University of Southern California Dissertations and Theses

The converb -j in Alashan Mongolian

(USC Thesis Other)

The converb -j in Alashan Mongolian

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content THE CONVERB -J IN ALASHAN MONGOLIAN
By
Ella Cerling
A Thesis Presented to the
FACULTY OF THE USC DORNSIFE COLLEGE OF LETTERS, ARTS, & SCIENCES
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF ARTS
(LANGUAGE SCIENCES)
May 2024
Copyright 2024 Ella Cerling

ii
TABLE OF CONTENTS
List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Section 1.2 Alashan Grammar Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
Section 1.3 Setup of the Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Chapter 2: Converb History and Clause-Chaining . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Section 2.1 The History of the Word ‘Converb’ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Section 2.2 Clause Chaining . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Chapter 3: Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Section 3.1 Adjunction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Section 3.2 Height . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Section 3.2.1 Major’s Argument . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Section 3.3 Monoclausality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Chapter 4: Main Argument and Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Section 4.1 -j: Member of a Clause Chain? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Section 4.2 Adjunct Status . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
Chapter 5: Crosslinguistic Variation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
Chapter 6: Suggestions for Further Research and Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . 33
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

iii
LIST OF FIGURES
Figure 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Figure 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
Figure 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

iv
ABSTRACT
This thesis is concerned with the converbal and aspectual suffix -j in the Alashan variety of
Mongolian. It covers the suffix’s adjunct, height, and clausal status, and explains the nature of
the relationship between its converbal and aspectual functions. I argue that -j heads an AspP that
Adjoins at or below vP, and which can create a monoclausal serial verb chain (as opposed to a
multiclausal clause chain), similar to the morpheme –(i)p in Uyghur. This thesis also briefly
covers -j and its functions as they appear in other varieties of Mongolian, specifically Khalkha,
Ordos, and Kalmuck. More generally, I argue that the claimed relationship between converbs and
clause chains is illusory in Mongolic languages.
Keywords: Mongolian, Alasha, converb, serial verb chain, clause chain, adjunct, aspectual suffix

1
Introduction
The goals of this thesis are the following: first, I hope to provide a brief overview of the
word ‘converb’ in the literature since it first appeared. This may seem rudimentary, but as a
number of scholars have noted, including Weisser (2015) and Sugar (2022), the term ‘converb’
has been used in the literature to describe a number of constructions which are functionally (if
not always morphologically) dissimilar. The ambiguity that this has created in the literature is
substantial enough to warrant some attention.
Second, I would like to look at a converb morpheme in Alasha Mongolian, the suffix -j,
and lay out the type of structure it heads—specifically, the structure’s argument/adjunct status, at
what height it attaches to the clausal spine, and whether or not it is monoclausal. I will argue that
the Alashan converb -j is monoclausal (in other words, it does not introduce an additional clause
into the matrix clause), an adjunct, and attaches between Voice and vP.
This paper is organized as follows: after this brief Introduction to the background of the
problem I am addressing, Chapter 2 deals with the history of the word ‘converb’ and its
relationship to clause chaining constructions. I will argue that despite their claimed relationship,
upon further investigation, this claimed relationship is illusory. Chapter 3 covers the various
diagnostics and arguments in the literature for determining converbal monoclausality, adjunct,
and height status. Chapter 4 applies the diagnostics from Chapter 3 to Alashan data, along with
my argument as to why -j should be treated as an adjunct that can form a monoclausal serial verb
chain, instead of as a multiclausal clause chaining construction. Chapter 5 provides a brief
crosslinguistic survey of similar constructions in other varieties of Mongolian, and Chapter 6
concludes.

2
Before moving to Chapter 2, this section will provide the background of the problem,
along with a brief explanation of the basic relevant grammar of Alasha Mongolian.
1.2 Alashan Grammar Background
Alashan is an SOV language, with a certain amount of word order flexibility; Tense is expressed
by suffixes attached to the lexical verb, which generally appears at the end of the sentence. 1
Voice is also expressed as a suffix on the lexical verb, and precedes the Tense suffix, as in the
example below:
1) en nom Baatr bich-ik-sn
a book Baatr write-pas-pst
“A book was written by Baatr.”
Aspect is expressed via suffixes on the lexical verb; however, Aspect and Tense suffixes are not
allowed to be affixed to the same verb. When an aspectual morpheme is present, it remains on
the lexical verb, while the auxiliary copula /baan/ appears to bear Tense, as seen in (2) and (3)
(note that the present tense in Alashan is phonologically null):
2) bi ombu-j ben
I swim-prg cop
“I am swimming.”
3) bi ombu-j be-sn
I swim-prg cop-pst
“I was swimming (yesterday).”
As (2) and (3) demonstrate, the morpheme -j is a suffix denoting instantaneous aspect.
1
Alashan is thought to be a dialect of Oirat Mongolian, spoken in the Alshan League in Inner Mongolia (Birtalan
2003).

3
1.3 Setup of the Problem
In looking at -j, one is presented with the following problem: -j seems to show up in
different environments and constructions which on the surface seem difficult to reconcile with
each other. Consider the following pieces of data:
4) bi ombu-j ben
I swim-prg cop
“I am swimming (right now).”
5) bi ombu-j ir-sn
I swim-cnv come-pst
“I arrived by swimming.”
6) bi ombu-j be-j tor-sn
I swim-cnv cop-cnv lose.weight-pst
“I lost weight by swimming.”
7) bi ombu-j gwi-j nis-j jolod-ed ir-sn
I swim-cnv run-cnv fly-cnv drive-fv come-pst
“I arrived by swimming, running, flying, and driving.”
On the surface, the first looks like a participle, the second looks like a serial verb
construction, and the third and fourth look like variations of a clause chain. This poses several
problems, as some characteristics of these constructions would seem to be mutually exclusive.
For example, (4) is certainly monoclausal, but does that monoclausality extend to (5), where the
copula is replaced with a fully lexical verb? Finally, are (6) and (7) examples of true clause
chaining, as some have argued is possible in Mongolian (Georg 2003, Haspelmath et al 2001), or
are they serial verb constructions? And what is the difference between clause chains and serial
verb constructions? Below is a brief description of the various problems one encounters when
beginning to answer these questions.

4
According to Weisser (2015) and Sugar (2022), both serial verb chains and clause chains
have a morphosyntactic dependency on the final verb—however, according to Sugar (2022), this
dependency demonstrates that serial verb chains are monoclausal, while Weisser (2015) claims
that this same dependency shows that clause chains are multiclausal. Given the fact that both
serial verb constructions and clause chains are reported to be present in Mongolic languages
(Georg 2003, Haspelmath et al 2001), this creates something of a diagnostic problem.
Another diagnostic problem is that there is a close link between converbs and aspectual
forms in Mongolic languages. These forms are often identical to one another; this is
unsurprising, as many converbs are fossilized versions of aspectual participles (Haspelmath et al,
2001). Converbs still commonly function as aspectual participles, without needing to introduce a
specifically converbal structure into the sentence.
In order to analyze the aspectual/converbal suffix -j well, we must take all of these
factors, as well as crosslinguistic similarities and differences, into consideration. Finally, as I
mentioned earlier, vague and conflicting terminology in the literature only adds to the
obfuscation surrounding the answers to these questions. In what follows, I will attempt to clarify
some of the terminology before turning to Alashan data.

