
semana “week” (4)
sur “south” (2)
tarde “afternoon, late” (1)
tiempo(h) “time” (39)
tonun (a coin) “money” (4)
veces “times, occasions” (2)
6. STATES (&PROPERTIES
alegria “joy, mirth” (1)
ansias “anxiety” (1)
borrachera “drunkenness” (1)
esperanza “hope” (1)
gratia “grace” (2)
gustoh “pleasure” (3)
261
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
necesidad (w/var. necesidan)
“necessity” (3)
novedad- “novelty” (2)
perjuicio “prejudice” (1)
picaidfa “mischievousness” (1)
respeto “respect” (3)
sabiduria “wisdom” (1)
tontito “foolishness” (1)
verguenzah “shame” (1)
vicio(-v) “vice, defect” (2)
7. ACTIVITIES:
accidente(h) “accident” (15)
acuerdo “accord, agreement” (2)
borracho- “drunk” (1)
briagos “drunks” (1)
cooperacidn “cooperation” (1)
cuidado “care, charge” (2)
cuidar “care, charge” (1)
-demanda “demand” (2)
dificultad “difficulty” (2)
educacidn “educadon” (2)
eleccidn “election” (2)
estudio(h) “study” (9)
favor “favor” (1)
gana “appetite, desire, will” (1)
guerra “war” (1)
-herencia “inheritance” (1)
-luchah “fight, battle” (1)
mandado “order, command” (1)
movimiento “movement” (1)
nacimiento “birth” (1)
patada “kick” (1)
probeza “trial” (1)
razdn “reason” (2)
recaudo “safeguard, precaution”
(1)
revolucidn “revolution” (3)
risa “laughter” (1)
traguitoh “gulp” (1)
visitah “visit” (2)
votacidn “vote” (2)
vueltah “turn, return” (1)
claraci6n(-c) “claim” (1)
cuenta(h) “account, story” (2)
cuento(h) “story, tale” (19)
cuesti6n “question, query” (1)
espanol “Spanish” (6)
grosena “rude remark” (3)
historias “history, account” (3)
idioma “language” (31)
Inglds “English” (4)
intonacidn “intonation” (1)
letrah “letter (of alphabet)” (3)
leyenda “legend” (4)
liccidn “lesson” (var. of leccidn)
(1)
mentxrah “lie, falsehood” (2)
nombramiento “naming,
nomination” (1)
nombre “name” (1)
oracion- “prayer, speech” (1)
palabra(h) “word” (20)
perddn “pardon, forgiveness” (1)
platica(h) “conversation” (3)
repaso “revision, re-examination”
( 1)
senas “signs, gestures” (2)
-tono “inflection, tone” (2)
-voz “voice” (2)
8. SPEECH ACTS:
burla “mockery, jest (2)
calunia “slandef” (2)
castella (var. of castilla” (1)
castellano(h) “Castilian Spanish”
( 100)
castilla(h) “Spanish” (49)
castillano (var. of castellano) (27)
262
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
VI— V e r b s
l . F u l l y in c o r po r a t e d verbs
-adivinaroah “divine, guess” (5)
-aguantar- “endure” (1)
-apaciguar- “pacify” (1)
-apenar- “grieve, afflict” (1)
-apuraroa “exhaust, drain” (6)
-arreglar- “arrange, put in order”
(3)
-atender- “understand” (1)
-cambiar- “change” (7)
-castigar- “punish” (2)
-cayer- “fall” (1)
-cenaroah “have dinner” (2)
-chotear- “fool around” (1)
-coliaroa “shake, wag” (1)
-comprender- “understand” (2)
-confesar- “confess” (1)
-consentiraoa “consent, agree” (2)
-contar- “count, recount” (1)
-contestar- “ask a question” (1)
-convenir- “agree, assemble” (1)
-costumbrar “become accustomed
to” (w/REFL) (2)
-crer- “believe” (1)
-cuatrear- “make mistakes” (2)
-cumplir- “fulfill, cany out” (1)
-defender- “defend” (2)
-depender- “depend” (1)
-desconoceroah “fail to recognize”
0 )
-despensar- “grant, distribute” (1)
-dirihgiroh “direct, regulate” (1)
-encamar- “be put in bed” (1)
-encontrar- “meet, find” (1)
-estudiar- “study” (24)
-invitar- “invite” (2)
-justicah- “judge, condemn” (1)
-juzgar- “judge, decide” (1)
-leer- “read” (1)
-linih- (from llenar) “fill” (1)
-maldecir- “speak ill of s.o.” (1)
-mandar- “order, send” (2)
-mencionar “mention” (1)
-negaroz “deny, refuse” (1)
-nes(z)tar- (from necesitar) “need,
be necessary” (2)
-nombrar- “name, nominate” (1)
-ntrigor- (from entregar)
“surrender, deliver” (1)
-obligar “compel, obligate” (2)
-ofrecer- “offer, promise” (1)
-osar- (from usar) “use” (2)
-padecer- (from parecer) “seem”
(3)
-paroa “stop, stand” (1)
-paxialo- (from pasar) “happen,
pass” (1)
-pensar- “think” (3)
-perdonar- “pardon, forgive” (1)
-persiguir “persecute, pursue” (1)
-planchar- “iron, smooth out” (1)
-quehjar- “complain” (1)
-quivocaroa “be miskaken” (2)
-recibiroa “receive” (1)
-reclamaroa “complain, protest”
(4)
-reganaroah “grumble” (1)
-rehgistrar- “register, inspect” (1)
-reinar “rule, prevail” (1)
-resolveroa “resolve, decide” (1)
-respetaroh “respect” (1)
-responder “answer, respond” (1)
-revolver- “revolve, mix up” (1)
-rezar- “pray, recite” (12)
-salvaroz “save (rescue)” (1)
-senelar- “point out, signal” (1)
-serviroa “serve, be of use” (1)
-su(h)frir- “suffer, allow” (5)
-suhjetar “subject, subdue” (1)
-tender- “understand” (58)
-tocar- “touch, play (musical
instrument)” (3)
-tomar- “drink, take” (1)
-torearoah (from corretear) “to
chase” (1)
-tratar- “treat, handle” (1)
-vivir- “Uve” (25)
-votar- “vote” (25)
-zpiraroh (fromespiar) “spy” (1)
2. Miscellaneous
depende (var. of -depender)
“depends” (1)
es que “it is that” (4)
parece (var. of -padecer-) “seems”
(3)
263
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
V II— ADJECTIVES
acostumbrados (1) mejor “better” (21)
ajena “another’s, alien” (2) -mero “real, true” (12)
anciano “ancient” (1) mezclado(h) “mixed” (1)
antigua/o “ancient, old” (4) mismo “same” (16)
atrasado “backward, slow” (1) modemo “modem” (1)
baraturah “cheap, inexpensive” nuevo “new” (2)
(2)
ordenada “milked” (1)
borrachito “drunk” (1) pare(h)jo “both, same” (6)
borracho(h) “drunk, intoxicated” pior (var. of peor) “worse” (1)
(3) -pobre “poor” (5)
bueno “good” (8) poco “little bit” (4)
capaz “capable” (1) poquito “very little bit” (1)
cargadoh “loaded (with)” (1) posible “possible” (1)
cayendo “falling” (1) practicado “practiced” (2)
ceicano “near, neighboring” (1) preferible “preferible” (2)
civilizados “civilized” (1) preparado “prepared” (1)
clemente “merciful” (1) prohibido “prohibited” (1)
colado “strained, sifted” (1) -proprioh “proper, one’s own”
completoh “complete” (2) (2)
contento(h) “happy” (13) puro “pure” (64)
correcto “correct” (1) raro “rare” (2)
criada “created, made” (1) rempujada “shoved, pushed” (1)
cualquier “whatever” (3) respetoso “respectful” (1)
cumplimiento “fulfilled, rado “rude, coarse” (1)
completed” (1) santa “holy” (2)
dicho “said” (1) solo “only” (26)
diferente “different” (3) suelta “paralyzed” (1)
divino “divine” (2) tonto “stupid, foolish” (2)
feliz “happy, lucky” (2) tranquilo “peaceful, quiet” (1)
feo “ugly” (3) tristeh “sad” (3)
finado “deceased” (1) util “useful” (2)
grosero “rude, impolite” (3) verdadero “real, true” (1)
huero “blond” (2) verde “green” (1)
igual “same, equal” (52)
impuesto “accustomed to” (2)
jodido “ruined” (1)
joven “young” (1)
juerteh (from fuerte) “strong” (3)
legitimo “legitimate, true” (5)
libre “free, liberated” (1)
listo “ready” (2)
maloh “bad” (5)
mayor “major” (1)
mediano2 “middle” (1)
medio/a “half, middle” (4)
2 Spoken by S49, a female, in reference to
herself; this illustrates the lack of gender
agreement in borrowed forms.
264
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
v m — a d v e r b i a ls
abajo “below” (1)
adentro “within” (1)
ahf “there” (1)
ahorita “now” (40)
algo “somewhat” (3)
alld “there” (3)
anteriormente “previously” (3)
antes “before” (12)
aparte “apart, aside” (2)
apenas “hardly, barely” (1)
apoco “hardly, unlikely” (4)
bastante “a lot, plenty” (2)
bien “well, good” (4)
cazi (from casi) “almost” (11)
cerca “around, nearby” (11)
ciertamente “certainly” (1)
comoquiera “however” (8)
completamente “completely” (1)
debajo “underneath” (1)
despuds “afterwards” (3)
dondequiera “wherever” (3)
entonces (w/var. tonces, tonz,
