Multifunctionality of non-finite verb forms in Turkic languages

Jonathan N. Washington¹

Francis M. Tyers²^,³

Ilnar Salimzianov⁴

1. Swarthmore College, USA
2. Indiana University, USA
3. Высшая Школа Экономики, Москва
4. Independent Scholar

Overview

Conventional delineation

• Participles/причастия - behave nominally and (especially) attributively
  (1) МағанmɑʁɑnI.dat кітапханағаkɘtɑpχɑnɑʁɑlibrary-dat барғанbɑɾʁɑngo-ptcp кісіkɘsɘperson ұнайды.⁠kazʊˈnɑjdəbe.liked-⁠npst-⁠3 "I like the/a person who went to the library."
• Converbs/деепричастия - pattern together with other verbs, "subordinate" (other terms used: adverbial particple, gerund, ...)
(2) МенmiɘnI кітапханағаkɘtɑpχɑnɑʁɑlibrary-dat барып,bɑɾəpgo-cvb селфиˈsɛɫfʲɪselfie жасадым.kazʒɑsɑdəmmake-⁠pst.def-⁠1sg "I went to the library and took a selfie."
• Sometimes infinitives - behaves as a subject (sometimes object, etc.)
(3) Соданsuʊdɑnthat-abl кейінkiɘjɘnafter кітапханағаkɘtɑpχɑnɑʁɑlibrary-dat баруbɑɾʊwgo-inf әдеткеædiɘtkiɘhabit-⁠dat айналады.⁠kazɑjnɑˈɫɑdəturn into-⁠npst-⁠3 "After that going to the library will turn into a habit."

Conventional delineation:
sporadic internal multifunctionality

• Multifunctionality of converbs
  (4) Аһаанɑhɑːneat-⁠cvb барда.⁠sah⁠—⁠from Popova, 2018bɑrdɑgo-⁠pst.3sg "S/he ate and left." "S/he started to eat."
  • Also cf. Johanson (1995) "On Turkic converb clauses":
    4 distinct "levels" of converb (syntactic functions)
• Multifunctionality of participles
  (5) Китепканадаkitepqɑnɑdɑlibrary-loc иштепiʃtepwork-⁠cvb жатканынʤɑtqɑnɯnprog-⁠ptcp-⁠poss.3-⁠acc бүгүнbyɣyntoday даdɑalso көрдүм.kirkørdymsee-⁠pst.⁠def-⁠1sg "Today too I saw the one who's working at the library." "Today too I saw him/her working at the library."
• We argue this is actually syncretism between distinct categories

novel delineation of Turkic NFV forms

• Based on morphological and syntactic criteria
• Some assumptions:
  • inflectional categories (not derivational)
  • a given form may fall in multiple categories (syncretism)
  • entire VP (and often subject) together cast to new syntactic function

Turkic verbal nouns

• VP functions as nominal head:
  subject, complement, adjunct relation to another V
• Take nominal morphology: case, possession
• Possession when agreeing in person/number with "extracted" subject (in genitive case)
• Do not (appear to) occur with dependent adjectives or determiners
• Restrictions regarding use of plural morphology
• Synchronic and diachronic interaction with other categories
• Some overlap with verbal adjective forms
• Adverbial with case morphology, so grammaticalisation into verbal adverb-forming affixes common
• Forms that don't overlap with verbal adjectives often referred to as "infinitives"

Turkic verbal adjectives

• VP functions attributively, modifies nominals
• One of the main strategies in Turkic for forming relative clauses
• modified noun understood to be "extracted" (or similar) from VP
• Morphological behaviour
• Through a null head can function nominally, take noun morphology
• Sometimes have person/number agreement with subject
• Some differences from adjectives
• (Often) may not be modified by adverbs
• Do not receive comparative morphology
• In some languages (incl. Turkish), agreement marking with subject

Turkic VN & VAdj: examples

Nominalisation of Vadj vs. VN

Verbal adverbs

• Head VPs subordinate to another V
• Adverbial relation to other V
• Semantically convey (usually) specific type of temporal or causal relationship between VPs
• Usually don't take additional morphology…
• Conditional vadv agrees in person/number with subj. (most Turkic lgs)
• Can share a subject with matrix VP or have a different subject

Infinitives

• Collocate with auxiliaries, single VP projection?
• Auxiliaries small lexical class for many Turkic languages
• Only one external argument, unlike for vadv
• Combination of specific aux and specific inf: conveys specific aspectual or modal information
• Auxiliary bears inflection for person/number, tense, etc.
• inf affixes almost always show syncretism with vadv

Verbal adverbs vs. Infinitives

Potential ambiguity: identical AUX and lexical verb

(10a) Аһаанɑhɑːneat-⁠inf барда.⁠sahbɑrdɑgo-⁠pst.3sg "S/he started to eat."
(10b) Аһаанɑhɑːneat-⁠vadv барда.⁠sahbɑrdɑgo-⁠pst.3sg "S/he ate and left."

