Morphophonological rules applying verb initially

Get Complete Project Material File(s) Now! »

Verbal valency cross-linguistically

This chapter provides an overview of the valency-changing constructions cross-linguistically. Basic notions are presented in §1.1, including semantic and grammatical roles, transitivity and valency, and the referentiality of the verb’s arguments. The main valency-increasing and valency-reducing operations are then described one by one in §1.2 and §1.3, respectively. In §1.4 I present the possible formal encodings of valency change: first the valency alterna-tions which are not encoded, then four types of encoding ordered from the most synthetic (lexical pairs) to the most analytic (multiclausal constructions). I conclude this chapter with some considerations regarding the description of valency change (§1.5): cases of syncretism, and more generally various blurred boundaries between the categories we intend to identify (continuums, stages of grammaticalization).

Basic concepts

In order to appropriately describe argument structure and argument marking, it is necessary to distinguish between a number of notions and to define them precisely. Distinctions between the semantic and the syntactic level are particularly important. In this section, I will present the notions of argument structure and semantic and grammatical roles (§1.1.1), the differences between arguments and adjuncts, and between core and peripheral expressions (§1.1.2), and the notions of transitivity, valency, and voice (§1.1.3). Finally, I will comment on some issues of argument referentiality (§1.1.4) and present the valency-changing operations by way of an introduction to the following sections (§1.1.5).

 Argument structure, semantic and grammatical roles

The lexical semantic structure1 refers to the number of arguments a given verb takes, and the semantic roles associated with each of them. The lexical semantic structure of a verb is mapped into a syntactic construction, with language-specific grammatical roles assigned to its arguments. It is a well-known fact that semantic roles and grammatical roles do not always coincide, both cross-linguistically and inside languages. Examples sentence (1) is a well-known illustration of how a given verb may map its arguments onto different syntactic structures.
Since Fillmore’s (1966, 1968) study of “(deep structure) cases” and Gruber’s (1965) study of “thematic relations”, the semantic roles and their mapping into various syntactic structures have been a central subject of research (among many others: Jackendoff 1972, 1976, Dowty 1986, 1991, Grimshaw 1990, Andrews 2007[1985])2. Semantic roles can be defined as “what the verb implies about the way each entity represented by a nominal constituent is involved in the process signified by the verb” (Creissels 2006a: 279)3. In reality, there is an unlimited number of possible semantic roles, according to the degree of specificity we choose to consider, since each verb assigns unique semantic roles to each of its arguments. More generally, labels such as “agent”, “patient”, “theme”, and “goal” have emerged in the literature, and serve to characterize various types of participants. As Andrews (2007[1985]) observes, semantic roles are generally not explicitly defined, but are rather given basic characterizations with exam-ples. Below is a reduced list of definitions found in Andrews (2007[1985]), Givón (2001a) and Creissels (2006a), that will be particularly relevant in this dissertation. Note that the labels and the particular characteristics associated with each can vary among scholars4. agent: “a participant which the meaning of the verb describes as doing something, or causing something to happen, possibly intentionally” (Andrews 2007[1985]: 137). It is typically animate. The hunter killed a deer. patient: “the participant, either animate or inanimate, that either is in a state or registers a change-of-state as a result of an event” (Givón 2001a: 107) The hunter killed a deer.; I ga e my textbook to Max.; The window broke.
Various names are commonly used in the literature, especially “thematic relations” and “thematic roles” (among others, works by Jackendoff 1972, 1976, Grimshaw 1990, Dowty 1986, 1991). Palmer (1994) calls them “notional roles”. Following Andrews (2007[1985]), Givón (2001a) Foley & Van Valin (1984) and others, I use the term “semantic roles”, as it highlights the syntax-semantic interface (this definition covers the semantic role of “theme” in other works: “an argument which undergoes a change of slate or location.” (Goldberg 1995: 111))
• beneficiary: “the participant, typically animate, for whose benefit the action is performed” He xed the roof for his mother. (Givón 2001a: 107)
recipient: “animate being toward which something or someone goes or is carried” (Creissels 2006a: 281)5 Sam ga e the piece of land to his son Sam recei ed/got/ac uired a package.
(Goldberg 1995: 111) (the recipient is sometimes named ‘goal’, e.g. by Marantz (1984)) cause: “inanimate entity which affects unconsciously and involuntarily a patient: The wind broke the branch.” (Creissels 2006a: 281)6 instrument: “a participant, typically inanimate, used by the agent to perform the action” (Givón 2001a: 107)
The syntactic construction of verbs, i.e. the way verbs encode their arguments, is referred to as the argument structure of the verb or its valency. The notion of “argument structure” (see among many others: Williams 1981, Marantz 1984, Grimshaw 1990, Goldberg 1995) has been developed in the formal-logical grammatical tradition. The notion of valency is an equivalent, more usually found in works taking a functional perspective (among others: Dik 1997, Lazard 1994, Dixon & Aikhenvald 2000b). I will present this notion in more detail in §1.1.3.