toz) “then, at that time” (20)
exactamente “exactly” (1)
fueras “outside” (1)
fuerza “hardly” (3)
mds “more” (44)
menos “less” (4)
muy “very” (13)
no “no, not” (75)
principalmente “principally (1)
pronto “soon, quickly” (3)
(de) repente “suddenly” (10)
siempre “always” (13)
sf “yes” (4)
solamente “only” (14)
solo “only” (11)
tal vez “perhaps” (2)
tambidn “also, as well” (43)
tampoco “either” (3)
tanto “so much” (9)
(al) tiro “a lot” (2)
todavia“yet” (8)
unicamente “only” (9)
ya “already” (2)
265
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
DC— INDEPENDENT FUNCTION WORDS AND AFFIXES
la. PREPOSITIONS (optional)
a “to, at” (17)
antes “before” (16)
cerca(h) “near” (9)
con “with” (5)
de (w / var. den3 )“of, from”
(1,273)
desde “since, from” (10)
despues “after” (7)
en “in, on” (28)
entre “between” (1)
hasta “until, towards, until” (56)
para “for, towards” (179)
por “for, by, through” (65)
segun “according to” (4)
sin “without” (3)
lb . Prepostions (obligatory)
(cargo) de (3)
(cerca) de (2)
(despuds) de (1)
(fin) de (1)
(segun) den (1)
2. COORDINATING CONJUNCTIONS
nion (from ni aun) “not even,
neither” (28)
o “or” (200)
pero “but” (289)
y “and” (9)
3. Subordinating conjunctions
como “like, how” (52)
cual “which” (1)
cuando “when” (76)
donde “where” (4)
lo que “that which” (1)
mas “though, even though” (22)
mazcamo (mazqui + amo)
“though not” (1)
mazque/i (from mas que) “even
though, though” (7)
para (pa’) que “so that” (6)
porque “because” (97)
que (complementizer) “that” (75)
quien “who” (3)
3 The variant den appears to be the Spanish
preposition de with cliticized Mexicano in (DEF).
quera (var. o f siquiera) “if only,
even i f ’ (2)
si “i f ’ (15)
siquiera “if only, even i f ’ (2)
4. MISCELLANEOUS
cada (DET) “each” (16)
eso (DET) “that” (9)
nada (PRO) “nothing” (6)
toda (DET) “all” (7)
5. CONVERSATIONAL PARTICLES (by
frequency)
pues (w. var. poz) “well” (392)
este (lit. “this”) “uh” (as hesitation
form) (79)
bueno “well” (43)
vaya(h) (pres, subjunctive form
o f ir “to go”) “I mean...” (25)
£ndale(h) “absolutely” (21)
aver (from a ver, lit. “to see”)
“let’s see, hmm” (20)
verda(d) “true/truth” (19)
claro(h) “o f course, clear” (11)
orale “right on!” (2)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
8. N u m e r a l s
cinco “five” (1)
cincuenta “fifty” (3)
cincuenta y nueve “twenty-five”
(1)
cuarenta “forty” (1)
cuarenta y cinco “forty-five” (1)
cuatro “four” (12)
diez y seis “sixteen” (4)
doce “twelve” (6)
dos “two” (1)
dosientos “two hundred” (1)
nueve "nine” (3)
ochenta “eighty” (2)
ochenta y cinco “eighty-four” (1)
ocho “eight” (4)
primero “first” (3)
quince “fifteen” (1)
quinto “fifth” (1)
segunda “second” (3
sesenta “seventy” (1)
sesenta y dos “sixty-two” (1)
sesenta y seis “sixty-six (2)
setenta y cinco “seventy-five” (1)
sexta “sixth” (4)
trece “thirteen” (1)
treinta seis “thiry-six” (1)
treinta “thirty” (1)
treinta y nueve “thirty-nine” (1)
tres “three” (3)
veinte “twenty” (2)
veinteicinco “twenty-five” (3)
9a. AFFIXES: DERIVATIONAL (by
frequency)
-ito “DIM and H O N ” (118)
-a/ido “past participial suffix used
to derive Adj” (32))
-ero4 “agentive” (24)
-mente “derives Adv from Adj—
akin to Eng. -/y” (23)
(c)idn “derives N from V” (18)
-ista “agentive” (12)
-dad “derives N from types of
Adj—akin to Eng. -fry” (10)
-a/ente “derives N from types of
Adj” (6)
-miento “N from types of V—
akin to Engl, -m enf' (5)
-ador “agentive” (4)
-arioh “derives Adj or N—akin to
Eng. -ary” (3)
-efios “agentive” (2)
-ible “derives Adj from V” (2)
-oso “agentive” (1)
en- “causative” (1)
9B. AFFIXES: INFLECTIONAL
-S “PLU” (164)
10. M ISCELLANEOUS P H R A SE S (not
included in counts of individual items)
adids “goodbye” (1)
jarreh! “rise (as from the dead)”
(3)
buenos dias “hello, good day” (7)
buenos tardes “hello, good
afternoon” (4)
i,como no? “why not” (19)
de acuerdo “in agreement” (1)
de veras “truly, really” (7)
mtis o menos “more or less” (2)
jojald! (lit. Would to Allah!”)
“God grant!; I hope so!” (2)
por eso “for that reason” (13)
por ejemplo “for example” (9)
^quien sabe? “who knows?” (5)
4 Also appears on native Mexicano forms, e.g..
cuah-tero “wood-cutter”.
267
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
X— w o r d C o u n t s b y v il l a g e
1. Ayometitla
S42 661
S43 1,624
S44 1,285
S45 727
S46 1,301
sub (1) 5,598
2. Acuamanala
S47 915
S48 550
S49 732
S50 1,139
S51 988
S52 868
sub (2) 5,192
10,790
3. Acxotla
S74 988
S75 1,522
S76 3,487
S77 2,057
S78 3,852
S79 576
sub (3) 12,482
10.790
23,272
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
REFERENCES
Aarsleff, Hans. 1972. From Locke to Saussure: Essays on the Study o f Language and
Intellectual History. Minneapolis, MN: University of Minnesota Press.
Albd, Xavier. 1970. Social constraints on Cochambamba Quechua. Doctoral dissertation,
Cornell University.
Alleyne, Mervyn C. 1980. “Introduction: Theoretical Orientations in Creole Studies.” In
Theoretical Orientations in Creole Studies. Albert Valdman and Arnold Highfield
(Eds.). New York: Academic Press.
Alleyne, Mervyn C. 1986. “Substratum Influences: Guilty until Proven Innocent” In
Substrata versus universals in Creole languages. Eds. Pieter Muysken and Norbert
Smith. Amsterdam: John Benjamins. Pp. 301-315.
Andersen, Roger W. 1989. “The “up” and “down” staircase in secondary language
development” In Investigating obsolescence: Studies in language contraction and
death. Nancy C. Dorian (Ed.). Cambridge: Cambridge University Press. 385—
394.
Anderson, Stephen R. 1992. A-Morphous Morphology. Cambridge University Press.
Andrews, J. Richard. 1975. Introduction to Classical Ndhuatl. Austin: University of
Texas Press.
Ard, Josh & Taco Homberg. 1993. “Verification of Language Transfer.” In Language
Transfer in Language Learning. Susan M. Gass & Larry Selinker (Eds.).
Amsterdam and Philadelphia: John Benjamins. 47— 70.
Arenas, Pedro de. 1611. Vocabulario manual de las lenguas caste liana y mexicana (edition
by Charles Romey, Paris, 1862).
Aronoff, Mark. 1994. Morphology by Itself. Cambridge, Mass.: The MIT Press.
Bakker, Peter and Maarten Mous (Eds.). 1994a. Mixed Languages: 15 Case Studies in
Language Intertwining. Amsterdam: IFOTT.
Bakker, Peter and Maarten Mous. 1994b. “Introduction.” In Peter Bakker and Maarten
Mous (Eds.), Mixed Languages: 15 Case Studies in Language Intertwining.
Amsterdam: IFOTT.
Bakker, Peter and Robert A. Papen. “Michif: A Mixed Language Based on Cree and
French. In Contact Languages: A Wider Perspective. Sarah G. Thomason (Ed.).
New York and Amsterdam: John Benjamins, pp. 295— 363.
Bakker, Peter. 1988. “Relexification: The case of Mdtif (French-Cree).” Vielfalt der
Kontakte: Beitrage zum 5. Essener Kolloquium uber “Grammatikalisierung:
Naturlichkeit und Systemomkonomie”. Essen: Universitatsverlag Dr. N.
Brockmeyer. 119— 137.
269
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Bakker, Peter. 1994. “Michif, the Cree-French mixed language of the Mdtis buffalo
hunters in Canada.” In Mixed Languages: 15 Case Studies in Language
Intertwining. Eds. Peter Bakker and Maarten Mous. Amsterdam: IFOTT. 13-33.
Bakker, Peter. 1997. A Language o f Our Own: The Genesis o f Michif, the M ixed Cree-
French Language o f the Canadian M itis. Oxford: Oxford University Press.
Bakker, Peter. 1998. “Some future challenges for pidgin and creole studies,” paper
presented January 9,1998 at the annual joint meeting of the Society of Pidgin and
Creole Linguistics with the Linguistic Society of America (LSA) at the Grand Hyatt
