Optimizing Display Other Formats:      PDF - 461k

Volume 7 Issue 1 (2009)        DOI:10.1349/PS1.1537-0852.A.334

Note: Linguistic Discovery uses Unicode characters to represent phonetic symbols. Please see Optimizing Display for requirements to accurately reproduce this page.

Participles and Finiteness: The Case of Akhvakh

Denis Creissels

Université Lumière (Lyon 2)

Akhvakh, a Nakh-Daghestanian language belonging to the Andic (sub-) branch of the Nakh-Daghestanian family, has participial relative clauses headed by verb forms that can also head independent clauses. Akhvakh data contradict the inflectional approach to finiteness according to which finiteness as a clausal feature necessarily correlates with the morphological structure of verb forms, and support a constructional approach to finiteness. In particular, the formulation of a general definition of participles must be compatible with the fact that forms found in relative clauses in which they behave at the same time as verbal heads and as adjectival dependents of a head noun may also head constructions having a different status with respect to finiteness.

1. Introduction

Akhvakh ( ašoƛ̱i mic̱’i , Russian axvaxskij jazyk ) ^[1] belongs to the Andic (sub-)branch of the Northeast Caucasian (or Nakh-Daghestanian) family. ^[2] Like the other Andic languages, Akhvakh has no writing tradition. According to Magomedova & Abdulaeva (2007), Akhvakh has approximately 20 000 speakers. Four dialects are traditionally recognized. One of them is designated as Northern Akhvakh, whereas the other three are grouped under the label of Southern Akhvakh.

Northern Akhvakh is spoken in four villages of the Axvaxskij Rajon in the western part of Daghestan (Tadmagitl’, Lologonitl’, Kudijab-Roso, and Izani), and in Axaxdərə near Zaqatala (Azerbaijan), where I carried field work on Akhvakh. ^[3] The Southern Akhvakh dialects are spoken in one village each (Cegob, Tljanub and Ratlub), all situated in the Šamil’skij Rajon of Daghestan.

The analysis proposed in this paper is entirely based on field work carried in Axaxdərə between June 2005 and April 2008. Judging from the data available on AR Akhvakh, there does not seem to be any contrast between AD Akhvakh and the other varieties of Northern Akhvakh in the aspects of grammar dealt with in this paper, but in the absence of more detailed information on AR Akhvakh I prefer to leave this question open.

Traditionally, finiteness is viewed as a morphological characterization of some verb forms correlating with their ability to combine with a canonical subject NP and to head independent clauses, and participles are viewed as a subtype of the more general type non-finite verb form. The aim of this paper is to show that Akhvakh data is incompatible with this position, and provides evidence supporting the approach to finiteness according to which finiteness is a feature of predicative constructions not necessarily correlated in a simple way with the morphological structure of the verb forms involved.

The paper is organized as follows. In section 2, I review the main theoretical approaches to finiteness, and I briefly comment Kalinina & Sumbatova’s proposal to recognize a Nakh-Daghestanian type of finiteness. In section 3, I give some basic information about Akhvakh morphosyntax. In section 4, I examine several aspects of Akhvakh morphosyntax relevant to the question of finiteness. Section 5 is devoted to participles. Section 6 puts the particularities of Akhvakh participles described in section 5 in their historical perspective.

2. Theoretical Approaches to Finiteness

The notion of finiteness originates in the traditional division found in Latin grammars between verbal forms inflected for person (‘verba finita’) and verbal forms devoid of person inflection (‘verba infinita’: infinitives, participles, gerunds, and supines).

According to what can be called the inflectional approach to finiteness, the finite vs. nonfinite distinction relies on the presence vs. absence of some inflectional characteristics, not only person, as in traditional Latin grammar, but also tense, and sometimes others too. The importance given to the classification of verb forms according to the presence vs. absence of some inflectional distinctions reflects the widespread view that “only finite verbs are able to form an independent utterance and that each independent utterance must have one and only one finite verb.” (Nikolaeva 2007a:3)

The limitations of this conception are well-known. The inflectional features posited as being responsible for finiteness are not universal, and counterexamples to the hypothesis of a universal correlation between reduced inflection and inability to head independent clauses are easy to find – see in particular (Nikolaeva 2007a), (Nikolaeva 2007b). This correlation is at best a tendency calling for functional explanations – see (Cristofaro 2007), (Bisang 2007).

Generative syntax developed a more abstract notion of finiteness viewed as a clausal category with the status of functional head, responsible for a variety of syntactic phenomena, in particular the presence of an overt subject in the nominative case in finite clauses, contrasting with its absence in nonfinite structures (control and raising structures, structures in which the subject of a dependent clause receives its Case from the main verb or from the complementizer). Generative syntax also developed the idea that finiteness is relevant to the distribution of referential expressions and anaphoric elements, in the sense that dependent finite clauses constitute opaque domains, not accessible to rules operating in the main clause, as opposed to the accessibility (or transparency) of nonfinite clauses.

However, until recently, the generative approach to finiteness maintained an essential element of the traditional approach, namely the hypothesis of a universal correlation between the syntactic properties of verb forms and the richness of specification of agreement and tense. Faced with data contradicting this assumption, some authors have explored solutions that make it possible to handle the individual cases without entirely dropping the basic tenets of the inflectional approach, but others, in line with the functional literature, have concluded that there is no universal correlation between finiteness as a clausal category and verbal morphology, although there are obvious cross-linguistic tendencies.

