Филологические науки
EMERGENCE AND EVOLUTION OF EXCEPTIONAL VA-LENCY PATTERNS IN ERGATIVE LANGUAGES: A CASE STUDY
Jérémy Pasquereau, Denis Creissels 1
1. University of Massachusetts, USA; University of Lyon, France
Резюме:
SLE – 44th annual meeting
Logroño, 8-11 September 2011
Emergence and evolution of exceptional valency patterns
in ergative languages: a case study
Jérémy Pasquereau
University of Massachusetts
jepasque@hotmail.com
Denis Creissels
University of Lyon
denis.creissels@univ-lyon2.fr
http://deniscreissels.fr
1. Introduction
This paper analyzes the possible origin and the evolution of a particular type of
exceptional valency pattern found in East Caucasian languages. It deals specifically
with a group of East Caucasian languages, the Andic languages (Andi, Akhvakh,
Bagvalal, Botlikh, Chamalal, Godoberi, Karata, and Tindi), spoken in the Western part
of Daghestan. The Andic languages constitute a genetic unit within the Avar-
Andic(-Tsezic) branch of the East Caucasian language family.
The data we present comes from our own work on two Andic languages, Akhvakh
and Karata – (Creissels 2010), (Pasquereau 2010), (Pasquereau 2011), and from the
consultation of dictionaries and grammars of Andic languages – (Kibrik 1996), (Kibrik
2001), (Magomedova 1999), (Magomedova 2003), (Magomedova 2004),
(Magomedova & Abdulaeva 2007), (Magomedova & Khalidova 2001), (Saidova 2006).
The paper is organized as follows. Section 2 provides basic information about the
grammatical structure of Andic languages. Section 3 provides illustrations of
exceptional case frames in Andic languages. Section 4 examines the question of the
emergence of the exceptional case frames <ERG, LOC> and <ERG, ALL>. Section 5
and 6 concentrate on the particular cases of ‘listen’ and ‘bite’, respectively. Section 7
puts forward some concluding remarks.
2. The basics of Andic morphosyntax
In Andic languages, the syntactic function of noun phrases is expressed by case
suffixes. Verbs agree in gender and number (not in person) but verb agreement is
redundant with case marking, since verbs consistently agree with their nominative
argument, and never express agreement with non-nominative noun phrases.
Constituent order is remarkably flexible and plays no role in the expression of
argument structure.
Andic languages have rich case systems, and in particular, very elaborate spatial
case systems. The nominative (alias absolutive), also used as the quotation form of
nouns, is characterized by a zero ending. The other cases are marked by suffixes
attached either to a stem coinciding with the nominative form, or to a special stem
traditionally called oblique stem (indicated by –ₒ in the glosses of examples). Spatial
case endings consist of two formatives, a topological marker (glossed TPL) and a
directionality marker.
J. Pasquereau & D. Creissels, Emergence and evolution of exceptional valency patterns, p. 2/8
As illustrated by Ex. (1), the coding of core syntactic roles in Andic languages is
consistently ergative:
– S (boʁosē ãdo in (1a)) and P (istaka in (1b)) are in the nominative case (alias
absolutive), contrasting with A (wašode in (1b)) in the ergative case.
– Transitive verbs do not agree with A, but agree with P in the same way as
intransitive verbs do with S: in (1a), the intransitive verb shows human plural
agreement; in (1b), A is masculine singular, whereas P is neuter singular, and the
verb shows neuter singular agreement.
(1) a. b-oʁosē ãd-o atobusī -ɬī-g-e b-eq’-īri. Akhvakh
H⁺-most person-PL bus-Nₒ-TPL-LOC H⁺-come-PF.H⁺
‘Most people came by bus.’
b. wašo-de istaka b-iq’ʷ-aj-ē godi.
boyₒ-ERG glass N-break-CAUS-ADV.N COP.N
‘The boy broke the glass.’
The valency frames of Andic verbs canonically include at least a slot for a nominative
NP (representing in particular the sole argument of monovalent verbs and the P
argument of prototypical action verbs). According to the account of verb valency
included in (Kibrik 2001: 369-376), in Bagvalal, this rule can be viewed as
exceptionless. However, most Andic languages have a restricted set of verbs used in
constructions in which no slot for a nominative NP can be posited.
3. Verbs without nominative arguments
In this section, we provide illustrations of Andic verbs used in case frames that do not
include a slot for a nominative NP. Note that, in the absence of a nominative
argument, the verb forms that express gender-number agreement show default neuter
singular agreement.
3.1. Monovalent verbs
Ex. (2) illustrates the case of a monovalent Akhvakh verb which, depending on the
situation referred to, may occur either in the canonical case frame <NOM>, as in
(2a), or in the exceptional case frame <LOC>, as in (2b).
(2) а. miq̄’i q̄’ʷaraɬ-ēhe godi. Akhvakh
road become_narrow-ADV.N COP.N
‘The road became narrow.’
b. miʕa-q̄-e q̄’ʷaraɬ-ēhe godi.
nose-TPL-LOC become_narrow-ADV.N COP.N
‘My nose is blocked.’ lit. ‘In the nose became narrow.’
J. Pasquereau & D. Creissels, Emergence and evolution of exceptional valency patterns, p. 3/8
Ex. (3) illustrates the Karata verb bibāɬa ‘cry’, used in the exceptional case frame
<ERG>.
(3) gugu-l ɬabc’e bib-ē. Karata
cuckoo-ERG three_times cry-PF
‘The cuckoo sang three times.’
3.2. Bivalent verbs
Ex. (4) illustrates a bivalent Akhvakh verb used in the exceptional case frame <ERG,
GEN>.
(4) hu-sʷ̄ -e daru-ɬī-ʟī ħalbix̄il-āri. Akhvakh
DIST-Mₒ-ERG medecine-Nₒ-GEN try-PF
‘He tried the medecine.’
3.3. Trivalent verbs
Ex. (5) & (6) illustrate a trivalent verb used in the exceptional case frame <ERG, ALL,
GEN>, in Karata and in Akhvakh.
(5) hu-ɬī-ʟī čela žo-ɬī mač-uwa du-g-a de-de Akhvakh
DIST-Nₒ-GEN other day-Nₒ[LOC] tell-POT 2SG-TPL-ALL 1SG-ERG
‘I will tell you about this another day.’
(6) k’̄ ʷāmsʷam hedela-ɬī-ʟ ̄ bas-imisē ! Karata
trivial.N thing-N-GEN tell-PROH
‘Don’t speak about trivial things!’
4. The emergence of the case frames <ERG, ALL> and <ERG, LOC>
In this section, we examine the case of verbs used in the case frames <ERG, ALL> or
<ERG, LOC>. In Andic languages these case frames are typically found with verbs
expressing the following meanings: ‘look at’ – Ex. (7), ‘listen’, ‘bite’, ‘pinch’ – Ex. (8),
‘sting’ – Ex. (9). For examples with ‘listen’ and ‘bite’, see Sections 5 & 6, where the
case of these verbs is examined in more detail.
(7) wašo-de di-g-a eq̄-ari. Akhvakh
boyₒ-ERG 1SG₀-TPL-ALL look_at-PF
‘The boy looked at me.’
(8) o-sʷ̄̌ -i č’un-o di-č’-i. Tindi
DIST-Mₒ-ERG pinch-PF 1SGₒ-TPL-LOC/ALL
‘He pinched me.’
J. Pasquereau & D. Creissels, Emergence and evolution of exceptional valency patterns, p. 4/8
(9) di-g-e ʟ̄’ižʷali-de č’in-āri Akhvakh
1SGₒ-TPL-LOC waspₒ-ERG sting-PF
‘A wasp stung me.’
Godoberi also has a verb obadi ‘kiss’ with the case frame <ERG, LOC>, but all the
other Andic languages for which we have data express this meaning by means of a
noun oba or ob̄a ‘kiss’ combined with a verb ‘do’, and consequently the exceptional
case frame of Godoberi obadi ‘kiss’ must probably be viewed as the result of the
univerbation of a do-compound.
The other cases are less easy to explain, but the variations observed in the
expression of the other meaning listed above in Andic languages support a hypothesis
already suggested by Charachidzé (1981) for Avar, according to which these
exceptional valency frames may result from the reduction of the regular frames <ERG,
NOM, ALL> or <ERG, NOM, LOC>, characteristic of verbs expressing meanings or
the type ‘X applies/holds Y on Z’ (and found in particular with the verbs expressing
‘hit’, with the hittee in the locative/allative and the instrument in the nominative).
The point is that not all Andic languages have bivalent verbs with meanings such as
‘look at’, ‘listen’, ‘bite’, ‘sting’, ‘pinch’. Among the Andic languages, it is also common
to find such meanings expressed by means of less specific verbs combined with three
NPs. For example, Tindi expresses ‘sting’ as eq̄ʷa k̄ʷēɬʲa, lit. ‘hit the sting (on someone)’,
and ‘bite’ as saldi bix̄ʲiɬʲa, lit. ‘hold the teeth (on someone)’, with respectively eq̄ʷa
‘sting (noun)’ and saldi ‘teeth’ occupying the nominative slot.
Starting from the hypothesis of valency frames originally including three slots, at
least two reduction scenarios can be imagined: conventionalization of the ellipsis of
the nominative argument, or fusion of the nominative argument with the verb. In
Andic languages, there is evidence for the conventionalization-of-ellipsis scenario in
the case of other verbs with exceptional case frames, but not for those examined here.
By contrast, some of them at least show evidence for the fusion scenario. The clearest
cases are those of ‘listen’ and ‘bite’, examined in Sections 5 and 6.
5. Emergence and evolution of verbs ‘listen’ in Andic languages
Three situations are found among Andic languages with respect to the expression of
‘listen’:
– Some Andic languages express ‘listen’ by means of a construction involving a
nominative noun phrase with the meaning ‘ear’ in addition to those encoding the
two participants. For example, Godoberi expresses ‘listen’ as hãt’uk’ja rik̄i, literally
‘fix the ear (on someone/something)’ – Ex. (10). Formally, this construction is an
instance of the regular valency pattern <ERG, NOM, ALL> with hãt’uk’ja ‘ear’
filling the nominative slot.
– Others have a verb ‘listen’ with the exceptional case frame <ERG, ALL>: Tindi
anix̄ʲiɬʲa – Ex. (11), Chamalal woɬuk’la;
– A verb ‘listen’ with the regular case frame <NOM, ALL> is found in three Andic
languages: Akhvakh hãdax̄uruʟa – Ex. (12), Karata ãdukaɬa – Ex. (13) , Bagvalal
aštila.
J. Pasquereau & D. Creissels, Emergence and evolution of exceptional valency patterns, p. 5/8
(10) wašu-di imu-q̄-i hãt’uk’ʲa r-ikk-i r-ukk-ida. Godoberi
sonₒ-ERG fatherₒ-TPL-LOC/ALL ear N⁺-hold-INF N⁺-must-IPF
‘The son must listen to his father.’
(11) di-q-̄ a anix̄ʲ-ā hik’ʲi o-sʷ̄̌ -ī Tindi
1SGₒ-TPL-LOC/ALL listen-IPF NEG DIST-Mₒ-ERG
‘He does not listen to me.’
(12) waša imo-g-a hãdax̄-ari. Akhvakh
boy fatherₒ-TPL-ALL listen-PF
‘The boy listened to his father.’
(13) waša imo-χa-r ãduk-e. Karata
boy fatherₒ-TPL- ALL listen-PF
‘The boy listened to his father.’
Interestingly, ‘fix the ear on’ is the obvious etymology of Akhvakh hãdax̄uruʟa
(compare with hãde ‘ear’, bix̄uruʟa ‘fix’), in spite of the fact that the nominative case
assigned to the NP representing the listener does not correspond to what could be
expected from this etymology.
The variation in the expression of ‘listen’ in Andic languages provides therefore
evidence supporting the reconstruction of the following evolution:
– at a first stage, the coalescence of a trivalent verb occurring in the regular frame
<ERG, NOM, ALL> with a noun occupying the nominative slot creates a bivalent
verb with the exceptional case frame <ERG, ALL>;
– at a second stage, attested by Akhvakh and Karata, the exceptional valency pattern
resulting from this evolution may be regularized into <NOM, ALL>.
6. Variations in the case frame of ‘bite’ in Andic languages
Among Andic languages, variations similar to those observed in the expression of
‘listen’ are attested for the expression of ‘bite’ too, with however a different case frame
in the languages in which the case frame of ‘bite’ has been regularized:
– Some Andic languages express ‘bite’ by means of a construction involving a noun
phrase with the meaning ‘tooth’ in addition to those encoding the two participants:
as mentioned above, Tindi expresses ‘bite’ as saldi bix̄ʲiɬʲa, lit. ‘hold the teeth (on
someone)’.
– Others have a verb ‘bite’ with the exceptional case frame <ERG, ALL>: Karata
q̄’ʷaraɬa – Ex. (14), Akhvakh q̄’eleč’uruʟa – Ex. (15), Chamalal q̄’āna – Ex. (16);
– A verb ‘bite’ with the regular case frame <ERG, NOM> is found in two Andic
languages: Godoberi q’̄ ami – Ex. (17),1 Bagvalal salīla – Ex. (18).
1 Note however that the two available sources on Godoberi give contradictory indications about this
verb. Saidova (2006) provides several examples of its use, all with the case frame <ERG, NOM>,
whereas Kibrik & al. mention it as a verb used in the case frame <ERG, LOC>, but give no example.
J. Pasquereau & D. Creissels, Emergence and evolution of exceptional valency patterns, p. 6/8
(14) χʷaj-ol q̄’ʷar-e di-č’o. Karata
dog-ERG bite-PF 1SG₀-TPL[LOC]
‘The dog bit me.’
(15) χʷe-de di-g-e q̄’eleč’-ari. Akhvakh
dog-ERG 1SGₒ-TPL-LOC bite-PF
‘The dog bit me.’
(16) χʷāj-d q’̄ ān-nida o-sū -č’. Chamalal
dog-ERG bite-PF DIST-Mₒ-TPL[LOC]
‘The dog bit me.’