Arguments vs. adjuncts, core and peripheral

A verb argument can be defined as a participant which is verb-specific, while an adjunct is not (Haspelmath & Hartmann 2015). Argumenthood can be recognized through coding-specificity, obligatoriness or specificity of occurrence7. A core syntactic role, or core argu-ment, is a nominal entity whose marking unambiguously identifies it as a verbal argument. A peripheral syntactic role is a nominal entity that receives a marking which is sometimes used for verbal arguments, and sometimes for adjuncts. Table 1.1 summarizes the correspon-dence between arguments and adjuncts at the semantic level, and between core and peripheral syntactic roles at the syntactic level. The distinction between peripheral arguments and ad-juncts has been an issue in grammatical theories at least since Chomsky (1965). Example (2) exemplifies how the same peripheral syntactic role can be a verb argument (2a) or an adjunct (2b).
Note that, unlike what is shown in Table 1.1, adjuncts can sometimes be formally identical to core arguments. This is often the case for time adjuncts, as in (3). This is a particularly dif-ficult issue in Umóⁿhoⁿ, where manner adjuncts and possibly also place adjuncts are expressed as NPs, and especially as bare nouns. See §2.5.3.

Transitivity, valency and voice

Research on transitivity, valency, and voice have lead to countless papers and books in linguis-tic research (among many others, Tesnière 1959, Hopper & Thompson 1980, Fox & Hopper 1994, Lazard 1994, Dixon & Aikhenvald 2000b, Creissels 2002, 2016), and transitivity has been a central issue in the development of linguistic theories (Fodiatou & Vassiliadou 2017). However, these terms have been used with different definitions. In this section, I will present them and specify in which way they will be used throughout this study.
Transitivity. Transitivity can be defined as a semantic concept, referring to clauses where an argument identified as an agent is performing a conscious and voluntary action that affects an identified patient, without being affected themself. This is, for example, the definition ap-plied by Lyons (1968) (cited in Hopper & Thompson 1982), and the starting point of Hopper Thompson’s analysis of the transitivity scale (1980). To kill and to break are well-known prototypical transitive verbs.
In particular language descriptions, “transitivity” also stands for the construction in which prototypical transitive verbs are inserted. Thus, in English, the semantically transitive verb to kill is inserted in a construction with two core arguments realized as NPs, without any particular marking on them. Many verbs that are not semantically transitive (like to hit, to see, to miss) have the same construction so that “transitivity” is also regularly defined with formal criteria: a construction with two core arguments (see for example Humphreys 1999: 391). In English and in many other languages, it refers to constructions with two arguments identified as the subject and the direct object. According to this definition, (4) gives three transitive clauses in English, despite the fact that only (4a) is semantically transitive.
Thus, “semantic transitivity” must be distinguished from “syntactic transitivity” (Creis-sels 2016: 18). Semantic transitivity is considered a scalar concept rather than an absolute category. Hopper & Thompson’s paper (1980) proposes different features according to which events are considered high or low on the transitivity scale. Conversely, syntactic transitivity can be seen as a binary concept. It can be extended with such labels as “extended intran-sitive” or “ditransitive” (Dixon & Aikhenvald 2000b). In this dissertation, I will use the standard labels of “intransitive”, “transitive” and “ditransitive” verbs to classify verb classes in Umóⁿhoⁿ (see Chapter 4). This refers to syntactic transitivity.