In the constructional approach to finiteness, developed in various non-transformational frameworks (Sells 2007), finiteness is a formal characterization of clauses accounting for their ability to constitute independent utterances with particular illocutionary forces and/or the way they can be inserted as constituents of complex structures, but there is no a priori limitation on the possible manifestations of finiteness. Situations where finiteness as a grammatical feature of clauses straightforwardly correlates with the choice between different morphological types of verb forms are viewed as only one of the possible ways of encoding finiteness, and the recognition of the status of a clause with respect to finiteness may also rely on a particular combination of words that, taken individually, cannot be analyzed as bearing finiteness markers. Note also that, in this conception, finiteness must not necessarily be conceived as a binary feature.

Before turning to the presentation and analysis of Akhvakh data, a remark is in order about the hypothesis of a Nakh-Daghestanian type of finiteness put forward by Kalinina & Sumbatova (2007). In spite of the fact that one of the three languages they analyze (Bagvala) is a close relative of Akhvakh, most of their generalizations are contradicted by the variety of Akhvakh analyzed in this paper. Two grammatical points are particularly crucial in their characterization of a Nakh-Daghestanian type of finiteness: the existence of a particular class of auxiliaries, termed ‘predicative particles’, that attach to the head of the focus phrase, and a tendency towards overt morphosyntactic marking of focused constituents. None of the examples they quote to illustrate these points can be transposed to AD Akhvakh. In particular, in AD Akhvakh, the copula in auxiliary function consistently follows the auxiliated verb, and therefore does not participate in focus marking. Note that the existence of this kind of contrast between closely related languages is not exceptional, since a similar contrast has been observed for example between Basque dialects: as shown by Rebuschi (1984:71-77), in most varieties of Basque, the auxiliary immediately follows the auxiliated verb, and moves from this position only in negative clauses, but in the dialects of the French Basque Country (Navarro-Labourdin), the auxiliary acts as a focus marker much in the same way as in the Daghestanian languages analyzed by Kalinina & Sumbatova.

3. General Remarks on Akhvakh Morphosyntax

3.1 Clause Structure

AD Akhvakh clause structure is characterized by flexible constituent order, without clear evidence of a preferred position for focalized constituents.

As illustrated by ex. (1), case marking and gender-number agreement between the verb and its core arguments are consistently ergative. In contrast, assertive agreement (see section 3.4.3) follows a split intransitive pattern.

Arguments whose identity is recoverable from the context can freely be omitted, and unexpressed arguments receiving an arbitrary interpretation are common too.

Causative is the only valency-changing mechanism systematically expressed via verb morphology or grammaticalized periphrases.

3.2 Nouns and Noun Phrases

Three semantically transparent agreement classes of nouns are distinguished in the singular: human masculine (M), human feminine (F), and non-human (N). ^[6] In the plural, the distinction masculine vs. feminine is neutralized, resulting in a binary opposition human plural (HPL) vs. non-human plural (NPL). Noun morphology shows only frozen vestiges of gender prefixes.

In canonical NPs, the head noun is in final position and is inflected for number and case. Noun dependents other than adjectives show no agreement mark and the agreement morphology of attributive adjectives is reduced in comparison with AR Akhvakh or the other Andic languages. ^[7]

As illustrated by ex. (2), in the absence of a head noun, the last word of the NP, whatever its nature, is marked for gender, number and case.

Number inflection of nouns is irregular and involves considerable free variation.

Except for 1st and 2nd person singular pronouns, whose absolute form is characterized by a non-void ending -ne , the absolute form of nominals (used in the extra-syntactic function of quotation or designation and in S or P roles) has no overt mark. Case suffixes may attach to a stem identical with the absolute form, or to a special oblique stem. In the singular, the formation of the oblique stem is very irregular and involves considerable free variation. The formation of the oblique stem is more regular in the plural. In particular, ‘oblique stem markers’ expressing class distinctions (M -s̱u- , F/N - λ ̱i- , HPL -lo- , NPL -le- ~ -li- ) are more systematically used in the plural than in the singular. Ex. (3) illustrates the variety in the possible relationships between the absolute forms and oblique stems of nouns, in the singular and in the plural.

Case inflection includes the following cases:

• ergative (-de),
• dative (-ƛa),
• genitive (Ø or -ƛ̱i),^[9]
• comitative (-k’ena),
• purposive (-ʁana),
• five series of spatial cases, which however tend to depart from the typical Daghestanian pattern in that one of the series (the -g- series) is a default series that does not encode a particular spatial configuration, and spatial configurations tend to be encoded by combining NPs showing default spatial case marking with locative adverbs showing parallel spatial case inflection, rather than via ‘traditional’ case marking.

Each series of spatial cases includes an essive (-i or -e), a lative (-a), and an elative (-u(ne)),^[10] and the distinction between 3 spatial cases applies to locative adverbs too.

3.3. Adjectives

Like verbs (see section 3.4.2), adjectives divide into those obligatorily including a class agreement prefix, and those devoid of it. Like nouns, they cannot bear TAM inflection and fulfill the predicate function by combining with the copula g‹o›di ^[11] or with the verb ‹b›ik’uruƛa ‘be’.^[12]

In Axaxdərə Akhvakh, adjectives in the role of noun dependent or in predicate function do not show suffixal inflection, whereas nominalized adjectives (i.e., adjectives occurring as the last word of a noun phrase) are inflected for plural and take suffixed class marks.^[13] In the absolute form, the class marks suffixed to nominalized adjectives are M -we , F -je , N -be , HPL -ji , NPL -re , whereas in combination with overt case markers, the class marks suffixed to nominalized adjectives are identical to the ‘oblique stem markers’ found in the case inflection of some nouns (M -s̱u- , F/N - λ ̱i- , HPL -lo- , NPL -le- ~ -li-).

