home edit page issue tracker

UD for Western Armenian

Tokenization and Word Segmentation

• In general, words are delimited by whitespace. Description of exceptions follows.
• According to typographical rules, punctuation marks written next to a neighbouring word are normally tokenized as separate syntactic words.
• A hyphenated compound is split into separate tokens, as in առասպելա-բանաստեղծական/aṙaspela-banasteġçakan “fabulo-poetic”, or into five tokens (three words and two hyphens), as in ռուսա-իրանա-սուրիական/ṙowsa-irana-sowriakan “Russian-Iranian-Syrian”. Compounds written without a hyphen, such as պաղեստինեւիսրայէլեան/paġestinewisrayēlean “Israeli-Palestinian” are not split.
• Inflectional bound morphemes attached to quoted names, phrases, numerals, or abbreviations with or without an hyphen are segmented as separate tokens. Thus «Ցեղին սիրտը»էն “from ‘The Heart of the Tribe’” is segmented as { «, Ցեղին, սիրտը, », էն }, 2020 թ.-ին “in the year 2020” as { 2020, թ, ., -, ին} and 1968էն “from 1968” as {1968, էն}.
• Numerical expressions with derivational endings, as well as adjectives and other forms containing digits, are split into separate tokens, e.g. 2-րդ/2-rd “2nd”, 44-օրեայ/44-oreay “44-day”, or 85-ամեայ “85 years old”. The preceding lexical token is the head, and the segmented inflectional element is attached to it with the dep relation.
• Time expressions and dates such as 19:45, 20.05.2000, and 20/05/2000 are split at the punctuation marks.
• Symbols occurring before or after numerical expressions are tokenized separately, e.g. $ + 250 and 4,81 + %.
• Decimal numbers such as 2.1 and 2,1 are single tokens.
• A whitespace separating digits in a large number is not treated as a word separator. For example, 1 000 000 (“1,000,000” by English rules) is one token. (However, such tokens do not occur in Western Armenian treebanks as of release 2.8.)
• There are several classes of “orthographic words” (sequences of letters without spaces inside) that are split into several syntactic tokens:
• Words containing infixed Armenian punctuation are represented as multi-word tokens. For example, ինչպէ՞ս is segmented as ինչպէս + ՞ “how?”.
• Orthographic words consisting of the indicative auxiliary կ՚/k՚ or the negative auxiliary չ՚/č՚՚ followed by a vowel-initial verb are also represented as multi-word tokens, e.g. կ՚այցելեմ = կ՚ + այցելեմ/k՚aycʼelem “(I am) visiting” and չ՚ուզեր = չի + ուզեր/č՚owzer “(He/She) does not want”.
• Email addresses, URLs, and social media handles are treated as single tokens.
• For more details, see tokenization.

Morphology

Tags

This is an overview only. For more detailed discussion and examples, see the list of Western Armenian POS tags and Western Armenian features.

• Western Armenian uses all 17 universal POS categories, including particles (PART).
• The tag DET is used for articles and pronominal words that modify a noun phrase, including possessive, demonstrative, interrogative, relative, indefinite, total, and negative determiners (the traditional grammar does not define determiners, but distinguishes pronominal modifiers). Pronominal quantifiers (which the traditional grammar includes in pronouns) such as քանի մը/kʼani më “several” and բոլոր/bolor “all” are DET as well.
• Independent possessive forms such as իմս/ims “mine”, քուկդ/kʼowkd “yours”, անորը/anorë “his/hers” etc., are personal pronouns with Poss=Yes; possessive determiners such as իմ/im “my”, քու/kʼow “your”, անոր/anor “his/her”, մեր/mer “our”, ձեր/jer “your”, and անոնց/anoncʼ “their” are DET.
• The tag PRON is reserved for pronouns functioning as the head of a noun phrase. Thus աս/as “this” is PRON in Աս ասանկ չ՚ըլլար/As asank č՚ëllar “This won’t work like this”, but DET when it modifies a noun, as in աս տեղէն կ՚սպասէին ճանբորդել/as teġēn k՚spasēin č̣anbordel “They were expecting to set off from here”.
• The main Western Armenian auxiliaries include:
• forms of եմ/em “be”, used as the copula and in compound tenses;
• կը/kë and its variants կ՚/k՚, կու/kow, which form the present and imperfect indicative with a finite lexical verb;
• պիտի/piti, which marks progressive aspect;
• ըլլամ/ëllam (ըլլալ/ěllal) “be”, ունիմ/ownim “have”, and կամ/kam “exist” in secondary compound constructions;
• postverbal կոր/kor, which occurs after the indicative form consisting of կը/kë and a finite (subjunctive) verb and marks progressive aspect, indicating that the situation is ongoing at the reference time;

• տալ/tal “cause / make someone perform an action”, used as the auxiliary of periphrastic causatives.