Valency. The notion of valency was first introduced by Tesnière (1959) to refer broadly to the number of arguments that a verb selects. While there is universal consensus over Tes-nière’s definition, scholars disagree on which elements should be included in the valency of a verb. Some scholars, like Dixon & Aikhenvald (2000a), relate valency to the number of core arguments, and mainly describe operations which add or remove core arguments to/from the verb. Humphreys (1999: 391) considers that valency designates the total number of ar-guments selected by the verb (core and peripheral), in contrast with syntactic transitivity which designates the number of core arguments only. Finally, many scholars include in the notion of valency the total number of arguments and the way they are encoded. I employ this final definition, which corresponds to the notion of “argument structure” (which itself is a central topic of interest in theoretical formal syntax; see, among many others, Williams 1981, Marantz 1984, Grimshaw 1990, Goldberg 1995). This definition also corresponds to the notion of “actance” described by Lazard (1994).
Example sentences (5) through (7) illustrate valency alternations according to this defi-nition, because each deals with different mappings of semantic roles onto grammatical roles. However, only examples (7) and (8) present a change in the number of arguments, and the valency alternation is overtly encoded on the verb only in examples (6) and (8).
Although valency is defined here as the number of arguments and the manner in which they are encoded, alternations can be described with labels such as “valency-increasing”, “valency-reducing” and “valency-rearranging” (Onishi 2000) to distinguish cases where the total number of arguments changes (examples (7) and (8) above), from those where only their grammatical encoding changes (examples (5) and (6)). Likewise, the terms “avalent” (no argument), “monovalent” (1 argument), “bivalent” (2 arguments), and “trivalent” (3 arguments) are sometimes used as equivalents to “impersonal”, “intransitive”, “transitive”, “ditransitive”, respectively.
Voice. The “grammatical voice” (i.e., voice as a grammatical category of verbal morphology) is a very ancient category in descriptive grammars, already identified in Sanskrit, in Ancient Greek and in Latin (Klaiman 1991). In Latin and Greek descriptive traditions, it was used as a verbal grammatical category along with tense, aspect and modality. Traditionally recognized voices are the active, the middle, and the passive.
Many scholars now use this term when describing of the different syntactic constructions affecting verb valency (the number of arguments and the way they are encoded). In short, the terms valency and voice generally cover the same concepts. (For instance, Dixon & Aikhen-vald (2000b) only use “valency”.) Constructions attested outside Indo-European languages are or are not considered voices depending on the precise definition chosen for “voice”, which may vary among scholars. These include the “antipassive” (Foley & Van Valin 1984, Shibatani 1988, Fox & Hopper 1994) and the “inverse” (Givón 1994, Fox & Hopper 1994, Croft 2001)8, for instance. Some scholars, like Creissels (2016) and Zúñiga & Kittilä (2019), consider voices to be formally marked valency frames9, and thus also include causative constructions and applicative constructions as voices. The term “voice” (from Latin vox) is linked to “diathesis” (from Greek). Scholars traditionally use either one or the other, although “voice” seems to be more common in recent literature, and generally includes more constructions than “diathesis”.
The “inverse” system, first described in Native American languages (Fox & Hopper 1994: ix), neither increases nor decreases the number of core arguments. It appears in languages that have a topicality hierarchy of arguments, and often occurs alongside the obviation sys-tem, which morphologically distinguishes topical (proximate) and non-topical (obviative) ar-guments10. The “inverse” marker appears to signal that the agent-like argument is lower in the hierarchy scale than the patient-like argument. An inverse construction is given in (9): The proximate argument (unmarked) is higher on the hierarchy than the obviative one (morphologically marked), but is the patient11.
Inverse construction in Plain Cree (Dahlstrom 1986: 77, cited in Givón (2001b)) osa m e -sa khikot ohta wiy-ah aw o skini kiw much love-obv.3.inv his.father-obv this young.man (for) his father[obv] loved this young man [prox] too much.