(17) χʷaji-di q̄’am-i ho-w. Godoberi
dog-ERG bite-PF₁ DIST-M
‘The dog bit him.’
(18) ʕamo-r salli di-b lela Bagvalal
donkeyₒ-ERG bite.PF₁ 1SGₒ[GEN]-N hand
‘The donkey bit my hand.’
A first interesting observation is that Godoberi q̄’ami and Chamalal q̄’āna ‘bite’ do not
have the same construction but come from a common root that can be reconstructed as
*q̄’am. Consequently, the difference observed in their case frames cannot be attributed
to their etymology, and a change must have occurred in the construction of one of
them.
A second interesting observation is that Bagvalal salīla ‘bite’ is quite obviously
cognate with Bagvalal & Chamalal salʷ, Karata sale, Tindi salu, Andi sol ‘tooth’. This
suggests that this verb results from the univerbation of a ‘noun + verb’ compound
similar to Tindi saldi bix̄ʲiɬʲa, in spite of the fact that the nominative case assigned to
the NP representing the bitee does not correspond to what could be expected from this
etymology.
The variation in the expression of ‘bite’ in Andic languages provides therefore
evidence supporting the reconstruction of the following evolution:
– at a first stage, the coalescence of a trivalent verb occurring in the regular frame
<ERG, NOM, ALL> with a noun occupying the nominative slot creates a bivalent
verb with the exceptional case frame <ERG, LOC>;
– at a second stage, attested by Bagvalal, the exceptional valency pattern resulting
from this evolution may be regularized into <ERG, NOM>.
The following observations can also be made about the verbs expressing ‘bite’ in Andic
languages:
– Several Andic languages express ‘eat’ by means of verbs that are reflexes of a root
*q̄’am: Akhvakh q̄’ōnuʟa ‘eat’, Karata q̄’amaɬa ‘eat’, Bagvalal q̄’anila ‘eat’. As
mentioned above, this root also has reflexes expressing ‘bite’ in Godoberi and
Chamalal. Interestingly, the irregular case frame observed with the reflexes of this
J. Pasquereau & D. Creissels, Emergence and evolution of exceptional valency patterns, p. 7/8
root expressing ‘bite’ is not found with the reflexes expressing ‘eat’, which always
occur in the case frame <ERG, NOM> characteristic of verbs encoding
prototypical transitive events.
– As illustrated by Ex. (15) above, the Akhvakh verb q̄’eleč’uruʟa ‘bite’ is used in the
case frame <ERG, LOC> to encode prototypical biting events that may cause pain
but do not result in a change of state of the bitee (for example, ‘The dog bit me’).
However, when referring to biting events that affect the physical integrity of the
bitee (bite off a piece of something and eat it), q̄’eleč’uruʟa is also attested with the
case frame <ERG, NOM> characteristic of the expression of prototypical
transitive events, as in Ex. (19).
(19) wašo-de ʕeče q̄’eleč’-ari.
boyₒ-ERG apple bite-PF
‘The boy bit off a chunk of apple.’
7. Conclusion
Andic languages are among the languages making a wide use of lexicalized
combinations ‘noun + verb’ in which the noun most commonly occupies the same
syntactic slot as the patient in prototypical transitive predication. Since Andic
languages are consistent ergative languages, the nominal element of such compounds
is in the nominative case and governs verb agreement. Consequently, the univerbation
of ‘noun + verb’ compounds in Andic languages yields verbs whose construction
includes no slot for a nominative NP governing verb agreement, creating thus potential
exceptions to the rule according to which, in Andic languages, the valency frames of
verbs must minimally include a nominative term controlling gender-number
agreement of the verb.2
The first conclusion that can be drawn from the data presented above is that, in the
evolution of languages, exceptional valency patterns that emerge as the accidental
result of lexicalization processes (in the case examined here, the univerbation of
lexicalized ‘noun + verb’ combinations in ergative languages) may subsequently
undergo a regularization process by means of a change in the encoding of one of the
arguments: in the case of ‘listen’, an argument originally encoded as an ergative NP
takes nominative marking, and in the case of ‘bite’, an argument originally encoded as
a locative NP takes nominative marking.
Moreover, the difference observed between ‘listen’ and ‘bite’ suggests that semantic
factors condition this regularization process. In the case of ‘bite’, the substitution of
nominative marking for locative marking results in the case frame <ERG, NOM>,
which in Andic languages is used to encode prototypical transitive events involving an
agent and a patient. By contrast, in the case of ‘listen’, regularization results in the case
frame <NOM, ALL>, typically used to encode movement towards a goal, and the
same is observed with ‘look at’. A plausible explanation is that the ergative argument
of ‘bite’ has more affinities with the prototypical agent than the ergative argument of
‘listen’ or ‘look at’, and the locative argument of ‘bite’ has more affinities with the
prototypical patient than the allative argument of ‘listen’ or ‘look at’. The ergative
2 See (Haspelmath 1993: 178–180) for a description of this process in Lezgi.
J. Pasquereau & D. Creissels, Emergence and evolution of exceptional valency patterns, p. 8/8
argument of ‘listen’ or ‘look at’ is clearly not a typical agent, whereas the allative
argument has clear affinities with the goal of movement. In other words, in the
regularization process, arguments whose initial case marking has clear semantic
motivations tend to maintain their case marking, whereas case marking is more likely
to be modified for arguments whose semantic roles can be viewed as relatively
ambiguous as regards their affinities with the prototypes underlying the use of cases.
This hypothesis is supported by the variation observed in the case frame of Akhvakh
q̄eleč’uruʟa ‘bite’, and by the fact that verbs glossed ‘eat’ cognate with verbs glossed
‘bite’ uniformly have the case frame <ERG, NOM>, since in the biting events of the
type illustrated by Ex. (19) as well as in eating events, in contrast with prototypical
biting events, the second participant shows a degree of affectedness similar to that of
prototypical patients.
Abbreviations
—ₒ : oblique stem / ADV: adverbial / ALL: allative / CAUS: causative / COP: copula /
DIST: distal demonstrative / ERG: ergative / F: feminine / FUT: future / GEN: genitive
/ H+: human plural / IMP: imperative / INF: infinitive / IPF: imperfective / LOC:
locative / M: masculine / nH+: non-human plural / N: neuter / NEG: negation / PF:
perfective / PL: plural / POT: potential / PROH: prohibitive / SG: singular / TPL:
topological marker
References
Charachidzé, G. 1981. Grammaire de la langue avar. Paris: Editions Jean-Favard.
Creissels, D. 2009. ‘Valency properties of Northern Akhvakh verbs.’
http://www.deniscreissels.fr/public/Creissels-valency_classes_project_Akhvakh.pdf
Haspelmath, Martin. 1993. A grammar of Lezgian. Berlin & New York: Mouton de Gruyter.
Kibrik, Alexandr (ed.). 1996. Godoberi. München: Lincom Europa.
Kibrik, Alexandr (ed.). 2001. Bagvalinskij jazyk: grammatika, teksty, slovari. Moskva: Nasledie.
Korjakov, Ju. B. 2006. Atlas of Caucasian languages. Moscow: Piligrim
Magomedova, P. 1999. Čamalinsko-russkij slovar’ [Chamalal-Russian dictionary].
Maxačkala: Dagestanskij Naučnyj Centr Rossiskoj Akademii Nauk.
Magomedova, P. 2003. Tindinsko-russkij slovar’ [Tindi-Russian dictionary]. Maxačkala:
Dagestanskij Naučnyj Centr Rossiskoj Akademii Nauk.
Magomedova, P. 2004. Bagvalinsko-russkij slovar’ [Bagvalal-Russian dictionary].
Maxačkala: Dagestanskij Naučnyj Centr Rossiskoj Akademii Nauk.
Magomedova, P & I. Abdulaeva. 2007. Axvaxsko-russkij slovar’ [Akhvakh-Russian
dictionary]. Maxačkala: Dagestanskij Naučnyj Centr Rossiskoj Akademii Nauk.
Magomedova, P & Š. Xalidova. 2001. Karatinsko-russkij slovar’ [Karata-Russian
dictionary]. Maxačkala: Dagestanskij Naučnyj Centr Rossiskoj Akademii Nauk.
Pasquereau, J. 2010. Spatial cases in Karata: spatial and non-spatial uses. First year’s
master thesis. University of Lyon
Pasquereau, J. 2011. Valency in Karata: a preliminary study. Second year’s master thesis.
University of Lyon
Saidova, P. 2006. Godoberinsko-russkij slovar’ [Godoberi-Russian dictionary].
Maxačkala: Dagestanskij Naučnyj Centr Rossiskoj Akademii Nauk.SLE – 44th annual meeting
Logroño, 8-11 September 2011
Emergence and evolution of exceptional valency patterns
in ergative languages: a case study
Jérémy Pasquereau
University of Massachusetts
jepasque@hotmail.com
Denis Creissels
University of Lyon
denis.creissels@univ-lyon2.fr
http://deniscreissels.fr
1. Introduction
This paper analyzes the possible origin and the evolution of a particular type of
exceptional valency pattern found in East Caucasian languages. It deals specifically
with a group of East Caucasian languages, the Andic languages (Andi, Akhvakh,
Bagvalal, Botlikh, Chamalal, Godoberi, Karata, and Tindi), spoken in the Western part
of Daghestan. The Andic languages constitute a genetic unit within the Avar-
Andic(-Tsezic) branch of the East Caucasian language family.
The data we present comes from our own work on two Andic languages, Akhvakh
and Karata – (Creissels 2010), (Pasquereau 2010), (Pasquereau 2011), and from the
consultation of dictionaries and grammars of Andic languages – (Kibrik 1996), (Kibrik
2001), (Magomedova 1999), (Magomedova 2003), (Magomedova 2004),
(Magomedova & Abdulaeva 2007), (Magomedova & Khalidova 2001), (Saidova 2006).
The paper is organized as follows. Section 2 provides basic information about the
grammatical structure of Andic languages. Section 3 provides illustrations of
exceptional case frames in Andic languages. Section 4 examines the question of the
emergence of the exceptional case frames <ERG, LOC> and <ERG, ALL>. Section 5
and 6 concentrate on the particular cases of ‘listen’ and ‘bite’, respectively. Section 7
puts forward some concluding remarks.
2. The basics of Andic morphosyntax
In Andic languages, the syntactic function of noun phrases is expressed by case
suffixes. Verbs agree in gender and number (not in person) but verb agreement is
redundant with case marking, since verbs consistently agree with their nominative
argument, and never express agreement with non-nominative noun phrases.
Constituent order is remarkably flexible and plays no role in the expression of
argument structure.
Andic languages have rich case systems, and in particular, very elaborate spatial
case systems. The nominative (alias absolutive), also used as the quotation form of
nouns, is characterized by a zero ending. The other cases are marked by suffixes
attached either to a stem coinciding with the nominative form, or to a special stem
traditionally called oblique stem (indicated by –ₒ in the glosses of examples). Spatial
case endings consist of two formatives, a topological marker (glossed TPL) and a
directionality marker.
J. Pasquereau & D. Creissels, Emergence and evolution of exceptional valency patterns, p. 2/8
As illustrated by Ex. (1), the coding of core syntactic roles in Andic languages is
consistently ergative:
– S (boʁosē ãdo in (1a)) and P (istaka in (1b)) are in the nominative case (alias
absolutive), contrasting with A (wašode in (1b)) in the ergative case.
– Transitive verbs do not agree with A, but agree with P in the same way as
intransitive verbs do with S: in (1a), the intransitive verb shows human plural
agreement; in (1b), A is masculine singular, whereas P is neuter singular, and the
verb shows neuter singular agreement.
(1) a. b-oʁosē ãd-o atobusī -ɬī-g-e b-eq’-īri. Akhvakh
H⁺-most person-PL bus-Nₒ-TPL-LOC H⁺-come-PF.H⁺
‘Most people came by bus.’
b. wašo-de istaka b-iq’ʷ-aj-ē godi.
boyₒ-ERG glass N-break-CAUS-ADV.N COP.N
‘The boy broke the glass.’
The valency frames of Andic verbs canonically include at least a slot for a nominative
NP (representing in particular the sole argument of monovalent verbs and the P
argument of prototypical action verbs). According to the account of verb valency
included in (Kibrik 2001: 369-376), in Bagvalal, this rule can be viewed as
exceptionless. However, most Andic languages have a restricted set of verbs used in
constructions in which no slot for a nominative NP can be posited.
3. Verbs without nominative arguments
In this section, we provide illustrations of Andic verbs used in case frames that do not
include a slot for a nominative NP. Note that, in the absence of a nominative
argument, the verb forms that express gender-number agreement show default neuter
singular agreement.
3.1. Monovalent verbs
Ex. (2) illustrates the case of a monovalent Akhvakh verb which, depending on the
situation referred to, may occur either in the canonical case frame <NOM>, as in
(2a), or in the exceptional case frame <LOC>, as in (2b).
(2) а. miq̄’i q̄’ʷaraɬ-ēhe godi. Akhvakh
road become_narrow-ADV.N COP.N
‘The road became narrow.’
b. miʕa-q̄-e q̄’ʷaraɬ-ēhe godi.
nose-TPL-LOC become_narrow-ADV.N COP.N
‘My nose is blocked.’ lit. ‘In the nose became narrow.’
J. Pasquereau & D. Creissels, Emergence and evolution of exceptional valency patterns, p. 3/8
Ex. (3) illustrates the Karata verb bibāɬa ‘cry’, used in the exceptional case frame
<ERG>.
(3) gugu-l ɬabc’e bib-ē. Karata
cuckoo-ERG three_times cry-PF
‘The cuckoo sang three times.’
3.2. Bivalent verbs
Ex. (4) illustrates a bivalent Akhvakh verb used in the exceptional case frame <ERG,
GEN>.
(4) hu-sʷ̄ -e daru-ɬī-ʟī ħalbix̄il-āri. Akhvakh
DIST-Mₒ-ERG medecine-Nₒ-GEN try-PF
‘He tried the medecine.’