Ex. (4) illustrates an adjective with a gender-number prefix ( ‹b›ašada ‘old’) and an adjective devoid of gender-number prefix ( č’ĩda ‘new’) in the role of noun dependent (a-b), in predicate function (c-d), and nominalized (e-h).

3.4. Verb Inflection

Independent verb forms are inflected for TAM, polarity, and gender-number agreement; TAM and polarity are conjointly expressed by portemanteau markers. Assertive agreement (see section 3.4.3 below) is morphologically distinct from gender-number agreement and occurs in one tense only.

In addition to the synthetic forms listed in section 3.4.1, AD Akhvakh has analytic verb forms with the copula g‹o›di or the verb ‹b›ik’uruƛa ‘be’ in auxiliary function.

3.4.1. TAM-Polarity Marking in Independent Verb Forms

TAM/polarity inflection of verbs heading independent clauses includes the following possibilities:^[14]

• perfective positive (-ari or -ada , according to ‘assertive agreement’ – see section 3.4.3),
• perfective negative (-iƛa),
• imperfective positive (-ida),
• imperfective negative (-ika),
• irrealis positive (-iri),
• irrealis negative (-iki),
• indirective past positive (M -uwi , F -iwi , N/NPL -awi),^[15]
• indirective past negative (M -iƛ-uwi , F -iƛ-iwi , N/NPL -iƛ-awi),
• potential (M/N/NPL -uwa , F -iwa , HPL -oji),
• imperative (-a),
• prohibitive (-uba),
• optative positive (-a-ƛ̱’a),
• optative negative (-uba-ƛ̱’a),
• interrogative positive (M -uwa , F -iwa , N/NPL -awa , HPL -aji),
• interrogative negative (M -uš-uwa , F -uš-iwa , N/NPL -uš-awa , HPL -uš-aji).

3.4.2. Class Agreement in Verbal Inflection

Class agreement of verbs involves both prefixes and suffixes, with two different kinds of conditioning:

• The presence of class prefixes in verb forms involves no grammatical conditioning. Verbs divide into two phonologically and semantically arbitrary morphological classes, those that cannot occur without a class prefix indexing the S or P argument (i.e., the argument encoded by an NP in the absolute form), and those that never take such a prefix.
• By contrast, the presence of class suffixes indexing the S or P argument is conditioned by the grammatical nature of the verb form. The rules governing the presence and the phonological realization of class suffixes in verb forms are complex. In some verb forms, class suffixes do not occur at all; in others, obligatory class agreement marks merge with TAM/polarity markers; in a third group of verb forms, class suffixes are optional, and when they are present they may appear as distinct segments, or merge with TAM/polarity markers.

The presence of class agreement marks in verb forms depends therefore on a complex combination of lexical and grammatical factors, but the agreement rule itself is always the same: whenever class marks are present in a verb form, they invariably index an absolutive argument, i.e. an argument that can be represented by an NP in the absolute form.

3.4.3. Assertive Agreement

The perfective positive is the only tense in which, in addition to class agreement with the absolutive argument, verbs express speech act role distinctions, which however cannot be straightforwardly formulated in terms of the traditional category of grammatical person (i.e., speaker vs. addressee vs. others).^[16] There are two possible endings for this tense, with basic allomorphs -ada (glossed ASSINV for ‘assertor’s involvement’) and -ari (glossed PFV). ^[17] In contexts other than reported speech, the choice between -ada and -ari expresses a 1st person (-ada) vs. 2nd/3rd person (-ari) contrast in declarative clauses, but 2nd person (-ada) vs. 1st/3rd person (-ari) contrast in questions, and follows a split intransitive pattern. In the following chart, S _A stands for ‘S argument of an intransitive verb triggering the choice of -ada in the same way as the A argument of a transitive verb’:

As indicated in this chart, transitive verbs invariably show agreement with A ( -ada with 1st person A and -ari with 2nd/3rd person A in statements, -ada with 2nd person A and -ari with 1st/3rd person A in questions), whereas intransitive verbs divide into two semantically motivated classes: S _A verbs agree with S in the same way as transitive verbs with A (accusative alignment), whereas S _P verbs do not agree, and invariably show the ending -ari (ergative alignment). Ex. (6) & (7) illustrate assertive agreement with transitive verbs.

In reported speech, the choice between -ari and -ada in the reported clause has no direct relation with the person value manifested by the NP in S or A role (which may depend on the deictic shifts occurring in reported speech), and exclusively depends on the fact that the A / S _A argument coincides or not with the person whose speech is reported. In ex. (8a), the speaker coincides with the agent of the reported event, but not with the person who asserted the reported clause, hence the choice of -ari  ; in ex. (8b), the person who asserted the reported clause is different from the speaker, but coincides with the agent of the reported event, hence the choice of -ada .

The division of Akhvakh intransitive verbs into two classes according to their compatibility with the ending -ada in the perfective positive transparently reflects the degree of control of the participant encoded as S. Consequently, the precise function of the ending -ada is to encode coincidence between the controller of the event and the assertor, i.e., the speech act participant responsible for the assertion (the speaker in declarative speech acts, the addressee in questions, the person whose speech is reported in reported clauses). This is the reason why I propose the term ‘assertive agreement’.^[18]

3.4.4. Participles

AD Akhvakh has four participles. Each of them is characterized by a stem homonymous with one of the independent verb forms listed above: perfective positive -ada ), perfective negative ( -iƛa ), imperfective positive (-ida), and imperfective negative (-ika). A more detailed presentation of the participles will be given in section 5.