in New York, N.Y.
Bavin, Edith. “Some lexical and morphological changes in Warlpiri.” Investigating
obsolescence: Studies in language contraction and death. Ed. Nancy C. Dorian.
Cambridge: Cambridge UP. 267-286.
Beard, R.. 1981. The Indo-European Lexicon: A Full Synchronic Theory. Amsterdam:
North Holland.
Berdan, Frances F. 1982. The Aztecs o f Central Mexico: An Imperial Society. Case
Studies in Cultural Anthropology. New York: Harcourt Brace.
Bickerton, Derek. 1981. Roots o f Language. Ann Arbor: Karoma.
Bickerton, Derek. 1984. “The language bioprogram hypothesis.” In The Behavioral and
Brain Sciences 7: 173— 221.
Bickerton, Derek. 1987. “The language bioprogram hypothesis and second language
acquisition. In Language Universals and Second Language Acquisition. William
Rutherford (Ed.). Amsterdam & Philadelphia: John Benjamins.
Bierwisch, Manfred and Robert Schreuder. 1992. “From Concepts to Names.” In
Lexical Access in Speech Production. Willem J. M. Levelt (Ed.). Cambridge,
MA: Blackwell. Pp. 23-60.
Blake, Barry J. 1994. Case. Cambridge: Cambridge University Press.
Bloomfield, Leonard. 1933. Language. New York: Henry Holt.
Boretzky, Norbert and Birgit Igla. 1994. “Romani Mixed Dialects.” In M ixed
Languages: 15 Case Studies in Language Intertwining. Peter Bakker and Maarten
Mous (Eds). Amsterdam: IFOTT. pp. 35-68.
Butterworth, Brian. 1992. “Disorders of Phonological Encoding.” In Lexical Access in
Speech Production. Willem J. M. Levelt (Ed.). Cambridge, MA: Blackwell. Pp.
Pp. 261-286.
By bee, Joan and William Pagliuca. 1985. “Cross-linguistic comparison an the
development of grammatical meaning. In Historical Semantics and Historical Word
Formation, Jacek Fisiak (Ed.). Berlin: deGruyter.
Bybee, Joan. 1985. Morphology. Amsterdam and Philadelphia: John Benjamins.
270
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Bybee, Joan. 1995. “Diachronic and typological properties of morphology and their
implications for representation.” In Morphological aspects o f language processing.
Ed. Laurie Beth Feldman. Hillsdale, NJ: Lawrence Erlbaum Associates.
Bynon, Theodora. 1977. Historical Linguistics. New York: Cambridge University
Press.
Campbell, Lyle and Martha C. Muntzel. 1989. “The structural consequences of language
death.” Investigating obsolescence: Studies in language contraction and death.
Ed. Nancy C. Dorian. Cambridge: Cambridge University Press. 197-210.
Caplan, David. 1992. Language: structure, processing, and disorders. Cambridge,
Mass.: The MIT Press.
Carr, Phillip. 1993. Phonology. New York: St. Martin’s Press.
Carstairs-McCarthy, Andrew. 1992. Current Morphology. New York, N.Y.:
Routledge.
Choi, Soonja and Alison Gopnik. 1993. “Nouns Are Not Always Learned Before Verbs:
An Early Verb Spurt in Korean.” The Proceedings o f the Twenty-fifth annual child
Language Research Forum. Ed. Eve Clark. Stanford University: Center for the
Study of Language and Information. 96-105.
Clark, Eve V. 1993. The Lexicon in Acquisition. Cambridge: Cambridge University
Press.
Clark, Eve V. 1997. “Conceptual Perspective and Lexical Choice in Acquisition.”
Cognition.
Comrie, Bernard. 1976. Aspect. Cambridge: Cambridge University Press.
Comrie, Bernard. 1985. Tense. Cambridge: Cambridge University Press.
Comrie, Bernard. 1989. Language Universals and Linguistic Typology. Chicago:
University of Chicago Press.
Corbett, Greville. 1991. Gender. New York: Cambridge University Press.
Corder, S.PiL 1993. “A Role for Mother Tongue.” la Language Transfer in Language
Learning. Susan M. Gass & Larry Selinker (Eds.). Amsterdam and Philadelphia:
John Benjamins. 18-31.
Croft, William. 1990. Typology and Universals. Cambridge: Cambridge University
Press.
Crystal, David. 1991. A Dictionary o f Linguistics and Phonetics. London: Blackwell.
Cutler, Anne. 1989. “Auditory Lexical Access: Where Do We Start?” In W. D. Marlsen-
Wilson (Ed.), Lexical representation and process. Cambridge, Mass.: The MIT
Press. 342-356.
271
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Cutler, Anne. 1994. “Segmentation problems, rhythmic solutions.” In L. Gleitman and
B. Landau (Eds), The Acquisition o f the Lexicon. Cambridge, Mass.: The MIT
Press. 81-104.
De Groot, Annette M. B. 1992. “Determinants of word translation.” Journal o f
Experimental Psychology: Learning, Memory, and Cognition, 1 8 ,1001— 1008,
De Groot, Annette M.B. 1993. “Word-Type Effects in Bilingual Processing Tasks:
Support for a Mixed-Representational System. In The Bilingual Lexicon. Robert
Schreuder & Bert Weltens (Eds.). Amsterdam and Philadelphia: John Benjamins.
27-52.
De Gruiter, Miel. 1994. “Javindo, a contact language in pre-war Semarang.” Mixed
Languages: 15 Case Studies in Language Intertwining. Eds. Peter Bakker and
Maarten Mous. Amsterdam: IFOTT. 151— 9.
Di Sciullo, Anna Maria and Edwin Williams. 1987. On the Definition o f Word.
Cambridge, Mass.: The MIT Press.
Dixon, Robert M. W. 1991. A New Approach to English Grammar, on Semantic
Principles. New York: Oxford University Press.
Dorian, Nancy. 1989. “Introduction.” In Investigating obsolescence: Studies in language
contraction and death. Nancy C. Dorian (Ed.). Cambridge: Cambridge University
Press. Pp. 1-10.
Field, Fredric. 1994a. “Caught in the Middle: the Case of Pocho and the Mixed
Language Continuum.” In General Linguistics, Vol. 34, No. 2. Binghamton,
N.Y.: Medieval and Renaissance Texts & Studies, State University of New York..
Field, Fredric. 1994b. “Implicational Universals and the Origins of Mixed-codes: When
Languages Combine.” Presented at the 10th Biennial Conference of the Society for
Caribbean Linguistics/Society for Pidgin & Creole Linguistics at the University of
Guyana, Turkeyen, August 26, 1994.
Field, Fredric. 1995. “When Languages Combine: Morphology and Language Mixing.”
Paper presented at the Annual Joint Meeting of the Linguistic Society of America
(LSA) and The Society for Pidgin & Creole Linguistics (SPCL) in New Orleans,
Louisiana, on January 7,1995.
Field, Fredric. 1997a. Review of Peter Bakker and Maarten Mous (eds.), Mixed
Languages: 15 Case Studies in Language Intertwining (Amsterdam: IFOTT) In
Journal o f Pidgin and Creole Languages, Vol. 12:2,1997.
Field, Fredric. 1997b. “Mixed systems: the determining role of a matrix (substrate)
system,” paper presented January 10,1997 at the annual joint meeting of the
Society of Pidgin and Creole Linguistics with the Linguistic Society of America
(LSA) at the Sheraton Chicago Hotel and Towers in Chicago, IL.
Field, Fredric. 1997c. “Patterns in language mixing: the effects of a matrix system,”
paper presented June 28,1997 at the annual meeting of the SPCL at the University
of Westminister in London, England.
272
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Field, Fredric. 1998. “Revealing contrasts: Function words and inflectional categories in
modem Mexicano and Palenquero,” paper presented January 10,1998 at the annual
joint meeting of the Society of Pidgin and Creole Linguistics with the Linguistic
Society of America (LSA) at the Grand Hyatt in New York, N.Y.
Fisher, Cynthia, D. Geoffrey Hall, Susan Rakowitz, and Lila Gleitman. 1994. “When it
is better to receive than to give: Syntactic and perceptual restraints on vocabulary
growth.” In The Acquisition o f the Lexicon. Lila Gleitman and Barbara Landau
(Eds.). Cambridge Mass.: the MIT Press. 333— 75.
Fortescue, Michael & Lise Lennert Olsen. 1992. “The Acquisition of West Greenlandic.”