5
Chapter 2: Converb History & Clause Chaining
The following section will sketch the use of the word ‘converb’ since it first appeared in
order to demonstrate the enormous breadth of the term’s use, which subsequently has resulted in
a significant ambiguity in the literature as to a converb’s properties and scope. This section will
also devote some special attention to the relationship between converbs and the clause chaining
constructions found in many Papuan and Austronesian languages; I will demonstrate that these
constructions are of a sufficiently different structure than the converbal constructions found in
Turkic languages, and so should not be thought of as fully ‘converbal’ constructions, nor equated
with the converb constructions analyzed in this paper.
2.1 The history of the word ‘converb’
The word “converb” made its first appearance in linguistic fieldwork in 1903, when the
Finish linguist Ramstedt wrote a grammar on Khalkha Mongolian. He used the term to describe
one of three observed verb forms; specifically, converbs referred to forms which “only occur as
predicate modifiers, never as predicates of complete sentences” (Ramstedt 1903). He
distinguished these from fully predicative verbs and verbal nouns. Since its first appearance in
1903, the term has been used to describe constructions in many languages, including those from
the Indo-European language family, but most especially languages from the Altaic and
Austronesian language families (Haspelmath et. al. 2001; Weisser 2015). Despite its
longstanding usage, the definition is still somewhat slippery; however, there are a few
characteristics of converbs that are generally agreed upon.
There is a consensus that the information typically conveyed by converbs is adverbial in
nature: Weisser (2015) argues that converbs are incontrovertibly adverbially subordinate, while
Nedjalkov (1998) discusses the adverbial functions of converbs at length.

6
Additionally, converbs cross-linguistically demonstrate both coordinative and subordinative
properties (Haspelmath et. al. 2001). I will follow Sugar (2022) in treating converbs as
subordinate in the following discussion and analysis.2
Over the last 120 years, however, the term “converb” has gradually begun to cover
increasingly distinct linguistic territories. For the last thirty years, researchers have used the term
in connection with the following constructions: clause chaining, serial verb constructions, clausal
conjunction (both that of coordination and subordination), participles of Aspect, and adverbial
functions/modifiers. From some definitions in the literature, it is unclear whether a converb is
merely a type of participle; indeed, many authors refer to converbs as either imperfective or
perfective, as converbs in Mongolic and Turkic languages are often formed from aspect stems,
fossilized verbal nouns, or past and present participles (Birtalan 2003; Bläsing 2003; Csató &
Johanson 2018; Georg 2003; Haspelmath et. al. 2001; Nedjalkov 1998; Svantesson 2003).
Unsurprisingly, this results in homophonous confusion between converbs and other verb
forms. In some cases, it can be almost impossible to distinguish between a postverbial
construction (a combination of a lexical verb and a semantically bleached verb) and a serial verb
construction (Csató & Johanson 2018; Haspelmath et. al. 2001).
Some authors emphasize a converb’s status as an adverbial modifier (Csató & Johanson
2018; Haspelmath 1995; Nedjalkov 1998), while still others focus exclusively on the clausal
coordinative properties involved in clause chaining and serial verb constructions (Weisser 2015).
However, because articles on converbs tend to exclusively focus on only one or two properties of
converbs, this creates a fractured and mutually exclusive picture of the abilities and properties of
2 Reconciling the coordinative and subordinative properties of converbs is outside the scope of this paper; for a
thorough discussion on the subject, see Weisser (2015). However, it is useful to note as an indication of the
conjunctive-participial dichotomy: we do not talk about participles (in a broader sense, we do not talk about
anything less than a clausal conjunct/coordinator) as having both coordinative and subordinative properties.

7
converbs, which is further obfuscated by the fact that there are legitimate cross-linguistic
variations in the properties of converbs.
Converbs largely function in lieu of individual clausal conjunctions in Mongolic and
Turkic languages (Nedjalkov 1998; Nedjalkov & Otaina 2013). It is this last property that has
most likely given rise to the use of the term within the context of Austronesian clause chaining.
2.2 Clause Chaining
Clause chaining is frequently mentioned in conjunction with converbs. We will see that
the topic of clause chaining intersects with that of converbs in two important ways. The first way
is via terminology; converbs can be associated with clause chaining because of imprecise
terminology, which can cause significant confusion in attempting to nail down the limits and
capabilities of converbs. This problem occurs with Papuan and Austronesian languages. The
second way is by true connection; there are several instances in the literature of clause chaining
being mentioned in conjunction with Mongolic-language varieties. Therefore, in Section 4.1, we
will examine the clause-chain-like construction in (7) above and determine whether or not it is a
true clause chain.
Mansfield and Barth (2021) define clause chains as long sequences of clauses where the
verbs are marked as either medial or final. Medial clauses are entirely dependent on the matrix
clause, and coordinating and subordinating conjunctions are not used (Weisser 2015). Often the
clauses within a clause chain consist only of the verb and its arguments. Since arguments can
often be dropped, the verb is often the only overt part of the clause (Weisser 2015).
However, it is important to note that medial clauses can be distinct from subordinate
clause structures even in the same language, so clause chains should not be equated or confused
with subordinate clause constructions (Mansfield and Barth, 2021). Both Mansfield and Barth

8
(2021) and Weisser (2015) identify the key characteristic of clause-chaining sequences to be the
morphosyntactic dependence of non-final (medial) verbs on the final verb in the chain. A wellattested morphosyntactic phenomenon in clause chains is that clauses get marked for switchreference. This is outside the scope of this paper, however, it is worth noting that this is a major
feature of clause chaining in Papuan languages (Weisser 2015). Some of the confusion and
overlap around clause chaining, serial verb constructions, and converbs might be further resolved
if more research was done to categorize the extent of the overlap between the abilities of Turkic
and Mongolic converbal switch-reference, and the switch-reference found in Papuan and
Austronesian clause chaining.
Weisser (2015) argues that there are two types of clauses within a clause chain (excluding
the final matrix clause): medial clauses and converb clauses. He also argues that the primary
distinction between these two types of clauses is that medial clauses are between coordination
and subordination, while converb clauses are always subordinate. He argues that medial clauses
are coordinate-subordinate because of their movement from a subordinate position into a
coordinate position, while converb clauses remain subordinate because they remain in situ.
Setting aside the relative merits of Weisser’s argument, this is a good example of the fluid
terminology that exists in the literature on converbs.
As discussed above, this paper will concern itself primarily with analyzing
monoclausality, adjunct status, and height—it will not look at converb movement, or the
possibility thereof.
Weisser bases his claim that Austronesian medial verbs in situ are converbs on the fact
that both Turkic converbs and Austronesian medial verbs rely morphosyntactically on the final
verb. While it is certainly true that Austronesian clause chaining and Turkic converbal

9
constructions share several of the same attributes—notably a morphosyntactic dependence on the
final verb—treating Turkic converbs and Austronesian clause chains as the same thing would be
premature.
There are several key differences between Austronesian ‘converbs’ and Turkic converbs:
coordination, same-subject/different-subject markers, and medial/final verb markers. Note the
following piece of data:
8) [Nono imai-te-pa] [mai mena-a-te] [pai aʔate-pa] [nono wi nenfe-pa]
[child 3.sg.carry-get-med.ss] [come.up stay-3.sg.-med.ds] [pig hit-med.ss] [child show
3.pl-tr-med.ss]
[yene wawai wi nen-fe-pa] [mene-pa] [pai aʔate-ti tefe-pa] [ʔeʔeripa]
[sacred flute show 3.pl-tr-med.ss] [stay-med.ss] [pig hit-conj put-med.ss] [dancemed.ss]
[toto-iʔ-a]
[cut-3.pl-ind]
‘She carried the child and came up and stayed; and they killed the pigs and showed them to the
children and they showed them the sacred flutes and stayed and they killed the pigs and put them
and they danced and cut (the pigs).’
Tauya: MacDonald (1990: 361)
The medial markers at the end of each verb are indicated by bold type. As can be seen,
the medial markers are glossed for both a medial/final distinction and a same-subject/differentsubject distinction. There is no evidence of same-subject/different-subject information being
encoded in Mongolian converbs; while they can exhibit a medial/final distinction, they do not do
so in the same way as the Tauyan data above. The example below demonstrates the main
difference: i.e., the ‘final’ marker only shows up within the serial verb chain; it is not present on
the linearly final verb, and as such is not present on the verb marked for tense.