3.3. Trivalent verbs
Ex. (5) & (6) illustrate a trivalent verb used in the exceptional case frame <ERG, ALL,
GEN>, in Karata and in Akhvakh.
(5) hu-ɬī-ʟī čela žo-ɬī mač-uwa du-g-a de-de Akhvakh
DIST-Nₒ-GEN other day-Nₒ[LOC] tell-POT 2SG-TPL-ALL 1SG-ERG
‘I will tell you about this another day.’
(6) k’̄ ʷāmsʷam hedela-ɬī-ʟ ̄ bas-imisē ! Karata
trivial.N thing-N-GEN tell-PROH
‘Don’t speak about trivial things!’
4. The emergence of the case frames <ERG, ALL> and <ERG, LOC>
In this section, we examine the case of verbs used in the case frames <ERG, ALL> or
<ERG, LOC>. In Andic languages these case frames are typically found with verbs
expressing the following meanings: ‘look at’ – Ex. (7), ‘listen’, ‘bite’, ‘pinch’ – Ex. (8),
‘sting’ – Ex. (9). For examples with ‘listen’ and ‘bite’, see Sections 5 & 6, where the
case of these verbs is examined in more detail.
(7) wašo-de di-g-a eq̄-ari. Akhvakh
boyₒ-ERG 1SG₀-TPL-ALL look_at-PF
‘The boy looked at me.’
(8) o-sʷ̄̌ -i č’un-o di-č’-i. Tindi
DIST-Mₒ-ERG pinch-PF 1SGₒ-TPL-LOC/ALL
‘He pinched me.’
J. Pasquereau & D. Creissels, Emergence and evolution of exceptional valency patterns, p. 4/8
(9) di-g-e ʟ̄’ižʷali-de č’in-āri Akhvakh
1SGₒ-TPL-LOC waspₒ-ERG sting-PF
‘A wasp stung me.’
Godoberi also has a verb obadi ‘kiss’ with the case frame <ERG, LOC>, but all the
other Andic languages for which we have data express this meaning by means of a
noun oba or ob̄a ‘kiss’ combined with a verb ‘do’, and consequently the exceptional
case frame of Godoberi obadi ‘kiss’ must probably be viewed as the result of the
univerbation of a do-compound.
The other cases are less easy to explain, but the variations observed in the
expression of the other meaning listed above in Andic languages support a hypothesis
already suggested by Charachidzé (1981) for Avar, according to which these
exceptional valency frames may result from the reduction of the regular frames <ERG,
NOM, ALL> or <ERG, NOM, LOC>, characteristic of verbs expressing meanings or
the type ‘X applies/holds Y on Z’ (and found in particular with the verbs expressing
‘hit’, with the hittee in the locative/allative and the instrument in the nominative).
The point is that not all Andic languages have bivalent verbs with meanings such as
‘look at’, ‘listen’, ‘bite’, ‘sting’, ‘pinch’. Among the Andic languages, it is also common
to find such meanings expressed by means of less specific verbs combined with three
NPs. For example, Tindi expresses ‘sting’ as eq̄ʷa k̄ʷēɬʲa, lit. ‘hit the sting (on someone)’,
and ‘bite’ as saldi bix̄ʲiɬʲa, lit. ‘hold the teeth (on someone)’, with respectively eq̄ʷa
‘sting (noun)’ and saldi ‘teeth’ occupying the nominative slot.
Starting from the hypothesis of valency frames originally including three slots, at
least two reduction scenarios can be imagined: conventionalization of the ellipsis of
the nominative argument, or fusion of the nominative argument with the verb. In
Andic languages, there is evidence for the conventionalization-of-ellipsis scenario in
the case of other verbs with exceptional case frames, but not for those examined here.
By contrast, some of them at least show evidence for the fusion scenario. The clearest
cases are those of ‘listen’ and ‘bite’, examined in Sections 5 and 6.
5. Emergence and evolution of verbs ‘listen’ in Andic languages
Three situations are found among Andic languages with respect to the expression of
‘listen’:
– Some Andic languages express ‘listen’ by means of a construction involving a
nominative noun phrase with the meaning ‘ear’ in addition to those encoding the
two participants. For example, Godoberi expresses ‘listen’ as hãt’uk’ja rik̄i, literally
‘fix the ear (on someone/something)’ – Ex. (10). Formally, this construction is an
instance of the regular valency pattern <ERG, NOM, ALL> with hãt’uk’ja ‘ear’
filling the nominative slot.
– Others have a verb ‘listen’ with the exceptional case frame <ERG, ALL>: Tindi
anix̄ʲiɬʲa – Ex. (11), Chamalal woɬuk’la;
– A verb ‘listen’ with the regular case frame <NOM, ALL> is found in three Andic
languages: Akhvakh hãdax̄uruʟa – Ex. (12), Karata ãdukaɬa – Ex. (13) , Bagvalal
aštila.
J. Pasquereau & D. Creissels, Emergence and evolution of exceptional valency patterns, p. 5/8
(10) wašu-di imu-q̄-i hãt’uk’ʲa r-ikk-i r-ukk-ida. Godoberi
sonₒ-ERG fatherₒ-TPL-LOC/ALL ear N⁺-hold-INF N⁺-must-IPF
‘The son must listen to his father.’
(11) di-q-̄ a anix̄ʲ-ā hik’ʲi o-sʷ̄̌ -ī Tindi
1SGₒ-TPL-LOC/ALL listen-IPF NEG DIST-Mₒ-ERG
‘He does not listen to me.’
(12) waša imo-g-a hãdax̄-ari. Akhvakh
boy fatherₒ-TPL-ALL listen-PF
‘The boy listened to his father.’
(13) waša imo-χa-r ãduk-e. Karata
boy fatherₒ-TPL- ALL listen-PF
‘The boy listened to his father.’
Interestingly, ‘fix the ear on’ is the obvious etymology of Akhvakh hãdax̄uruʟa
(compare with hãde ‘ear’, bix̄uruʟa ‘fix’), in spite of the fact that the nominative case
assigned to the NP representing the listener does not correspond to what could be
expected from this etymology.
The variation in the expression of ‘listen’ in Andic languages provides therefore
evidence supporting the reconstruction of the following evolution:
– at a first stage, the coalescence of a trivalent verb occurring in the regular frame
<ERG, NOM, ALL> with a noun occupying the nominative slot creates a bivalent
verb with the exceptional case frame <ERG, ALL>;
– at a second stage, attested by Akhvakh and Karata, the exceptional valency pattern
resulting from this evolution may be regularized into <NOM, ALL>.
6. Variations in the case frame of ‘bite’ in Andic languages
Among Andic languages, variations similar to those observed in the expression of
‘listen’ are attested for the expression of ‘bite’ too, with however a different case frame
in the languages in which the case frame of ‘bite’ has been regularized:
– Some Andic languages express ‘bite’ by means of a construction involving a noun
phrase with the meaning ‘tooth’ in addition to those encoding the two participants:
as mentioned above, Tindi expresses ‘bite’ as saldi bix̄ʲiɬʲa, lit. ‘hold the teeth (on
someone)’.
– Others have a verb ‘bite’ with the exceptional case frame <ERG, ALL>: Karata
q̄’ʷaraɬa – Ex. (14), Akhvakh q̄’eleč’uruʟa – Ex. (15), Chamalal q̄’āna – Ex. (16);
– A verb ‘bite’ with the regular case frame <ERG, NOM> is found in two Andic
languages: Godoberi q’̄ ami – Ex. (17),1 Bagvalal salīla – Ex. (18).
1 Note however that the two available sources on Godoberi give contradictory indications about this
verb. Saidova (2006) provides several examples of its use, all with the case frame <ERG, NOM>,
whereas Kibrik & al. mention it as a verb used in the case frame <ERG, LOC>, but give no example.
J. Pasquereau & D. Creissels, Emergence and evolution of exceptional valency patterns, p. 6/8
(14) χʷaj-ol q̄’ʷar-e di-č’o. Karata
dog-ERG bite-PF 1SG₀-TPL[LOC]
‘The dog bit me.’
(15) χʷe-de di-g-e q̄’eleč’-ari. Akhvakh
dog-ERG 1SGₒ-TPL-LOC bite-PF
‘The dog bit me.’
(16) χʷāj-d q’̄ ān-nida o-sū -č’. Chamalal
dog-ERG bite-PF DIST-Mₒ-TPL[LOC]
‘The dog bit me.’
(17) χʷaji-di q̄’am-i ho-w. Godoberi
dog-ERG bite-PF₁ DIST-M
‘The dog bit him.’
(18) ʕamo-r salli di-b lela Bagvalal
donkeyₒ-ERG bite.PF₁ 1SGₒ[GEN]-N hand
‘The donkey bit my hand.’
A first interesting observation is that Godoberi q̄’ami and Chamalal q̄’āna ‘bite’ do not
have the same construction but come from a common root that can be reconstructed as
*q̄’am. Consequently, the difference observed in their case frames cannot be attributed
to their etymology, and a change must have occurred in the construction of one of
them.
A second interesting observation is that Bagvalal salīla ‘bite’ is quite obviously
cognate with Bagvalal & Chamalal salʷ, Karata sale, Tindi salu, Andi sol ‘tooth’. This
suggests that this verb results from the univerbation of a ‘noun + verb’ compound
similar to Tindi saldi bix̄ʲiɬʲa, in spite of the fact that the nominative case assigned to
the NP representing the bitee does not correspond to what could be expected from this
etymology.
The variation in the expression of ‘bite’ in Andic languages provides therefore
evidence supporting the reconstruction of the following evolution:
– at a first stage, the coalescence of a trivalent verb occurring in the regular frame
<ERG, NOM, ALL> with a noun occupying the nominative slot creates a bivalent
verb with the exceptional case frame <ERG, LOC>;
– at a second stage, attested by Bagvalal, the exceptional valency pattern resulting
from this evolution may be regularized into <ERG, NOM>.
The following observations can also be made about the verbs expressing ‘bite’ in Andic
languages:
– Several Andic languages express ‘eat’ by means of verbs that are reflexes of a root
*q̄’am: Akhvakh q̄’ōnuʟa ‘eat’, Karata q̄’amaɬa ‘eat’, Bagvalal q̄’anila ‘eat’. As
mentioned above, this root also has reflexes expressing ‘bite’ in Godoberi and
Chamalal. Interestingly, the irregular case frame observed with the reflexes of this
J. Pasquereau & D. Creissels, Emergence and evolution of exceptional valency patterns, p. 7/8
root expressing ‘bite’ is not found with the reflexes expressing ‘eat’, which always
occur in the case frame <ERG, NOM> characteristic of verbs encoding
prototypical transitive events.
– As illustrated by Ex. (15) above, the Akhvakh verb q̄’eleč’uruʟa ‘bite’ is used in the
case frame <ERG, LOC> to encode prototypical biting events that may cause pain
but do not result in a change of state of the bitee (for example, ‘The dog bit me’).
However, when referring to biting events that affect the physical integrity of the
bitee (bite off a piece of something and eat it), q̄’eleč’uruʟa is also attested with the
case frame <ERG, NOM> characteristic of the expression of prototypical
transitive events, as in Ex. (19).
(19) wašo-de ʕeče q̄’eleč’-ari.
boyₒ-ERG apple bite-PF
‘The boy bit off a chunk of apple.’
7. Conclusion
Andic languages are among the languages making a wide use of lexicalized
combinations ‘noun + verb’ in which the noun most commonly occupies the same
syntactic slot as the patient in prototypical transitive predication. Since Andic
languages are consistent ergative languages, the nominal element of such compounds
is in the nominative case and governs verb agreement. Consequently, the univerbation
of ‘noun + verb’ compounds in Andic languages yields verbs whose construction
includes no slot for a nominative NP governing verb agreement, creating thus potential
exceptions to the rule according to which, in Andic languages, the valency frames of
verbs must minimally include a nominative term controlling gender-number
agreement of the verb.2
The first conclusion that can be drawn from the data presented above is that, in the
evolution of languages, exceptional valency patterns that emerge as the accidental
result of lexicalization processes (in the case examined here, the univerbation of
lexicalized ‘noun + verb’ combinations in ergative languages) may subsequently
undergo a regularization process by means of a change in the encoding of one of the
arguments: in the case of ‘listen’, an argument originally encoded as an ergative NP
takes nominative marking, and in the case of ‘bite’, an argument originally encoded as
a locative NP takes nominative marking.
Moreover, the difference observed between ‘listen’ and ‘bite’ suggests that semantic
factors condition this regularization process. In the case of ‘bite’, the substitution of
nominative marking for locative marking results in the case frame <ERG, NOM>,
which in Andic languages is used to encode prototypical transitive events involving an
agent and a patient. By contrast, in the case of ‘listen’, regularization results in the case
frame <NOM, ALL>, typically used to encode movement towards a goal, and the
same is observed with ‘look at’. A plausible explanation is that the ergative argument
of ‘bite’ has more affinities with the prototypical agent than the ergative argument of
‘listen’ or ‘look at’, and the locative argument of ‘bite’ has more affinities with the
prototypical patient than the allative argument of ‘listen’ or ‘look at’. The ergative
2 See (Haspelmath 1993: 178–180) for a description of this process in Lezgi.
J. Pasquereau & D. Creissels, Emergence and evolution of exceptional valency patterns, p. 8/8
argument of ‘listen’ or ‘look at’ is clearly not a typical agent, whereas the allative
argument has clear affinities with the goal of movement. In other words, in the
regularization process, arguments whose initial case marking has clear semantic
motivations tend to maintain their case marking, whereas case marking is more likely
to be modified for arguments whose semantic roles can be viewed as relatively
ambiguous as regards their affinities with the prototypes underlying the use of cases.
This hypothesis is supported by the variation observed in the case frame of Akhvakh
q̄eleč’uruʟa ‘bite’, and by the fact that verbs glossed ‘eat’ cognate with verbs glossed
‘bite’ uniformly have the case frame <ERG, NOM>, since in the biting events of the
type illustrated by Ex. (19) as well as in eating events, in contrast with prototypical
biting events, the second participant shows a degree of affectedness similar to that of
prototypical patients.