In The Crosslinguistic Study o f Language Acquisition. Vol. 3. Dan I. Slobin
(Ed.). Hillsdale, N.J.: Lawrence Erlbaum. 111-219.
Frauenfelder, U. H. and R. Schreuder. 1992. “Constraining Psycholinguistic Models of
Morphological Processing and Representation.” Yearbook o f Morphology 1991.
G. Booij and J. van Marie (Ed.). The Netherlands: Kluwer.
Frazier, Lyn & Charles Clifton, Jr. 1996. Construal. Cambridge, Mass.: The MIT
Press.
Friederici, Angela. 1985. “Levels of processing and vocabulary types: Evidence from on
line comprehension in normals and agrammatics.” Cognition, 19,133-66.
Gal, Susan. 1989. “Lexical innovation and loss: The use and value of restricted
Hungarian.” Investigating obsolescence: Studies in language contraction and
death. Ed. Nancy C. Dorian. Cambridge: Cambridge Univ. Press. 313-31.
Gass, Susan M. & Larry Selinker. 1993. “Afterword.” In Language Transfer in
Language Learning. Susan M. Gass & Larry Selinker (Eds.). Amsterdam and
Philadelphia: John Benjamins. 233-236.
Gass, Susan M. 1995. “Universals, SLA, and language pedagogy: 1984 revisited.” In
The Current State o f Interlanguage: Studies in Honor o f William E. Rutherford.
Lynn Eubank, Larry Selinker, and Michael Sharwood Smith (Eds.). Amsterdam
and Philadelphia: John Benjamins. 31-42.
Gillman, Charles. 1993. Black Identity, Homeostasis, and Survival: African and
Metropolitan Speech Varieties in the New World. In Africanisms in Afro-American
Language Varieties. Ed. Salikoko S. Mufwene. Athens, Ga.: University of
Georgia Press. Pp. 388-402.
Gimate-Welsh, Adrian. 1980. Lenguaje y Sociedad. Puebla, Mexico: Universidad
Autonoma de Puebla.
Gleitman, Lila. 1993. “The Structural Sources of Verb Meanings.” Language
Acquisition: core readings. Ed. Paul Bloom. Cambridge, Mass.: The MIT Press
(First MIT Press edition 1994). 174-221.
Golovko, Evgenij. 1994. “Mednyj Aleut or Copper Island Aleut: an Aieut-Russian mixed
language.” Mixed Languages: 15 Case Studies in Language Intertwining. Eds.
Peter Bakker and Maarten Mous. Amsterdam: IFOTT. 113-21.
273
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Green, David W.. 1993. ‘Towards a Model of L2 Comprehension and Production.” In
The Bilingual Lexicon. Robert Schreuder & Bert Weltens (Eds.). Amsterdam and
Philadelphia: John Benjamins. 249-278.
Greenberg, Joseph H. 1966. Some Universals of Grammar with Particular Reference to
the Order of Meaningful Elements. In Universals o f Language. J. H. Greenberg
(Ed.) Cambridge, Mass.: MIT Press. 73-113.
Greenberg, Joseph H. 1974. Language Typology: A Historical and Analytic Overview.
The Hague: Mouton.
Greenberg, Joseph. 1991. “The last stages of grammatical elements; contractive and
expansive desemanticization.” In Approaches to Grammaticalization, Vol. 1 .
Elizabeth Closs Traugott & Bemd Heine (Eds.). Amsterdam and Philadelphia:
John Benjamins. Pp. 301-314.
Grosjean, Francois and James Paul Gee. 1987. “Prosodic structure and spoken word
recognition.” Cognition, 25: 135-155.
Grosjean, Frangois. 1982. Life with Two Languages: An Introduction to Bilingualism.
Cambridge, Mass.: Harvard University Press.
Grosjean, Frangois. 1995. A Psycholinguistic Approach to Code-switching: the
recognition of guest words by bilinguals. In One speaker, two languages: Cross-
disciplinary perspectives on code-switching. Lesley Milroy and Pieter Muysken
(Eds.). New York: Cambridge University Press.
Grosjean, Frangois. 1997. “Processing Mixed Language: Issues, Findings, and Models.
In Tutorials in Bilingualism: Psycholinguistic Perspectives. Annette M. B. de
Groot & Judith F. Kroll (Eds.). Mahwah, New Jersey: Lawrence Erlbaum
Associates. Pp. .225-254.
Grosjean, Frangois. In press. “The Bilingual Individual.” Interpreting: International
Journal o f Research and Practice in Interpreting.
Gumperz, John J. and Robert Wilson. 1971. Convergence and Creolization: A Case
from the Indo-Aryan/Dravidian Border in India. In Pidginization and Creolization
o f Languages, Dell Hymes (Ed.). 151-69.
Gumperz, John J., (Ed.). 1982. Language and Social Identity. New York, N.Y.:
Cambridge University Press.
Gutknecht, Christoph and Lutz J. Rolle. 1996. Translating by Factors. Albany, N.Y.:
State University of New York Press.
Hall, Christopher. 1988. “Integrating Diachronic and Processing Principles in Explaining
the Suffixing Preference.” In Explaining Language Universals. Cambridge,
Mass.: Blackwell. 321-349.
Harris, John. 1985. Phonological variation and change: studies in Hiberno-English.
New York: Cambridge University Press.
274
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Hartmann, R. R. K. and F. C. Stork. 1972. Dictionary o f Language and Linguistics.
New York and Toronto: John Wiley & Sons.
Haugen, Einar. 1950. “The Analysis of Linguistic Borrowing.” In Language 26: 210-
231.
Haugen, Einar. 1953 (reprinted 1969). The Norwegian language in America: A study in
bilingual behavior. Philadelphia: University of Pennsylvania Press/Bloomington:
Indian University Press.
Haugen, Einar. 1989. “The rise and fall of an immigrant language: Norwegian in
America.” In Investigating obsolescence: studies in language contraction and
death. Nancy Dorian (Ed.). New York: Cambridge University Press.
Hawkins, John A. 1980. “On implicational and distributional universals of word order.
Journal o f Linguistics 16: 193-235.
Hawkins, John A. 1983. Word Order Universals. Philadelphia, Penn.: John Benjamins.
Hawkins, John A. 1988. Explaining Language Universals. Cambridge, Mass.:
Blackwell.
Hawkins, John A. 1990. “Seeking motives for change in typological variation.” In
Studies in Typology and Diachrony: Papers presented to Joseph Greenberg on the
occasion o f his 75 th birthday. William Croft, Keith Denning and Suzanne Kemmer
(Eds.). Amsterdam/Philadelphia: John Benjamins. Pp. 95-128.
Hawkins, John A. 1991. “Language Universals in Relation to Acquisition and Change: a
Tribute to Roman Jakobson.” In New Vistas in Grammar Invariance and Variation.
Linda R. Waugh and Stephen Rudy (Eds.). John Benjamins,
Amsterdam/Philadelphia 473-493.
Hawkins, John A. 1994. A performance theory o f order and constituency. Cambridge:
Cambridge Univ. Press.
Hawkins, John A. and Anne Cutler 1988. “Psycholinguistic Factors in Morphological
Asymmetry.” In Explaining Language Universals. Cambridge, Mass.: Blackwell.
280-317.
Heath, Jeffrey. 1981. “A case of intensive lexical diffusion.” Language 57, 335-367.
Heine, Bemd & Mechthild Reh. 1984. Grammaticalization and Reanalysis in African
Languages. Chicago: University of Chicago Press.
Heine, Bemd, Ulrike Claudi, and Friederike Hunnemeyer. 1991a. Grammaticalization: A
Conceptual Framework. Chicago: University of Chicago Press.
Heine, Bemd, Ulrike Claudi, and Friederike Hunnemeyer. 1991b. “From cognition to
grammar: Evidence from African languages.” In Approaches to
Grammaticalization, Vol. 1. Elizabeth Closs Traugott & Bemd Heine (Eds.).
Amsterdam and Philadelphia: John Benjamins. Pp. 149-187.
275
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Herring, Susan. 1991. “The Grammaticalization of Rhetorical Questions in Tamil.” In
Approaches to Grammaticalization, Vol. 1 . Elizabeth Closs Traugott & Bemd
Heine (Eds.). Amsterdam and Philadelphia: John Benjamins. Pp. 253— 284.
Hill, Jane H. & Kenneth Hill. 1986. Speaking Mexicano: Dynamics o f Syncretic
Language in Central Mexico. Tucson: University of Arizona Press.
Hill, Jane H. & Kenneth Hill. MS. Mexicano Corpus. University of Arizona.