Argument referentiality

In all the above section’s examples, an increase or reduction in valency was visible both in syntax (the number of arguments expressed in NPs and/or realized by indexation markers on the verb) and in semantics. But the number of arguments involved in an activity or event at the semantic level does not always correspond to the number of arguments that are realized. There are cases where arguments are not realized but nonetheless are semantically implied, and conversely there are cases where a formal realization serves a grammatical purpose, but does not correspond to any referent at the semantic level. This last case involves the question of the referentiality of the nominal expressions and pronouns.
Note that conversely, some scholars like Palmer (1994) or Creissels (2006b: 51) do not consider inversion as a voice. Lazard (1994) uses the term “diathesis” , but he includes in it the antipassive and makes no reference to the inverse.
9Creissels writes that voice is a “morphologically coded valency alternation”. Zúñiga & Kittilä (2019: 10) write: “[The present study] defines diatheses as mappings of the roles of the semantic arguments of predicates
onto grammatical relations in clauses, and voices as diatheses formally marked on predicates”.
See §2.6.1 for a description of the obviation system in Umóⁿhoⁿ.
See Jacques (2010) for a description of the inverse system in Japhug Rgyalrong (Gyalrongic, China), where it serves as a means for recognizing who is acting on whom.
Missing arguments. It is often the case that verb arguments are not overtly expressed. This occurs with greater or lesser systematicity across languages. Languages where verbs encode the person and number of the subject often do not express it overtly, like all “pro-drop” languages. Umóⁿhoⁿ and other Siouan languages are pro-drop languages, and additionally they never encode 3rd person arguments on the verb (except for O3pl in Umóⁿhoⁿ and some others;
see §3.1.3 and §7.7). Thus, 3rd persons can only be overtly realized by noun phrases (NPs), and these NPs are often missing when they are recoverable from the context. In (10) the transitive verb thixóⁿ ‘to break {x}’ is a clause in itself, and is often used with no NPs when the subject and object are both already known. The high frequency of argument omission in Umóⁿhoⁿ makes it difficult to establish from the corpora whether or not a given verb involves an object argument. (See Chapter 7 for discussion of the antipassive.)
The missing 3rd person subjects in pro-drop languages also lead to difficult interpretations related to the analysis of passive or passive-like constructions; see §4.8 for an analysis of the passive reading in Umóⁿhoⁿ.
In all languages that do not index the object on the verb, non-expression of the object can be interpreted in different ways depending on the verb and the context12. In French, object omission commonly occurs with many transitive verbs to direct focus to the process and defocus the object, which is unimportant (and often indefinite or nonspecific). This is the “absolutive construction” of transitive verbs. On the other hand, a definite and contextually prominent object can be omitted yet remain referential. Both cases are illustrated in (11)13. Note that Umóⁿhoⁿ bare nouns have the same distinct referential statuses: either indefinite or nonspecific, or attention-central (§8.4).
Omission of a referential object at the center of attention Alors, tu ach tes? So, are you buying (it)
Note that even in absolutive uses, omission of the object does not necessarily mean that the object is completely deleted. In French, for example, the missing object of absolutive
For English, see Humphreys (1999), which provides an overview of valency-changing alternations in English, many of them unmarked, and different semantic interpretations for the missing arguments.
Haspelmath & Hartmann (2015) propose to include the interpretation of omitted argument among the possible tests to recognize argumenthood. They note: “Often one can make a clear distinction between an anaphoric and an existential interpretation of argument omission, and when argument absence implies an anaphoric interpretation, this could be taken as evidence of verb-specificity and argumenthood”.
This example comes from Creissels (2006a: 274). As he points out, it is the kind of sentence that can be heard in a shop, where a prominent object is the topic of the conversation constructions can be modified: J’ai mangé, et c’était bon! ‘I ate, and it was good!’ Object omission resulting in the object’s complete non-referentiality is more likely to occur with imperfective aspects, in order to describe activities or properties, as in (11a). Croft (2001:
317-8) comments that “the situation denoted by a detransitivized verb form such as drink, or its antipassivized counterpart in other languages, can be analyzed as either a two-participant situation type with an obligatory non-salient participant, or as one participant situation type.”
Empty arguments. Verbs are sometimes used in syntactic constructions involving a noun or pronoun as a syntactic argument, but with empty reference. In Turkish, non-expression of the object is regularly interpreted as anaphoric (Creissels 2006b: 3). On the other hand, Turkish transitive verbs can sometimes be used with what is called the “internal object” (“objet interne”; Creissels 2006b: 3): “to sew sewing”, “to knit knitwear”, etc.15. In such constructions, the syntactic transitivity remains the same, but the semantic transitivity clearly decreases, as the object is not referential. Göksel & Kerslake (2005: 329) mention that constructions with non-referential objects of this type are sometimes found as lexical entries in dictionaries. Two examples of Turkish internal objects are given in (12).
Dummy pronouns are another typical example of a discrepancy between the number of syntactic constituents in a construction and the number of entities involved in the event. Dummy pronouns are used in French and English to fill the subject argument in impersonal constructions: Il pleut (‘It rains’). Meteorological verbs are usually considered avalent (having no argument), irrespective of the presence or absence of a pronoun as a grammatical subject.
In a similar way, reflexive pronouns in Romance languages take the same form as object pro-nouns (except in the 3rd person): Je me regarde ‘I look at myself’ (lit. ‘I look at me’). Thus, on the surface this construction looks like a bivalent construction, but semantically only one argument is involved in the process.
As can be seen in the last examples, the concept of valency primarily takes into account the arguments involved in a process at the semantic level. Dummy subjects are not considered parts of the verb valency because they are semantically empty. Likewise, a given verb’s valency does not necessarily decrease each time one of its arguments is not realized.
Levin (1993: 95) calls the English equivalent a “cognate object”: Sarah sang a song. In English, however, cognate objects do not have the same valency-reduction value as they do in Turkish. Often, they are only possible when the object is modified by relevant information: Sarah sang a beautiful song. See also Boons et al. (1976: 64 ff.) for an analysis of French “internal objects” and its implication for the verb transitivity.
Reference of NPs. Between the non-existence of an argument and the presence of a defi-nite, highly individualized one, there is a continuum of indefinite, plural and nonspecific argu-ments. Indefiniteness and plurality are two factors of decreasing individuality, and thus they are used in the less referential expressions. This is the reason why they are included among the features surveyed in Hopper & Thompson’s (1980) study of the semantic transitivity scale.
The distinction between “specific” and “nonspecific” nominal expressions can be observed in the different possible interpretations of a sentence such as I want to marry a Tahitian woman (Kleiber 1981: 146; my translation)16, where the speaker can have a specific person in mind that they call “a Tahitian woman” (specific reference), or just mean “any Tahitian woman” (nonspecific reference)17. Generic expressions are often opposed to specific expressions (e.g., Corblin 1987: 82), and in fact they can be considered a particular subtype of nonspecific expressions (Corblin 1987: 47).