• The aorist and the present and imperfect subjunctive forms of ըլլալ/ěllal (եղաւ/eġaw, ըլլայ/ëllay, ըլլար/ëllar) are analyzed as lexical VERB when they occur with nominal predicates, and as AUX only when they combine with participles. In compound forms containing both _եմ_/_em_ and _ըլլալ_/_ěllal_ with a nominal predicate, _եմ_/_em_ is AUX` and ըլլալ/ěllal is the lexical VERB.

• Western Armenian distinguishes three values of VerbForm:
  • Infinitive Inf, tagged VERB or AUX.
  • Finite verb Fin, tagged VERB or AUX.
  • Participle Part, tagged VERB or AUX.

Case-marked infinitives retain VerbForm=Inf even when they have an adverbial, converb-like function.

• Participles used in compound tenses or adjectivally are normally tagged VERB; their syntactic function is expressed by the dependency relation.

Nominal Features

• Nominal words (NOUN, PROPN, and relevant PRON forms) are annotated with Animacy, mainly Hum or Nhum.
• The main values of Number are Sing and Plur. The lexical values Coll and Assoc are used where appropriate.
• Case has seven values in the Western Armenian documentation: Nom, Gen, Dat, Acc, Abl, Ins, and Loc.
  • Nouns do not morphologically distinguish nominative and accusative; direct objects normally have nominative form. Acc is used where a pronominal paradigm makes the distinction relevant.
  • For nouns, the genitive-dative distinction is closely connected with definiteness. Possessive personal pronouns distinguish the genitive separately.
  • Case is also an inflectional feature of the closed class of postpositional localizers (ADP), e.g. forms based on մէջ/mēǰ, վրայ/vray, տակ/tak, մօտ/mòt, ետեւ/etew, and առջեւ/aṙǰew.
• Definite has the values Def and Ind and is used with nominal categories and nominalized verb forms where appropriate.
• Degree applies to adjectives (ADJ) and some adverbs (ADV). Its values include Pos, Cmp, Sup, and Abs.

Verbal Features

• Subcat distinguishes intransitive (Intr) and transitive (Tran) verbs.
• Aspect has the values Dur, Imp, Iter, Perf, Prog, and Prosp.
  • The auxiliary կը/kë is annotated with Aspect=Imp; պիտի/piti is annotated with Aspect=Prosp; postverbal կոր/kor marks Aspect=Prog.
  • Participles and other verb forms receive the aspect value appropriate to their morphological form.
• Finite verbs have one of three Mood values: Ind, Imp, or Sub.
  • The present and imperfect indicative are formed analytically with կը/kë (կ՚/k՚, կու/kow) plus a present or imperfect subjunctive form of the lexical verb.

Prospective forms are built with պիտի or պէտք է plus a subjunctive form.

Tense has the values Pres, Imp, and Past. The present and imperfect values used with subjunctive forms are formal morphological labels and may have future or hypothetical interpretations in context.

Voice distinguishes Act, Mid, Pass, Rcp, Cau, and CauPass.

Synthetic causatives in -ցն- are annotated Voice=Cau when they retain a transparent causative analysis. Lexicalized transitive verbs derived from middle verbs are treated as active.

In periphrastic causatives, only the auxiliary տալ receives Voice=Cau; the infinitive retains its own voice value.

CauPass is reserved for passive forms of causative verbs.

Polarity

Pronouns, Determiners, Quantifiers

Other Features

• The following universal features are not used in Armenian: Gender, Evident.

Syntax

Core Arguments, Oblique Arguments and Adjuncts

Non-verbal Clauses

Relations Overview

Treebanks

There is just one Western Armenian UD treebank.