Abbreviations
—ₒ : oblique stem / ADV: adverbial / ALL: allative / CAUS: causative / COP: copula /
DIST: distal demonstrative / ERG: ergative / F: feminine / FUT: future / GEN: genitive
/ H+: human plural / IMP: imperative / INF: infinitive / IPF: imperfective / LOC:
locative / M: masculine / nH+: non-human plural / N: neuter / NEG: negation / PF:
perfective / PL: plural / POT: potential / PROH: prohibitive / SG: singular / TPL:
topological marker
References
Charachidzé, G. 1981. Grammaire de la langue avar. Paris: Editions Jean-Favard.
Creissels, D. 2009. ‘Valency properties of Northern Akhvakh verbs.’
http://www.deniscreissels.fr/public/Creissels-valency_classes_project_Akhvakh.pdf
Haspelmath, Martin. 1993. A grammar of Lezgian. Berlin & New York: Mouton de Gruyter.
Kibrik, Alexandr (ed.). 1996. Godoberi. München: Lincom Europa.
Kibrik, Alexandr (ed.). 2001. Bagvalinskij jazyk: grammatika, teksty, slovari. Moskva: Nasledie.
Korjakov, Ju. B. 2006. Atlas of Caucasian languages. Moscow: Piligrim
Magomedova, P. 1999. Čamalinsko-russkij slovar’ [Chamalal-Russian dictionary].
Maxačkala: Dagestanskij Naučnyj Centr Rossiskoj Akademii Nauk.
Magomedova, P. 2003. Tindinsko-russkij slovar’ [Tindi-Russian dictionary]. Maxačkala:
Dagestanskij Naučnyj Centr Rossiskoj Akademii Nauk.
Magomedova, P. 2004. Bagvalinsko-russkij slovar’ [Bagvalal-Russian dictionary].
Maxačkala: Dagestanskij Naučnyj Centr Rossiskoj Akademii Nauk.
Magomedova, P & I. Abdulaeva. 2007. Axvaxsko-russkij slovar’ [Akhvakh-Russian
dictionary]. Maxačkala: Dagestanskij Naučnyj Centr Rossiskoj Akademii Nauk.
Magomedova, P & Š. Xalidova. 2001. Karatinsko-russkij slovar’ [Karata-Russian
dictionary]. Maxačkala: Dagestanskij Naučnyj Centr Rossiskoj Akademii Nauk.
Pasquereau, J. 2010. Spatial cases in Karata: spatial and non-spatial uses. First year’s
master thesis. University of Lyon
Pasquereau, J. 2011. Valency in Karata: a preliminary study. Second year’s master thesis.
University of Lyon
Saidova, P. 2006. Godoberinsko-russkij slovar’ [Godoberi-Russian dictionary].
Maxačkala: Dagestanskij Naučnyj Centr Rossiskoj Akademii Nauk.SLE – 44th annual meeting
Logroño, 8-11 September 2011
Emergence and evolution of exceptional valency patterns
in ergative languages: a case study
Jérémy Pasquereau
University of Massachusetts
jepasque@hotmail.com
Denis Creissels
University of Lyon
denis.creissels@univ-lyon2.fr
http://deniscreissels.fr
1. Introduction
This paper analyzes the possible origin and the evolution of a particular type of
exceptional valency pattern found in East Caucasian languages. It deals specifically
with a group of East Caucasian languages, the Andic languages (Andi, Akhvakh,
Bagvalal, Botlikh, Chamalal, Godoberi, Karata, and Tindi), spoken in the Western part
of Daghestan. The Andic languages constitute a genetic unit within the Avar-
Andic(-Tsezic) branch of the East Caucasian language family.
The data we present comes from our own work on two Andic languages, Akhvakh
and Karata – (Creissels 2010), (Pasquereau 2010), (Pasquereau 2011), and from the
consultation of dictionaries and grammars of Andic languages – (Kibrik 1996), (Kibrik
2001), (Magomedova 1999), (Magomedova 2003), (Magomedova 2004),
(Magomedova & Abdulaeva 2007), (Magomedova & Khalidova 2001), (Saidova 2006).
The paper is organized as follows. Section 2 provides basic information about the
grammatical structure of Andic languages. Section 3 provides illustrations of
exceptional case frames in Andic languages. Section 4 examines the question of the
emergence of the exceptional case frames <ERG, LOC> and <ERG, ALL>. Section 5
and 6 concentrate on the particular cases of ‘listen’ and ‘bite’, respectively. Section 7
puts forward some concluding remarks.
2. The basics of Andic morphosyntax
In Andic languages, the syntactic function of noun phrases is expressed by case
suffixes. Verbs agree in gender and number (not in person) but verb agreement is
redundant with case marking, since verbs consistently agree with their nominative
argument, and never express agreement with non-nominative noun phrases.
Constituent order is remarkably flexible and plays no role in the expression of
argument structure.
Andic languages have rich case systems, and in particular, very elaborate spatial
case systems. The nominative (alias absolutive), also used as the quotation form of
nouns, is characterized by a zero ending. The other cases are marked by suffixes
attached either to a stem coinciding with the nominative form, or to a special stem
traditionally called oblique stem (indicated by –ₒ in the glosses of examples). Spatial
case endings consist of two formatives, a topological marker (glossed TPL) and a
directionality marker.
J. Pasquereau & D. Creissels, Emergence and evolution of exceptional valency patterns, p. 2/8
As illustrated by Ex. (1), the coding of core syntactic roles in Andic languages is
consistently ergative:
– S (boʁosē ãdo in (1a)) and P (istaka in (1b)) are in the nominative case (alias
absolutive), contrasting with A (wašode in (1b)) in the ergative case.
– Transitive verbs do not agree with A, but agree with P in the same way as
intransitive verbs do with S: in (1a), the intransitive verb shows human plural
agreement; in (1b), A is masculine singular, whereas P is neuter singular, and the
verb shows neuter singular agreement.
(1) a. b-oʁosē ãd-o atobusī -ɬī-g-e b-eq’-īri. Akhvakh
H⁺-most person-PL bus-Nₒ-TPL-LOC H⁺-come-PF.H⁺
‘Most people came by bus.’
b. wašo-de istaka b-iq’ʷ-aj-ē godi.
boyₒ-ERG glass N-break-CAUS-ADV.N COP.N
‘The boy broke the glass.’
The valency frames of Andic verbs canonically include at least a slot for a nominative
NP (representing in particular the sole argument of monovalent verbs and the P
argument of prototypical action verbs). According to the account of verb valency
included in (Kibrik 2001: 369-376), in Bagvalal, this rule can be viewed as
exceptionless. However, most Andic languages have a restricted set of verbs used in
constructions in which no slot for a nominative NP can be posited.
3. Verbs without nominative arguments
In this section, we provide illustrations of Andic verbs used in case frames that do not
include a slot for a nominative NP. Note that, in the absence of a nominative
argument, the verb forms that express gender-number agreement show default neuter
singular agreement.
3.1. Monovalent verbs
Ex. (2) illustrates the case of a monovalent Akhvakh verb which, depending on the
situation referred to, may occur either in the canonical case frame <NOM>, as in
(2a), or in the exceptional case frame <LOC>, as in (2b).
(2) а. miq̄’i q̄’ʷaraɬ-ēhe godi. Akhvakh
road become_narrow-ADV.N COP.N
‘The road became narrow.’
b. miʕa-q̄-e q̄’ʷaraɬ-ēhe godi.
nose-TPL-LOC become_narrow-ADV.N COP.N
‘My nose is blocked.’ lit. ‘In the nose became narrow.’
J. Pasquereau & D. Creissels, Emergence and evolution of exceptional valency patterns, p. 3/8
Ex. (3) illustrates the Karata verb bibāɬa ‘cry’, used in the exceptional case frame
<ERG>.
(3) gugu-l ɬabc’e bib-ē. Karata
cuckoo-ERG three_times cry-PF
‘The cuckoo sang three times.’
3.2. Bivalent verbs
Ex. (4) illustrates a bivalent Akhvakh verb used in the exceptional case frame <ERG,
GEN>.
(4) hu-sʷ̄ -e daru-ɬī-ʟī ħalbix̄il-āri. Akhvakh
DIST-Mₒ-ERG medecine-Nₒ-GEN try-PF
‘He tried the medecine.’
3.3. Trivalent verbs
Ex. (5) & (6) illustrate a trivalent verb used in the exceptional case frame <ERG, ALL,
GEN>, in Karata and in Akhvakh.
(5) hu-ɬī-ʟī čela žo-ɬī mač-uwa du-g-a de-de Akhvakh
DIST-Nₒ-GEN other day-Nₒ[LOC] tell-POT 2SG-TPL-ALL 1SG-ERG
‘I will tell you about this another day.’
(6) k’̄ ʷāmsʷam hedela-ɬī-ʟ ̄ bas-imisē ! Karata
trivial.N thing-N-GEN tell-PROH
‘Don’t speak about trivial things!’
4. The emergence of the case frames <ERG, ALL> and <ERG, LOC>
In this section, we examine the case of verbs used in the case frames <ERG, ALL> or
<ERG, LOC>. In Andic languages these case frames are typically found with verbs
expressing the following meanings: ‘look at’ – Ex. (7), ‘listen’, ‘bite’, ‘pinch’ – Ex. (8),
‘sting’ – Ex. (9). For examples with ‘listen’ and ‘bite’, see Sections 5 & 6, where the
case of these verbs is examined in more detail.
(7) wašo-de di-g-a eq̄-ari. Akhvakh
boyₒ-ERG 1SG₀-TPL-ALL look_at-PF
‘The boy looked at me.’
(8) o-sʷ̄̌ -i č’un-o di-č’-i. Tindi
DIST-Mₒ-ERG pinch-PF 1SGₒ-TPL-LOC/ALL
‘He pinched me.’
J. Pasquereau & D. Creissels, Emergence and evolution of exceptional valency patterns, p. 4/8
(9) di-g-e ʟ̄’ižʷali-de č’in-āri Akhvakh
1SGₒ-TPL-LOC waspₒ-ERG sting-PF
‘A wasp stung me.’
Godoberi also has a verb obadi ‘kiss’ with the case frame <ERG, LOC>, but all the
other Andic languages for which we have data express this meaning by means of a
noun oba or ob̄a ‘kiss’ combined with a verb ‘do’, and consequently the exceptional
case frame of Godoberi obadi ‘kiss’ must probably be viewed as the result of the
univerbation of a do-compound.
The other cases are less easy to explain, but the variations observed in the
expression of the other meaning listed above in Andic languages support a hypothesis
already suggested by Charachidzé (1981) for Avar, according to which these
exceptional valency frames may result from the reduction of the regular frames <ERG,
NOM, ALL> or <ERG, NOM, LOC>, characteristic of verbs expressing meanings or
the type ‘X applies/holds Y on Z’ (and found in particular with the verbs expressing
‘hit’, with the hittee in the locative/allative and the instrument in the nominative).
The point is that not all Andic languages have bivalent verbs with meanings such as
‘look at’, ‘listen’, ‘bite’, ‘sting’, ‘pinch’. Among the Andic languages, it is also common
to find such meanings expressed by means of less specific verbs combined with three
NPs. For example, Tindi expresses ‘sting’ as eq̄ʷa k̄ʷēɬʲa, lit. ‘hit the sting (on someone)’,
and ‘bite’ as saldi bix̄ʲiɬʲa, lit. ‘hold the teeth (on someone)’, with respectively eq̄ʷa
‘sting (noun)’ and saldi ‘teeth’ occupying the nominative slot.
Starting from the hypothesis of valency frames originally including three slots, at
least two reduction scenarios can be imagined: conventionalization of the ellipsis of
the nominative argument, or fusion of the nominative argument with the verb. In
Andic languages, there is evidence for the conventionalization-of-ellipsis scenario in
the case of other verbs with exceptional case frames, but not for those examined here.
By contrast, some of them at least show evidence for the fusion scenario. The clearest
cases are those of ‘listen’ and ‘bite’, examined in Sections 5 and 6.
5. Emergence and evolution of verbs ‘listen’ in Andic languages
Three situations are found among Andic languages with respect to the expression of
‘listen’:
– Some Andic languages express ‘listen’ by means of a construction involving a
nominative noun phrase with the meaning ‘ear’ in addition to those encoding the
two participants. For example, Godoberi expresses ‘listen’ as hãt’uk’ja rik̄i, literally
‘fix the ear (on someone/something)’ – Ex. (10). Formally, this construction is an
instance of the regular valency pattern <ERG, NOM, ALL> with hãt’uk’ja ‘ear’
filling the nominative slot.
– Others have a verb ‘listen’ with the exceptional case frame <ERG, ALL>: Tindi
anix̄ʲiɬʲa – Ex. (11), Chamalal woɬuk’la;
– A verb ‘listen’ with the regular case frame <NOM, ALL> is found in three Andic
languages: Akhvakh hãdax̄uruʟa – Ex. (12), Karata ãdukaɬa – Ex. (13) , Bagvalal
aštila.
J. Pasquereau & D. Creissels, Emergence and evolution of exceptional valency patterns, p. 5/8
(10) wašu-di imu-q̄-i hãt’uk’ʲa r-ikk-i r-ukk-ida. Godoberi
sonₒ-ERG fatherₒ-TPL-LOC/ALL ear N⁺-hold-INF N⁺-must-IPF
‘The son must listen to his father.’
(11) di-q-̄ a anix̄ʲ-ā hik’ʲi o-sʷ̄̌ -ī Tindi
1SGₒ-TPL-LOC/ALL listen-IPF NEG DIST-Mₒ-ERG
‘He does not listen to me.’
(12) waša imo-g-a hãdax̄-ari. Akhvakh
boy fatherₒ-TPL-ALL listen-PF
‘The boy listened to his father.’
(13) waša imo-χa-r ãduk-e. Karata
boy fatherₒ-TPL- ALL listen-PF
‘The boy listened to his father.’