Describing valency-changing operations

Thus far, I have used the terminology “valency-changing alternations”, which suggests no directionality between a basic construction and a derived, or more marked, construction. In-deed, it is sometimes difficult to clearly determine directionality, as evidenced in (5) and (7), on p. 49 above.
Many valency alternations, however, have a clear directionality; in (8), a basic construc-tion undergoes a valency-changing operation which adds an argument in the subject position. In (8b), the derived construction is marked by means of an affix on the verb. There are many such valency-changing operations, where formal marking indicates a change in the number of core arguments encoded by the verb, and how they are encoded18.
In this dissertation, I focus on formally-marked valency-changing operations that either add or remove a core argument to or from the verb. This choice reflects the valency al-ternation found in Umóⁿhoⁿ: there are numerous valency-increasing and valency-reducing operations that are formally marked on the verb or in the syntactic structure. Conversely, there are no valency-rearranging constructions (with the exception, arguably, of the possessive prefix briefly described in §4.4.3), and unmarked valency changes are marginal. There seems to be a restricted class of labile verbs (§4.1.6), and transitive verbs can receive a passive in-terpretation, which I do not consider to be a case of valency reduction (see discussion in §4.8).
16The original example is: Je eux épouser une Tahitienne. Kleiber argues that the “specific” vs. “nonspecific distinction” is different from the “known” vs. “unknown” distinction, which is a relevant distinction within the specific reference.
There is one famous exception in the identification of directionality. In Philippine languages, it is not possible to clearly identify a basic, unmarked transitive construction to be identified as the default “active” voice. See Lazard (1994: 180), Zúñiga & Kittilä (2019: 122-7), or Creissels (2006b: 17) for commented examples on Tagalog.
Valency-increasing and valency-reducing operations receive different labels. The most well-known ones are the passive, the causative, the reflexive and reciprocal, the antipassive and the applicative. Each implies a different kind of operation on the verbal arguments, represented in Table 1.2. Valency-increasing operations are presented in section 1.2, and valency-reducing operations are presented in section 1.3.