• Western_Armenian-ArmTDP

---

UD for Armenian (Eastern)

Morphology

Tags

• Eastern Armenian has one auxiliary verb (AUX), եմ (“to be”), but the lemma լինել is also possible. This is in fact just an aspectual variant of եմ, but this is a separate lemma because the morphological process that relates it to եմ is considered derivational. There is another auxiliary տալ (“cause/make someone to perform action”) for periphrastic causatives. Auxiliaries are all verbal in Eastern Armenian and can be grouped into three types:
  • The copula with non-verbal predicates.
  • Periphrastic present tense (present form of եմ + resultative participle of the main verb).
  • Periphrastic past tense (present form of եմ + perfect of the main verb; imperfect form of եմ + imperfective, perfect, future-I and resultative participles of the main verb).
  • Periphrastic future tense (present form of եմ + future-I participle of the main verb).
  • Periphrastic negated conditional (negated present or imperfect form of եմ + connegative form of the main verb).
  • Periphrastic “secondary compound tenses” (any form of լինել, including periphrastic forms, + processual, resultative and future-I participles of the main verb).
  • Periphrastic causative (any form of տալ, including periphrastic forms, + infinitve of the main verb).
• In other words, եմ, լինել and տալ are the only lemmas that occur with the AUX tag (Exception: finite existentials կամ and ունեմ in combination with resultative participle). Note, that եմ and լինել may also occur as normal VERB if they are used in purely existential sentences (i.e. they don’t indicate location because if they do, then եմ and լինել will be treated as copula).
• Verbs with modal meaning are not considered to be auxiliaries in Armenian.
• There are five main (de)verbal forms, distinguished by the UPOS tag and the value of the VerbForm feature:
  • Infinitive Inf, tagged VERB or AUX.
  • Finite verb Fin, tagged VERB or AUX.
  • Participle Part, tagged VERB or AUX (in so-called “secondary compound tenses”).
  • Converb Conv, tagged VERB or AUX.
  • Gerundive Gdv, tagged VERB or AUX.
  • Verbal noun Vnoun, tagged NOUN.
• Though the resultative, subject and future-I participles can be used adjectivally and can be negated they are generally tagged VERB. The only exception is future-II participles, they are tagged ADJ.

Nominal Features

• Nominal words (NOUN, PROPN and some of PRON) have an inherent Animacy feature with one of three values: Hum, Nhum. Note that this may be changed in future. Nominal words could be treated as having inherent Hum, Nhum and further would be subclassified by the layered Animacy[gram] values Anim vs. Inan.
• The two main values of the Number feature are Sing and Plur. The following parts of speech inflect for number: NOUN, PROPN, PRON, VERB, AUX (finite).
  • Selected nouns are plurale tantum (Ptan) or singulare tantum (Coll). These two values are lexical, and cannot be used with the agreeing verbs. They also never occur with pronouns. Coll occurs with gerundives.
  • There is a language-specific value Assoc (associative plural). This is also lexical and occurs with NOUN and PROPN. Some of pronouns (մերոնք, ձերոնք, իմոնք, քոնոնք) are also associative.
• Case has 6 possible values: Nom, Gen, Dat, Abl, Ins, Loc. It occurs with the nominal words, i.e., NOUN, PROPN, PRON, DET, and gerundives. Note, that Gen occurs only with pronouns and determiners.
  • The Case feature also occurs with some of adpositions, subclassified as “localizers” (ADP). It is an inflectional feature here.
• The two main values of the Definite feature are Def and Ind. The following parts of speech inflect for definitness: NOUN, PROPN, PRON. With gerundives, resultative and subject participles the feature sometimes encodes the lexical person of the possessor, although they can be almost interpreted as the 3rd person. We mark them as Def (see the layered features below).
• Degree applies to adjectives (ADJ) and some adverbs (ADV) and has one of four possible values: Pos, Cmp, Sup, Abs.