3.4.5. Dependent Verb Forms

Strictly dependent verb forms include an infinitive (-u(ruƛa)), a general converb (M -o(ho), F -e(he), N -e, HPL -i, NPL -ere), a progressive converb (M -ero, HPL -eri, F/N/NPL -ere), and several specialized converbs expressing various semantic types of adverbial subordination.^[19]

AD Akhvakh also has a verbal noun (-e), which however is rarely found with dependents treated like dependents of a verbal head. Contrary to the verbal noun (or ‘masdar’) of other Andic languages – see (Kibrik (ed.) 1996:180-2) on Godoberi, (Kibrik (ed.) 2001:523-9) on Bagvala), the verbal noun of Akhvakh tends to behave like a de-verbal noun rather than an inflected verb form, and is not productively used as the head of subordinate clauses.

4. Akhvakh Morphosyntax and the Domains of Finiteness

4.1. Finiteness and Tense

In some languages, the presence of tense markers in verb forms clearly correlates with finiteness as a grammatical feature of predicative constructions. The problem in Akhvakh is that all verb forms are overtly marked by an inflectional suffix, and inflectional suffixes are portemanteau markers conflating aspecto-temporal and modal meanings, polarity, and sometimes agreement too. The structure of Akhvakh verb forms does not involve a morphological slot that could be straightforwardly characterized as reserved for the expression of tense to the exclusion of other inflectional distinctions. There is in Akhvakh no clear morphological evidence supporting a division of verb forms into two subsets on the basis of tense marking. The semantic specifications carried by the inflectional suffixes of dependent verb forms are different from those carried by suffixes characteristic of independent verb forms, but it is not possible to characterize them globally as ‘reduced’ or ‘more rudimentary’. Note in particular that (a) the ada-, iƛa-, ida- and ika- forms used as participles express the same perfective vs. imperfective distinction as when used as heads of independent assertive clauses, and (b) the temporal converbs, which are strictly dependent verb forms, express relative tense distinctions more detailed than the tense specifications carried by independent verb forms.

4.2. Finiteness and Agreement

A striking particularity of Akhvakh is that there is no simple correlation between the agreement morphology of verb forms and the ability of predicative constructions to be used as independent clauses with particular illocutionary values and/or to be involved in particular types of complex structures.

The speech act role distinctions found in the perfective positive do not correlate with a particular behavior that would distinguish the clauses headed by a verb marked for this tense. Concerning class agreement, there is no simple relationship between the way Akhvakh verb forms agree or not with their absolutive argument, and their ability to head independent clauses, as can be seen from the classification of AD Akhvakh verb forms according to the presence/absence of suffixed class marks given in (9). In this chart, verb forms are divided into those that have the ability to head independent clauses, and those that are found in clauses involved in complex constructions only.

4.3. The Expression of the S/A Argument

4.3.1. The Expression of the S/A Argument of Verbs in the Imperative

Incompatibility with canonical S/A NPs is a common manifestation of nonfiniteness, and in many languages, imperatives pattern from this point of view with dependent verb forms (Nikolaeva 2007b). Akhvakh has an imperative and a prohibitive that are strictly addressee-oriented, but do not differ from independent assertive verb forms with respect to the expression of the S/A argument. Akhvakh is a ‘pro-drop’ language, in which NPs representing arguments are not required syntactically, and pronouns occur in argumental roles only if there is a possibility of ambiguity, or to express emphasis. Similarly, imperatives and prohibitives do not necessarily combine with second person pronouns in A/S role, but there is no ban on their presence – ex. (4).

Note in particular that imperatives and prohibitives of transitive verbs are compatible with a second person pronoun in the ergative case, which excludes analyzing the second person pronoun accompanying imperatives or prohibitives as a kind of vocative.

4.3.2. The Expression of the S/A Argument of Verbs in the Infinitive

The infinitive of Akhvakh may occur in control constructions in which its S/A argument is obligatorily left unexpressed, and is semantically identified to an argument of the main verb, as in ex. (11).

However, this behavior is triggered by some of the verbs taking infinitival complements (in ex. (5), ĩdunuƛa ‘be able’ and ‹b›ašlōruƛa ‘begin’), and does not constitute an intrinsic property of the infinitive itself. For example, kwĩλuruƛa ‘want’ and bužuruƛa ‘believe’ combine with infinitival complement clauses in which all the arguments of the verb in the infinitive can be expressed without any co-reference constraint – ex. (12). Note in particular that, with infinitives of transitive verbs, the fact that the S/A argument occurs in the ergative case excludes an analysis in terms of ‘raising to object’.

The same applies to infinitives heading adverbial clauses of purpose – ex. (13).

It is also worth noting that, with the exception of a very limited number of verbs forming with their infinitival complement a control construction, as in ex. (11) above, even when the S/A argument of the infinitive is not overtly expressed, it is not necessarily identified to an argument of the main verb. Depending on the context, an arbitrary interpretation is always possible. For example, the sentence in ex. (14) has two possible readings; most of the time, an unexpressed argument in the construction of an infinitive complement of kwĩλuruƛa ‘want’ is identified with the dative argument of kwĩλuruƛa, but in the text from which this sentence has been extracted, it is clear from the context that an arbitrary interpretation was intended.

In other words, in general, the interpretation of missing arguments in infinitival clauses is a pragmatic mechanism which does not differ from the interpretation of missing arguments in independent clauses.