10
9) bi ombu-j gwi-j nis-j jolod-ed ir-sn
I swim-prg run-prg fly-prg drive-fv come-pst
“I arrived by (a combination of) swimming, running, flying, and driving.”
Given that: 1) the clause chain in (8) is treated as a multi-clausal construction; 2) there seems to
be little doubt in the literature that these types of Austronesian clause chains are multiclausal;
and 3) Alashan serial verb chain constructions are monoclausal (as I will demonstrate); therefore
4) it is reasonable to conclude that using the ‘converb’ terminology to describe both of these
constructions adds unnecessary confusion to an already complex topic.

11
Chapter 3: Literature Review
This chapter is organized in the following way: Section 3.1 deals briefly with adjunction,
and presents diagnostics relevant to determining the argument, complement, or adjunct status of
a particular structure; Section 3.2 covers height; Section 3.3 covers monoclausality, and briefly
introduces the internal structure that I will be adopting for -j.
Section 3.1 Adjunction
In his discussion of adjuncts, complements, and arguments, Privoznov (2022) makes two
sets of distinctions. First, he distinguishes between an argument and a modifier; second, between
a complement and an adjunct. For the sake of clarity, what follows is a short discussion of these
terms.
There are several key differences between an argument and a modifier; an argument has
received a theta-role from the main predicate, and is therefore necessary to complete the meaning
of the expression.3 A modifier, on the other hand, has neither been assigned a theta-role, nor is
necessary for the completion of the expression. In other words, modifiers are optional and
subsequently iterative. Modifiers differ structurally from arguments in that the sister of a
modifier’s maximal projection is a phrase, which makes the modifier an adjunct. If the maximal
projection’s sister is a head, then structurally, the object is a complement, even if it is not an
argument. We will see in the next section that Alashan converbs fulfill Privoznov’s criteria to be
considered adjuncts.
3 Without getting into the murky depths of the differences between complements and arguments, this paper will
generally treat them as the same, with the general exception that if a complement has not received a theta-role from
any given predicate (but still is necessary for completing the given expression), then it is not an argument, and will
be noted and treated as such.

12
In order to demonstrate that Uyghur serial verb constructions are adjuncts rather than
complements, Sugar looks at data involving iteration and case marking.
4
Sugar presents evidence for an adjunction analysis by looking at V1’s properties of
iteration. He argues that the V1 in -(i)p constructions is iterable and demonstrates this by taking
two verbs that can both function as V1 to the monotransitive verb ‘to flatten’ and putting them in
the same construction, as seen below:
10) Ahmat mitalni urup bésip tüzliwetti.
Ahmat mital-ni uru-(i)p bas-(i)p tüzle-iwet-di-0.
Ahmat metal-acc pound-(I)P press-(I)P flattend-compl-pst-3.
“Ahmat pounded and pressed the metal flat (flattened it by pounding and pressing).”
Although it could be argued that the two –(i)p elements are covertly coordinated, Sugar
circumvents this by showing that when the overt coordination marker is introduced into the same
sentence, it significantly changes the sentence’s meaning by creating two events instead of just
one. The fact that there are two events can be seen by the fact that two different time adverbials
can modify each event. His data is reproduced below:
11) Iskender mitalni chüshte urup hem chüshtin kéyin bésip tüzliwetti.
Iskender mital-ni chüsh-da uru-(i)p hem chüsh-din kéyin bas-(i)p tüzle-iwet-di-0.
Iskender metal-acc noon-loc hit-(I)P and noon-abl after press-(I)P flatten-compl-pst-3
“Iskender pounded the metal at noon and pressed it flat in the afternoon.”
In this section, we have looked briefly at Privoznov’s key characteristics for distinguishing
between arguments, modifiers, complements and adjuncts; we have also looked at Sugar’s
reasons for arguing against a complementation and coordination analysis of –(i)p structures.
These will be relevant diagnostics for Chapter 4.
4 Sugar (2022) subdivides Uyghur serial verb chains into two types: Inner Aspect Serial Verb Chains (IASVCs) and
Event SVCs. The argument presented here is taken from his section on IASVCs.

13
Section 3.2 Height
Sugar (2022), Major (in press), and Privoznov (2022) have proposed different landing
sites for converb constructions; these include at VP, vP, TP, and CP. The problem with
presenting a single landing site for converbs in the tree is that the same converb has
demonstrably different properties not based in any internal differences, but solely based on
height.
Sugar posits that converbs can attach at two different sites in the tree: one directly above
the vP, and one within the TP (between AuxP and PassP). Major argues that –(i)p clauses are
adjuncts that can merge at VP and TP. Privoznov argues that there are three types of converbs
that can attach at 3 different levels: CP, T’ and vP, and within the vP5. Interestingly, Privoznov
requires that T’-level converbs be adjuncts, while vP-converbs be complements.
3.2.1 Major’s Argument
Major argues that Uyghur converbial –(i)p clauses attach at two heights, VP and TP,
based on the fact that the same converb has different properties depending on where it adjoins.
Major demonstrates that VP-level –(i)p clauses can be answers to “how” questions, effectively
making them VP-manner modifiers. VP-manner modifiers are elements which provide additional
information about how an action is performed. Below is an example of an –(i)p construction
modifying the matrix VP.
12) a. Q: Abliz qandaq kel-d-i
Abliz how come-pst-3
“How did Abliz come (here)?” (adapted from Sugar, 2019, 77)
b. A: (U) méng-ip/téz kel-d-i.
5Privoznov’s evidence for arguing for three types of converbs attaching at three different heights is based on scope.
Essentially, the Balkar converb -p can both be able and unable to be interpreted under the scope of a causative
marker, negation, and SpecTP. He explains this duality by positing three different heights which are available for -p
to attach at. (Prizovnov 2022)

