Classical Armenian Valency Lexicon

CAVaL provides access to sentences from a treebank of Classical Armenian, which is annotated accoding to the guidelines of Universal Dependencies (UD)

Grammatical glosses, converted from the UD tagset to the Leipzig Glossing Rules, are given for each word of output sentences.

Syntactic relations of dependencies

The verbal frames are construed with nominals tagged by the following syntactic relations:

nsubj (nominal subject)
nsubj:caus (nominal subject of the causative construction)
nsubj:pass (nominal subject of the passive construction)
csubj (clausal subject))
csubj:caus (clausal subject of the causative construction)
obj (object)
ccomp (clausal complement)
iobj (indirect object)
obl (oblique nominal)
obl:agent (an agent of the passive construction)
obl:arg (1. an agent expressed by the genitive case in the periphrastic perfect tenses; 2. a recipient-like participant expressed by the cՙ= + accusative)

Note that Classical Armenian is a pro-drop language and the subject dependency is often omitted.

Note also that the UD annotation does not distinguish between arguments and adjuncts. CAVaL does not impose any additional specification of oblique dependencies except for the enhanced tags obl:agent and obl:arg. This allows users to explore to which extent the dependencies fulfil the requirements of an argument without constraints imposed by the annotation.

Predicates and dependencies connected by a conj (conjunct) or appos (appositional modifier) relations are considered as separate dependencies of a verb.

Morphological encoding of dependencies

The verbal dependencies are encoded by seven cases and their combinations with a wide array of adpositions, see the table below.

	Nominative	Accusative	Locative	Ablative	Dative	Instrumental	Genetive
Bare cases
	nsubj nsubj:caus nsubj:pass	obj	obl		iobj	obl obl:agent	obl:arg
Adpositional constructions
z=		obj	obl	obl		obl
cʽ=		obl obl:arg
aṙ		obl	obl			obl
i		obl	obl obl:agent
ǝnd		obl	obl	obl		obl	obl
ǝst				obl	obl
handerj						obl
aṙaǰi							obl
etc.

The combinations of adpositions with cases have different degree of grammaticalization. Thus, for example, z= functions as a definite article in conjunction with the accusative case and almost completely lost its prepositional semantics. Yet, z= is annotated as an adposition for the sake of consistency with expressions where z= is used in combination with other cases.

Adpositional phrases are predominantly built with prepositions (e.g. aṙ + acc. ‘towards’) and only rarely with postpositions (e.g. instr. + handerj ‘together with’). The contrast between pre- and postpositions is indicated by the relative order to an adposition and a case form in the selector of the “Encoding” filter.

A dependency can be encoded with the help of a complex adposition, e.g. ի վերայ / i very ‘above’. According to the UD annotation, subsequent words of a complex adposition are attached to the first one with the fixed relation. Complex prepositions can be occasionally disrupted by the head nominal, in which case they are annotated as circumpositions, both parts of which are linked to the head by the case relation, cf. յ=որոյ վերայ քաղաքն նոցա շինեալ էր / y=oroy veray kʻałakʻn nocʻa šineal ēr “on which their city had been built” (Lk. 4:29). A special case of complex adpositions are combinations of prepositional construction with demonstrative local adverbs that reinforce the meaning of the preceding construction, cf. էջ ի լեռնէ անտի / ēǰ i leṙnē anti “he came down from the mountain” (Mt. 8:1). When used in this function, local adverbs are linked to the modified nominal by the case relation, and function as constituents of a circumposition. Again, the position of constituents in a circumpositional construction is reflected in relative order of adpositions and a case form in the selector of the “Encoding” filter.