Verbal Features

• Verbs have a lexical Subcat, either intransitive (Intr) or transitive (Tran).
• Verbs have one of six values of Aspect: Dur, Imp, Iter, Perf, Prog or Prosp.
  • Note, that in Armenian iterative is considered as a lexical feature of verbs, thus they have morphologically related not iterative counterparts, but it is not a regular system and the two verbs are represented by different lemmas. We mark them as biaspectual.
  • The Aspect feature should be also used with the corresponding deverbatives, if they have the VerbForm feature.
• Finite verbs always have one of five values of Mood: Cnd, Imp, Ind, Nec of Sub.
  • The necessitative mood is only used with necessitative particle պիտի or the impersonal predicative պետք է. The subjuncitve finite form of the main verb, that is needed to form a periphrastic necessitative, is not marked with this feature.
  • The negated conditional mood is only used with indicative auxiliaries (եմ, էի). The connegative of the main verb, that is needed to form a periphrastic negated conditional, is also marked with this feature.
• Verbs in indicative mood always have one of three values of Tense: mainly Past and for auxilaries and some content verbs Pres or Imp.
  • Imperative and necessitative forms do not have the Tense feature (note that imperfect and present necessitatives are distinguished analytically).
  • Subjunctve and conditional forms have one of the Tense=Imp or Tense=Pres features, which are formally imperfect or present, but semantically future.
  • There are five values of the Voice feature: Act, Cau, Mid, Pass, Rcp. Active and causative verbs have Subcate=Tran, the other three Subcate=Intr.

Polarity

• Polarity has two values, Pos and Neg, and applies primarily to verbs (VERB, AUX) that can be negated using the bound morpheme չ-.
  • Occasionally ոչ occurs as an independent negation particle (PART) and is marked with Polarity=Neg. The Polarity=Neg feature is also used with necessitative mood particles (չ)պիտի, (չ)պետք է .
  • Negating nouns are usually limited to those derived from verbs (չունեցողի, չգրվածները).
  • The Polarity feature is not used with pronouns and determiners, although there is a subset of negative pronouns and determiners. The PronType=Neg feature is used there instead.
  • The Polarity=Neg feature is not used with negated conditional mood, the Connegative=Yes feature is used there instead.

Pronouns, Determiners, Quantifiers

• PronType is used with pronouns (PRON), determiners (DET) and adverbs (ADV).
• NumType is used with numerals (NUM), adjectives (ADJ) and adverbs (ADV).
• The Poss feature marks possessive personal determiners (e.g. իմ “my”), possessive interrogative, relative determiners (e.g. ում, որի “whose”), possessive nouns (e.g. հայրիկինը “father’s”) and possessive adjectives (e.g. հայոց “armenian, armenians’, refer to armenians”).
• The Reflex feature marks reflexive pronouns (ինձ, քեզ, իրեն, մեզ, ձեզ, իրենց) and determiners (իր, իրենց). In Armenian it is always used together with PronType=Emp or PronType=Prs.
• Person is a lexical feature of personal pronouns (PRON) and has three values, 1, 2 and 3. With personal possessive determiners (DET), the feature actually encodes the person of the possessor. Person is not marked on other types of pronouns and on nouns, although they can almost always be interpreted as the 3rd person.
  • As a cross-reference to subject, person is also marked on finite verbs (VERB, AUX).
• The Polite feature distinguishes informal second-person pronouns (դու, դուք, Polite=Infm) from the formal Դուք (Polite=Form). The formal pronoun is phonologically equivalent in all its case forms to the second-person plural դուք but it is distinguished in orthography by the capital letter Դ. We tag it as second person (because that is its meaning) and we tag also its number (it is used for singular addressees) despite the fact that it combines with second-person plural verbs. The parser must learn that Number=Sing|Person=2|Polite=Form subject attaches to Number=Plur|Person=2 verbs, while Number=Sing|Person=2|Polite=Infm subject attaches to Number=Sing|Person=2 verbs.
• There are two layered features, Person[psor] and Number[psor]. They appear with nouns, gerundives, certain determiners and adpositions and encode the lexical person(deixis)/number of the possessor. The extra layer is needed to distinguish these lexical features from the inflectional person and number that mark agreement with the modified (possessed) noun.

Other Features

• Besides the layered features listed above, there are several other language-specific features:
  • NumForm
  • NameType
  • AdpType
  • ConjType
  • Hyph
  • Style
  • Deixis
  • Echo
  • Typo
• The following universal features are not used in Armenian: Gender, Evident.

Syntax

This is an overview only. For more detailed discussion and examples, see the list of Armenian relations, as well as Armenian-specific examples scattered across the documentation of constructions.