Interestingly, ‘fix the ear on’ is the obvious etymology of Akhvakh hãdax̄uruʟa
(compare with hãde ‘ear’, bix̄uruʟa ‘fix’), in spite of the fact that the nominative case
assigned to the NP representing the listener does not correspond to what could be
expected from this etymology.
The variation in the expression of ‘listen’ in Andic languages provides therefore
evidence supporting the reconstruction of the following evolution:
– at a first stage, the coalescence of a trivalent verb occurring in the regular frame
<ERG, NOM, ALL> with a noun occupying the nominative slot creates a bivalent
verb with the exceptional case frame <ERG, ALL>;
– at a second stage, attested by Akhvakh and Karata, the exceptional valency pattern
resulting from this evolution may be regularized into <NOM, ALL>.
6. Variations in the case frame of ‘bite’ in Andic languages
Among Andic languages, variations similar to those observed in the expression of
‘listen’ are attested for the expression of ‘bite’ too, with however a different case frame
in the languages in which the case frame of ‘bite’ has been regularized:
– Some Andic languages express ‘bite’ by means of a construction involving a noun
phrase with the meaning ‘tooth’ in addition to those encoding the two participants:
as mentioned above, Tindi expresses ‘bite’ as saldi bix̄ʲiɬʲa, lit. ‘hold the teeth (on
someone)’.
– Others have a verb ‘bite’ with the exceptional case frame <ERG, ALL>: Karata
q̄’ʷaraɬa – Ex. (14), Akhvakh q̄’eleč’uruʟa – Ex. (15), Chamalal q̄’āna – Ex. (16);
– A verb ‘bite’ with the regular case frame <ERG, NOM> is found in two Andic
languages: Godoberi q’̄ ami – Ex. (17),1 Bagvalal salīla – Ex. (18).
1 Note however that the two available sources on Godoberi give contradictory indications about this
verb. Saidova (2006) provides several examples of its use, all with the case frame <ERG, NOM>,
whereas Kibrik & al. mention it as a verb used in the case frame <ERG, LOC>, but give no example.
J. Pasquereau & D. Creissels, Emergence and evolution of exceptional valency patterns, p. 6/8
(14) χʷaj-ol q̄’ʷar-e di-č’o. Karata
dog-ERG bite-PF 1SG₀-TPL[LOC]
‘The dog bit me.’
(15) χʷe-de di-g-e q̄’eleč’-ari. Akhvakh
dog-ERG 1SGₒ-TPL-LOC bite-PF
‘The dog bit me.’
(16) χʷāj-d q’̄ ān-nida o-sū -č’. Chamalal
dog-ERG bite-PF DIST-Mₒ-TPL[LOC]
‘The dog bit me.’
(17) χʷaji-di q̄’am-i ho-w. Godoberi
dog-ERG bite-PF₁ DIST-M
‘The dog bit him.’
(18) ʕamo-r salli di-b lela Bagvalal
donkeyₒ-ERG bite.PF₁ 1SGₒ[GEN]-N hand
‘The donkey bit my hand.’
A first interesting observation is that Godoberi q̄’ami and Chamalal q̄’āna ‘bite’ do not
have the same construction but come from a common root that can be reconstructed as
*q̄’am. Consequently, the difference observed in their case frames cannot be attributed
to their etymology, and a change must have occurred in the construction of one of
them.
A second interesting observation is that Bagvalal salīla ‘bite’ is quite obviously
cognate with Bagvalal & Chamalal salʷ, Karata sale, Tindi salu, Andi sol ‘tooth’. This
suggests that this verb results from the univerbation of a ‘noun + verb’ compound
similar to Tindi saldi bix̄ʲiɬʲa, in spite of the fact that the nominative case assigned to
the NP representing the bitee does not correspond to what could be expected from this
etymology.
The variation in the expression of ‘bite’ in Andic languages provides therefore
evidence supporting the reconstruction of the following evolution:
– at a first stage, the coalescence of a trivalent verb occurring in the regular frame
<ERG, NOM, ALL> with a noun occupying the nominative slot creates a bivalent
verb with the exceptional case frame <ERG, LOC>;
– at a second stage, attested by Bagvalal, the exceptional valency pattern resulting
from this evolution may be regularized into <ERG, NOM>.
The following observations can also be made about the verbs expressing ‘bite’ in Andic
languages:
– Several Andic languages express ‘eat’ by means of verbs that are reflexes of a root
*q̄’am: Akhvakh q̄’ōnuʟa ‘eat’, Karata q̄’amaɬa ‘eat’, Bagvalal q̄’anila ‘eat’. As
mentioned above, this root also has reflexes expressing ‘bite’ in Godoberi and
Chamalal. Interestingly, the irregular case frame observed with the reflexes of this
J. Pasquereau & D. Creissels, Emergence and evolution of exceptional valency patterns, p. 7/8
root expressing ‘bite’ is not found with the reflexes expressing ‘eat’, which always
occur in the case frame <ERG, NOM> characteristic of verbs encoding
prototypical transitive events.
– As illustrated by Ex. (15) above, the Akhvakh verb q̄’eleč’uruʟa ‘bite’ is used in the
case frame <ERG, LOC> to encode prototypical biting events that may cause pain
but do not result in a change of state of the bitee (for example, ‘The dog bit me’).
However, when referring to biting events that affect the physical integrity of the
bitee (bite off a piece of something and eat it), q̄’eleč’uruʟa is also attested with the
case frame <ERG, NOM> characteristic of the expression of prototypical
transitive events, as in Ex. (19).
(19) wašo-de ʕeče q̄’eleč’-ari.
boyₒ-ERG apple bite-PF
‘The boy bit off a chunk of apple.’
7. Conclusion
Andic languages are among the languages making a wide use of lexicalized
combinations ‘noun + verb’ in which the noun most commonly occupies the same
syntactic slot as the patient in prototypical transitive predication. Since Andic
languages are consistent ergative languages, the nominal element of such compounds
is in the nominative case and governs verb agreement. Consequently, the univerbation
of ‘noun + verb’ compounds in Andic languages yields verbs whose construction
includes no slot for a nominative NP governing verb agreement, creating thus potential
exceptions to the rule according to which, in Andic languages, the valency frames of
verbs must minimally include a nominative term controlling gender-number
agreement of the verb.2
The first conclusion that can be drawn from the data presented above is that, in the
evolution of languages, exceptional valency patterns that emerge as the accidental
result of lexicalization processes (in the case examined here, the univerbation of
lexicalized ‘noun + verb’ combinations in ergative languages) may subsequently
undergo a regularization process by means of a change in the encoding of one of the
arguments: in the case of ‘listen’, an argument originally encoded as an ergative NP
takes nominative marking, and in the case of ‘bite’, an argument originally encoded as
a locative NP takes nominative marking.
Moreover, the difference observed between ‘listen’ and ‘bite’ suggests that semantic
factors condition this regularization process. In the case of ‘bite’, the substitution of
nominative marking for locative marking results in the case frame <ERG, NOM>,
which in Andic languages is used to encode prototypical transitive events involving an
agent and a patient. By contrast, in the case of ‘listen’, regularization results in the case
frame <NOM, ALL>, typically used to encode movement towards a goal, and the
same is observed with ‘look at’. A plausible explanation is that the ergative argument
of ‘bite’ has more affinities with the prototypical agent than the ergative argument of
‘listen’ or ‘look at’, and the locative argument of ‘bite’ has more affinities with the
prototypical patient than the allative argument of ‘listen’ or ‘look at’. The ergative
2 See (Haspelmath 1993: 178–180) for a description of this process in Lezgi.
J. Pasquereau & D. Creissels, Emergence and evolution of exceptional valency patterns, p. 8/8
argument of ‘listen’ or ‘look at’ is clearly not a typical agent, whereas the allative
argument has clear affinities with the goal of movement. In other words, in the
regularization process, arguments whose initial case marking has clear semantic
motivations tend to maintain their case marking, whereas case marking is more likely
to be modified for arguments whose semantic roles can be viewed as relatively
ambiguous as regards their affinities with the prototypes underlying the use of cases.
This hypothesis is supported by the variation observed in the case frame of Akhvakh
q̄eleč’uruʟa ‘bite’, and by the fact that verbs glossed ‘eat’ cognate with verbs glossed
‘bite’ uniformly have the case frame <ERG, NOM>, since in the biting events of the
type illustrated by Ex. (19) as well as in eating events, in contrast with prototypical
biting events, the second participant shows a degree of affectedness similar to that of
prototypical patients.
Abbreviations
—ₒ : oblique stem / ADV: adverbial / ALL: allative / CAUS: causative / COP: copula /
DIST: distal demonstrative / ERG: ergative / F: feminine / FUT: future / GEN: genitive
/ H+: human plural / IMP: imperative / INF: infinitive / IPF: imperfective / LOC:
locative / M: masculine / nH+: non-human plural / N: neuter / NEG: negation / PF:
perfective / PL: plural / POT: potential / PROH: prohibitive / SG: singular / TPL:
topological marker
References
Charachidzé, G. 1981. Grammaire de la langue avar. Paris: Editions Jean-Favard.
Creissels, D. 2009. ‘Valency properties of Northern Akhvakh verbs.’
http://www.deniscreissels.fr/public/Creissels-valency_classes_project_Akhvakh.pdf
Haspelmath, Martin. 1993. A grammar of Lezgian. Berlin & New York: Mouton de Gruyter.
Kibrik, Alexandr (ed.). 1996. Godoberi. München: Lincom Europa.
Kibrik, Alexandr (ed.). 2001. Bagvalinskij jazyk: grammatika, teksty, slovari. Moskva: Nasledie.
Korjakov, Ju. B. 2006. Atlas of Caucasian languages. Moscow: Piligrim
Magomedova, P. 1999. Čamalinsko-russkij slovar’ [Chamalal-Russian dictionary].
Maxačkala: Dagestanskij Naučnyj Centr Rossiskoj Akademii Nauk.
Magomedova, P. 2003. Tindinsko-russkij slovar’ [Tindi-Russian dictionary]. Maxačkala:
Dagestanskij Naučnyj Centr Rossiskoj Akademii Nauk.
Magomedova, P. 2004. Bagvalinsko-russkij slovar’ [Bagvalal-Russian dictionary].
Maxačkala: Dagestanskij Naučnyj Centr Rossiskoj Akademii Nauk.
Magomedova, P & I. Abdulaeva. 2007. Axvaxsko-russkij slovar’ [Akhvakh-Russian
dictionary]. Maxačkala: Dagestanskij Naučnyj Centr Rossiskoj Akademii Nauk.
Magomedova, P & Š. Xalidova. 2001. Karatinsko-russkij slovar’ [Karata-Russian
dictionary]. Maxačkala: Dagestanskij Naučnyj Centr Rossiskoj Akademii Nauk.
Pasquereau, J. 2010. Spatial cases in Karata: spatial and non-spatial uses. First year’s
master thesis. University of Lyon
Pasquereau, J. 2011. Valency in Karata: a preliminary study. Second year’s master thesis.
University of Lyon
Saidova, P. 2006. Godoberinsko-russkij slovar’ [Godoberi-Russian dictionary].
Maxačkala: Dagestanskij Naučnyj Centr Rossiskoj Akademii Nauk.SLE – 44th annual meeting
Logroño, 8-11 September 2011
Emergence and evolution of exceptional valency patterns
in ergative languages: a case study
Jérémy Pasquereau
University of Massachusetts
jepasque@hotmail.com
Denis Creissels
University of Lyon
denis.creissels@univ-lyon2.fr
http://deniscreissels.fr
1. Introduction
This paper analyzes the possible origin and the evolution of a particular type of
exceptional valency pattern found in East Caucasian languages. It deals specifically
with a group of East Caucasian languages, the Andic languages (Andi, Akhvakh,
Bagvalal, Botlikh, Chamalal, Godoberi, Karata, and Tindi), spoken in the Western part
of Daghestan. The Andic languages constitute a genetic unit within the Avar-
Andic(-Tsezic) branch of the East Caucasian language family.
The data we present comes from our own work on two Andic languages, Akhvakh
and Karata – (Creissels 2010), (Pasquereau 2010), (Pasquereau 2011), and from the
consultation of dictionaries and grammars of Andic languages – (Kibrik 1996), (Kibrik
2001), (Magomedova 1999), (Magomedova 2003), (Magomedova 2004),
(Magomedova & Abdulaeva 2007), (Magomedova & Khalidova 2001), (Saidova 2006).
The paper is organized as follows. Section 2 provides basic information about the
grammatical structure of Andic languages. Section 3 provides illustrations of
exceptional case frames in Andic languages. Section 4 examines the question of the
emergence of the exceptional case frames <ERG, LOC> and <ERG, ALL>. Section 5
and 6 concentrate on the particular cases of ‘listen’ and ‘bite’, respectively. Section 7
puts forward some concluding remarks.
2. The basics of Andic morphosyntax
In Andic languages, the syntactic function of noun phrases is expressed by case
suffixes. Verbs agree in gender and number (not in person) but verb agreement is
redundant with case marking, since verbs consistently agree with their nominative
argument, and never express agreement with non-nominative noun phrases.
Constituent order is remarkably flexible and plays no role in the expression of
argument structure.
Andic languages have rich case systems, and in particular, very elaborate spatial
case systems. The nominative (alias absolutive), also used as the quotation form of
nouns, is characterized by a zero ending. The other cases are marked by suffixes
attached either to a stem coinciding with the nominative form, or to a special stem
traditionally called oblique stem (indicated by –ₒ in the glosses of examples). Spatial
case endings consist of two formatives, a topological marker (glossed TPL) and a
directionality marker.
J. Pasquereau & D. Creissels, Emergence and evolution of exceptional valency patterns, p. 2/8
As illustrated by Ex. (1), the coding of core syntactic roles in Andic languages is
consistently ergative:
– S (boʁosē ãdo in (1a)) and P (istaka in (1b)) are in the nominative case (alias
absolutive), contrasting with A (wašode in (1b)) in the ergative case.
– Transitive verbs do not agree with A, but agree with P in the same way as
intransitive verbs do with S: in (1a), the intransitive verb shows human plural
agreement; in (1b), A is masculine singular, whereas P is neuter singular, and the
verb shows neuter singular agreement.