Hill, Jane H. 1988. “Ambivalent language attitudes in modem Nahuatl.” in
Sociolinguistica Latinoamericana. Rainer Engrique Hamel, Yolanda Lastra de
Sudrez, and Hdctor Munoz (Eds.). Mexico City: Universidad Nacional Autonoma
de Mdxico.
Hock, Hans Henrich & Brian D. Joseph. 1996. Language History, Language Change,
and Language Relationship: An Introduction to Historical and Comparative
Linguistics. Berlin and New York: Mouton de Gruyter.
Hoffmann, Charlotte. 1991. An Introduction to Bilingualism. New York, N.Y.:
Longman.
Holm, John. 1988. Theory and Structure. Vol. 1 of Pidgins and Creoles. Cambridge:
Cambridge Univ. Press.
Holm, John. 1989. Reference Survey. Vol. 2. of Pidgins and Creoles. Cambridge:
Cambridge Univ. Press.
Hopper, Paul J. 1987. “Emergent grammar.” In Proceedings o f the Thirteenth Berkeley
Linguistic Society Meeting. Pp. 139-157.
Hopper, Paul J. 1991. “On some principles of grammaticalization.” In Approaches to
Grammaticalization, Vol. 1. Elizabeth Closs Traugott & Bemd Heine (Eds.).
Amsterdam and Philadelphia: John Benjamins. Pp. 17— 36.
Hopper, Paul J. and Elizabeth Closs Traugott 1993. Grammaticalization. Cambridge:
Cambridge University Press.
Hudson, R. A. 1980. Sociolinguistics. Cambridge: Cambridge University Press.
Huffines, Marion Lois. 1989. “Case usage among the Pennsylvania German sectarians
and nonsectarians.” Investigating obsolescence: Studies in language contraction
and death. Ed. Nancy C. Dorian. Cambridge: Cambridge University Press. 211-
226.
Huttenlocher Janellen and Patricia Smiley. 1987. “Early word meanings: The case of
object names.” In Cognitive Psychology, 19: 1987. Reprinted (1993) in
Language Acquisition: core readings. Ed. Paul Bloom. Cambridge, Mass.: The
MIT Press. 222-247.
Jackendoff, Ray S. 1983. Semantics and Cognition. Current Studies in Linguistics
Series 8. Cambridge, MA: MIT Press.
276
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Jake, Janice & Carol Myers-Scotton. 1998. “How to build a creole: Splitting and
recombining lexical structure.” Paper presented January 10,1998 at the annual
joint meeting of the Society of Pidgin and Creole Linguistics with the Linguistic
Society of America (LSA) at the Grand Hyatt in New York, N.Y.
Jake, Janice L. 1996. “Constructing Interlanguage: Building a Composite Matrix
Language”, unpublished mss.
Johanson, Lars. 1992. Strukturelle Faktoren in tuerkischen Sprachkontakten. Stuttgart,
Germany: F. Steiner.
Karttunen, Frances E. and James Lockhart 1976. Nahuatl in the Middle Years:
Language Contact Phenomena in Texts o f the Colonial Period. Berkeley:
University of California Press.
Karttunen, Frances. 1983. An Analytical Dictionary o f Nahuatl. Norman, Okla.:
University of Oklahoma Press.
Katamba, Francis. 1993. Morphology. New York: S t Martin’s Press.
Keesing, Roger M. 1991. “Substrates, calquing and grammaticalization in Melanesian
Pidgin.” In Approaches to Grammaticalization, Vol. 1. Elizabeth Closs Traugott &
Bemd Heine (Eds.). Amsterdam and Philadelphia: John Benjamins. Pp. 315—
342.
Kroll, Judith F. and Annette M. B. de Groot 1997. “Lexical and Conceptual Memory n
the Bilingual: Mapping Form to Meaning in Two Languages.” In Tutorials in
Bilingualism: Psycholinguistic Perspectives. Annette M. B. de Groot & Judith F.
Kroll (Eds.). Mahwah, New Jersey: Lawrence Erlbaum Associates. Pp. 169— 199.
Labov, William. 1972. Sociolinguistic Patterns. Philadelphia, PA: University of
Pennsylvania Press.
Landau, Barbara and Lila Gleitman. 1985. Language and experience: Evidence fro the
blind child. Cambridge, MA: Harvard University Press.
Landau, Barbara. 1993. “Ontology and Perception, Object Kind and Object Naming.”
The proceedings o f the twenty-fifth annual Child Language Research Forum. Ed.
Eve Clark. Stanford University: Center for the Study of Language and
Information. 191-6.
Landau, Barbara. 1994. “Where’s what and what’s where: The language of objects in
space.” In L. Gleitman and B. Landau (Eds), The Acquisition o f the Lexicon.
Cambridge, Mass.: The MIT Press. Pp. 259-296.
Langacker, Ronald W. 1977. “Syntactic reanalysis. In Mechanisms o f Syntactic Change.
Charles Li (Ed.). Austin: University of Texas Press.
Lastra, Yolanda. 1968. Cochabamba Quechua Syntax. The Hague: Mouton.
Lefebvre Claire & John Lumsden. 1989. “Les langues creoles et la theorie linguistique.
In The Canadian Journal o f Linguistics, 34:249— 272.
277
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Lefebvre, Claire. 1986. “Relexification and creole genesis revisited.” In Substrata versus
Universals in Creole Genesis, Pieter Muysken and Norval Smith (Eds.).
Amsterdam and Philadelphia: John Benjamins.
Lefebvre, Claire. 1993. “The Role of Relexification and Syntactic Reanalysis in Haitian
Creole: Methodological Aspects of a Research Program.” Africanisms in Afro-
American Language Varieties. Ed. Salikoko S. Mufwene. Athens, Ga.:
University of Georgia Press. 254— 78.
Lehmann, Christian. 1986. “Grammaticalization and Linguistic Typology.” in General
Linguistics, 26 (1), 3-23.
Lehmann, Winfred P. 1992. Historical Linguistics, An Introduction. New York, N.Y.:
Holt, Rinehart and Winston, Inc..
Levelt, Willem J. M. 1989. Speaking: From Intention to Articulation. Cambridge,
Mass.: The MIT Press.
Levelt, Willem J. M. Levelt 1992. Accessing Words in Speech Production: Stages,
processes and representations. In Lexical Access in Speech Production. Willem J.
M. Levelt (Ed.). Cambridge, MA: Blackwell. Pp. 1-22.
Leon-Portilla, Ascenci6n H. de. 1972. “Bibliograffa lingiiistica nahua.” Estudios de
Cultura Ndhuatl 10: 409-41.
Li, Charles. 1983. “Languages in contact in western Chine. Papers in East Asian
Languages 1:31— 51.
Lichtenberk, Frantisek. 1991. “On the gradualness of grammaticalization.” In
Approaches to Grammaticalization, Vol. 1. Elizabeth Closs Traugott & Bemd
Heine (Eds.). Amsterdam and Philadelphia: John Benjamins. Pp. 37-80.
Lipski, John. 1994, Latin American Spanish. London and New York: Longman, 1994..
Lumsden, John S. 1994. Possession: Substratum semantics in Haitian Creole. Journal
o f Pidgin and Creole Languages. Vol. 9:1. Amsterdam: John Benjamins. 25-50.
Lyons, John. 1968. Introduction to Theoretical Linguistics. New York, N.Y.:
Cambridge University Press.
Lyons, John. 1995. Linguistic Semantics: An introduction. Cambridge: Cambridge
University Press.
Markman, Ellen M. 1989. Categorization and Naming in Children: Problems o f
Induction. Cambridge, Mass.: The MIT Press.
Markman, Ellen M. 1993. “Constraints Children Place on Word Meanings.” In Language
Acquisition: core readings. Cambridge, Mass.: The MTT Press. 154-73.
Markman, Ellen M. 1994. “Constraints on word meaning in early language acquisition.”
In L. Gleitman and B. Landau (Eds), The Acquisition o f the Lexicon. Cambridge,
Mass.: The MIT Press. Pp. 199-227.
278
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Marisen-Wilson, William and L. K. Tyler. 1980. “The temporal structure of spoken
language understanding. Cognition 8: 1-71.
Marisen-Wilson, William. 1989. “Access and integration: Projecting sound onto
meaning.” In W. D. Marisen-Wilson (Ed.), Lexical representation and process.
Cambridge, Mass.: MIT Press. 3— 24.
Matthews, P. H. 1991. Morphology. Cambridge Univ. Press.
Mertz, Elizabeth. 1989. “Sociolinguistic creativity: Cape Breton Gaelic’s linguistic “tip”.”
In Investigating obsolescence: Studies in language contraction and death. Nancy
C. Dorian (Ed.). Cambridge: Cambridge University Press. 117-137.