14
he walk-cnv/quickly come-pst-3
“He came by walking/came quickly.” (adapted from Sugar, 2019, 77)
Not only is (12b) a perfectly workable answer to the question, it can also be substituted for the
more recognizable VP-manner modifier quickly. This is not possible for all –(i)p converb
clauses. The data in (13) demonstrates that in the same environment, TP-level –(i)p clauses
cannot fulfill that role. Importantly, manner modifiers must Adjoin at the VP-level if they are to
actually modify the VP. In the data below, Major demonstrates that a TP-level –(i)p cannot
modify the VP.
13) a. Q: Ahmat polu-ni qandaq yé-d-i?
Ahmat pilaf-acc how eat-pst-3
“How did Ahmat eat the pilaf?”
b. A: #(U-ni) ét-ip yé-d-i.
it-acc make-cnv make-compl-pst-3
Intended: “Having made it, he ate it.”
Given their low position in the clause, VP-(i)p phrases can also function as the antecedent for
“like this/that” constructions. In the construction below, VP-level –(i)p can be shown to function
as an antecedent to shundaq, while TP-level –(i)p cannot be shown to do so.
14) a. Ahmat métal-ni ur-up tüzli-we-t-i, men=mu shundaq tüzli-wet-t-im.
Ahmat metal-acc hit-cnv flatten-compl-pst-3 1sg=add like.this flatten-compl-pst-1sg.
“Ahmat flattened the metal by pounding it, and I flattened [=metal by pounding] like this,
too.”
b. Tursun polu ét-ip yé-d-i, #men=mu shundaq yé-d-im.
Tursun pilaf make-cnv eat-pst-3 1sg=add like.this eat-pst-1sg
“Tursun, having made pilaf, ate it and #I ate [pilaf having cooked it] like this too.”
This is important evidence that clearly illustrates the different abilities of VP- and TP-level-(i)p.
Major also demonstrates that at the TP-level, any two predicates can combine. No
relationship (either semantic or causal) is necessary between these predicates at this level. There

15
also can be a voice and aspect mismatch between the TP-adjoined converb clause and the matrix
clause; this can be seen in the data in (15).
15) [[Girim qil-in-ip] mengz-ingiz qizir-ip ket-t-i].
makeup do-pass-conv cheek-2sg.poss redden-cnv ket-pst.-3
“Makeup was done and your cheeks reddened.”
This is a crucial difference between VP-level and TP-level –(i)p—the latter does not necessarily
set up a semantic, causal relationship between V1 and V2, while the former does. I will
demonstrate in Chapter 4 that the Alashan -j demonstrates very similar characteristics to VPlevel-(i)p in this regard. In Chapter 5, we will also see that a TP-level -j might be evident in other
Mongolian variations.
Finally, NPIs contained within VP-(i)p phrases can be licensed by negation in the matrix
clause—as we will see in the following section, this indicates that VP-level –(i)p phrases are part
of the matrix clause, not clauses themselves.
Section 3.3 Monoclausality
The following section will present Sugar’s (2022) argument on Uyghur serial verb constructions
involving the converb –(i)p. Specifically, I will present his reasons for arguing that these
constructions are monoclausal. I will use the same diagnostics in order to determine whether
Alashan converb constructions demonstrate the same features.
Serial verb chains look like the following: [(Obj) V1-(ip) V2]. Both V1 and V2 are fully
lexical, and V1 provides manner information about how V2 was performed.
Sugar bases his argument for the monoclausality of these serial verb chains on the
following theoretical and empirical observations. He first begins by briefly laying out the
composition of a clause: a clause contains the locus for voice morphology, located between the
verbal and inflection domains (Kratzer 1996, Cinque 1999, Alexiadou et al. 2006) and the locus

16
of aspect morphology; it is also the domain which permits licensing by negative markers (Choe
1988, Progovac 1988, 1993, Zanuttini 1991, Déprez 2000, Giannakidou 1998, 2000, 2006).
This means that if the matrix verb is inflected for voice, aspect, or negation, and the –
(i)p-suffixed verb is also interpreted for the same voice/aspect/negation without bearing those
inflectional markers, it strongly suggests that the –(i)p verb is part of the matrix clause. If, on the
other hand, the –(i)p-suffixed verb is not interpreted as having the same VAN as the matrix verb,
or can be shown to independently bear different VAN marking from the matrix verb, then this
would be very good evidence that the –(i)p-suffixed verb contains its own locus for voice,
aspect, and negation morphology, making it its own clause. Accordingly, I will demonstrate in
Section 4 that the structure I propose is merely a part of the aspectual locus of morphology for
the matrix clause, and will not prove to be the head of a structure that contains its own locus of
aspectual morphology.
Taking all this into account, Sugar tests whether Uyghur serial verb chains are
monoclausal using diagnostics that involve passive morphology, apsect morphology, and the
possibility of negation. He concludes, based on the following data, that Uyghur serial verb chains
are monoclausal.
Uyghur has a passive morpheme -il, which can passivize any verbs within its clause;
however, the morpheme cannot passivize any verbs outside that clause. This means that a passive
morpheme in a matrix clause cannot passivize the verb in an embedded clause, and vice versa.
This can be seen in the data in (16—18).
16) Tursun [roman-ning yézilghanliquini] dédi.
Tursun [roman-ning yaz-il-gan-lik-i-ni] de-di-0
Tursun [novel-gen write-pass-rel-nmlz-3.poss-acc] say-pst-3.
“Tursun said a novel was written.”
(Sugar: 2022: 79)

17
17) [Tursunning roman yazghanliqi] déyildi.
[Tursun-ning roman yaz-gan-lik-i] de-il-di-0
[Tursun-gen novel write-rel-nmlz-3.poss] say-pass-pst-3
“It was said that Tursun wrote a novel.” (*”a novel was written by Tursun”)
(Sugar: 2022: 79)
18) [Romanning yézilghanliqi] diyildi.
[Roman-ning yaz-il-gan-lik-i] de-il-di-0
[Novel-gen write-pass-rel-nmlz-3.poss] say-pass-pst-3
“It was said that a novel was written.”
(Sugar: 2022: 79)
Consequently, if a verb within a serial verb chain is passivized by the passive morpheme,
and the other verb is also interpreted as passive, the two verbs are within the same clause.
Furthermore, it is not possible to have the passive morpheme on both verbs without changing the
meaning of the sentence. In (20), the presence of -il on both verbs indicates that two separate
events occurred, and the –(i)p-suffixed verb no longer provides manner information about how
V2’s action was performed.
19) Mital urup tüzliwetildi.
Mital uru-(i)p tüzle-iwet-il-di-0
Mital pound-(I)P flatten-compl-pass-pst-3
“The metal was pounded flat.”
(Sugar 2022: 80)
20) Mital urulup tüzliwetildi.
Mital uru-il-(i)p tüzle-iwet-il-di-0
Mital pound-pass-(I)P flatten-compl-pass-pst-3
“The metal was pounded and flattened (by some other means).”
(Sugar 2022: 80)
The same is true for the progressive morpheme -iwat; like -il, it cannot assign Progressive
aspect outside of its own clause. When it is present on the matrix verb, V1 is also interpreted for
Progressive aspect, and (perhaps most importantly) it is ungrammatical to suffix -iwat to V1.
This strongly supports Sugar’s claim that Inner Aspect serial verb chains do not house voice or

18
aspectual morphology without changing the interpretation or becoming ungrammatical, which
points to the conclusion that these serial verb chains are monoclausal. In order to explain this
data, Sugar proposes the following structure:
Fig. 1
Here, an InnerAspectP, headed by –(i)p, Adjoins at v2P and which then takes v1P as its
complement. This is the same structure I will propose to explain the data from -j.
Sugar also relies on data involving NPI licensing to demonstrate that these serial verb
chains are monoclausal. Put briefly, he first demonstrates that negation markers can only license
NPIs within their clause, and then presents the following data:
21) Abliz héchyerge méngip ketmidi
Abliz héchyer-ga mang-(i)p ket-ma-di-0
Abliz nowhere-DAT walk-(I)P leave-neg-pst-3
“Abliz didn’t go walking anywhere.”
This is similar data to that shown in (19), and it demonstrates that a negative marker on the
matrix verb can license an NPI within the –(i)p-headed phrase, which means that the –(i)p phrase
is part of the matrix clause.
In conclusion, Sugar uses the confined scope of passive and negative markers in order to
demonstrate that the Uyghur –(i)p converb is monoclausal. I will attempt to do the same, using
similar methods.

