ADP: adposition
Definition
Turkish only has postpositions. They occur after a complement noun phrase (or a nominal subordinate clause) and they form a single structure with the complement to express its grammatical and semantic relation to another unit within a clause.
A number of postpositions in Turkish are complex,
derived from a closed set of nouns (see Göksel & Kerslake 2005, ch.17).
Examples include, alt-ın-da “below”, arka-sı-nda “behind”, konu-su-nda “about”.
We also mark these expressions as ADP when they are used as postpositions
(TODO: this requires discussion).
Examples
- Ali’ye rağmen “despite Ali”
- Ali için “for Ali”
- Kalabalığın arasında “between/within the crowd”
- Ali’nin dışında “except Ali”
References
Aslı Göksel and Celia Kerslake. Turkish: A Comprehensive Grammar. London: Routledge, 2005.
Treebank Statistics (UD_Turkish)
There are 44 ADP lemmas (1%), 112 ADP types (1%) and 1884 ADP tokens (4%).
Out of 14 observed tags, the rank of ADP is: 7 in number of lemmas, 9 in number of types and 5 in number of tokens.
The 10 most frequent ADP lemmas: ki, li, için, gibi, diye, kadar, _, olarak, sonra, ce
The 10 most frequent ADP types: ki, için, gibi, diye, kadar, li, lı, olarak, sonra, önce
The 10 most frequent ambiguous lemmas: ki (ADP 297, CCONJ 100), _ (ADP 89, ADJ 8, NOUN 1), sonra (ADP 70, ADV 43, NOUN 4), önce (ADP 51, ADV 32, NOUN 17), siz (PRON 62, ADP 36), ci (ADP 30, NOUN 1), doğru (ADJ 47, ADP 29), birlikte (ADP 19, ADV 16), ile (CCONJ 85, ADP 19), beri (ADP 10, ADV 2)
The 10 most frequent ambiguous types: ki (ADP 261, CCONJ 99), için (ADP 201, ADJ 1), diye (ADP 140, VERB 3), sonra (ADP 66, ADV 34), önce (ADP 49, ADV 22, NOUN 7), doğru (ADP 29, ADJ 22), siz (ADP 20, PRON 11), birlikte (ADP 18, ADV 15, NOUN 1), ile (CCONJ 85, ADP 19), lar (ADP 14, AUX 5)
- ki
- için
- diye
- sonra
- önce
- doğru
- siz
- birlikte
- ADP 18: Babası her hafta düzenli olarak kardeşleriyle birlikte tıraş ederdi onu .
- ADV 15: Onunla daha fazla birlikte olabilmek .
- NOUN 1: Bilimde bu tür mucizevi elatmalara yer olmadığı gibi , varlığı , Tanrı’ya inanan lar için tehlike yaratır : diferansiyel eşitliklerinizde ki matematiksel tekillikleri daha iyi bir modelle ortadan kaldırdığınızda Tanrınız da tekillikle birlikte ortadan kalkar .
- ile
- lar
Morphology
The form / lemma ratio of ADP is 2.545455 (the average of all parts of speech is 2.690133).
The 1st highest number of forms (24) was observed with the lemma “_”: a, dan, den, e, i, in, lar, lara, lardan, larla, ları, ların, le, ler, lerden, leri, lerin, leriyle, si, ya, ı, ın, ını, ıyla.
The 2nd highest number of forms (15) was observed with the lemma “ki”: aki, aveau’daki, eki, ithatpaşa’daki, ki, kiler, kilere, kileri, kilerin, kinde, kinden, kine, kini, kü, nki.
The 3rd highest number of forms (14) was observed with the lemma “lik”: ‘lik, ‘luk, anlışlanabilirlik, anlışlanabilirliği, irmibeş’lik, lik, liğin, lüğünün, lık, lıklar, lığı, lığıma, lığının, lığınız.
ADP occurs with 7 features: Number (168; 9% instances), Person (168; 9% instances), Case (163; 9% instances), Number[psor] (16; 1% instances), Person[psor] (16; 1% instances), Mood (3; 0% instances), Tense (2; 0% instances)
ADP occurs with 17 feature-value pairs: Case=Abl, Case=Acc, Case=Dat, Case=Gen, Case=Ins, Case=Loc, Case=Nom, Mood=Cnd, Number=Plur, Number=Sing, Number[psor]=Plur, Number[psor]=Sing, Person=3, Person[psor]=1, Person[psor]=2, Person[psor]=3, Tense=Past
ADP occurs with 26 feature combinations.
The most frequent feature combination is _ (1716 tokens).
Examples: ki, için, gibi, diye, kadar, li, lı, olarak, sonra, önce
Relations
ADP nodes are attached to their parents using 13 different relations: case (1827; 97% instances), compound (20; 1% instances), root (14; 1% instances), flat (9; 0% instances), tr-dep/obj (3; 0% instances), cc (2; 0% instances), conj (2; 0% instances), nmod:poss (2; 0% instances), advmod:emph (1; 0% instances), ccomp (1; 0% instances), tr-dep/compound:lvc (1; 0% instances), det (1; 0% instances), nsubj (1; 0% instances)
Parents of ADP nodes belong to 12 different parts of speech: NOUN (915; 49% instances), VERB (440; 23% instances), ADJ (251; 13% instances), PRON (106; 6% instances), PROPN (59; 3% instances), NUM (49; 3% instances), ADV (21; 1% instances), DET (19; 1% instances), ROOT (14; 1% instances), CCONJ (4; 0% instances), ADP (3; 0% instances), INTJ (3; 0% instances)
1808 (96%) ADP nodes are leaves.
47 (2%) ADP nodes have one child.
22 (1%) ADP nodes have two children.
7 (0%) ADP nodes have three or more children.
The highest child degree of a ADP node is 5.
Children of ADP nodes are attached using 20 different relations: punct (18; 16% instances), cop (15; 13% instances), tr-dep/obl (12; 10% instances), conj (9; 8% instances), det (9; 8% instances), nmod (6; 5% instances), nmod:poss (6; 5% instances), nummod (6; 5% instances), advmod:emph (5; 4% instances), tr-dep/obj (5; 4% instances), advmod (4; 3% instances), amod (4; 3% instances), cc (4; 3% instances), case (3; 3% instances), acl (2; 2% instances), tr-dep/compound:lvc (2; 2% instances), nsubj (2; 2% instances), compound (1; 1% instances), compound:redup (1; 1% instances), parataxis (1; 1% instances)
Children of ADP nodes belong to 12 different parts of speech: NOUN (19; 17% instances), PUNCT (18; 16% instances), VERB (17; 15% instances), AUX (15; 13% instances), NUM (10; 9% instances), CCONJ (9; 8% instances), ADJ (7; 6% instances), ADV (5; 4% instances), DET (5; 4% instances), PRON (4; 3% instances), ADP (3; 3% instances), PROPN (3; 3% instances)
ADP in other languages: [am] [ar] [bg] [bxr] [ca] [ckb] [cop] [cs] [cu] [da] [de] [el] [en] [es] [et] [eu] [fa] [fi] [fo] [fr] [ga] [gl] [got] [grc] [he] [hi] [hr] [hu] [id] [it] [ja] [kk] [kmr] [ko] [la] [lv] [mr] [nl] [no] [pl] [pt] [ro] [ru] [sa] [sk] [sla] [sl] [so] [sr] [sv] [swl] [ta] [tr] [ug] [uk] [u] [urj] [ur] [vi] [yue] [zh]