4.3.3. The Expression of the S/A Argument of Verbs in a Converbal Form

With the only exception of the progressive converb (which differs from all the other converbs of Akhvakh in its syntactic properties and mainly occurs as an element of analytic verb forms), a general property of the converbs is that their arguments can always be expressed exactly in the same way as in independent clauses. The ability of the converbs to combine with an NP in S/A role whose referent does not coincide with any of the arguments of the main verb is illustrated in ex. (15) by complex sentences involving the negative form of the posterior converb (a), the immediate converb (b), the imminent converb (c), the conditional converb (d), and the explicative converb (e).

A missing argument in a clause headed by a specialized converb may be identified to an argument or the main verb, but the progressive converb is the only one whose construction obligatorily involves a missing argument that must be identified to an argument of the main verb. The general rule is that the identification of missing arguments in the construction of specialized converbs is a purely pragmatic phenomenon, which does not put into play syntactic constraints. Depending on the context, a missing argument in the construction of a converb can always be identified to a speech act participant, or to any other salient referent, as illustrated by ex. (16).

The translation given in ex. (16) corresponds to the interpretation of this sentence in the context in which I have found it, but the same sentence in different contexts could equally be interpreted as ‘While sleeping I saw a ghost’, ‘While I slept he saw a ghost’, ‘While he slept I saw a ghost’, ‘While he_i slept he_j saw a ghost’, etc. The only constraint is that the missing argument of ‘sleep’ must be masculine singular, since the converbal form ƛ̱’ũk’ideλ̱i shows masculine singular agreement; the interpretation of the missing argument of ‘see’ is totally open, since the form harigwiri includes no agreement mark, and there is no co-reference or disjoint reference constraint between the missing argument of the converb and any of the arguments of the main verb.

5. Participles

5.1. The Notion of Participle in Traditional Grammar

In languages in which the inflectional approach to finiteness is not problematic, the forms traditionally labeled ‘participles’ have the following properties:

• participles are verb forms in the sense that, with respect to their ‘internal syntax’ (i.e., the internal structure of the phrases they head), they have the same properties as verb forms heading independent clauses;
• participles are non-finite verb forms, i.e., they cannot head independent clauses, and this inability is correlated with the lack of morphological distinctions characteristic of the independent verb forms of the same language;
• participles have the ‘external syntax’ of adjectives: taken as a whole, clauses headed by participles are syntactically equivalent to adjective phrases; they can fulfill the roles of noun dependent and non-verbal predicate, or undergo nominalization, in the same way as adjective phrases;
• in all of the roles accessible to adjective phrases, the verb form heading a participial clause shows the same behavior (in particular, the same inflectional characteristics) as the head of an adjective phrase fulfilling the same role;
• semantically, participial clauses modify the noun they depend on by identifying it to an unexpressed constituent of the participial clause.

5.2. Participles and Independent Verb Forms in Akhvakh

If participles are defined as verb forms involved in a particular type of relativization strategy,^[29] there is no difficulty in recognizing four participles in Akhvakh (perfective positive, perfective negative, imperfective positive, and imperfective negative). This means that these four verb forms occur in pre-nominal relative clauses, showing at the same time the same characteristics as attributive adjectives with respect to their relation to a head noun.^[30] Such relative clauses can be used in predicate function or nominalized in the same way as adjective phrases, and the verb forms that head them take agreement suffixes and case inflection exactly like adjectives. What is particular in the case of Akhvakh is that none of the verb forms found in participial relatives is specialized in participle function. In Akhvakh, the set of verb forms occurring as heads of relative clauses with a typically participial behavior is a proper subset of the set of verb forms occurring as heads of independent clauses. Moreover, when used as heads of independent clauses, the verb forms in question do not show properties that would distinguish them from the independent verb forms that cannot function as participles.

Note however that, in constructional terms, participial relative clauses are not entirely identical to independent clauses headed by the same verb forms, in spite of the fact that they may include the same NPs with the same case marking. The point is that relative clauses are strictly head-final, whereas the verbal head of an independent clause has no fixed position.

In addition to that, with one of the four participles of Akhvakh (the perfective positive, see section 5.5) the participle is formally identical to an independent verb form carrying the same TAM and polarity specifications, but is not involved in the mechanism of assertive agreement characterizing the same form in independent clauses.^[31] Here again, the necessity to complete the morphological observations by a constructional approach is patent.

5.3. The Imperfective Positive Participle

Verb forms marked by the imperfective positive suffix -ida occur as heads not only of independent assertive clauses, as in ex. (17a), but also of relative clauses, as in ex. (17b). There is no overt mark of the dependent status of the relative clause, and there is no overt indication of the relativized role either. The only difference between such a relative clause and an independent clause with a missing term lending itself to an anaphorical or arbitrary reading is the obligatory final position of the verb in the relative clause, as illustrated by the fact that a sequence such as du-ƛa kw-ī̃da b-eχ-uruƛa is acceptable as an independent clause with a missing argument anaphorically identified to a discursively salient entity – ex. (17c), but not as a relative clause – ex. (17d).

Ex. (18) & (19) compare relative clauses in the imperfective positive modifying a noun with the corresponding free relatives. In ex. (18b), the free relative fulfills a role requiring the zero-marked absolute case, whereas in ex. (19b), it fulfills a role requiring an overt case mark. Comparison with ex. (4) above shows that, in both cases, the suffixes that attach to the participle (a class suffix in (18b), an oblique stem marker followed by the case marker in (19b)) are identical to those that would attach to nominalized adjectives in the same contexts.