19
Chapter 4: Main Argument and Analysis
In this chapter, I will demonstrate that the construction we saw in (7) is not an example of a
clause chain, but rather is part of a monoclausal serial verb chain. I will also argue in favor of an
analysis of -j in which it heads a ConverbP and which Adjoins at the vP level. I will also present
an argument demonstrating its monoclausality, along with a short discussion on height. In
Chapter 5, we will see how this analysis sheds light on similar data from other Mongolian
variations.
4.1 -j: Member of a Clause Chain?
Consider the following pieces of data:
22) Bi ombu-j gwi-j be-j ir-sn
I swim-prg run-prg cop-cnv come-pst
“I arrived by swimming and running.”
23) Bi ombu-j gwi-j nis-j jolod-ed ir-sn
I swim-prg run-prg fly-prg drive-fv come-pst
“I arrived by swimming, running, flying, and driving.”
Given the preponderance of literature linking converbs and clause chains, the
construction above in (23) would seem to support that relationship. (23) looks like a clause
chain; there are five verbs in the construction, with only one marked for Tense and one marked
with a final-verb marker. Given this last morpheme’s presence, a cursory glance might lead one
to assume that the -j morpheme is a medial-verb marker. Additionally, although final-verb
markers often appear on the final linear verb, this is easily accounted for by Alashan’s strict rules
governing which suffixes, and how many, may appear on a tensed verb. However, as we shall
see, the construction in (23) is not a clause chain, but a monoclausal serial verb chain of the same
type as Sugar’s Inner Aspect Serial Verb Chain construction discussed above. I will also show

20
that -j is not a medial-verb marker, but a converbal suffix that introduces a Manner relationship
between its own verb and the matrix verb.
Consider the following data:
24) bi ombu-j tor-sn
I swim-prg lost.weight-pst
“I lost weight by swimming.”
25) bi ombu-j ir-sn
I swim-prg come-pst
“I came by swimming.”
This is a fairly straightforward piece of data. We can see that ombu-j is adding additional
information to the main clause, and that information is about the manner in which the action was
performed. This Manner relationship between the two verbs is due to the presence of -j.
Interestingly, this structure is identical to Sugar’s example of an InnerAspect Serial Verb Chain,
repeated below:
26) Ahmat métal-ni ur-up tüzli-wet-t-i.
Ahmat metal-acc pound-cnv flatten-compl-pst-3
“Ahmat pounded the metal flat (flattened by pounding).”
(Sugar, 2019, 14)
27) Abliz méng-ip kel-d-i.
Abliz walk-cnv come-pst-3.
“Abliz walked here (came by walking).”
(Sugar, 2019, 14)
In both (26) and (27), we can see that the morpheme –(i)p introduces the same type of
relationship between the two verbs.6 Now consider the data in (28):
28) bi ombu-j be-j tor-sn
I swim-prg cop-cnv lose.weight-pst
“I lost weight by swimming.”
6 Note that Uyghur –(i)p constructions can also consist of multiple lexical verbs, resulting in a clause-chain-like
appearance.

21
Here, we have the same elements as in (24), with the somewhat puzzling addition of the word
be-j. According to native speaker judgments, this construction does not add any semantic
information to the structure in (25). Instead, the presence of be-j reinforces the Manner
relationship between the two lexical verbs. This raises the following question: is the presence of -
j on baan only possible because of its presence on V1? In other words, to what extent is be-j’s
interpretation as a Manner reinforcer dependent on ombu-j? Additionally, since there is no
semantic difference between (25) and (28), this reinforcement initially seems to be unnecessary.
It is also worthwhile to point out that at this point, the structure is beginning to look like a clause
chain. However, as I said earlier, we will see that this is not the case. The data in (29) easily
solves this problem:
29) bi om-sar be-j tor-sn
I swim-rep cop-cnv lose.weight-pst
“I lost weight by swimming (consistently over a period of time).”
This is a crucial piece of data. The morpheme -sar is another aspectual suffix, and it
denotes continuity. The fact that it can substitute for -j in an otherwise identical clause
demonstrates first that -j is functioning as an aspectual morpheme on ombu- in this sentence, and
second, that the Manner relationship that be-j reinforced in (6) is not predicated on -j’s
appearance on V1. Indeed, as we can see, the Manner relationship between V1 and V2 is now
borne solely by be-j. Given this, and given that the same Manner relationship can be produced by
-j on V1, this proves that -j is the morpheme responsible for introducing this specific Manner
relationship. It also indicates that when -j is present on a lexical verb, there is no meaningful
distinction between the aspectual form and the converbal one. One final note: be-j—the
combination of the copular and -j—is an interesting construction, which I will discuss in Section
4.2. One final piece of data to note is the data in (30) below.

22
30) Baatr tumurlgig cho-xut be-j tikshit-sn, bi bas tig-sn
Baatr metal.object pound-? cop-cnv flatten-pst, I again like.this-pst
“Baatr flattened the metal by pounding it, and I did it like this, too.”
This is the same structure that we see in (25) and (26). Clearly there is nothing in this structure
that resembles a clause chain. Here, -xut is yet another aspect marker; the Manner relationship
introduced by -j is preserved by be-j, but there is nothing else in this structure that resembles the
clause-chain-like structure in (23) and (24).
One final note before concluding: it is unclear why the final verb marker -ed must appear.
Native speaker judgments were quite firm that it must appear on the final verb in a four-verb
construction, while in a three-verb construction, it is preferred but optional. However, the data in
(31) indicates that this is by no means a requirement in every such construction:
31) bi ombu-j gwi-j nis-j mashin beɹe-t ir-sn
I swim-prg run-prg fly-prg car drive-dat? come-pst
“I came by swimming, running, flying, and driving a car.”
It is unclear what about this particular construction either absolves or blocks -ed from appearing.
One possibility is that the suffix -t on beɹex, which appears to license mashin, blocks -ed from
appearing. However, there is the question of whether -t only blocks it, or whether it possesses
some additional quality that fulfills -t’s role in the construction.
Given all this, let us re-examine the data in (23) and (24). Instead of analyzing it as a
clause chain in which -j functions as a medial verb marker, we can clearly see that it is nothing of
the sort; instead, it seems to be an example of the iterativity displayed by adjuncts, which I will
discuss in the next section.

