TRmorph: A Turkish morphological analyzer

TRmorph has a new version with lots of improvents. Unless you want to replicate analyses you got in the past, you should consider using the new version here.

TRmorph is a relatively complete morphological analyzer for Turkish. It is implemented using SFST, and uses a lexicon based on (but heavily modified) the word list from Zemberek spell checker. The morphological analyzer is distributed under the GPL.

Latest tested version can be downloaded here. As well as the full source code, a compiled fsa, suitable to be used with SFST tools, is included in this package. The latest development version can be downloaded from github. To use the analyzer you need SFST. The analyzer can also be compiled and used with HFST tools. A UNIX makefile is provided for easy compilation from the sources (see the included README file for details.

The analyzer is fairly complete, however, it may not be easy on unaccustomed eyes. Documentation and cleanup work is going on, you may want to visit soon to get a newer version. I'm always interested in comments, corrections or improvements from others. Please feel free to contact me.

If you use this analyzer in your research, and want to cite it, please cite the following paper:

Çağrı Çöltekin (2010). A Freely Available Morphological Analyzer for Turkish In Proceedings of the 7th International Conference on Language Resources and Evaluation (LREC2010), Valletta, Malta, May 2010. (pdf).

You can try the current development version (as of 2011-06-21) of TRmorph here. If you have javascript enabled you can use the buttons below the input box to enter special Turkish characters. The analysis symbols are linked to their descriptions in this page. The stems are linked to their Wiktionary definitions.

Some characters in the input string are filtered out in the web demo. If the demo fails to accept the input you provide, or if you need to analyze large amount of data, please download and use an off-line version.

If things do not looks as they should, please let me know.

Type the word to analyze:

The following is a (partial) list of analysis symbols used in TRmorph. The documentation below does not yet include the derivational morphemes that start with D_.

