Characteristics of the Turkish language in AX Semantics

Fundamentals

Turkish has no genders for nouns. There are two numbers: singular and plural. Additionally, Turkish has six cases for nouns.

grammatical name	values	examples
number	singular	eski araba (the old car)
	plural	eski arabalar (the old cars)
cases (noun)	nominative	ev (the house)
	accusative	Evi boyuyorum. (I paint the house.)
	dative	Eve yeni bir renk kattım. (I put new color on the house.)
	genitive	evin rengi (the house's color)
	ablative	Bir evden diğerine yürüdüm. (I walked from the house to another.)
	locative	Evde kalıyorum. (I stay at house/home.)
adjectives (noun)	before noun	kırmızı ev (red house)
verb tenses	present (imperfective)	o satın alıyor (he is buying)
	present (aorist)	o satın alır (he buys)
	past	o satın aldı (he bought)

The standard order of a noun phrase in Turkish is the following:

determiner + numeral + adjective + noun + preposition

Note that in the below example, the noun (kitap) stays in singular form because it is proceeded by a cardinal numeral:

bu             üç     popüler  kitap         hakkında
these[pl,dem]  three  popular  book[sg,nom]  about
DET            NUM    ADJ      NOUN          PREP
"about these three popular books"

Lexicon

Nouns

Turkish nouns are inflected for number and case. Lexicon entries for nouns may also be necessary for inflecting determiners, adjectives and pronouns correctly. They are omitted, if a lexicon entry is required, but missing.

Examples

The basic lexicon entry for ev (house) contains:

• inflection table for case and number:

	Singular	Plural
Nominative	ev	evler
Accusative	evi	evleri
Dative	eve	evlere
Genitive	evin	evlerin
Ablative	evden	evlerden
Locative	evde	evlerde

Note

If you need lexicon entries for countries, write to the support about that and you will get them for Turkish with automatic handling of prepositions.

Adjectives

Turkish adjectives do not inflect, therefore no lexicon entries for them need to be created. For adjective position, the default is "before noun".

Verbs

Turkish verbs inflect for person, number, and tense. The most common verbs are encoded in our software. If a verb inflects incorrectly, you should add it to the lexicon.

Container settings

Determiner

The AX NLG platform supports the following determiners for Turkish: indefinite, demonstrative, possessive, and quantifier (every).

Possessive determiners are expressed by adding certain possessive suffixes to the possessor (noun), which are put between the plural suffix and the case suffix. The AX NLG platform will automatically inflect nouns with a possessive suffix, when a possessive determiner is chosen. Take ev (house) for example:

evim [possessive 1st sg] -> "my house"
evl-er-im-i [pl, possessive 1st sg, acc] -> "my houses"

Additionally, when using the indefinite determiner "bir", the word order will change. The AX NLG platform changes the word order automatically, for example:

bu güzel günde [det adj noun] -> "this nice day"
güzel bir gün [adj det noun] -> "a nice day"

Numerals

The noun will automatically agree with the numeral number when a numeral variable is used. Four types of numerals are possible on the AX NLG platform: cardinal, cardinal as digit, ordinal, and ordinal as digit.

As the example below shows, note that nouns (e.g, gün) stay in singular form even when the cardinal numeral is more than 1:

	cardinal	ordinal
text	dokuz gün (nine days)	dokuzuncu gün (the ninth day)
digit	9 gün (9 days)	9. gün (the 9th day)

For Turkish, both cardinal and ordinal numerals are written out until 100 on the platform, otherwise (above 100) the output is in digit form. Take cardinal numerals for example:

yüz araba
(one hundred cars)
vs.
101 araba
(101 cars)

Preposition/Postposition

Turkish rather uses postpositions than prepositions. Therefore, the AX NLG platform automatically puts prepositions behind the noun in a container for this language. Prepositions function as postpositions despite being called prepositions in the container, since they are analogous to prepositions in English (both are adpositions).

araba ile [noun postposition] -> "with the car"

Vowel harmony

The AX platform applies vowel harmonies automatically. Vowel harmony means that many inflection suffixes have different variants, that are used depending on what kind of vowels a word contains.

For Turkish vowel harmony depends on whether the last vowel of the word‘s stem or the preceding suffix is a front or back vowel and whether it is rounded or unrounded.

Take Turkish noun pluralization, for example, where the suffix can be either -ler or -lar: if the last vowel of the stem is an undotted/back vowel (a, ı, o, u ), the plural suffix -lar is used. Vice-versa, if the last vowel of the stem is a dotted/front vowel (e, i, ö, ü), -ler is used as the plural suffix.

Singular	Plural
göz (eye)	gözler (eyes)
kuş (bird)	kuşlar (birds)

Consonant Assimilation

The AX platform also applies consonant assimilation automatically. This involves that consonants change from unvoiced consonants (e.g. k/p/ç) to voiced consonants (e.g. ğ/b/c). Consonant assimilation is not applicable for one-syllable words, foreign words and proper nouns.

The following example shows a consonant change from p to b, when adding the possessive suffix to the noun kitap (book):

kitap (book) -> Benim kitabım (my book)

Buffer letters

Turkish makes use of three buffer letters [n/y/s] to keep vowels apart when adding suffixes. The AX platform automatically adds them as need during inflection.

The following example shows a buffer letter being added to a word ending in a vowel, when putting it into the accusative form:

kedi (cat, lemma) -> kediyi (cat, acc. case)