Meyer, Michael G., and William L. Sherman. 1987. The Course o f Mexican History.
New York: Oxford Univ. Press.
Milroy, Lesley and Li Wei. 1995. “A social network approach to code-switching: the
example o f a bilingual community in Britain.” In One speaker, two languages:
Cross-disciplinary perspectives on code-switching. Lesley Milroy and Pieter
Muysken (Eds.). Cambridge: Cambridge University Press. Pp. 136-157.
Milroy, Lesley and Pieter Muysken. 1995. “Introduction: code-switching and bilingual
research.” In One speaker, two languages: Cross-disciplinary perspectives on
code-switching. Lesley Milroy and Pieter Muysken (Eds.). Cambridge:
Cambridge University Press. Pp. 1-14.
Milroy, Lesley and Sue Magrain. 1980. “Vernacular Language Loyalty and Social
Network”. Language In Society. Cambridge, Mass.: Cambridge University
Press. 43— 70.
Milroy, Lesley. 1987. Language and social networks (2nd edition). Oxford: Basil
Blackwell.
Mithun, Marianne. 1984. “The evolution of noun incorporation.” In Language, 60: 847-
894.
Mithun, Marianne. 1989. “The incipient obsolescence of polysynthesis: Cayuga in
Ontario and Oklahoma. In Investigating obsolescence: Studies in language
contraction and death. Nancy C. Dorian (Ed.). Cambridge: Cambridge Univ.
Press. 243-257.
Molina Alonso de. 1571b. Vocabulario en Lengua Castellano y Mexicana. Mexico:
Antonio de Spinos. Facsimile edition: Coleccidn de Incunables Americanos, Signo
XVI, Vol. IV. Madrid: Ediciones Cultura Hispdnica, 1944.
Molina, Alonso de. 1571a. Arte de la lengua Mexicana y Castellano. Mdxico: Pedro
Ocharte. Facsimile edition: Coleccidn de Incunables Americanos, Siglo XVI, Vol.
VI. Madrid: Ediciones Cultura Hispdnica, 1945.
279
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Mougeon, Raymond and Edouard Beniak. 1989. “Language contraction and linguistic
change: The case of Welland French.” In. Investigating obsolescence: Studies in
language contraction and death. Nancy C. Dorian (Ed.). Cambridge: Cambridge
University Press. 287-312.
Mufwene, Salikoko. 1986. “The Universalist and Substrate Hypotheses Complement One
Another.” Substrata versus universals in Creole languages. Eds. Pieter Muysken
and Norbert Smith. Amsterdam: John Benjamins.
Muysken, Pieter. 1981. “Halfway between Quechua and Spanish: the case for
relexification.” Historicity and Variation in Creole Studies. Eds. Valdman and
Highfield. Ann Arbor, Mich.: Karoma. 52-78.
Muysken, Pieter. 1988. “Media Lengua and Linguistic Theory.” Canadian Journal o f
Linguistics, 33 (4), 409-422.
Muysken, Pieter. 1994. “Media Lengua.” In Mixed Languages: 15 Case Studies in
Language Intertwining. Eds. Peter Bakker and Maarten Mous. Amsterdam:
IFOTT. Pp. 207-211.
Muysken, Pieter. 1995. “Code-switching and Grammatical Theory.” In One speaker,
two languages: Cross-disciplinary perspectives on code-switching. Lesley Milroy
and Pieter Muysken (Eds.). Cambridge: Cambridge University Press. Pp. 177-
198
Muysken, Pieter. 1997. “Media Lengua.” In. Contact Languages: A Wider Perspective.
Sarah G. Thomason (Ed.). New York and Amsterdam: John Benjamins, pp.
365-426.
Muhlhaiisler, Peter. 1980. “Structural Expansion and the Process of Creolization.”
Theoretical Orientations in Creole Studies. Albert Valdman and Arnold Highfield
(Eds.). New York: Academic Press.
Miihlhausler, Peter. 1990. “Tok Pisin: Model or special case.” In Melanesian Pidgin and
Tok Pisin. John W. M. Verhaar (Ed.). Philadelphia and Amsterdam: John
Benjamins. Pp. 171— 186.
Myers-Scotton, Carol. 1993a. Duelling Languages: Grammatical Structure in
Codeswitching. New York: Oxford Univ. Press.
Myers-Scotton, Carol. 1993b. Social Motivations fo r Codeswitching. New York:
Oxford Univ. Press.
Myers-Scotton, Carol. 1995. A lexically based model of code-switching. In One speaker,
two languages: Cross-disciplinary perspectives on code-switching. Lesley Milroy
& Pieter Muysken (Eds.). New York: Cambridge University Press. Pp. 233-
256.
Myers-Scotton, Carol. 1998. “A way to dusty death: the Matrix Language turnover
hypothesis.” In Endangered Languages: Language Loss and Community
Response. Lenore A. Grenoble and Lindsay J. Whaley (Eds.) Cambridge:
Cambridge University Press.
280
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Paradis, Michel. 1997. “The cognitive neuropsychology of bilingualism. In Tutorials in
Bilingualism: Psycholinguistic Perspectives. Annette M. B. de Groot & Judith F.
Kroll (Eds.). Mahwah, New Jersey: Lawrence Erlbaum Associates. Pp. .331-
354.
Penfield, Joyce and Jacob L. Omstein-Galicia. 1985. Chicano English: An Ethnic
Contact Dialect. Philadelphia: John Benjamins.
Pfaff, Carol W.. 1979. “Constraints on language mixing.” Language. Vol. 55 (2): 291-
318.
Pinker, Steven. 1996. Language Learnability and Language Development. Cambridge,
MA: Harvard University Press.
Poplack, Shana, and Maijory Meecham. 1995. “Patterns of language mixture: nominal
structure on Wolof-French and Fongbe-French bilingual discourse. In One
speaker, two languages: Cross-disciplinary perspectives on code-switching.
Lesley Milroy and Pieter Muysken (Eds.). Cambridge: Cambridge University
Press. Pp. 199-232.
Poplack, Shana, David Sankoff, and Christopher Miller. 1988. “The social correlates and
linguistic processes of lexical borrowing and assimilation. Linguistics, 26: 47—
104.
Poplack, Shana. 1982. “Sometimes I’ll start a sentence in Spanish y termino en espanol”:
toward a typology of code-switching. Spanish in the United States. Jon Amastae
and Lucia Elfas-Olivares. New York, N.Y.: Cambridge University Press. 230-
263.
Poulisse, Nanda. 1997. “Language production in bilinguals. In Tutorials in Bilingualism:
Psycholinguistic Perspectives. Annette M. B. de Groot & Judith F. Kroll (Eds.).
Mahwah, New Jersey: Lawrence Erlbaum Associates. Pp. 201-224.
Pullum, Geoffrey K. & Arnold M. Zwicky. 1988. “The syntax-phonology interface.” In
Linguistics: The Cambridge Survey: Vol. I: Linguistic Theory: Foundations.
Frederick J. Newmeyer (Ed.). New York: The Cambridge University Press.
281-302.
Radford, Andrew. 1988. Transformational Grammar. New York: Cambridge University
Press.
Reesink, Ger P. 1990. “Mother tongue and Tok Pisin.” In Melanesian Pidgin and Tok
Pisin. John W. M. Verhaar (Ed.). Philadelphia and Amsterdam: John Benjamins.
Pp. 189-306.
Relinger, W. & Park, T.-Z. 1980. “Language mixing in young bilinguals.” In Journal o f
Child Language 7(2):337-352.
Rickford, John. 1988. Dimensions o f a Creole Continuum. Palo Alto, CA: Stanford
University Press.
Romaine, Suzanne. 1988. Pidgin and Creole Languages. London: Longman.
281
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Romaine, Suzanne. 1989. “Pidgins, Creoles, Immigrant, and Dying languages.”
Investigating obsolescence: Studies in language contraction and death. Ed. Nancy
C. Dorian. Cambridge: Cambridge UP. 369-383
Romaine, Suzanne. 1995. Bilingualism. 2nd Edition. Vol. 13 of Language in Society.
Cambridge, Mass.: Blackwell.
Rosch, Eleanor & Carolyn B. Mervis. 1975 “Family Resemblances: Studies in the
Internal Structure of Categories.” In Cognitive Psychology, 7:573-605.
Rosch, Eleanor, Carolyn B. Mervis, Wayne D. Gray, David M. Johnson, and Penny
Boyes-Braem. 1976. “Basic Objects in Natural Categories.” In Cognitive
Psychology, 8:382-439.
Rosch, Eleanor. 1975. “Cognitive Reference Points.” In Cognitive Psychology, 7:533-
547.
Rosetti, Al. 1945/49. “Langue mixte et melange des langues.” Acta Linguistica, 5:73-79.
Sankoff, D., S. Poplack, & S. Vanniarajan. 1990. “The case of the nonce loan in Tamil.