(1) a. b-oʁosē ãd-o atobusī -ɬī-g-e b-eq’-īri. Akhvakh
H⁺-most person-PL bus-Nₒ-TPL-LOC H⁺-come-PF.H⁺
‘Most people came by bus.’
b. wašo-de istaka b-iq’ʷ-aj-ē godi.
boyₒ-ERG glass N-break-CAUS-ADV.N COP.N
‘The boy broke the glass.’
The valency frames of Andic verbs canonically include at least a slot for a nominative
NP (representing in particular the sole argument of monovalent verbs and the P
argument of prototypical action verbs). According to the account of verb valency
included in (Kibrik 2001: 369-376), in Bagvalal, this rule can be viewed as
exceptionless. However, most Andic languages have a restricted set of verbs used in
constructions in which no slot for a nominative NP can be posited.
3. Verbs without nominative arguments
In this section, we provide illustrations of Andic verbs used in case frames that do not
include a slot for a nominative NP. Note that, in the absence of a nominative
argument, the verb forms that express gender-number agreement show default neuter
singular agreement.
3.1. Monovalent verbs
Ex. (2) illustrates the case of a monovalent Akhvakh verb which, depending on the
situation referred to, may occur either in the canonical case frame <NOM>, as in
(2a), or in the exceptional case frame <LOC>, as in (2b).
(2) а. miq̄’i q̄’ʷaraɬ-ēhe godi. Akhvakh
road become_narrow-ADV.N COP.N
‘The road became narrow.’
b. miʕa-q̄-e q̄’ʷaraɬ-ēhe godi.
nose-TPL-LOC become_narrow-ADV.N COP.N
‘My nose is blocked.’ lit. ‘In the nose became narrow.’
J. Pasquereau & D. Creissels, Emergence and evolution of exceptional valency patterns, p. 3/8
Ex. (3) illustrates the Karata verb bibāɬa ‘cry’, used in the exceptional case frame
<ERG>.
(3) gugu-l ɬabc’e bib-ē. Karata
cuckoo-ERG three_times cry-PF
‘The cuckoo sang three times.’
3.2. Bivalent verbs
Ex. (4) illustrates a bivalent Akhvakh verb used in the exceptional case frame <ERG,
GEN>.
(4) hu-sʷ̄ -e daru-ɬī-ʟī ħalbix̄il-āri. Akhvakh
DIST-Mₒ-ERG medecine-Nₒ-GEN try-PF
‘He tried the medecine.’
3.3. Trivalent verbs
Ex. (5) & (6) illustrate a trivalent verb used in the exceptional case frame <ERG, ALL,
GEN>, in Karata and in Akhvakh.
(5) hu-ɬī-ʟī čela žo-ɬī mač-uwa du-g-a de-de Akhvakh
DIST-Nₒ-GEN other day-Nₒ[LOC] tell-POT 2SG-TPL-ALL 1SG-ERG
‘I will tell you about this another day.’
(6) k’̄ ʷāmsʷam hedela-ɬī-ʟ ̄ bas-imisē ! Karata
trivial.N thing-N-GEN tell-PROH
‘Don’t speak about trivial things!’
4. The emergence of the case frames <ERG, ALL> and <ERG, LOC>
In this section, we examine the case of verbs used in the case frames <ERG, ALL> or
<ERG, LOC>. In Andic languages these case frames are typically found with verbs
expressing the following meanings: ‘look at’ – Ex. (7), ‘listen’, ‘bite’, ‘pinch’ – Ex. (8),
‘sting’ – Ex. (9). For examples with ‘listen’ and ‘bite’, see Sections 5 & 6, where the
case of these verbs is examined in more detail.
(7) wašo-de di-g-a eq̄-ari. Akhvakh
boyₒ-ERG 1SG₀-TPL-ALL look_at-PF
‘The boy looked at me.’
(8) o-sʷ̄̌ -i č’un-o di-č’-i. Tindi
DIST-Mₒ-ERG pinch-PF 1SGₒ-TPL-LOC/ALL
‘He pinched me.’
J. Pasquereau & D. Creissels, Emergence and evolution of exceptional valency patterns, p. 4/8
(9) di-g-e ʟ̄’ižʷali-de č’in-āri Akhvakh
1SGₒ-TPL-LOC waspₒ-ERG sting-PF
‘A wasp stung me.’
Godoberi also has a verb obadi ‘kiss’ with the case frame <ERG, LOC>, but all the
other Andic languages for which we have data express this meaning by means of a
noun oba or ob̄a ‘kiss’ combined with a verb ‘do’, and consequently the exceptional
case frame of Godoberi obadi ‘kiss’ must probably be viewed as the result of the
univerbation of a do-compound.
The other cases are less easy to explain, but the variations observed in the
expression of the other meaning listed above in Andic languages support a hypothesis
already suggested by Charachidzé (1981) for Avar, according to which these
exceptional valency frames may result from the reduction of the regular frames <ERG,
NOM, ALL> or <ERG, NOM, LOC>, characteristic of verbs expressing meanings or
the type ‘X applies/holds Y on Z’ (and found in particular with the verbs expressing
‘hit’, with the hittee in the locative/allative and the instrument in the nominative).
The point is that not all Andic languages have bivalent verbs with meanings such as
‘look at’, ‘listen’, ‘bite’, ‘sting’, ‘pinch’. Among the Andic languages, it is also common
to find such meanings expressed by means of less specific verbs combined with three
NPs. For example, Tindi expresses ‘sting’ as eq̄ʷa k̄ʷēɬʲa, lit. ‘hit the sting (on someone)’,
and ‘bite’ as saldi bix̄ʲiɬʲa, lit. ‘hold the teeth (on someone)’, with respectively eq̄ʷa
‘sting (noun)’ and saldi ‘teeth’ occupying the nominative slot.
Starting from the hypothesis of valency frames originally including three slots, at
least two reduction scenarios can be imagined: conventionalization of the ellipsis of
the nominative argument, or fusion of the nominative argument with the verb. In
Andic languages, there is evidence for the conventionalization-of-ellipsis scenario in
the case of other verbs with exceptional case frames, but not for those examined here.
By contrast, some of them at least show evidence for the fusion scenario. The clearest
cases are those of ‘listen’ and ‘bite’, examined in Sections 5 and 6.
5. Emergence and evolution of verbs ‘listen’ in Andic languages
Three situations are found among Andic languages with respect to the expression of
‘listen’:
– Some Andic languages express ‘listen’ by means of a construction involving a
nominative noun phrase with the meaning ‘ear’ in addition to those encoding the
two participants. For example, Godoberi expresses ‘listen’ as hãt’uk’ja rik̄i, literally
‘fix the ear (on someone/something)’ – Ex. (10). Formally, this construction is an
instance of the regular valency pattern <ERG, NOM, ALL> with hãt’uk’ja ‘ear’
filling the nominative slot.
– Others have a verb ‘listen’ with the exceptional case frame <ERG, ALL>: Tindi
anix̄ʲiɬʲa – Ex. (11), Chamalal woɬuk’la;
– A verb ‘listen’ with the regular case frame <NOM, ALL> is found in three Andic
languages: Akhvakh hãdax̄uruʟa – Ex. (12), Karata ãdukaɬa – Ex. (13) , Bagvalal
aštila.
J. Pasquereau & D. Creissels, Emergence and evolution of exceptional valency patterns, p. 5/8
(10) wašu-di imu-q̄-i hãt’uk’ʲa r-ikk-i r-ukk-ida. Godoberi
sonₒ-ERG fatherₒ-TPL-LOC/ALL ear N⁺-hold-INF N⁺-must-IPF
‘The son must listen to his father.’
(11) di-q-̄ a anix̄ʲ-ā hik’ʲi o-sʷ̄̌ -ī Tindi
1SGₒ-TPL-LOC/ALL listen-IPF NEG DIST-Mₒ-ERG
‘He does not listen to me.’
(12) waša imo-g-a hãdax̄-ari. Akhvakh
boy fatherₒ-TPL-ALL listen-PF
‘The boy listened to his father.’
(13) waša imo-χa-r ãduk-e. Karata
boy fatherₒ-TPL- ALL listen-PF
‘The boy listened to his father.’
Interestingly, ‘fix the ear on’ is the obvious etymology of Akhvakh hãdax̄uruʟa
(compare with hãde ‘ear’, bix̄uruʟa ‘fix’), in spite of the fact that the nominative case
assigned to the NP representing the listener does not correspond to what could be
expected from this etymology.
The variation in the expression of ‘listen’ in Andic languages provides therefore
evidence supporting the reconstruction of the following evolution:
– at a first stage, the coalescence of a trivalent verb occurring in the regular frame
<ERG, NOM, ALL> with a noun occupying the nominative slot creates a bivalent
verb with the exceptional case frame <ERG, ALL>;
– at a second stage, attested by Akhvakh and Karata, the exceptional valency pattern
resulting from this evolution may be regularized into <NOM, ALL>.
6. Variations in the case frame of ‘bite’ in Andic languages
Among Andic languages, variations similar to those observed in the expression of
‘listen’ are attested for the expression of ‘bite’ too, with however a different case frame
in the languages in which the case frame of ‘bite’ has been regularized:
– Some Andic languages express ‘bite’ by means of a construction involving a noun
phrase with the meaning ‘tooth’ in addition to those encoding the two participants:
as mentioned above, Tindi expresses ‘bite’ as saldi bix̄ʲiɬʲa, lit. ‘hold the teeth (on
someone)’.
– Others have a verb ‘bite’ with the exceptional case frame <ERG, ALL>: Karata
q̄’ʷaraɬa – Ex. (14), Akhvakh q̄’eleč’uruʟa – Ex. (15), Chamalal q̄’āna – Ex. (16);
– A verb ‘bite’ with the regular case frame <ERG, NOM> is found in two Andic
languages: Godoberi q’̄ ami – Ex. (17),1 Bagvalal salīla – Ex. (18).
1 Note however that the two available sources on Godoberi give contradictory indications about this
verb. Saidova (2006) provides several examples of its use, all with the case frame <ERG, NOM>,
whereas Kibrik & al. mention it as a verb used in the case frame <ERG, LOC>, but give no example.
J. Pasquereau & D. Creissels, Emergence and evolution of exceptional valency patterns, p. 6/8
(14) χʷaj-ol q̄’ʷar-e di-č’o. Karata
dog-ERG bite-PF 1SG₀-TPL[LOC]
‘The dog bit me.’
(15) χʷe-de di-g-e q̄’eleč’-ari. Akhvakh
dog-ERG 1SGₒ-TPL-LOC bite-PF
‘The dog bit me.’
(16) χʷāj-d q’̄ ān-nida o-sū -č’. Chamalal
dog-ERG bite-PF DIST-Mₒ-TPL[LOC]
‘The dog bit me.’
(17) χʷaji-di q̄’am-i ho-w. Godoberi
dog-ERG bite-PF₁ DIST-M
‘The dog bit him.’
(18) ʕamo-r salli di-b lela Bagvalal
donkeyₒ-ERG bite.PF₁ 1SGₒ[GEN]-N hand
‘The donkey bit my hand.’
A first interesting observation is that Godoberi q̄’ami and Chamalal q̄’āna ‘bite’ do not
have the same construction but come from a common root that can be reconstructed as
*q̄’am. Consequently, the difference observed in their case frames cannot be attributed
to their etymology, and a change must have occurred in the construction of one of
them.
A second interesting observation is that Bagvalal salīla ‘bite’ is quite obviously
cognate with Bagvalal & Chamalal salʷ, Karata sale, Tindi salu, Andi sol ‘tooth’. This
suggests that this verb results from the univerbation of a ‘noun + verb’ compound
similar to Tindi saldi bix̄ʲiɬʲa, in spite of the fact that the nominative case assigned to
the NP representing the bitee does not correspond to what could be expected from this
etymology.
The variation in the expression of ‘bite’ in Andic languages provides therefore
evidence supporting the reconstruction of the following evolution:
– at a first stage, the coalescence of a trivalent verb occurring in the regular frame
<ERG, NOM, ALL> with a noun occupying the nominative slot creates a bivalent
verb with the exceptional case frame <ERG, LOC>;
– at a second stage, attested by Bagvalal, the exceptional valency pattern resulting
from this evolution may be regularized into <ERG, NOM>.
The following observations can also be made about the verbs expressing ‘bite’ in Andic
languages:
– Several Andic languages express ‘eat’ by means of verbs that are reflexes of a root
*q̄’am: Akhvakh q̄’ōnuʟa ‘eat’, Karata q̄’amaɬa ‘eat’, Bagvalal q̄’anila ‘eat’. As
mentioned above, this root also has reflexes expressing ‘bite’ in Godoberi and
Chamalal. Interestingly, the irregular case frame observed with the reflexes of this
J. Pasquereau & D. Creissels, Emergence and evolution of exceptional valency patterns, p. 7/8
root expressing ‘bite’ is not found with the reflexes expressing ‘eat’, which always
occur in the case frame <ERG, NOM> characteristic of verbs encoding
prototypical transitive events.
– As illustrated by Ex. (15) above, the Akhvakh verb q̄’eleč’uruʟa ‘bite’ is used in the
case frame <ERG, LOC> to encode prototypical biting events that may cause pain
but do not result in a change of state of the bitee (for example, ‘The dog bit me’).
However, when referring to biting events that affect the physical integrity of the
bitee (bite off a piece of something and eat it), q̄’eleč’uruʟa is also attested with the
case frame <ERG, NOM> characteristic of the expression of prototypical
transitive events, as in Ex. (19).
(19) wašo-de ʕeče q̄’eleč’-ari.
boyₒ-ERG apple bite-PF
‘The boy bit off a chunk of apple.’
