Morphological categories

Various different systems of morpho-syntactic categories are used for Estonian. There are two reasons for this: first, a linguist may prefer a certain grammatical tradition, which in turn may be tied to the traditions of concrete "model" languages, and which may change over time; and second, the language used in a corpus may force one to modify the system of categories: to add new ones or to give up theoretical distinctions which are hard to maintain in practice.

The following systems will be considered below:

• FS
  • starting point: Viks, Ülle (1992). "Väike vormisõnastik I: Sissejuhatus & grammatika." Tallinn: Eesti Teaduste Akadeemia Keele ja Kirjanduse Instituut.
  • software: github.com/Filosoft/vabamorf/
  • corpus: www.sketchengine.eu/ettenten-estonian-corpus/
• CG
  • starting point:
    • conventions for representation: MULTEXT-East
    • level of detail for (sub)categories: "Eesti keele grammatika II. Süntaks." Peatoim. Mati Erelt, toim. Tiiu Erelt, Henn Saari, Ülle Viks. Tallinn: Eesti Teaduste Akadeemia Keele ja Kirjanduse Instituut (1993). dspace.ut.ee
  • software: github.com/Filosoft/vabamorf + github.com/EstSyntax/EstCG
  • corpus: github.com/EstSyntax/EDT
• giellalt
  • starting point:
    • conventions for representation: github.com/giellalt/
    • verb categories: Kaalep, Heiki-Jaan (2015) "Eesti verbi vormistik." Keel ja Kirjandus 1, lk. 1–15 pdf fail
    • categories for declinables: Kaalep, Heiki-Jaan (2009) "Kuidas kirjeldada lühikest sisseütlevat kasutusandmetega kooskõlas?" Keel ja Kirjandus, 6, 411 - 425 pdf fail
  • software: github.com/giellalt/lang-est-x-utee
  • corpus: --
• GT
  • starting point:
    • parts of speech: FS
    • other categories: giellalt
  • software: github.com/Filosoft/vabamorf
  • corpus: www.keeleveeb.ee
• UD 2.10
  • starting point: universaldependencies.org/guidelines.html
  • software: neural models trained on a corpus Stanza, UDPipe, EstSpacy
  • corpus: github.com/EstSyntax/EstUD

Square brackets in the CG and UD columns denote category values that are in free variation.

In addition to morphosyntactic categories, UD has also categories for other phenomena:

• Typo=Yes
• Foreign=Yes
• In the presence of NumType=, the representational form of a number may be given as NumForm=[Word, Digit, Roman]
• Words mitte, ega and ei have Polarity=Neg in addition to part-of-speech.

Parts of speech

Parts of speech and morphological/lexical features are formally independent in UD. For example, if a word is part of a named entity, then its part-of-speech tag might be PROPN, while other features remain unchanged, e.g. Teine (the Second) might be PROPN NumForm=Word NumType=Ord

¹ Genitive attribute is a proper noun in the genitive case that is written in lowercase, because it has generalized into representing a type or kind, e.g. dooria (sammas).

² Verb particle is a non-verbal constituent of a multi-word verb, having no part-of-speech tag in "Väike vormisõnastik", e.g. luuslanki (lööma)

³ Symbol is a single non-alpha-numerical character or character sequence that acts as a syntactic constituent in a sentence.

⁴ Emoticons are handled inconsistently by different taggers and in corpora.

⁵ A word may be unanalyzable because of foreign language, sequence of symbols, typo etc.

Punctuation in corpora

Corpus tagging usually involves separating punctuation marks from the words and giving them analyses of their own. Usually punctuation marks do not get individual special category names, but CG singles out the following ones:

Declinables

Grammatical categories

Paradigm

Verbs

Grammatical categories

In UD, Mood= always entails VerbForm=Fin

This system of categories does not extend to complex forms of the predicate like olen elanud, olin elanud, oleksin elanud, ei olnud elanud, ei olnuks elanud, olgu elatud etc., if they were taken as single units. All the individual words are covered, though.

Paradigm

The following table presents the full paradigm for verbs, i.e. all the possible bundles of grammatical category values that are representable as wordforms. (The forms of ära and pole are listed for the completeness of the set of possible combinations.)

Numbers denote homonymous slots of the paradigm.

NB! In UD, the present participles are not part of the verb paradigm; they are ADJ with some features specific to verbs only.

Asterisk (*) denotes combinations of categories which are represented by wordforms that are neither reckognized nor generated by tools that follow the GT or giellalt system.

---

Webmaster    Last modified: November 25 2024 14:01:05.