Language Variation and Change, 2: 71-101.
Sapir, Edward. 1921. Language: An Introduction to the Study o f Speech. New York,
N.Y.: Harcourt Brace Jovanovich,.
Sdnchez, Rosaura. 1982. “Our linguistic and social context” Spanish in the United
States. Jon Amastae and Lucia Elfas-Olivares (Eds.). Cambridge: Cambridge
University Press. 9-46.
Schacter, Jacqueline. 1993. “A New Account of Language Transfer.” In Language
Transfer in Language Learning. Susan M. Gass & Larry Selinker (Eds.).
Amsterdam and Philadelphia: John Benjamins. 32-46.
Schreuder, Robert and Bert Weltens. 1993. “The Bilingual Lexicon: An Overview.” In
The Bilingual Lexicon. Robert Schreuder and Bert Weltens (Eds.). Amsterdam
and Philadelphia: John Benjamins. Pp. 1-10.
Selinker, Larry & Usha Lakshmanan. 1993. “Language Transfer and Fossilization.” In
Language Transfer in Language Learning. Susan M. Gass & Larry Selinker
(Eds.). Amsterdam and Philadelphia: John Benjamins. 197-216.
Seuren, Pieter and Herman Wekker. 1986. “Semantic Transparency as a Factor in Creole
Genesis.” Substrata versus universals in Creole genesis. Eds. Pieter Muysken and
Norval Smith. Philadelphia: John Benjamins. 57— 70.
Sharwood Smith, Michael. 1994. Second Language Learning: Theoretical Foundations.
London and New York: Longman.
Siegel, Jeff. 1997. “Dialectal difference and substrate reinforcement in Melanesian
Pidgin.” Paper presented at the annual meeting of the Society for Pidgin and Creole
Linguistics, London, June 26, 1997.
282
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Silva-Corvaldn, Carmen. 1994. Language Contact and Change: Spanish in Los Angeles.
New York: Oxford University Press.
Singh, Rajendra. 1981. ’’Aspects of language borrowing: English loans in Hindi.” In
Sprachkontakt und Sprachkonflikt. P. H. Nelde (Ed.). Wiesbaden: Steiner. 113—
116.
Singler, John Victor. 1993. “African Influence upon Afro-American Language Varieties:
A Consideration of Sociohistorical Facts. Africanisms in Afro-American Language
Varieties. Ed. Salikoko S. Mufwene. Athens, Ga.: University of Georgia Press.
235-53.
Slabbert, Sarah, and Carol Myers-Scotton. 1997. “The Structure of Totsitaal and
Iscamtho: Codeswitching and Ingroup Identity in South African Townships.” In
Linguistics 35, pp. 317-342.
Slobin, Daniel I. 1985. “Crosslinguistic Evidence for the Language-Making Capacity.”
In The Crosslinguistic Study o f Language Acquisition. Vol. 2: Theoretical Issues.
Dan I. Slobin (Ed.). Hillsdale, N.J.: Lawrence Erlbaum. 1157-1256.
Smith, Marilyn Chapnik. 1997. “How Do Bilinguals Access Lexical Information?” In
Tutorials in Bilingualism: Psycholinguistic Perspectives. Annette M. B. de Groot
& Judith F. Kroll (Eds.). Mahwah, New Jersey: Lawrence Erlbaum Associates.
Pp. 145-168
Spencer, Andrew. 1991. Morphological theory: an introduction to word structure in
generative grammar. Cambridge, MA: Blackwell Publishers.
Sullivan, Thelma D. 1988. Thelma D. Sullivan’s compendium o f Nahuatl grammar.
Translated from Spanish by Thelma D. Sullivan and Neville Stiles. Wick R. Miller
and Karin Dakin (Eds.). Salt Lake City: University of Utah Press.
Sweetser, Eve. 1988. “Grammaticalization and semantic bleaching.” In Berkeley
Linguistics Society: General Session and Parasession on Grammaticalization
(Berkeley Linguistics Society 14). Berkeley: University of California.
Thomason, Sarah Grey and Terrence Kaufman. 1988. Language Contact, Creolization,
and Genetic Linguistics. Berkeley: University of California Press.
Thomason, Sarah Grey. 1997a. “Introduction.” In Contact Languages: A Wider
Perspective. Sarah G. Thomason (Ed.). New York and Amsterdam: John
Benjamins, pp. 1-8.
Thomason, Sarah Grey. 1997b. “Mednyj Aleut” in Contact Languages: A Wider
Perspective. Sarah G. Thomason (Ed.). New York and Amsterdam: John
Benjamins, pp. 449— 468.
Thomason, Sarah Grey. 1997c. “Ma’a (Mbugu).” In Contact Languages: A Wider
Perspective. Sarah G. Thomason (Ed.). New York and Amsterdam: John
Benjamins, pp. 469-487.
283
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Traugott, Elizabeth C. & Bemd Heine. 1991. “Introduction.” hi Approaches to
Grammaticalization, Vol. 1. Elizabeth Closs Traugott & Bemd Heine (Eds.).
Amsterdam and Philadelphia: John Benjamins. Pp. 1-13.
Traugott, Elizabeth C. & Ekkehard Konig. 1991. “The semantics-pragmatics of
grammaticalization revisited.” In Approaches to Grammaticalization, Vol. 1.
Elizabeth Closs Traugott & Bemd Heine (Eds.). Amsterdam and Philadelphia:
John Benjamins. Pp. 189-218.
Traugott, Elizabeth C. 1982. “From prepositional to textual and expressive meanings;
some semantic-pragmatic aspects of grammaticalization.” In Perspectives on
Historical Linguistics, Winfred Lehmann and Yakov Malkiel (Eds). Amsterdam:
John Benjamins. Pp. 245— 271.
Traugott, Elizabeth C. 1988. “Pragmatic strengthening and grammaticalization.” In
Berkeley Linguistics Society, Proceedings o f the Fourteenth Annual Meeting. Pp.
406-416.
Van Hout, Roeland & Pieter Muysken.. 1994. “Modeling lexical borrowability.” In
Language Variation and Change, 6 (1), 39-62.
Van Rheeden, Hadewych. 1994. “Petjo: the mixed language of the Indos in Batavia.” In
Mixed Languages: 15 Case Studies in Language Intertwining. Eds. Peter Bakker
and Maarten Mous. Amsterdam: IFOTT. Pp. 223— 237.
Vigil, James Diego. 1980. From Indians to Chicanos: The dynamics o f Mexican
American culture. Prospect Heights, 1 1 1 . : Waveland Press.
Volterra, V. & Taeschner, R. 1978. “The acquisition and development of language by
bilingual children. In Journal o f Child Language 5:311-326.
Wang, William S-Y and Chinfa Lien. 1993. “Bidirectional diffusion in sound change.”
In Historical linguistics: problems and perspectives. Charles Jones (Ed.). New
York: Longman Publishing. Pp. 345400
Watson, Seosamh. 1989. “Scottish and Irish Gaelic: the giant’s bedfellows.” In
Investigating obsolescence: Studies in language contraction and death. Nancy C.
Dorian (Ed.). Cambridge: Cambridge University Press. 41-59.
Waxman, Sandra R. 1994. “The development of an appreciation of specific linkages
between linguistic and conceptual organization.” In L. Gleitman and B. Landau
(fids), The Acquisition o f the Lexicon. Cambridge, Mass.: The MTT Press. Pp.
229-257.
Weinreich, Uriel. 1953 (reprinted in 1968). Languages in Contact. The Hague: Mouton.
West, Michael. 1953. A General Service List o f English Words. London: Longman,
Green and Co.
Whitney, William D. 1881. “On Mixture in Language.” TAPA 12.5-26.
284
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Woolard, Kathryn A. 1989. “Language convergence and language death as social
processes.” Investigating obsolescence: Studies in language contraction and death.
Ed. Nancy C. Dorian. Cambridge: Cambridge UP. 355— 368.
Wright, Ronald. 1992. Stolen Continents: The Americas through Indian Eyes since
1492. New York: Houghton Mifflin.
Zobl, Helmut 1993. “Prior Linguistic Knowledge and the Conservation of the Learning
Procedure: Grammaticality Judgments of Unilingual and Multilingual Learners.”
In Language Transfer in Language Learning. Susan Gass and Larry Selinker
(Eds). Amsterdam and Philadelphia: John Benjamins. 176-196.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
IMAGE EVALUATION
TEST TARGET (Q A -3 )
150mm
IM/4GE. Inc
1653 East Main Street
Rochester. NY 14609 USA
Phone: 716/482-0300
Fax: 716/288-5989
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Causativity in Korean: Syntactic causative, control, and purpose