Core Arguments, Oblique Arguments and Adjuncts

• Nominal subject (nsubj) is a noun phrase in the nominative case, without preposition.
  • If the noun is quantified in partitive meaning, it may be in the ablative: Հնչեցին Կոմիտասի երգերից “Sounded from songs of Komitas.”
  • An infinitive verb may serve as the subject and is labeled as clausal subject, csubj. On the other hand, verbal nouns or gerundives as subjects are just nsubj.
  • A finite subordinate clause may serve as the subject and is labeled csubj.
• Objects defined in the Armenain grammar may be bare noun phrases in nominative or dative (as direct objects or “voice objects”) and in dative, ablative, instrumental or locative, or adpositional phrases mainly in dative (as indirect / “objects of nature” or adpositional objects). For the purpose of UD the objects are divided to core objects, labeled obj or iobj. Oblique objects are labeled obl.
  • Bare nominative and dative objects are considered core. Verbs that subcategorize for a single dative object are also considered core. Example: մոտենում էր քաղաքին “He was approaching the city.”
  • All adpositional objects are considered oblique.
  • Nominative objects of some verbs alternate with finite clausal complements, which are labeled ccomp.
  • If a verb subcategorizes for the infinitive (e.g. modal verbs or verbs of control), the infinitival complement is labeled xcomp.
  • If a verb subcategorizes for two core objects, one of them nominative (or ccomp) and the other non-nominative (mainly dative), then the non-nominative object is labeled iobj. Core nominal objects in other situations are labeled just obj.
• Adjuncts are usually adpositional phrases, but they can be bare noun phrases as well (following the Armenian grammar, adverbial modifiers are realized as noun phrases). They are labeled obl:
  • Temporal or locational modifiers realized as noun phrases: կեսգիշերին եկավ “He came at midnight.”
  • Dative noun phrases with benefactive or possessive role (i.e. if the verb does not subcategorize for a single dative object and if it is not a verb of giving (or similar), where the dative could be interpreted as the recipient. Example: նրան սուրճ եփեց “he made (for) him coffee.”
  • Instrumental or directional noun phrases expressing the way or means with which something was done or direction from some point. Example: հաճույքով լսում էր “He was listening with pleasure.”
  • All adpositional phrases (i.e., their role and form is not defined lexically by the predicate) are adjuncts.
• In passive clauses (both reflexive and reciprocal), the subject is labeled with nsubj:pass or csubj:pass, respectively.
  • The auxiliary verb in periphrastic passive is labeled aux.
  • If the demoted agent is present, it has the form of a bare instrumental or adpositional phrase and its relation is labeled obl:agent.
• In causative clauses (both bare and periphrastic causative), the subject is labeled with nsubj:caus.
  • The auxiliary verb in periphrastic causative is labeled aux:caus.
  • The demoted agent of the action (if present) has the form of a bare dative and is labeled [iobj:agent].

Non-verbal Clauses

• The copula verb եմ (be) is used in equational, attributional, locative, possessive and benefactory nonverbal clauses.
  • Purely existential clauses (without indicating location) normally use different lemmas, լինել or կամ, and they are treated as the head of the clause and tagged VERB.

Relations Overview

• The following relation subtypes are used in Armenian:
  • acl:relcl for relative clauses
  • advmod:emph for adverbs or particles that modify noun phrases and emphasize or negate them
  • aux:caus for causative auxiliaries
  • aux:ех for existentials as auxiliary
  • case:loc for postpositional localizers
  • compound:lvc for light compound verbs
  • compound:redup for reduplicated compounds
  • compound:svc for serial compound verbs
  • csubj:pass for clausal subjects of passive verbs
  • det:poss for possessive determiners
  • iobj:agent for agentive indirect objects of causative verbs
  • nmod:npmod for noun phrases
  • nmod:poss for possessive modifiers
  • nsubj:caus for nominal subjects of causative verbs
  • nsubj:pass for nominal subjects of passive verbs
  • obl:agent for agents of passive verbs
• The following relation types are not used in Armenian at all: clf

Treebanks

There is just one Eastern Armenian UD treebank.

• Armenian-ArmTDP

© 2014–2024 Universal Dependencies contributors. Site powered by Annodoc and brat

.