7. Conclusion
Andic languages are among the languages making a wide use of lexicalized
combinations ‘noun + verb’ in which the noun most commonly occupies the same
syntactic slot as the patient in prototypical transitive predication. Since Andic
languages are consistent ergative languages, the nominal element of such compounds
is in the nominative case and governs verb agreement. Consequently, the univerbation
of ‘noun + verb’ compounds in Andic languages yields verbs whose construction
includes no slot for a nominative NP governing verb agreement, creating thus potential
exceptions to the rule according to which, in Andic languages, the valency frames of
verbs must minimally include a nominative term controlling gender-number
agreement of the verb.2
The first conclusion that can be drawn from the data presented above is that, in the
evolution of languages, exceptional valency patterns that emerge as the accidental
result of lexicalization processes (in the case examined here, the univerbation of
lexicalized ‘noun + verb’ combinations in ergative languages) may subsequently
undergo a regularization process by means of a change in the encoding of one of the
arguments: in the case of ‘listen’, an argument originally encoded as an ergative NP
takes nominative marking, and in the case of ‘bite’, an argument originally encoded as
a locative NP takes nominative marking.
Moreover, the difference observed between ‘listen’ and ‘bite’ suggests that semantic
factors condition this regularization process. In the case of ‘bite’, the substitution of
nominative marking for locative marking results in the case frame <ERG, NOM>,
which in Andic languages is used to encode prototypical transitive events involving an
agent and a patient. By contrast, in the case of ‘listen’, regularization results in the case
frame <NOM, ALL>, typically used to encode movement towards a goal, and the
same is observed with ‘look at’. A plausible explanation is that the ergative argument
of ‘bite’ has more affinities with the prototypical agent than the ergative argument of
‘listen’ or ‘look at’, and the locative argument of ‘bite’ has more affinities with the
prototypical patient than the allative argument of ‘listen’ or ‘look at’. The ergative
2 See (Haspelmath 1993: 178–180) for a description of this process in Lezgi.
J. Pasquereau & D. Creissels, Emergence and evolution of exceptional valency patterns, p. 8/8
argument of ‘listen’ or ‘look at’ is clearly not a typical agent, whereas the allative
argument has clear affinities with the goal of movement. In other words, in the
regularization process, arguments whose initial case marking has clear semantic
motivations tend to maintain their case marking, whereas case marking is more likely
to be modified for arguments whose semantic roles can be viewed as relatively
ambiguous as regards their affinities with the prototypes underlying the use of cases.
This hypothesis is supported by the variation observed in the case frame of Akhvakh
q̄eleč’uruʟa ‘bite’, and by the fact that verbs glossed ‘eat’ cognate with verbs glossed
‘bite’ uniformly have the case frame <ERG, NOM>, since in the biting events of the
type illustrated by Ex. (19) as well as in eating events, in contrast with prototypical
biting events, the second participant shows a degree of affectedness similar to that of
prototypical patients.
Abbreviations
—ₒ : oblique stem / ADV: adverbial / ALL: allative / CAUS: causative / COP: copula /
DIST: distal demonstrative / ERG: ergative / F: feminine / FUT: future / GEN: genitive
/ H+: human plural / IMP: imperative / INF: infinitive / IPF: imperfective / LOC:
locative / M: masculine / nH+: non-human plural / N: neuter / NEG: negation / PF:
perfective / PL: plural / POT: potential / PROH: prohibitive / SG: singular / TPL:
topological marker
References
Charachidzé, G. 1981. Grammaire de la langue avar. Paris: Editions Jean-Favard.
Creissels, D. 2009. ‘Valency properties of Northern Akhvakh verbs.’
http://www.deniscreissels.fr/public/Creissels-valency_classes_project_Akhvakh.pdf
Haspelmath, Martin. 1993. A grammar of Lezgian. Berlin & New York: Mouton de Gruyter.
Kibrik, Alexandr (ed.). 1996. Godoberi. München: Lincom Europa.
Kibrik, Alexandr (ed.). 2001. Bagvalinskij jazyk: grammatika, teksty, slovari. Moskva: Nasledie.
Korjakov, Ju. B. 2006. Atlas of Caucasian languages. Moscow: Piligrim
Magomedova, P. 1999. Čamalinsko-russkij slovar’ [Chamalal-Russian dictionary].
Maxačkala: Dagestanskij Naučnyj Centr Rossiskoj Akademii Nauk.
Magomedova, P. 2003. Tindinsko-russkij slovar’ [Tindi-Russian dictionary]. Maxačkala:
Dagestanskij Naučnyj Centr Rossiskoj Akademii Nauk.
Magomedova, P. 2004. Bagvalinsko-russkij slovar’ [Bagvalal-Russian dictionary].
Maxačkala: Dagestanskij Naučnyj Centr Rossiskoj Akademii Nauk.
Magomedova, P & I. Abdulaeva. 2007. Axvaxsko-russkij slovar’ [Akhvakh-Russian
dictionary]. Maxačkala: Dagestanskij Naučnyj Centr Rossiskoj Akademii Nauk.
Magomedova, P & Š. Xalidova. 2001. Karatinsko-russkij slovar’ [Karata-Russian
dictionary]. Maxačkala: Dagestanskij Naučnyj Centr Rossiskoj Akademii Nauk.
Pasquereau, J. 2010. Spatial cases in Karata: spatial and non-spatial uses. First year’s
master thesis. University of Lyon
Pasquereau, J. 2011. Valency in Karata: a preliminary study. Second year’s master thesis.
University of Lyon
Saidova, P. 2006. Godoberinsko-russkij slovar’ [Godoberi-Russian dictionary].
Maxačkala: Dagestanskij Naučnyj Centr Rossiskoj Akademii Nauk.SLE – 44th annual meeting
Logroño, 8-11 September 2011
Emergence and evolution of exceptional valency patterns
in ergative languages: a case study
Jérémy Pasquereau
University of Massachusetts
jepasque@hotmail.com
Denis Creissels
University of Lyon
denis.creissels@univ-lyon2.fr
http://deniscreissels.fr
1. Introduction
This paper analyzes the possible origin and the evolution of a particular type of
exceptional valency pattern found in East Caucasian languages. It deals specifically
with a group of East Caucasian languages, the Andic languages (Andi, Akhvakh,
Bagvalal, Botlikh, Chamalal, Godoberi, Karata, and Tindi), spoken in the Western part
of Daghestan. The Andic languages constitute a genetic unit within the Avar-
Andic(-Tsezic) branch of the East Caucasian language family.
The data we present comes from our own work on two Andic languages, Akhvakh
and Karata – (Creissels 2010), (Pasquereau 2010), (Pasquereau 2011), and from the
consultation of dictionaries and grammars of Andic languages – (Kibrik 1996), (Kibrik
2001), (Magomedova 1999), (Magomedova 2003), (Magomedova 2004),
(Magomedova & Abdulaeva 2007), (Magomedova & Khalidova 2001), (Saidova 2006).
The paper is organized as follows. Section 2 provides basic information about the
grammatical structure of Andic languages. Section 3 provides illustrations of
exceptional case frames in Andic languages. Section 4 examines the question of the
emergence of the exceptional case frames <ERG, LOC> and <ERG, ALL>. Section 5
and 6 concentrate on the particular cases of ‘listen’ and ‘bite’, respectively. Section 7
puts forward some concluding remarks.
2. The basics of Andic morphosyntax
In Andic languages, the syntactic function of noun phrases is expressed by case
suffixes. Verbs agree in gender and number (not in person) but verb agreement is
redundant with case marking, since verbs consistently agree with their nominative
argument, and never express agreement with non-nominative noun phrases.
Constituent order is remarkably flexible and plays no role in the expression of
argument structure.
Andic languages have rich case systems, and in particular, very elaborate spatial
case systems. The nominative (alias absolutive), also used as the quotation form of
nouns, is characterized by a zero ending. The other cases are marked by suffixes
attached either to a stem coinciding with the nominative form, or to a special stem
traditionally called oblique stem (indicated by –ₒ in the glosses of examples). Spatial
case endings consist of two formatives, a topological marker (glossed TPL) and a
directionality marker.
J. Pasquereau & D. Creissels, Emergence and evolution of exceptional valency patterns, p. 2/8
As illustrated by Ex. (1), the coding of core syntactic roles in Andic languages is
consistently ergative:
– S (boʁosē ãdo in (1a)) and P (istaka in (1b)) are in the nominative case (alias
absolutive), contrasting with A (wašode in (1b)) in the ergative case.
– Transitive verbs do not agree with A, but agree with P in the same way as
intransitive verbs do with S: in (1a), the intransitive verb shows human plural
agreement; in (1b), A is masculine singular, whereas P is neuter singular, and the
verb shows neuter singular agreement.
(1) a. b-oʁosē ãd-o atobusī -ɬī-g-e b-eq’-īri. Akhvakh
H⁺-most person-PL bus-Nₒ-TPL-LOC H⁺-come-PF.H⁺
‘Most people came by bus.’
b. wašo-de istaka b-iq’ʷ-aj-ē godi.
boyₒ-ERG glass N-break-CAUS-ADV.N COP.N
‘The boy broke the glass.’
The valency frames of Andic verbs canonically include at least a slot for a nominative
NP (representing in particular the sole argument of monovalent verbs and the P
argument of prototypical action verbs). According to the account of verb valency
included in (Kibrik 2001: 369-376), in Bagvalal, this rule can be viewed as
exceptionless. However, most Andic languages have a restricted set of verbs used in
constructions in which no slot for a nominative NP can be posited.
3. Verbs without nominative arguments
In this section, we provide illustrations of Andic verbs used in case frames that do not
include a slot for a nominative NP. Note that, in the absence of a nominative
argument, the verb forms that express gender-number agreement show default neuter
singular agreement.
3.1. Monovalent verbs
Ex. (2) illustrates the case of a monovalent Akhvakh verb which, depending on the
situation referred to, may occur either in the canonical case frame <NOM>, as in
(2a), or in the exceptional case frame <LOC>, as in (2b).
(2) а. miq̄’i q̄’ʷaraɬ-ēhe godi. Akhvakh
road become_narrow-ADV.N COP.N
‘The road became narrow.’
b. miʕa-q̄-e q̄’ʷaraɬ-ēhe godi.
nose-TPL-LOC become_narrow-ADV.N COP.N
‘My nose is blocked.’ lit. ‘In the nose became narrow.’
J. Pasquereau & D. Creissels, Emergence and evolution of exceptional valency patterns, p. 3/8
Ex. (3) illustrates the Karata verb bibāɬa ‘cry’, used in the exceptional case frame
<ERG>.
(3) gugu-l ɬabc’e bib-ē. Karata
cuckoo-ERG three_times cry-PF
‘The cuckoo sang three times.’
3.2. Bivalent verbs
Ex. (4) illustrates a bivalent Akhvakh verb used in the exceptional case frame <ERG,
GEN>.
(4) hu-sʷ̄ -e daru-ɬī-ʟī ħalbix̄il-āri. Akhvakh
DIST-Mₒ-ERG medecine-Nₒ-GEN try-PF
‘He tried the medecine.’
3.3. Trivalent verbs
Ex. (5) & (6) illustrate a trivalent verb used in the exceptional case frame <ERG, ALL,
GEN>, in Karata and in Akhvakh.
(5) hu-ɬī-ʟī čela žo-ɬī mač-uwa du-g-a de-de Akhvakh
DIST-Nₒ-GEN other day-Nₒ[LOC] tell-POT 2SG-TPL-ALL 1SG-ERG
‘I will tell you about this another day.’
(6) k’̄ ʷāmsʷam hedela-ɬī-ʟ ̄ bas-imisē ! Karata
trivial.N thing-N-GEN tell-PROH
‘Don’t speak about trivial things!’
4. The emergence of the case frames <ERG, ALL> and <ERG, LOC>
In this section, we examine the case of verbs used in the case frames <ERG, ALL> or
<ERG, LOC>. In Andic languages these case frames are typically found with verbs
expressing the following meanings: ‘look at’ – Ex. (7), ‘listen’, ‘bite’, ‘pinch’ – Ex. (8),
‘sting’ – Ex. (9). For examples with ‘listen’ and ‘bite’, see Sections 5 & 6, where the
case of these verbs is examined in more detail.
(7) wašo-de di-g-a eq̄-ari. Akhvakh
boyₒ-ERG 1SG₀-TPL-ALL look_at-PF
‘The boy looked at me.’
(8) o-sʷ̄̌ -i č’un-o di-č’-i. Tindi
DIST-Mₒ-ERG pinch-PF 1SGₒ-TPL-LOC/ALL
‘He pinched me.’
J. Pasquereau & D. Creissels, Emergence and evolution of exceptional valency patterns, p. 4/8
(9) di-g-e ʟ̄’ižʷali-de č’in-āri Akhvakh
1SGₒ-TPL-LOC waspₒ-ERG sting-PF
‘A wasp stung me.’
Godoberi also has a verb obadi ‘kiss’ with the case frame <ERG, LOC>, but all the
other Andic languages for which we have data express this meaning by means of a
noun oba or ob̄a ‘kiss’ combined with a verb ‘do’, and consequently the exceptional
case frame of Godoberi obadi ‘kiss’ must probably be viewed as the result of the
univerbation of a do-compound.
The other cases are less easy to explain, but the variations observed in the
expression of the other meaning listed above in Andic languages support a hypothesis
already suggested by Charachidzé (1981) for Avar, according to which these
exceptional valency frames may result from the reduction of the regular frames <ERG,
NOM, ALL> or <ERG, NOM, LOC>, characteristic of verbs expressing meanings or
the type ‘X applies/holds Y on Z’ (and found in particular with the verbs expressing
‘hit’, with the hittee in the locative/allative and the instrument in the nominative).
The point is that not all Andic languages have bivalent verbs with meanings such as
‘look at’, ‘listen’, ‘bite’, ‘sting’, ‘pinch’. Among the Andic languages, it is also common
to find such meanings expressed by means of less specific verbs combined with three
NPs. For example, Tindi expresses ‘sting’ as eq̄ʷa k̄ʷēɬʲa, lit. ‘hit the sting (on someone)’,
and ‘bite’ as saldi bix̄ʲiɬʲa, lit. ‘hold the teeth (on someone)’, with respectively eq̄ʷa
‘sting (noun)’ and saldi ‘teeth’ occupying the nominative slot.