5.4. The Imperfective Negative Participle

As illustrated by ex. (20) & (21), verb forms marked by the imperfective negative suffix -ika share this ability to occur both in independent clauses and in participial relatives.

5.5. The Perfective Positive Participle

The suffix of the perfective positive participle -ada is identical to one of the two suffixes marking the perfective positive in independent clauses. The difference is however that, in independent clauses, -ada implies a 1st person A/S_A argument in declarative clauses and a 2nd person A/S_A argument in questions, and the same TAM value is marked by the suffix -ari if this condition is not met, whereas relative clauses ignore this agreement mechanism, and uniformly mark the perfective positive with -ada – ex. (22).

5.6. The Perfective Negative Participle

As illustrated by ex. (23), the situation with perfective negative -iƛa is exactly the same as with imperfective positive -ida or imperfective negative -ika .

5.7. Accessibility to Relativization

Ex. (24) illustrates the fact that the participial relatives presented in sections 5.3 to 5.6 can be used to relativize any term (argument or adjunct) of the construction of the verb, and also genitives.

5.8. Non-Restrictive Participial Relatives

Ex. (25) shows that, in Akhvakh, participial relatives are not necessarily restrictive.

5.9. Discussion

Akhvakh has participial clauses, i.e. clauses headed by a verb form behaving like an adjective with respect to the insertion of the phrase it heads into a broader construction. Participial clauses share with clauses headed by strictly dependent verb forms (infinitive or converbs) the obligatory final position of the verb, but none of the forms found as heads of participial clauses is specialized in this function: all of them also occur in independent assertive or interrogative clauses.

The case of the perfective positive is particularly interesting. As in the other cases, the verb form heading a participial clause in the perfective positive is not morphologically marked as specialized in participle function, but in constructional terms, there is a clear distinction between independent clauses in the perfective positive, in which the inflectional suffix of the verb can be -ada or -ari according to the rule of assertive agreement (see section 3.4.3), and participial clauses in the perfective positive, in which -ada is the only possibility.

In order to avoid contradictions and/or circularity in the description of such situations, it is crucial to admit that the definitions of construction types are logically anterior to the definitions of form types . Very often, the recognition of a type of construction is ensured by the morphological nature of its head, but the formal identification of a construction does not necessarily rely on the presence of a word belonging to a given morphological type. In Akhvakh, defining participial clauses as clauses headed by a participle would not be correct, since Akhvakh has no form specialized in participle function. The definition of participial clause must be posited first, and participles can be defined then as forms fulfilling the predicate function in participial clauses, which does not exclude that the same forms may occur in other functions too.

6. Conclusion

There is nothing exceptional in the existence of verb forms fulfilling the predicate function both in independent assertive or interrogative clauses and in participial clauses, and the historical source of such situations is well-known. A scenario attested in many languages whose history is documented or reconstructed with a high degree of plausibility is that such forms originally were specialized participles. Given their adjectival nature, participles can be used in adjectival predication, and adjectival predications involving participles tend to undergo evolutions blurring the distinction with verbal predication: if a copula is originally present, it may be deleted, or fuse with the participle, becoming thus a TAM/agreement affix; if the case marking of the arguments of a participle used as an adjectival predicate differs from that found in verbal predication proper, it may be readjusted; a similar readjustment may concern constraints on constituent order too, if adjectival predication with a participle in predicate function originally involves constraints different from those observed in verbal predication proper, etc.

Nakh-Daghestanian languages provide ample evidence that such processes have been very active in the history of this language family. What makes the case of Akhvakh particularly interesting is that, in its present state, this language has no specialized participles at all, and at the same time clearly possesses a clause type identifiable as a participial clause.

I have tried to show in sections 4 & 5 that, in several respects, Akhvakh has particularities hardly compatible with the traditional approach to finiteness. The participial clauses analyzed in section 5 are a clear case of nonfinite clauses including no nonfinite form. The Akhvakh data shows that a very common type of historical process (the integration of participles into the paradigm of verb forms heading independent assertive or interrogative clauses) may result in a situation that necessitates positing the notion of participial clause (defined in constructional terms) as logically anterior to the notion of participial form.

Acknowledgements

I am grateful to the Linguistic Discovery readers for very useful comments on previous versions of this article.

Abbreviations

*A very limited number of Akhvakh verbs have a discontinuous root the two segments of which are separated by a class agreement mark. Eight such verbs occur in my data. In the glosses, the English equivalent of their lexical meaning is given as the gloss of the second segment, and the first segment is glossed prev, but this is purely conventional, and ‘preverb’ must be understood here as ‘first segment of a discontinuous verb root’. Akhvakh has nothing comparable to Russian or Georgian preverbation.

References

Bisang, W. 2007. Categories that make finiteness: discreteness from a functional perspective and some of its repercussions. In Nikolaeva, I. (ed.), Finiteness, 115-137. Oxford: Oxford University Press.

Comrie, B. & M. Polinsky. 1999. Form and function in syntax: Relative clauses in Tsez. In Darnell M., E. Moravcsik, F. Newmeyer, M. Noonan, & K. Wheatley (eds.), Functionalism and Formalism in Linguistics, 77-91. Amsterdam / Philadelphia: John Benjamins.

Creissels, D. 2008a. Person variations in Akhvakh verb morphology: functional motivation and origin of an uncommon pattern. Sprachtypologie und Universalienforchung 61-4.309-325. doi:10.1524/stuf.2008.0027

-----. 2008b. Remarks on so-called “conjunct/disjunct” systems. Paper delivered at the Syntax of the World’s Languages III conference. Berlin.

