home edit page issue tracker

UD for Pashto

For the principles of transliteration of Pashto used in UD see the Transliteration page.

Tokenization and Word Segmentation

• The words are delimited by whitespaces and punctuation.
• Multiword tokens are used in these cases:
  • separable light verb verbs in forms, when the two parts are joined, e.g. بندوم bandawë́m “I close” → کوم band + بند kawë́m
  • contractions of pronouns and adpositions, e.g. پرې pre “on it/him/her/them” → پر یې për ye (see ADP for details)

Morphology

Lemmatization

• Direct (nominative) singular masculine (if applicable) form is used as lemma for nominals.
• Infinitive (in the direct case) is usually used as lemma for verbs.
  • The existential word شته šta “there is / there are”, tagged as VERB, has only one form, so it is used as the lemma.

Tags

• The overview of part-of-speech tags used in Pashto can be found here.
• Pashto uses all 17 universal tags.
• Several words are tagged as PART:
  • Negative particles نه në and مه ma “no, not” and the affirmative particle هو ho “yes”.
  • Modal particles: باید bấyad (necessity “must, have to, should”) and دې de (order or desire “let, should”).
• These verbs may be tagged as AUX:
  • verb ول wël “to be”, used as copula and auxiliary verb for perfect tenses
  • verb کېدل kedë́l “to become”, used for passive voice or potential forms
  • verb کول kawë́l “to do”, used for longer potential forms
  • word بۀ bë, used for future tense, habitual past forms and in conditional sentences
  • word ونۀ wënë́ joining a perfective prefix (in affirmative forms connected to the verb) with a negation particle, this may hold also for some other prefixes.
• Modal verbs, such as غوښتل ġox̌të́l “to want to”, are tagged as VERB (besides that, some modal meanings are expressed using modal particles or using the mentioned potential verb forms). Light verbs are tagged also as VERB, with the nominal part depending on them with compound:lvc relation.
• Pronouns that depend on nouns and behave similarly like their attributes (some of them even agree with the nouns in number, case and gender) are tagged as DET; possessive pronouns are mostly treated this way. Pronouns used individually (often as arguments of a verb) are tagged as PRON. This includes e.g. relative or non-possessive personal pronouns. Various interrogative, demonstrative or indefinite pronouns can be tagged both ways depending on the situation. Enclitic weak pronouns, used as unstressed core arguments or as alternative possessive pronouns, are always tagged as PRON, even when marking possession, because they do not have the attributive relation to the noun, e.g. they follow the noun, while all other pronouns tagged as DET precede it. Directional pronouns, merged with several prepositions, are separated from them with a PRON tag.
• The deverbal forms like infinitive or participles (sometimes behaving like verbal noun and verbal adjectives) are usually tagged as VERB. Only nouns and adjectives originally derived from infinitives or participles, but now perceived clearly as nouns and adjectives, are tagged as NOUN and ADJ.
• Adjectives and adverbs derived from adjectives have often the same form. Their tagging as ADJ or ADV depends on the context.

Features

• The overview of all features used in Pashto can be found here.
• There are three VerbForm values used in Pashto: finite Fin, infinitive Inf and participle Part.
• An important feature of Pashto verbs is Aspect, which strictly divides verb forms to imperfect Imp and perfecgt Perf.
• The finite verb forms inflect for Mood feature with indicative Ind, imperative Imp, subjunctive Sub and potential Pot values.
• The finite verb forms conjugate for Tense feature taking present Pres or past Past mark, with an auxiliary with marking the future Fut.
• The finite verb forms inflect also for the Person feature with the common three values, which is also an inherent feature of many personal pronouns.
• Generally all inlfectional parts of speech inflect for Number taking a singular Sing or a plural Plur value. There is a few collective nouns having the collective Coll number. Infinitives always behave like plural, so they do not have the number tagged. Non-past finite verb forms do not have the number feature in the third person, since the forms for both numbers are always identical.
• Nominals, participles and infinitives inflect for Case feature. There are five cases tagged in UD for Pashto: direct (marked as nominative) Nom, oblique (marked as accusative) Acc, locative Loc, ablative Abl and vocative Voc.
• Nouns and some pronouns have inherent Gender feature with two possible values: masculine Masc and feminine Fem. Adjectives, other pronouns and participles inflect for the gender in order to agree with nouns. Finite verb forms inflect for the gender only in the past forms in the third person (both singular and plural).
• Other important features of Pashto include PronType, Animacy, Deixis or Variant.

Syntax

• The overview of all dependency relations used in Pashto can be found here.

Core Arguments

• Core arguments (subjects and objects) in Pashto are mostly nouns, pronouns or infinitives in either bare direct case Nom or bare oblique case Acc. The exact use of these cases depends on the inherent transitivity of the verb and the voice and tense used (language phenomenon called split ergativity occurring also in other Indo-Iranian langages)
• The only argument (i.e. the subject) of intransitive verbs or of transitive verbs used in the passive voice are always in the direct case Nom.
• For transitive verbs in the active voice holds:
  • The subject in non-past tenses is always in the direct case Nom.
  • The subject in past tenses is always in the oblique case Acc.
  • The object in all tenses is almost always in the direct case Nom.
    • The only exceptions are personal pronouns for the first and the second person singular in non-past tenses, where oblique forms ما mâ “me”, تا tâ “you” are used instead of the direct رۀ zë “I”, تۀ të “you”.
  • The subject usually comes before the object regardless of the case.
  • The verb agrees in Person and Number (and also Gender in the third person of past tenses) with the subject in non-past tenses and with the object in past tenses.

	intransitive	transitive
	nsubj	nsubj	obj
non-past	direct	direct	direct *
past	direct	oblique	direct

• exceptions: ما mâ, تا tâ (see above)

Non-verbal Clauses

• The copula verb ول wël “to be” is used in most non-verbal clauses.
• The nominal part of the predicate is usually in the direct case Nom
• In the existential clauses the word شته šta “there is / there are” is used, but it is tagged VERB

Relations Overview

• The following relation subtypes are used in Pashto:

  Auxiliaries (see aux):

  • aux:pass for passive voice
  • aux:pot for potential mood
  • aux:perf for perfect tenses (do not confuse with the perfective aspect)
  • aux:fut for future tense
  • aux:hab for habitual past tense
  • aux:cnd for conditional mood

  Orphan constructions (see orphan):

  • orphan:nsubjobj for orphan objects dependent on orphan subjects
  • orphan:nsubjobl for orphan obliques dependent on orphan subjects
  • orphan:objobl for orphan obliques dependent on orphan objects

  Other:

  • nsubj:pass for nominal subjects of passive
  • compound:lvc for nominal part of light verb construction

Treebanks

There is currently one Pashto UD treebank:

• Pashto-Sikaram

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

.