23
4.2 Adjunct Status
In this section, I will argue that -j heads an adjunctive structure which selects either the
V1 or the copula as its complement; if it selects the copula, then this selects the V1 as its
complement.
7
As I claimed above, if the construction in (23) and (24) is not a clause chain, then there is
a strong argument to be made that it is demonstrating iteration of a non-clausal phrase headed by
-j. Consider the data in (23) and (24) again; any of these -j phrases can be taken out without
causing problems in the rest of the clause. In fact, all of these phrases can be taken out without
making the clause ungrammatical. This is true for all of the Alashan data we have looked at thus
far. Conversely, we can add multiple -j phrases to a sentence without any difficulty, as the
juxtaposition of (24) and (26) demonstrates. This indicates that the structure headed by -j can be
iterated, which in turn means that the structure is a modifier (as per Privoznov [2022]).
However, we still need to clarify whether the structure headed by -j Adjoins as an adjunct
or a complement, and we need to determine whether the -j in be-j possesses the same internal
structure. We will then be able to begin answering the questions around –j’s height and clausal
status.
I propose that -j Adjoins as an adjunct and which can select either V1 or the copula as its
complement. The entire construction is therefore an adjunct, although there can be multiple
elements within it which are complements.
I propose the following structure, depicted in the tree below:
7A note on labeling: if -j selects the copula, I have labeled it as ConverbP, and if it selects a lexical verb, I have
labeled it as AspP. The reason for this is that when -j selects baan, the Aspectual reading is unavailable to it. There
are no syntactic differences between ConverbP and AspP.

24
Fig. 2
Here, -j heads an AspP which Adjoins at the v2P level, and which selects the V1 as its
complement. This is a fairly straightforward way of solving this problem—we can see that it
results in the correct surface level word order, and it allows multiple lexical verbs to Adjoin in
the same way. This accounts for the data in (23). There is good reason for adopting this structure
for -j. Sugar (2022) uses the same structure to explain the data involving vP-level –(i)p.
Additionally, it accommodates what we know thus far about the size of -j, its status as a
modifier, and (as I will discuss below) its clausal status.
Fig. 3
However, this still leaves us with be-j. Does be-j also adjoin as an adjunct, or does it
merge as a complement? I argue that be-j Adjoins as an adjunct, but which then selects the V1 as
its complement in order to fulfill the semantic relationship that be-j introduces between V2 and
V1. This means that while be-j also adjoins as an adjunct, its presence makes V1 into a
complement. We can see this complementary relationship between V1 and V2 in the tree below:

25
Fig. 4
The reason that be-j ought to be analyzed as selecting V1 as its complement is because it
introduces a semantic relationship between the V1 and V2. This relationship is not any different
from the added information that -j provides when affixed to a lexical verb—it still provides
Manner information about how an action was accomplished. The only difference is that it
provides the flexibility to use another Aspectual suffix while preserving the Manner relationship
between the two verbs.
Thus the V1 becomes necessary to complete the expression, whereas the V1 is
completely optional in the sentence and structure in (25). There is additionally some precedent
for arguing that V1 verbs merge as complements in this environment; Privoznov (2022) argues
the same thing, although there is no interceding equivalent of be-j, and vP-converbs can only
merge as complements to verbs of position and motion.
We can see that the copula + -j construction introduces a specific semantic relationship
by looking at the data in (32):

26
32) bi toguli-j be-rat shitau
I play-prg cop-? burn
“I was playing around and burned (my hand)”
Here, we can see the same copular converb construction with a different converb. The
suffix -rat introduces a different semantic relationship between V1 and V2 than -j does. In fact, -
j’s presence on baan would introduce an unacceptable relationship between the two lexical
verbs. This demonstrates that the semantic relationship between V1 and V2 changes depending
on the converb suffix. As has already been noted, -j has available to it both an Aspectual and a
Manner reading. The only difference between baan-suffixed -j and lexical-verb-suffixed -j is that
when it is affixed to a lexical verb, it has both the Manner reading and Aspectual reading
available to it, but when it affixes to baan, it only has access to the Manner reading. This is
unsurprising since a copula can’t bear Aspect in the same way a lexical verb can. Therefore,
since we see that -j creates the same Manner relationship between V1 and V2 when it is present
on the lexical verb and the copula, there is evidence in favor of analyzing -j as a single structure
which remains unchanged regardless of the verb it affixes to.
Notes on a Coordinating Analysis for be-j
Since be-j introduces a complementation structure, what are we to do with a structure that
introduces multiple lexical verbs affixed with -j, as in the data below?
33) bi ombu-j gwi-j be-j ir-sn
I swim-prg run-prg cop-cnv come-pst
“I came by swimming and running.”
It seems reasonable to assume that if be-j does indeed introduce a complementation structure,
both ombu-j and gwi-j are selected as its complements under a coordinating analysis. This makes
more sense than to have one lexical verb selected as its complement, and the other Adjoined as
an adjunct. Given the evidence in the literature of converbs demonstrating both coordinative and

27
subordinative properties, it would be unsurprising to find that coordination plays a role in be-j
structures involving multiple lexical non-final verbs. However, that is beyond the scope of this
paper, and further research is needed to determine the extent to which coordination is involved in
be-j structures.
A Note on Height:
As I have shown in the Literature Review, height can be an important factor in explaining
the capabilities of converbs; we have also seen the necessity of proposing different attachment
sites. It remains to be seen whether an explanation of -j’s capabilities requires a similar proposal
of multiple attachment sites.
Here, I have proposed that -j heads an AspP that Adjoins below Voice as the daughter
and sister of the v2P, and which contains the v1P as part of the adjunct. I have proposed that the
ConverbP structure Adjoins at vP. To reiterate, both Sugar and Major posit different landing sites
for this construction, the former at vP and the latter at VP. Further research and analysis is
needed to determine whether -j attaches at vP or VP. I have, however, tried to eliminate TP as a
possible landing site because of the data in (33):
34) Baatr tumurlgig cho-xut be-j tikshit-sn, bi bas tig-sn
Baatr metal.object pound-? cop-cnv flatten-pst, I again like.this-pst
“Baatr flattened the metal by pounding it, and I did it like this, too.”
We can see that at the very least, the structure Adjoins below TP, given the data in (31). Here,
we can see that the phrase ‘tumurlgig cho-xut be-j’ functions as the antecedent for the ‘like.this’
construction in the following clause. Given this, we can infer that the phrase must Adjoin at least
below TP, since Tense is present in the following clause.
However, this problem needs further analysis. It could be that -j can Adjoin at two
different heights depending on whether it selects the copula or a lexical verb as its complement. I

28
mentioned earlier that be-j only has the converbal Manner reading available to it, while a lexical
verb can access both Manner and Aspect. It would be interesting to determine whether height
plays a role in this—specifically, if this is because be-j Adjoins higher than AspP—since it is
clearly possible to have both be-j and another aspectual marker present in the same clause, as the
data above in (29) demonstrates.
Monoclausality
Finally, we come to the question of monoclausality. There are three main points to this
section: first, the difficulty of getting one-to-one data similar to Sugar’s; second, the size of the
structure headed by -j as an indication about its monoclausality; third, NPI licensing as stronger
evidence for -j’s monoclausality.
Sugar is able to demonstrate the monoclausality of Uyghur –(i)p-suffixed verbs because it
is possible in Uyghur to have multiple suffixes on the –(i)p-suffixed verb. He uses this to
demonstrate that a passive morpheme present on the –(i)p verb indicates that there is only one
locus for passive morphology available in the entire construction, and if two are present (on V1
and V2, respectively) this changes the entire meaning and structure of the sentence, making it
multiclausal. However, this is not possible to do in Alashan. In Alashan, when the suffix -j has
been affixed to the lexical verb, no other suffixes are allowed to be present. This makes it
impossible to establish via suffix presence whether there are multiple loci of voice present in the
construction.
However, if the structure headed by -j is so small that it cannot house any other
morphology, including Tense and Voice, this would seem to point to the conclusion that the
structure is not a clause. There is also additional evidence for this from NPI licensing. Consider
the data in (35—37):