-----. To appear. Specialized converbs and adverbial subordination in Axaxdərə Akhvakh. In Bril, I. (ed.), Clause-hierarchy and clause-linking, syntax and pragmatics.

Cristofaro, S. 2007. Deconstructing categories: finiteness in a functional-typological perspective. In Nikolaeva, I. (ed.), Finiteness, 91-114. Oxford: Oxford University Press.

Curnow, T. 2002a. Conjunct/disjunct marking in Awa Pit. Linguistics 40-3.611-627. doi:10.1515/ling.2002.025

-----. 2002b. Conjunct/disjunct systems in Barbacoan languages. Santa Barbara Papers in Linguistics 11. 3-12.

DeLancey S. 1986. Evidentiality and volitionality in Tibetan. In Chafe, W. & J. Nichols (eds.), Evidentiality: the Linguistic Coding of Epistemology, 203-213. Norwood NJ: Academic Press.

-----. 1990. Ergativity and the cognitive model of event structure in Lhasa Tibetan. Cognitive Linguistics 1-3.289-321. doi:10.1515/cogl.1990.1.3.289

-----. 1992. The Historical status of the conjunct/disjunct pattern. Acta Linguistica Hafniensa 25.39-42. doi:10.1080/03740463.1992.10412277

Dickinson, C. 2000. Mirativity in Tsafiki. Studies in Language 24-2.379-421. doi:10.1075/sl.24.2.06dic

Hale, A. 1980. Person markers: finite conjunct and disjunct forms in Newari. In Trail R. (ed.), Papers in Southeast Asian Linguistics 7.95-106. Canberra: Pacific Linguistics.

Hargreaves, D. 2005. Agency and intentional action in Kathmandu Newar. Himalayan Linguistics 5.1-48.

Kalinina, E. & N. Sumbatova. 2007. Clause structure and verbal forms in Nakh-Daghestanian languages. In Nikolaeva, I. (ed.), Finiteness, 183-249. Oxford: Oxford University Press.

Kibrik, A. E. 1985. Towards a typology of ergativity. In Nichols, J. & A. Woodbury (eds.), Grammar inside and outside the clause, 268-324. Cambridge: Cambridge University Press.

-----. (ed.). 1996. Godoberi. München: LINCOM EUROPA.

-----. (ed.). 2001. Bagvalinskij jazyk. Grammatika, teksty, slovari. Moskva: IMLI RAN, “Nasledie”.

Loughnane, R. 2007. Expanding the typology of evidentiality: the participatory/factual in Oksapmin. Paper delivered at the ALT 7 conference. Paris.

Magomedbekova, Z.M. 1967. Axvaxskij jazyk (grammatičeskij analiz, teksty, slovar’). Tbilisi: Mecniereba.

Magomedova, P & I. Abdulaeva. 2007. Axvaxsko-russkij slovar’. Maxačkala: Dagestanskij Naučnyj Centr Rossiskoj Akademii Nauk.

Magometov, A. 1982. Megebskij dialekt darginskogo jazyka. Tbilisi: Mecniereba.

Nikolaeva, I. 2007a. Introduction. In Nikolaeva, I. (ed.), Finiteness, 1-19. Oxford: Oxford University Press.

-----. 2007b. Constructional economy and nonfinite independent clauses. In Nikolaeva, I. (ed.), Finiteness, 138-180. Oxford: Oxford University Press.

Nikolaeva, I. (ed.). 2007. Finiteness. Oxford: Oxford University Press.

Rebuschi, G. 1984. Structure de l’énoncé en basque. Paris: SELAF.

Sells, P. 2007. Finiteness in non-transformational syntactic frameworks. In Nikolaeva, I. (ed.), Finiteness, 59-88.

^[1] In general, I use Akhvakh as the term most commonly found in the literature for the language in question, but when I quote Russian terms, axvaxskij occurs as the transliteration of Russian ахвахский.

^[2] The other Andic languages are Andi, Bagvala, Botlikh, Chamala, Godoberi, Karata, and Tindi. None of them has a particularly close relationship to Akhvakh. Andic languages are traditionally grouped with Avar and Tsezic languages into a single branch of the Northeast Caucasian family. The other branches of the Northeast Caucasian family are Lak, Dargi (or Dargwa), Lezgi, Khinalug (sometimes considered a marginal member of the Lezgi branch), and Nakh.

^[3] Judging from (Magomedbekova 1967) and (Magomedova & Abdulaeva 2007), the variety of Akhvakh spoken in Axaxdərə does not differ significantly from the varieties of Northern Akhvakh spoken in the Axvaxskij Rajon, and this judgment was confirmed without any reservation by Indira Abdulaeva, co-author of the Akhvakh-Russian dictionary and a native speaker of Northern Akhvakh herself, who spent one week in Axaxdərə while I was carrying field work there in April 2008. I have been able to find no precision about the time when Akhvakh migrants began to settle in Axaxdərə, but there are still in Axaxdərə old people who were born in Daghestan, and whose parents migrated to Azerbaijan at the end of the second world war, when the economic situation in Daghestan was particularly difficult.

^[4] The underlying structure of verb forms involving morphophonological processes is given in footnotes. In this example, the underlying form of w-oq’-ari is |w-eq’-ari|.

^[5] The underlying form of w-oƛ-ari is |w-eƛ-ari|.