Table of contents :

Introduction 
I The Umóⁿhoⁿ language and the valency-changing operations 
1 Verbal valency cross-linguistically
1.1 Basic concepts
1.1.1 Argument structure, semantic and grammatical roles
1.1.2 Arguments vs. adjuncts, core and peripheral
1.1.3 Transitivity, valency and voice
1.1.4 Argument referentiality
1.1.5 Describing valency-changing operations
1.2 Valency-increasing operations
1.2.1 Adding an A: Causative
1.2.1.1 Syntactic variations and restrictions
1.2.1.2 Semantic variations and restrictions
1.2.2 Adding a P: Applicatives
1.2.2.1 Syntactic variations and restrictions
1.2.2.2 Semantic variations
1.2.2.3 Functions
1.2.3 Other valency-increasing operations
1.3 Valency-reducing operations
1.3.1 Demoting A: Passive
1.3.1.1 Passive constructions in a canonical perspective
1.3.1.2 The functions of passive constructions
1.3.2 Demoting P: Antipassive
1.3.3 Merging A and P: Reflexive / reciprocal
1.3.4 Other valency-reducing operations
1.4 The formal realizations of valency change
1.4.1 No formal marking
1.4.2 Lexical pairs
1.4.3 Verbal morphology
1.4.4 Complex predicates
1.4.4.1 Compounds
1.4.4.2 Noun incorporation
1.4.4.3 Periphrastic CPs
1.4.5 Open syntax: multiclausal constructions
1.5 Syncretism and blurred boundaries around valency-changing operations
2 Grammatical sketch of Umoⁿhoⁿ 
2.1 The Umóⁿhoⁿ language: available documentation
2.1.1 Available documentation
2.1.2 Typology of the documentation
2.1.3 Corpora and databases used in this dissertation
2.2 Phonology
2.2.1 Consonants
2.2.2 Vowels
2.2.3 Accent and prosody
2.3 Morphology
2.3.1 Typological aspects
2.3.2 Nominal morphology
2.3.3 Verbal morphology
2.4 Parts of speech
2.4.1 Verbs
2.4.2 Nouns
2.4.3 Articles
2.4.4 Postpositions
2.4.5 Adverbs
2.4.6 Demonstratives, possessives, quantifiers
2.4.7 Remarks on predication
2.4.8 Remarks on the noun-verb distinction
2.5 Syntax
2.5.1 Word order and NPs
2.5.1.1 Basic argument order
2.5.1.2 Double object constructions
2.5.1.3 Appositive NPs
2.5.1.4 Internal structure of Noun Phrases (NPs)
2.5.2 Grammatical roles and morphological alignment
2.5.2.1 Grammatical roles indexed by the verb
2.5.2.2 Split intransitivity
2.5.2.3 Nominative/accusative features
2.5.3 Core arguments, peripheral arguments, adjuncts
2.5.4 Relative clauses (RCs)
2.5.4.1 Relativized arguments
2.5.4.2 Head nouns are never followed by articles
2.5.4.3 Headless relative clauses
2.5.4.4 Some RC lack a relativizer
2.5.5 Clausal complements
2.5.6 Verb sequences
2.5.7 Possession
2.6 Discourse and pragmatics
2.6.1 Obviation system
2.6.1.1 Formal marking
2.6.1.2 Pragmatic functions
2.6.2 Gendered speech
2.7 Comparison of “19th-century Umóⁿhoⁿ” with “Contemporary Umóⁿhoⁿ”
3 Verbal morphology 
3.1 Verb inflection: introduction to person and number marking
3.1.1 The person category of first dual and plural
3.1.2 Plural marking
3.1.3 The specificity of third person plural object
3.2 Verb inflection: post-verbal markers
3.2.1 Suffixes, enclitics or free forms?
3.2.2 Markers =(a)zhi and -xti
3.2.3 Enclitic =i􀁡bi􀁡b
3.2.4 Auxiliaries and evidential markers
3.3 Verb derivation
3.4 Prefixal template
3.4.1 Template of derivational prefixes
3.4.2 Template with person markers
3.4.3 Detailed template of person markers
3.4.4 Detailed template of person markers with oblique prefixes
3.4.5 Multiple exponence
3.4.5.1 Types of ME with examples
3.4.5.2 Analysis
3.4.5.3 Combinations and exuberant exponence
3.4.6 Conclusion
3.5 Morphophonology
3.5.1 Morphophonological rules applying verb initially
3.5.1.1 Rules affecting particular morphemes
3.5.1.2 Rules affecting vowel sequences