29
35) bi ter jamrch hool id-sn=gwi ge-le
I she any food eat-pst=neg say-pst
“I said she did not eat any food.”
36) *bi ter jamrch hool id-sn ge-le=gwi
I she any food eat-pst say-pst=neg
Intended: *“I did not say she ate any food”
37) bi Japon-hil-ig jamrch dorok-t uzhi-j be-j sor-sn=gwi
I Japan-tonge-acc any TV-dat? watch-prg cop-cnv learn-pst=neg
“I did not learn Japanese by watching any television.”
Here, we have an NPI marker, jamrch, as part of the complement phrase to be-j, which
establishes the Manner relationship between uzhi-j and sor-sn. We can also see that the negation
marker gwi is modifying the matrix verb sor-sn. As we can see from the data in (35) and (36),
and as Sugar (2022) affirms, the licensing of n-words is clause-bound—in other words, a
negative marker in one clause cannot license an NPI in another clause. Since the negation marker
gwi is modifying the matrix verb sor-sn, if the structure headed by -j was clausal, it would be
impossible for the negation marker in the matrix clause to license an NPI in an embedded clause.
This is further evidence that the structure headed by -j is part of the main clause, and does not,
indeed cannot, house sufficient clausal morphology to be considered a clause.

30
Chapter 5: Crosslinguistic Variation
There are several Mongolian varieties that possess the -j morpheme: Khalkha, Ordos,
Oirat, and Kalmuck (Svantesson 2003; Georg 2003; Birtalan 2003; Bläsing 2003). In most
varieties, it functions as both a converbal suffix and as an aspectual suffix. As an aspectual
suffix, it creates the progressive construction by combining with the copular axuiliary baan
(Svantesson 2003; Georg 2003), which is the same in Alashan. In three of the four varieties, -j is
taken to be an imperfective converb that denotes simultaneity. Below I will discuss the Khalkha
and Ordos varieties, and I will touch on the Kalmuck variety briefly.
Below is data involving -j from Khalkha, the most widely-spoken variety of Mongolian:
38) ter or-j ir-sen
he enter-cnv come-pst (gloss mine)
‘He entered and came.’
“He came in.”
(Svantesson 2003: 169-70)
39) Ter sandal deer suu-j nom unshi-j baina
that chair upon sit-cv:imperf book read-prg-dur
‘He is sitting in a chair reading a book.’
(Svantesson 2003: 173)
Here, we can see that (38) follows exactly the same structure as in Alashan (25) and Uyghur
(12b). Interestingly, (39) seems to demonstrate a more truly coordinated structure than any we
have yet seen. There does not seem to be a Manner reading present at all; there also doesn’t seem
to be any kind of semantic or causal relationship between suu-j and unshi-j. It’s possible that this
is equivalent to either Major’s TP-adjoined –(i)p proposal, or Sugar’s bieventive-butmonoclausal approach. In either case, given the lack of causal relationship between the two
verbs, it’s reasonable to tentatively assume that the data in (39) is illustrating a higher level of
Adjoinment than VP.

31
There is little data on Ordos in English sources—however, two pieces of data from a
short grammar on Ordos (Georg, 2003) are below:
Progressive
40) kara budaa ide-ji bai-ji karada-ji bain
millet eat-prg cop-cnv sick-prg cop (gloss mine)
‘he has been eating plain millet and (now) he is sick’
(Georg 2003: 206)
Clause Chaining
41) öglööni bos-ci ündür mudundaaraan gar-ci . . .
‘(he) rose in the morning, climbed his high tree and . . .’
(Georg 2003: 206)
Georg (2003) claims that, of the two constructions in which -j is present, one is a clause-chaining
construction, while the other is a progressive construction. There is limited information to be
gleaned from the data in (40) and (41); however, some cursory observations may be noted, both
from the data itself and what the author has said about the data. First, he makes a distinction
between a progressive construction and a clause-chaining construction. We can see that the
progressive construction in (40) is very similar to the Alashan construction in (28). We can also
see from the translation that there may even be a similar Manner relationship between the verbs:
it is implied that as a result of eating plain millet, he is now sick. The same resultative
relationship is present in the Manner relationships that -j can establish. Consider the fact that one
can translate (28) as “He lost weight as the result of swimming.” Ideally, we would have more
data from Ordos to corroborate this, but it seems apparent that the construction in (40) is
identical to the one in Alashan. In any case, we can see that more is happening in (40) than just a
simple progressive construction.
The data in (41) presents several interesting points of note. In order to show that (41) is a
proper clause chain, in the sense that there are multiple clauses present, one would need to begin

32
by demonstrating that the Tense of the final verb is interpreted on verbs that are non-tensed.
Georg certainly implies this in the translation, where the progressive aspect seems to be
swallowed up by a past tense, but given that he does not include either the entire sentence (with
the final verb) or the gloss, it is difficult to compare this data either to clause chaining
constructions or Alashan -j Manner constructions.
It may be that the data in (41) is demonstrating a different attachment site for -j, at TP or
CP. Sugar states that one of the characteristics of his vP-(i)p serial verb chain is that there is no
intervening object between the –(i)p-suffixed verbs. Sugar’s argument for a TP-level –(i)p is that
it is monoclausal but bieventive. Likewise, Major (in press) demonstrates that at the TP-level, -
(i)p-suffixed verbs do not enter into any sort of semantic or causal relationship with each other.
The data in (41) could be displaying a similar construction.
In the Kalmuck variety alone is -j presented as a perfective converb. Bläsing (2003), in
her brief treatment of the suffix, notes that -j puts the predicates involved in a more concrete
relationship with one another. She illustrates this with the following example:
42) küükd inäld-j xääkrlkw
‘The girls shouted laughing.’
(Bläsing 2003: 243)
We can see that this seems identical to the Manner relationship -j introduces in Alashan. It is
unclear why Bläsing identifies the morpheme here as perfective.

33
Chapter 6: Suggestions for Further Research & Conclusion
Further study of the case system as it relates to case marking and object control could
prove insightful, as Sugar (2022) relies on evidence from both to make his argument for
monoclausality more forceful. There are also several morphemes that are inadequately
documented and understood, particularly -xut, -rat and -ed. Further analysis of the verb tig and
its function as a verbal anaphor would also prove interesting in its own right. Additionally, it
would be useful in providing a unified theory of -j to look at the different attachment heights that
seem available to it in other dialects.
In conclusion, we have seen that -j is able to introduce a Manner relationship between
two lexical verbs, either by affixing to a lexical verb (in which case it keeps its aspectual
interpretation) or to the copula. We’ve also seen how it compares to the Uyghur –(i)p
construction and to other -j morphemes in other Mongolian dialects. I have demonstrated that
there is no reason to treat -j constructions as clause chains, and demonstrated that there is good
reason to treat them as ConverbP structures that Adjoin as adjuncts but do not have enough
structure to be considered clauses.