Methodology

• Chose 6 geographically diverse Turkic languages

• Consultation with grammars, native speakers, corpora
• For each NFV-forming suffix identified
• examined morphological and syntactic behaviour
• categorised uses
• Output: near-exhaustive typology of NFV-forming suffixes

Methodology

• Morphological behaviour, example tests:
• Take possession and/or case morphemes?
  → Verbal noun or verbal adjective reading (rely on semantics)
• Take plural morpheme?
  → Probably verbal adjective
• Some forms with limited morphology
• Syntactic behaviour (here with some "participle" forms):

	complement clause	adnominal modifier	clausal subject	nominal subject	nominal object
gördüğüm	✔	✔	✔	✔	–
gören	–	✔	–	–	–
gördüklerim	–	–	✔	✔	–
görenler	–	–	✔	✔	–
gittiğimi	✔	–	–	–	✔
gideni	–	–	–	–	✔

Results Kazakh

Results Kyrgyz

Results Sakha

Results Tatar

Results Turkish

Results Tuvan

Generalisations

• No Turkic language clusters NFV-forming suffixes into two groups
• (Some tendency in that direction)
• Tuvan is the only language where conventional patterning is broken (grammaticalistion)
• Forms often called infinitives are VN-only participles (but not always limited to certain nominal inflectional)

Historical perspective

• vn + case → vadv
• Verbal nouns with cases are adverbial, same function
• Can/do grammaticalise to verbal adverbs
• when one of the components otherwise becomes less productive
• when semantics of the combination are no longer compositional
• Sometimes hard to tell when fully grammaticalised
• -/GAnʧA/kir vadv (until _-ing) as -/GAn-ʧA/ vn-manner
• -/GAlAK/tuv vn, vadj, as vn only takes loc, so -/GAlAKtA/ as vadv
• vn ↔ vadj
• Nouns and adjectives often used as overt heads of other's phrase in Turkic (via null head)
• vadv → inf?
• Accompanied v → aux?
• vn-nom/acc → inf
• And sometimes hard to tell if vn is limited to nom/acc

Outstanding issues: Tatar infinitive

• Appears to be used as all four categories?
(11a) МиңаmiɴɑI.dat китапханәгәkitɑpχɑnæɣælibrary-dat барыргаbɒɾəɾʁɑgo-vn кирәкkiɾæknecessary иде.⁠tatiðɘbe-⁠pst.3 "I needed to go to the library."
(11b) Китапханәгәkitɑpχɑnæɣælibrary-dat барыргаbɒɾəɾʁɑgo-vadj кыюлыгымqəjuɫəʁəmnerve-⁠poss.1sg җитмәде.⁠tatʑitmæðɘbe.sufficient-⁠neg-⁠pst.3 "I didn’t have the nerve to go to the library."
(11c) Китапханәгәkitɑpχɑnæɣælibrary-⁠dat барыргаbɑɾəɾʁɑgo-⁠vadv вакытwɒqəttime кирәк.⁠tatkiræknecessary "Time is needed in order to go to the library."
(11d) МинminI китапханәгәkitɑpχɑnæɣælibrary-dat барыргаbɒɾəɾʁɑgo-inf(?) яратам.⁠tatjɒɾɑtɑmnecessary "I like going to the library."

Outstanding issues: volationals

Outstanding issues: compound/serial verbs

• Not clear how compound verbs fit our typology
(13) МенmiɘnI⁠ кітаптыkɘtɑptəbook-⁠acc мектепкеmiɘktiɘpkiɘschool-⁠dat алыпɑɫəptake-⁠vadv? бардым.⁠kazbɑɾdəmgo-⁠pst⁠-⁠1sg "I brought the book to school."
• алып can't be inf because бар isn't auxiliary (& contributes lexical meaning to sentence)
• алып can't be vadv because it doesn't form contiguous constituent with object кітапты
• Together the two verb forms appear to form a single verb, but have internal structure

Outstanding issues: non-finite forms with copulas

• Single predicate formed from NFV + copula?
(14a) МенmiɘnI кітапханағаkɘtɑpχɑnɑʁɑlibrary-⁠dat барғанbɑɾʁɑngo-⁠vn? едім.⁠kaziɘdɘmcop-⁠pst-⁠1sg "I had gone to the library."
(14b) МенmenI китепканагаkitepqɑnɑʁɑlibrary-⁠dat барсамbɑɾsɑmgo-⁠vadv?⁠-⁠1sg экен.⁠kirekenmir⁠-⁠cop.npst.3 "I wish I could go to the library."

• Finite verb form grammaticalised from NFV + copula?
(15) МенmiɘnI кітапханағаkɘtɑpχɑnɑʁɑlibrary-⁠dat барғанмын.⁠kazbɑɾʁɑnməngo-⁠past⁠-⁠1sg "I went to the library."

Conclusions

• In-depth study of Turkic NFV-forming morphology
• Traditionally assumed categories not coherent
• Syncretism and historical factors led to incorrect groupings
• Also focus on similarity of functions (syntax) without considering morphology or semantics
• New approach takes into account morphology, syntax, and semantics