^[6] The only exceptions to the semantic rule of class assignment are ãde ‘person’ and mik’e ‘child’, which in the singular trigger N agreement, whereas the corresponding plural forms ãdo and mik’eli regularly trigger HPL agreement.

^[7] It seems that in AR Akhvakh, all noun dependents in canonical NPs optionally take class suffixes agreeing with the head noun, but in the data I have collected in Axaxdərə, noun dependents other than adjectives never occur with agreement marks in canonical NPs, and suffixal agreement of adjectives never occurs in classes other than HPL.

^[8] The underlying form of gw-ēri is |gwi(j)-ari|.

^[9] In principle, zero-marked genitive characterizes M and HPL NPs, whereas -ƛ̱i  is used with F, N or NPL NPs, but this rule is not very strict, and variations are observed.

^[10] In AR Akhvakh -u has been identified as ablative proper, and -une as perlative, but in AD Akhvakh, these two endings are in free variation.

^[11] Words obligatorily including a class marker are conventionally quoted in the non-human singular (N) form, with the class marker between small angle brackets.

^[12] In Akhvakh, non-verbal predications involving neither the copula nor the verb ‹b›ik’uruƛa ‘be’ are exceptional in statements. By contrast, the omission of the copula regularly occurs in questions.

^[13] In AR Akhvakh, attributive or predicative adjectives optionally show suffixal inflection.

^[14] In this enumeration, each suffix is given in the form of its default allomorph, i.e., the allomorph occurring in the absence of any morphophonological process triggered by the stem.

^[15] There is no specific form of the indirective past in the HPL class, and this gap is filled by the HPL form of the perfect, an analytic tense consisting of the general converb of the auxiliated verb and the copula in auxiliary function.

^[16] For more detailed presentation/discussion of speech act role distinctions in Akhvakh verb morphology, see (Creissels 2008a), (Creissels2008b).

^[17] In addition to variations due to morphophonological interaction with the stem, these suffixes show (partly optional) variations involving class agreement : -ari takes the form -iri if the absolutive argument belongs to the HPL class, but invariably appears as -ari in the other classes, whereas -ada obligatorily appears as -idi if the absolutive argument belongs to the HPL class and undergoes optional class agreement in the other classes, resulting in the variants M -ada ~ -adawe ~ -ado , F -ada ~ -adaje ~ -ade , N -ada ~ -adabe ~ -ade , NPL -ada ~ -adare ~ -ade .

^[18] Assertive agreement has not been identified in previous studies of Akhvakh – (Magometbekova 1967), (Kibrik 1985). Magomedbekova (1967) describes an ‘optional’ 1st vs. 2nd/3rd person contrast, but does not provide examples of interrogative clauses that could reveal the existence of an assertor’s involvement marking pattern. In the grammatical sketch included in the Akhvakh-Russian dictionary, the two verbal endings expressing distinctions in assertive agreement are just mentioned as two possible marks of the same tense (prošedšee očevidnoe, i.e. ‘past of direct knowledge’), without any indication about their distribution or difference in meaning. However, the dictionary itself includes many examples suggesting a pattern similar to that of Axaxdərə Akhvakh, and this was confirmed by the discussions I had with Indira Abdulaeva. The only Caucasian language in which assertive agreement has been recognized so far is the Mehweb dialect of Dargwa – (Magometov 1982). Similar patterns (more commonly termed ‘conjunct/disjunct systems’) have been first described for Tibetan, Newari, and a few other Tibeto-Burman languages – (Hale 1980), (DeLancey 1986), (DeLancey 1990), (DeLancey 1992), (Hargreaves 2005), and have also been signaled in the Barbacoan languages spoken in Colombia and Ecuador – (Dickinson 2000), (Curnow 2002a), (Curnow 2002b), and in the Papuan language Oksapmin – (Loughnane 2007).

^[19] On the specialized converbs of Akhvakh, see (Creissels To appear).

^[20] The underlying form of w-ašl-ēri is |w-ašla(j)-iri|.

^[21] The underlying form of m-ič-unuƛa is |b-ĩč-uruƛa|.

^[22] The underlying form of gw-īri is |gwi(j)-iri|.

^[23] The underlying form of r-eq̄-ōruƛa is |r-eq̄a(j)-uruƛa|.

^[24] The underlying form of c̱’-ēdaλ̱a is |c̱’a(b)-idaλ̱a|.

^[25] The underlying form of b-oč’il-āri is |b-oč’ila(j)-ari|.

^[26] The underlying form of m-ič-ala is |b-ĩč-ala|.

^[27] The underlying form of m-īda is |b-(ãʔ-)ida|.

^[28] The underlying form of gw-ēre is |gwi(j)-ere|.

^[29] On participial constructions as a relative clause formation strategy, see in particular (Comrie & Polinsky 1999).

^[30] Akhvakh also has a correlative relative clause construction, but it is much less usual than the participial construction.

^[31] For a historical explanation of this situation, see (Creissels 2008a).

^[32] The underlying form of kw-ī̃da is |kwĩ(λ)-ida|.

^[33] The underlying form of m-ač-ika is |b-ãč-ika|.

^[34] The underlying form of w-ošq̄-iƛa is |w-ešq̄-iƛa |.

^[35] The underlying form of j-ā̃da is |j-(ãʔ-)ada|.

^[36] The underlying form of gw-ēda is |gwi(j)-ada|.

^[37] The underlying form of w-ī̃ƛa is |w-(ãʔ-)iƛa|.

^[38] The underlying form of w-oƛ-ari is |w-eƛ-ari|.

^[39] The underlying form of w-ux̱-uwi is |w-ix̱-uwi |.