34
References
Alexiadou, A., Anagnostopoulou, E., and Schäfer, F. (2006). The properties of anticausatives
crosslinguistically. Phases of interpretation, 91:187-211.
Birtalan, Á. (2003). Oirat. In J. Janhunen (Ed.), The Mongolic languages. (pp. 210—228).
Routledge.
Bläsing, U. (2003). Kalmuck. In J. Janhunen (Ed.), The Mongolic languages. (pp. 229—247).
Routledge.
Cinque, G. (1999). Adverbs and functional heads: A cross-linguistic perspective. Oxford
University Press, Incorporated.
Choe, H. S. (1988). Restructuring parameters and complex predicates—A transformational
approach. Dissertation, Massachusetts Institute of Technology.
Csató, É. Á., & Johanson, L. (2018). Grammaticalization in Turkic. In Grammaticalisation from
a typological perspective (Vol. 31, pp. 146–165). Oxford University Press.
https://doi.org/10.1093/oso/9780198795841.003.0008
Déprez, V. (2000). Parallel (a) symmetries and the internal structure of negative expressions.
Natural Language & Linguistic Theory, 18(2):253-342.
Georg, S. (2003). Ordos. In J. Janhunen (Ed.), The Mongolic languages. (pp. 193—209).
Routledge.
Giannakidou, A. (1998). Polarity sensitivity as (non) veridical dependency, volume 23. John
Benjamins Publishing.
Giannakidou, A. (2000). Negative . . . concord? Natural Language & Linguistic Theory,
18(3):457-523.
Giannakidou, A. (2006). N-words and negative concord. The Blackwell companion to syntax,
3:327-391.
Haspelmath, M., & König, E. (1995). Converbs in cross-linguistic perspective. Structure and
meaning of adverbial verb forms—Adverbial participles, gerunds. De Gruyter Mouton.
Haspelmath, M., König, E., Oesterreicher, W., & Raible, W. (Eds.). (2001). Language typology
and language universals 2. teilband. De Gruyter, Inc.
Kratzer, A. (1996). Severing the external argument from its verb. In Phrase structure and the
lexicon, pp. 109-137. Springer.

35
Major, T. (to appear). Re-analyzing “say” complementation: implications for case theory and
beyond.
Mansfield, J. & Barth, D. (2021). Clause chaining and the utterance phrase: Syntax–prosody
mapping in Matukar Panau. Open Linguistics, 7(1), 423-447. https://doi.org/10.1515/opli2021-0023 (published by De Gruyter Open Access, July 17, 2021)
MacDonald, L. (1990). A grammar of tauya. Mouton de Gruyter.
Nedjalkov, I. (1998). Converbs in the languages of Eastern Siberia. Language Sciences
(Oxford), 20(3), 339–351. https://doi.org/10.1016/S0388-0001(98)00008-4
Nedjalkov, V. P. (1995). Some typological parameters of converbs. In: Haspelmath & König
(eds.), 97-136.
Nedjalkov, V. P., & Otaina, G. A. (2013). Converbs. Syntax of the nivkh language: The Amur
dialect (Geniušienè, E., Gruzdeva, E.) John Benjamins Publishing Company.
Privoznov, D. K., (2022) “Adjunct islands are configurational”, Glossa: a journal of general
linguistics 7(1). doi: https://doi.org/10.16995/glossa.5863
Progovac, L. (1988). A binding approach to polarity sensitivity. PhD thesis, University of
Southern California.
Progovac, L. (1993). Negative polarity: Entailment and binding. Linguistics and Philosophy,
16(2): 149-180.
Ramstedt, G. J. (1903). Über die Konjugation des Khalkha-Mongolischen. (ChatGPT 3.5,
Trans.) https://babel.hathitrust.org/cgi/pt?id=uc1.ax0003920212&seq=7 . Accessed
January 7 2024.
Sugar, A. D. (2019). Verb-Linking and events in syntax: The case of Uyghur –(i)p constructions.
University of Washington. Dissertation.
Svantesson, J. (2003). Khalka. In J. Janhunen (Ed.), The Mongolic languages. (pp. 154—176).
Routledge.
van Gelderen, E. (2013). The clause: a description. In Clause Structure (pp. 39–74). chapter,
Cambridge: Cambridge University Press.
Weisser, P. (2015). Derived Coordination: A minimalist perspective on clause chains, converbs
and asymmetric coordination. Dissertation.
Zanuttini, Raffaella. (1997). Negation and verb movement. The New Comparative Syntax. 214-
245.

Abstract (if available)

Abstract This thesis is concerned with the converbal and aspectual suffix -j in the Alashan variety of Mongolian. It covers the suffix’s adjunct, height, and clausal status, and explains the nature of the relationship between its converbal and aspectual functions. I argue that -j heads an AspP that Adjoins at or below vP, and which can create a monoclausal serial verb chain (as opposed to a multiclausal clause chain), similar to the morpheme –(i)p in Uyghur. This thesis also briefly covers -j and its functions as they appear in other varieties of Mongolian, specifically Khalkha, Ordos, and Kalmuck. More generally, I argue that the claimed relationship between converbs and clause chains is illusory in Mongolic languages.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

The grammar of individuation, number and measurement

PDF

Subject and adjunct island effects through the lens of selective opacity in Bangla subjects and adjuncts

PDF

Silence in answers: a study of ellipsis in Hindi

PDF

Building phrase structure from items and contexts

PDF

Syntax-prosody interactions in the clausal domain: head movement and coalescence

PDF

The morphosyntax of states: deriving aspect and event roles from argument structure

PDF

Reference time in the dynamics of temporal dependency in Korean

PDF

Decomposing Slavic aspect: the role of aspectual morphology in Polish and other Slavic languages

PDF

Exploring the effects of Korean subject marking and action verbs’ repetition frequency: how they influence the discourse and the memory representations of entities and events

PDF

Copy theory of movement and PF conditions on spell-out

PDF

What I learned from 28 years of drawing (from) whatever

PDF

Bright portals: illness & the environment in contemporary poetry

PDF

Factors affecting relative clause processing in Mandarin

PDF

On the Kronecker product of Schur functions

PDF

Between literary training and literary service: school writing instruction in the Russian Empire, 1780-1860

PDF

A curriculum to teach innovation in K-4 gifted classrooms

PDF

Number marking and definiteness in Bangla

PDF

Syntactic and non-syntactic factors in reflexive pronoun resolution in Mandarin Chinese

PDF

Dynamics of multiple pronoun resolution

PDF

Comparative iIlusions at the syntax-semantics interface

Asset Metadata

Creator Cerling, Ella (author)

Core Title The converb -j in Alashan Mongolian

School College of Letters, Arts and Sciences

Degree Master of Arts

Degree Program Linguistics

Degree Conferral Date 2024-05

Publication Date 06/13/2024

Defense Date 06/03/2024

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

Tag Adjunct,Alasha,aspectual suffix,clause chain,converb,Mongolian converb,OAI-PMH Harvest,serial verb chain

Format theses (aat)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Major, Travis (committee chair), Disner, Sandra (committee member), Simpson, Andrew (committee member)

Creator Email ellacerling@icloud.com,ercerlin@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC1139963AM

Unique identifier UC1139963AM

Identifier etd-CerlingEll-13098.pdf (filename)

Legacy Identifier etd-CerlingEll-13098

Document Type Thesis

Format theses (aat)

Rights Cerling, Ella

Internet Media Type application/pdf

Type texts

Source 20240614-usctheses-batch-1169 (batch), 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 author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright.

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Repository Email cisadmin@lib.usc.edu