Symbol	Gloss	Notes/Example
POS Tags
adj	Adjective
adv	Adverb
bch	Backchannel
cnjadv	Conjunctive adverb (or Discourse connective)
cnjcoo	Coordinating Conjunction
cnjsub	Subordination Conjunction
exist	Existential particle `var'
ij	Interjection
n	Noun
nexist	Negative Existential particle `yok'
not	Negative marker `değil'
np	Proper Name
num	Number
pau	Pause	This is for special for the STD project
pnct	Punctuation
postp	Postposition
prn	Pronoun
ui	Utterance Initializer	This is for special for the STD project
v	Verb
vaux	Auxiliary Verb
Subcategories
ngtv	Negative	This is currently used only with the POS tag
Subcategories of pronouns
dem	Demonstrative pronoun
locp	Locative pronoun	these are only three pronouns `ora', `bura' and `şura'.
pers	Personal pronoun
qst	Qeustion particles that act like pronouns	These are words like `nere', `ne', `kim' that make wh- type questions. They act like pronouns, and commonly classifeid as such. For example Göksel&Kerlaske (2005) names them `interrogative pronouns'
Nominal morphemes
Plural Suffix
pl	Plural suffix	ev-ler `houses
Case suffixes
abl	Ablative case	ev-den `from the house'
acc	Accusative case	ev-i
dat	Dative case	ev-e `to the house'
gen	Genitive case	ev-in `the one that belongs the house'
ins	Instrumental/comitative case	This one also has a clitic equivalent `ile', and is not considered as a case for most grammar books. Probably this has three different functions. (1) and: `araba-yla evi sattık' (we sold the car and the house). (2) instrumental case: `araba-yla eve gittik' (we went home with the car). (3) comitative case: `Ali-yle eve gittik' (we went home with Ali). Current version of TRmorph uses single symbol for all three senses. ev-le `with the house'
loc	Locative case	ev-de `in/on/at the house'
Possessive Suffixes
p1p	1st person pulural possessive	ev-imiz `our house'
p1s	1st person singular possessive	ev-im `my house'
p2p	2nd person pulural possessive	ev-iniz `your house'
p2s	2nd person singular possessive	ev-in `your house'
p3p	3rd person pulural possessive	ev-leri `their house'
p3s	3rd person singular possessive	ev-i `his/her/its house'
ki
ki	ki	-ki attaches to the nominals in locative and genitive case and forms "attributive adjectival phrases or pronominal expressions" (G&K). It can attach to the same word multiple times. ev-de-ki-nin-ki `the one that belongs to the person in the house'
Morphemes that attach to (mostly) verbs
Compound verb morphemes
abil	Ability	gör-ebil `to be able to see'
adur	Continuously	This is not very productive.
agel	Agel	This is not very productive.
agor	Agor	This is not very productive.
akal	Akal	This is not very productive.
akoy	Akal	This is not very productive.
ayaz	Ayaz	This is not very productive.
iver	Quickly	yıka-yıver `wash it quickly'
Copular suffixes
cpl_di	Past copula	gel-iyor `he/she/it is coming', gel-iyor-du `he/she/it was coming'
cpl_mis	Evidential copula	gel-iyor `he/she/it is coming', gel-iyor-muş `(it is said that) he/she/it was coming'
cpl_sa	Conditional copula	gel-iyor `he/she/it is coming', gel-iyor-sa `if he/she/it is coming'
dir	DIr	This suffix (except a few exceptions) follows person agreement, serves a number of purposes including introducing supposition, nominal predicates (especially in formal language). G&K calls this suffix 'generalized modality marker', Kornfilt treats it (more or less) as a copula. doktor-dur `(supposedly) s/he is a doctor' (note that this -dur is generally not used in spoken language if supposition is not implied)
Negative marker(s)
neg	Negative marker	We do not distinguish between the `regular' negative marker -me/-ma and -e/-a before -abil.
Person Agreement
1p	1st person plural	gör-dü-k `We saw'
1s	1st person singular	gör-dü-m `I saw'
2p	2nd person plural/formal	gör-dü-nüz `you saw'
2s	2nd person singular	gör-dü-n `you saw'
3p	3rd person plural	gör-dü-ler `they saw'
3s	3rd person singular	gör-dü `he/she/it saw' (null morpheme)
Sobordinating Suffixes
cv	Converb markers	These markers form subordinating clauses with adverbial functions. Most of the suffixes that form converbs can also form other forms of subordination (verbal nouns and participles). Normally, the converbial markers has a restricted context, i.e., they serve as cv only if they are followed by certain morphemes or words (particles). Current version of TRmorph does not fully restrict these. gör-mek için gittim `I went in order to see'.
part_acak	Participle markers	This is one of the morphemes that make non-finite verbs of relative clauses.
part_dik	Participle markers	This is one of the morphemes that make non-finite verbs of relative clauses. gör-düğ-üm film `the movie that I saw'
part_yan	Participle markers	This is one of the morphemes that make non-finite verbs of relative clauses.
vn_acak	Verbal Noun Markers	This is one of the morphemes that form noun clauses from non-finite verbs. Together with Participles and Converbs, some of the previous versions of TRmorph used to assign a single symbol for all verbal noun markers. adam ol-acak çocuk
vn_dik	Verbal Noun Markers	This is one of the morphemes that form noun clauses from non-finite verbs. Together with Participles and Converbs, some of the previous versions of TRmorph used to assign a single symbol for all verbal noun markers.
vn_ma	Verbal Noun Markers	This is one of the morphemes that form noun clauses from non-finite verbs. Together with Participles and Converbs, some of the previous versions of TRmorph used to assign a single symbol for all verbal noun markers.
vn_mak	Verbal Noun Markers	This is one of the morphemes that form noun clauses from non-finite verbs. Together with Participles and Converbs, some of the previous versions of TRmorph used to assign a single symbol for all verbal noun markers. gör-mey-e gittim `I went to see'.
vn_yis	Verbal Noun Markers	This is one of the morphemes that form noun clauses from non-finite verbs. Together with Participles and Converbs, some of the previous versions of TRmorph used to assign a single symbol for all verbal noun markers.
Tense/aspect/modality markers
t_aor	Aorist	gör-ür `he/she/it sees (something)'
t_cond	Conditional	It can also give the optative mood. gör-se `if he/she/it sees'
t_cont	Continuous tense	gör-üyor `he/she/it is seeing (something)'
t_fut	Future tense	gör-ecek `he/she/it will se (something)'
t_makta	??Continuos/progressive/imperfective	This is similar to t_cont, but used less frequently. Most of the time it is used in formal situations, and has a more definite progressive sense (-yor can be used for future events as well). gör-mekte `he/she/it is seeing (something)'
t_narr	Narrative (or evidential) past tense	gör-müş `it is evident/said that he/she/it saw (something)'
t_obl	Obligation	gör-meli `he/she/it must see (something)'
t_opt	Optative	Indicates wish and hope, it can also have imperative meaning (archaic). Note that t_cond below may also indicate the same mood. bitir-e `(I) wish/hope/order that he/she/it finishes'
t_past	Past tense	gör-dü `he/she/it saw (something)'
Voice
caus	Causative	Causative suffix can be attached to the same stem multiple times. yika-t-tır `to make someone have (something) washed'
pass	Passive	sev-il `to be loved'
rec	Reciprocal	sev `to love' -> sev-iş `to love each other' or `make love'
ref	Reflexive	yıka `to wash' -> yıka-n `to wash oneself'