Starting from the hypothesis of valency frames originally including three slots, at
least two reduction scenarios can be imagined: conventionalization of the ellipsis of
the nominative argument, or fusion of the nominative argument with the verb. In
Andic languages, there is evidence for the conventionalization-of-ellipsis scenario in
the case of other verbs with exceptional case frames, but not for those examined here.
By contrast, some of them at least show evidence for the fusion scenario. The clearest
cases are those of ‘listen’ and ‘bite’, examined in Sections 5 and 6.
5. Emergence and evolution of verbs ‘listen’ in Andic languages
Three situations are found among Andic languages with respect to the expression of
‘listen’:
– Some Andic languages express ‘listen’ by means of a construction involving a
nominative noun phrase with the meaning ‘ear’ in addition to those encoding the
two participants. For example, Godoberi expresses ‘listen’ as hãt’uk’ja rik̄i, literally
‘fix the ear (on someone/something)’ – Ex. (10). Formally, this construction is an
instance of the regular valency pattern <ERG, NOM, ALL> with hãt’uk’ja ‘ear’
filling the nominative slot.
– Others have a verb ‘listen’ with the exceptional case frame <ERG, ALL>: Tindi
anix̄ʲiɬʲa – Ex. (11), Chamalal woɬuk’la;
– A verb ‘listen’ with the regular case frame <NOM, ALL> is found in three Andic
languages: Akhvakh hãdax̄uruʟa – Ex. (12), Karata ãdukaɬa – Ex. (13) , Bagvalal
aštila.
J. Pasquereau & D. Creissels, Emergence and evolution of exceptional valency patterns, p. 5/8
(10) wašu-di imu-q̄-i hãt’uk’ʲa r-ikk-i r-ukk-ida. Godoberi
sonₒ-ERG fatherₒ-TPL-LOC/ALL ear N⁺-hold-INF N⁺-must-IPF
‘The son must listen to his father.’
(11) di-q-̄ a anix̄ʲ-ā hik’ʲi o-sʷ̄̌ -ī Tindi
1SGₒ-TPL-LOC/ALL listen-IPF NEG DIST-Mₒ-ERG
‘He does not listen to me.’
(12) waša imo-g-a hãdax̄-ari. Akhvakh
boy fatherₒ-TPL-ALL listen-PF
‘The boy listened to his father.’
(13) waša imo-χa-r ãduk-e. Karata
boy fatherₒ-TPL- ALL listen-PF
‘The boy listened to his father.’
Interestingly, ‘fix the ear on’ is the obvious etymology of Akhvakh hãdax̄uruʟa
(compare with hãde ‘ear’, bix̄uruʟa ‘fix’), in spite of the fact that the nominative case
assigned to the NP representing the listener does not correspond to what could be
expected from this etymology.
The variation in the expression of ‘listen’ in Andic languages provides therefore
evidence supporting the reconstruction of the following evolution:
– at a first stage, the coalescence of a trivalent verb occurring in the regular frame
<ERG, NOM, ALL> with a noun occupying the nominative slot creates a bivalent
verb with the exceptional case frame <ERG, ALL>;
– at a second stage, attested by Akhvakh and Karata, the exceptional valency pattern
resulting from this evolution may be regularized into <NOM, ALL>.
6. Variations in the case frame of ‘bite’ in Andic languages
Among Andic languages, variations similar to those observed in the expression of
‘listen’ are attested for the expression of ‘bite’ too, with however a different case frame
in the languages in which the case frame of ‘bite’ has been regularized:
– Some Andic languages express ‘bite’ by means of a construction involving a noun
phrase with the meaning ‘tooth’ in addition to those encoding the two participants:
as mentioned above, Tindi expresses ‘bite’ as saldi bix̄ʲiɬʲa, lit. ‘hold the teeth (on
someone)’.
– Others have a verb ‘bite’ with the exceptional case frame <ERG, ALL>: Karata
q̄’ʷaraɬa – Ex. (14), Akhvakh q̄’eleč’uruʟa – Ex. (15), Chamalal q̄’āna – Ex. (16);
– A verb ‘bite’ with the regular case frame <ERG, NOM> is found in two Andic
languages: Godoberi q’̄ ami – Ex. (17),1 Bagvalal salīla – Ex. (18).
1 Note however that the two available sources on Godoberi give contradictory indications about this
verb. Saidova (2006) provides several examples of its use, all with the case frame <ERG, NOM>,
whereas Kibrik & al. mention it as a verb used in the case frame <ERG, LOC>, but give no example.
J. Pasquereau & D. Creissels, Emergence and evolution of exceptional valency patterns, p. 6/8
(14) χʷaj-ol q̄’ʷar-e di-č’o. Karata
dog-ERG bite-PF 1SG₀-TPL[LOC]
‘The dog bit me.’
(15) χʷe-de di-g-e q̄’eleč’-ari. Akhvakh
dog-ERG 1SGₒ-TPL-LOC bite-PF
‘The dog bit me.’
(16) χʷāj-d q’̄ ān-nida o-sū -č’. Chamalal
dog-ERG bite-PF DIST-Mₒ-TPL[LOC]
‘The dog bit me.’
(17) χʷaji-di q̄’am-i ho-w. Godoberi
dog-ERG bite-PF₁ DIST-M
‘The dog bit him.’
(18) ʕamo-r salli di-b lela Bagvalal
donkeyₒ-ERG bite.PF₁ 1SGₒ[GEN]-N hand
‘The donkey bit my hand.’
A first interesting observation is that Godoberi q̄’ami and Chamalal q̄’āna ‘bite’ do not
have the same construction but come from a common root that can be reconstructed as
*q̄’am. Consequently, the difference observed in their case frames cannot be attributed
to their etymology, and a change must have occurred in the construction of one of
them.
A second interesting observation is that Bagvalal salīla ‘bite’ is quite obviously
cognate with Bagvalal & Chamalal salʷ, Karata sale, Tindi salu, Andi sol ‘tooth’. This
suggests that this verb results from the univerbation of a ‘noun + verb’ compound
similar to Tindi saldi bix̄ʲiɬʲa, in spite of the fact that the nominative case assigned to
the NP representing the bitee does not correspond to what could be expected from this
etymology.
The variation in the expression of ‘bite’ in Andic languages provides therefore
evidence supporting the reconstruction of the following evolution:
– at a first stage, the coalescence of a trivalent verb occurring in the regular frame
<ERG, NOM, ALL> with a noun occupying the nominative slot creates a bivalent
verb with the exceptional case frame <ERG, LOC>;
– at a second stage, attested by Bagvalal, the exceptional valency pattern resulting
from this evolution may be regularized into <ERG, NOM>.
The following observations can also be made about the verbs expressing ‘bite’ in Andic
languages:
– Several Andic languages express ‘eat’ by means of verbs that are reflexes of a root
*q̄’am: Akhvakh q̄’ōnuʟa ‘eat’, Karata q̄’amaɬa ‘eat’, Bagvalal q̄’anila ‘eat’. As
mentioned above, this root also has reflexes expressing ‘bite’ in Godoberi and
Chamalal. Interestingly, the irregular case frame observed with the reflexes of this
J. Pasquereau & D. Creissels, Emergence and evolution of exceptional valency patterns, p. 7/8
root expressing ‘bite’ is not found with the reflexes expressing ‘eat’, which always
occur in the case frame <ERG, NOM> characteristic of verbs encoding
prototypical transitive events.
– As illustrated by Ex. (15) above, the Akhvakh verb q̄’eleč’uruʟa ‘bite’ is used in the
case frame <ERG, LOC> to encode prototypical biting events that may cause pain
but do not result in a change of state of the bitee (for example, ‘The dog bit me’).
However, when referring to biting events that affect the physical integrity of the
bitee (bite off a piece of something and eat it), q̄’eleč’uruʟa is also attested with the
case frame <ERG, NOM> characteristic of the expression of prototypical
transitive events, as in Ex. (19).
(19) wašo-de ʕeče q̄’eleč’-ari.
boyₒ-ERG apple bite-PF
‘The boy bit off a chunk of apple.’
7. Conclusion
Andic languages are among the languages making a wide use of lexicalized
combinations ‘noun + verb’ in which the noun most commonly occupies the same
syntactic slot as the patient in prototypical transitive predication. Since Andic
languages are consistent ergative languages, the nominal element of such compounds
is in the nominative case and governs verb agreement. Consequently, the univerbation
of ‘noun + verb’ compounds in Andic languages yields verbs whose construction
includes no slot for a nominative NP governing verb agreement, creating thus potential
exceptions to the rule according to which, in Andic languages, the valency frames of
verbs must minimally include a nominative term controlling gender-number
agreement of the verb.2
The first conclusion that can be drawn from the data presented above is that, in the
evolution of languages, exceptional valency patterns that emerge as the accidental
result of lexicalization processes (in the case examined here, the univerbation of
lexicalized ‘noun + verb’ combinations in ergative languages) may subsequently
undergo a regularization process by means of a change in the encoding of one of the
arguments: in the case of ‘listen’, an argument originally encoded as an ergative NP
takes nominative marking, and in the case of ‘bite’, an argument originally encoded as
a locative NP takes nominative marking.
Moreover, the difference observed between ‘listen’ and ‘bite’ suggests that semantic
factors condition this regularization process. In the case of ‘bite’, the substitution of
nominative marking for locative marking results in the case frame <ERG, NOM>,
which in Andic languages is used to encode prototypical transitive events involving an
agent and a patient. By contrast, in the case of ‘listen’, regularization results in the case
frame <NOM, ALL>, typically used to encode movement towards a goal, and the
same is observed with ‘look at’. A plausible explanation is that the ergative argument
of ‘bite’ has more affinities with the prototypical agent than the ergative argument of
‘listen’ or ‘look at’, and the locative argument of ‘bite’ has more affinities with the
prototypical patient than the allative argument of ‘listen’ or ‘look at’. The ergative
2 See (Haspelmath 1993: 178–180) for a description of this process in Lezgi.
J. Pasquereau & D. Creissels, Emergence and evolution of exceptional valency patterns, p. 8/8
argument of ‘listen’ or ‘look at’ is clearly not a typical agent, whereas the allative
argument has clear affinities with the goal of movement. In other words, in the
regularization process, arguments whose initial case marking has clear semantic
motivations tend to maintain their case marking, whereas case marking is more likely
to be modified for arguments whose semantic roles can be viewed as relatively
ambiguous as regards their affinities with the prototypes underlying the use of cases.
This hypothesis is supported by the variation observed in the case frame of Akhvakh
q̄eleč’uruʟa ‘bite’, and by the fact that verbs glossed ‘eat’ cognate with verbs glossed
‘bite’ uniformly have the case frame <ERG, NOM>, since in the biting events of the
type illustrated by Ex. (19) as well as in eating events, in contrast with prototypical
biting events, the second participant shows a degree of affectedness similar to that of
prototypical patients.
Abbreviations
—ₒ : oblique stem / ADV: adverbial / ALL: allative / CAUS: causative / COP: copula /
DIST: distal demonstrative / ERG: ergative / F: feminine / FUT: future / GEN: genitive
/ H+: human plural / IMP: imperative / INF: infinitive / IPF: imperfective / LOC:
locative / M: masculine / nH+: non-human plural / N: neuter / NEG: negation / PF:
perfective / PL: plural / POT: potential / PROH: prohibitive / SG: singular / TPL:
topological marker
References
Charachidzé, G. 1981. Grammaire de la langue avar. Paris: Editions Jean-Favard.
Creissels, D. 2009. ‘Valency properties of Northern Akhvakh verbs.’
http://www.deniscreissels.fr/public/Creissels-valency_classes_project_Akhvakh.pdf
Haspelmath, Martin. 1993. A grammar of Lezgian. Berlin & New York: Mouton de Gruyter.
Kibrik, Alexandr (ed.). 1996. Godoberi. München: Lincom Europa.
Kibrik, Alexandr (ed.). 2001. Bagvalinskij jazyk: grammatika, teksty, slovari. Moskva: Nasledie.
Korjakov, Ju. B. 2006. Atlas of Caucasian languages. Moscow: Piligrim
Magomedova, P. 1999. Čamalinsko-russkij slovar’ [Chamalal-Russian dictionary].
Maxačkala: Dagestanskij Naučnyj Centr Rossiskoj Akademii Nauk.
Magomedova, P. 2003. Tindinsko-russkij slovar’ [Tindi-Russian dictionary]. Maxačkala:
Dagestanskij Naučnyj Centr Rossiskoj Akademii Nauk.
Magomedova, P. 2004. Bagvalinsko-russkij slovar’ [Bagvalal-Russian dictionary].
Maxačkala: Dagestanskij Naučnyj Centr Rossiskoj Akademii Nauk.
Magomedova, P & I. Abdulaeva. 2007. Axvaxsko-russkij slovar’ [Akhvakh-Russian
dictionary]. Maxačkala: Dagestanskij Naučnyj Centr Rossiskoj Akademii Nauk.
Magomedova, P & Š. Xalidova. 2001. Karatinsko-russkij slovar’ [Karata-Russian
dictionary]. Maxačkala: Dagestanskij Naučnyj Centr Rossiskoj Akademii Nauk.
Pasquereau, J. 2010. Spatial cases in Karata: spatial and non-spatial uses. First year’s
master thesis. University of Lyon
Pasquereau, J. 2011. Valency in Karata: a preliminary study. Second year’s master thesis.
University of Lyon
Saidova, P. 2006. Godoberinsko-russkij slovar’ [Godoberi-Russian dictionary].
Maxačkala: Dagestanskij Naučnyj Centr Rossiskoj Akademii Nauk.
Библиографическая ссылка
Jérémy Pasquereau, Denis Creissels EMERGENCE AND EVOLUTION OF EXCEPTIONAL VA-LENCY PATTERNS IN ERGATIVE LANGUAGES: A CASE STUDY // . – . – № ;
kontensivtyp.csrae.ru/ru/0-27 (дата обращения:
15.05.2024).
Код для вставки на сайт или в блог
Просмотры статьи
Сегодня: 438 | За неделю: 438 | Всего: 438
Комментарии (0)