3.5.1.3 Reanalysis
3.5.1.4 Summary and examples
3.5.2 Ablaut
3.5.3 The origin of the athematic paradigms
3.6 D and B: between derivation and inflection
3.7 Conjugational paradigms
3.7.1 Initial consonant alternation
3.7.2 Morphophonological changes induced by derivational prefixes
3.7.3 Combination of different features
3.7.4 Main conjugational paradigms
4 Valency alternations in Umóⁿhoⁿ: an overview 
4.1 Verb classes of Umóⁿhoⁿ: syntactic categories
4.1.1 Avalent verbs
4.1.2 Monovalent verbs
4.1.3 Bivalent verbs
4.1.3.1 Verbs encoding A/P
4.1.3.2 Verbs encoding A/D
4.1.3.3 Bivalent stative verbs
4.1.4 Trivalent verbs
4.1.4.1 Verbs encoding A/P
4.1.4.2 Verbs encoding A/D
4.1.4.3 Verbs encoding A/B
4.1.4.4 Double object marking
4.1.5 Quadrivalent verbs
4.1.6 Accusative and ergative lability
4.1.6.1 Accusative lability
4.1.6.2 Ergative lability of stative verbs
4.2 Causative constructions
4.2.1 Causative bound stems
4.2.2 Causation through instrumental prefixes
4.2.3 Periphrastic causative constructions
4.3 Applicative constructions in Umóⁿhoⁿ
4.4 Dative and benefactive-possessive applicatives
4.4.1 “Dative” applicative
4.4.1.1 The semantic roles introduced by the dative applicative
4.4.1.2 Polysemy of the dative applicative
4.4.1.3 The dative prefix as a non-applicative derivational prefix
4.4.2 Benefactive-possessive applicative
4.4.3 An etymologically related prefix: the possessive gi-
4.5 Oblique applicatives
4.6 Antipassive
4.6.1 Antipassive marker wa-
4.6.2 Object incorporation
4.7 Reflexive/ reciprocal
4.8 Passive reading of transitive clauses
4.9 Derivation, lexicalization, semantic demotivation
4.9.1 Degrees of semantic demotivation
4.9.2 The impact of demotivation on morphology and syntax
4.10 Summary of valency-changing operations
II Case studies 
5 Causative constructions in morphology and syntax 
5.1 The causative bound root -the and its derived forms
5.1.1 Semantic analysis: -the vs. -kʰithe
5.1.2 Other derived causative stems
5.1.3 Positioning and scope of inflection and derivation
5.1.4 Recursivity
5.1.5 Morphological and syntactic features of causative bound roots
5.2 Case study: semantics and syntax of ga- ‘by force’
5.2.1 Instrumental prefixes: overview
5.2.2 Database and methodology
5.2.3 Prefix ga- : meanings
5.2.4 Prefix ga- : syntactic functions
5.2.5 Prefix ga- : semantic analysis of complex predicates
5.3 Other instrumental prefixes
5.3.1 Generalization of the properties of ga- to the other prefixes
5.3.2 Semantic analysis of instrumental verbs
5.3.2.1 Causative construction: instrumental expressing manner, base expressing result
5.3.2.2 Instrumental expressing non-agentive force, base expressing result
5.3.2.3 Instrumental with predicative function, base with an adverbial or circumstantial function
5.3.2.4 Both prefix and base having predicative functions
5.3.3 The causative function of the instrumental prefixes
5.3.3.1 Causative derivation of verbs
5.3.3.2 Causative derivation of non-verbal bases
5.3.3.3 Evolution of thi- towards a neutral causative marker
5.3.4 Instrumental prefixes expressing non-agentive forces
5.4 Periphrastic causative gáxe
5.4.1 Meanings of gáxe and corresponding syntactic constructions
5.4.2 Causative constructions using gáxe
5.4.3 Derivations from gáxe
5.5 Other causative or causative-like constructions
5.6 Comparison of the different causative constructions
5.6.1 Semantic comparison
5.6.2 Recursivity
5.6.3 Formal comparison
5.7 Summary
6 Oblique applicatives 
6.1 Meanings of the applicative prefixes
6.1.1 Locative applicative constructions
6.1.2 Figurative locations