PDF

Correspondence and faithfulness constraints in optimality theory: A study of Korean consonantal phonology

PDF

Communicative rights and responsibilities in an East Los Angeles barrio: An analysis of epistemic modal use

PDF

Arthur Machen’s treatment of the occult and a consideration of its reception in England and America

PDF

An analysis of United States overseas English language policy, 1938-1990

PDF

'Beowulf' and 'the hobbit': elegy into fantasy in j. R. R. Tolkien's creative technique

PDF

Andrej Belyj's "Petersburg" and James Joyce's "Ulysses": A comparative study

PDF

Conditions on null objects in Basque Spanish and their relation to leismo and clitic doubling

PDF

Allegories of dispersal: Nation and participation in Indian cinema, 1947-1977

PDF

A descriptive grammar of San Bartolome Zoogocho Zapotec

PDF

A rhetoric of the short story: A study of the realistic narratives of Flaubert, Maupassant, Joyce, and Hyon Chin'gon

PDF

A case grammar of the parker manuscript of the "Anglo-Saxon chronicle" from 734 to 891

PDF

Educational development aid and the role of language: A case study of AIT -Danida and the Royal University of Agriculture in Cambodia

PDF

A comparative study of focus constructions

PDF

A critical edition of the Escorial manuscript of 'historia de los indios de la neuva espana' of fray toribio de benavente (motolinia) (spanish text)

PDF

A comparative study of the realm of meaning of four child prostitutes in Taiwan: A hermeneutic approach

PDF

A Descriptive Analysis Of The Gascon Dialect Spoken At Donzac (Tarn-Et-Garonne), France

PDF

Automated postediting of documents

PDF

An Investigation Of Some Critical Factors In Language Synthesis And The Implications Of These Factors For Linguists As Language Engineers

PDF

Beyond words and phrases: A unified theory of predicate composition

Asset Metadata

Creator Field, Fredric William (author)

Core Title Borrowing and borrowability

Degree Doctor of Philosophy

Degree Program Linguistics

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

Tag anthropology, cultural,language, linguistics,OAI-PMH Harvest

Language English

Contributor Digitized by ProQuest (provenance)

Advisor Hawkins, John (committee chair), Comrie, Bernard (committee member), Eskey, David (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c17-388273

Unique identifier UC11350348

Identifier 9919036.pdf (filename),usctheses-c17-388273 (legacy record id)

Legacy Identifier 9919036.pdf

Dmrecord 388273

Document Type Dissertation

Rights Field, Fredric William

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