6.1.3 Instrumental and other meanings associated with í-
6.1.4 Applicative í- yielding complex prefixes
6.2 Syntactic properties of the applicative constructions
6.2.1 Verb classes undergoing applicative derivation
6.2.2 Object properties of monotransitive applicative verbs
6.2.3 Object properties of ditransitive applicative verbs
6.2.4 Applicative objects: NPs vs. PPs
6.2.5 Applicative í- introducing clausal complements
6.2.6 Instrumental/causative syncretism: instrumental prefix í- introducing inanimate causers
6.3 Optionality/obligatoriness of the oblique applicative constructions
6.3.1 Locative applicatives
6.3.2 Instrumental applicative
6.3.3 “Reason” applicative
6.4 Other functions and demotivation of the oblique prefixes
6.4.1 Valency-preserving derivational prefixes
6.4.2 Demotivation of the applicative prefixes
6.5 Summary
7 Prefix wa􀀐 : antipassive and other functions 
7.1 The homonymous morphemes wa- in Umóⁿhoⁿ
7.1.1 Third person plural object marker – O3pl
7.1.2 First person plural patientive marker – P1pl
7.1.3 Marker of an underspecified argument
7.1.4 Opaque formations and unknown function
7.2 Formal resemblances and distinctions
7.2.1 Common formal behavior
7.2.2 Formal distinctions
7.3 The subfunctions of wa- as an underspecified argument marker
7.3.1 Generic interpretation: Antipassive marker – antip
7.3.2 Indefinite object – indef
7.3.3 Nominalizer – nmlz
7.3.4 Evolution towards an aspectual marker?
7.4 Antipassive constructions and verbs in Umóⁿhoⁿ
7.4.1 Difficulties in interpreting wa- as an antipassive marker
7.4.2 Basics of morpho-syntax
7.4.2.1 Antipassive on inherent ditransitive verbs
7.4.2.2 Combination of wa- with valency-increasing morphemes
7.4.3 Typology of the Umóⁿhoⁿ antipassive and indefinite object
7.4.4 Antipassive vs. indefinite object: derivation and syntax
7.5 The ambiguities between the functions of wa-
7.5.1 Antipassive vs. O3pl
7.5.2 Antipassive vs. indefinite object
7.6 Wa- : a network of functions
7.6.1 Classification of the functions of wa- in the database
7.6.2 Network and examples
7.6.3 Two possible sources for the antipassive
7.6.4 Wa- as a marker of transitivity decrease
7.7 Comparative data
7.7.1 O3pl
7.7.2 Underspecified argument marker
7.7.3 Wa- attested on nouns
7.7.4 Other functions
7.7.5 Possible sources of the antipassive marker
7.8 Summary
8 Nominal Incorporation 
8.1 Methodology: how to recognize Noun Incorporation?
8.1.1 NI in Siouan languages
8.1.2 Does Umóⁿhoⁿ have Noun incorporation?
8.1.3 Why NI is difficult to recognize in Umóⁿhoⁿ
8.1.4 The database
8.2 Formal evidence of N-V coalescence
8.2.1 Bound nominal elements [1]
8.2.2 Inflection and derivation on the left edge of NI [2 & 3]
8.2.3 Conjugation forms [4 & 5]
8.2.4 Accentual patterns [6 & 7]
8.2.5 Loss of final vowel [8]
8.2.6 Semantic opacity [9]
8.2.7 Summary and discussion
8.3 Morphosyntactic features of NI
8.3.1 Conjugation of incorporating verbs
8.3.2 Nature of incorporated and incorporating elements
8.3.3 Syntactic functions incorporated
8.3.3.1 Core arguments
8.3.3.2 Possessed objects
8.3.3.3 Other syntactic functions and dubious cases
8.3.4 Case study: gthóⁿ vs. míⁿgthóⁿ
8.3.4.1 Míⁿgthóⁿ
8.3.4.2 Gthóⁿ
8.3.5 Summary
8.4 Bare nouns
8.4.1 Absence of determiner as a grammatically determined feature
8.4.2 Absence of determiner in an information-structural variation
8.4.3 Bare nouns losing their syntactic status
8.4.4 From definite syntactic object to NI by juxtaposition
8.5 Possible NI relics
8.6 Summary
Conclusion 
A. Grammatography
B. The interpretation of non-realized arguments
C. Morphology vs. syntax
Bibliography

GET THE COMPLETE PROJECT