Open Issue: Serial Verb Constructions

Problem Statement

Serial verb constructions (roughly, constructions where a clause contains two or more ‘main’ verbs, which typically share some or all of their arguments) are common in the world’s languages, but not that prominent in the initial set of DELPH-IN languages. I would love to see an extension to the Grammar Matrix which allowed users to add serial verb constructions through the customization system. This would require first determining the target semantic representation, then working from the typological literature to map out the range of kinds of serial verb constructions, developing analyses for each kind, and relating them to the customization questionnaire.

Short Description of Serial Verb Constructions

A serial verb construction consists of at least two main verbs or verb phrases, strung together without any connective markers. The verbs in an SVC form together a single clause. Serial Verb Constructions fall within two main categories: clause chaining and integrated SVCs. Under clause chaining, the number of VPs have no upward bound and each verb has its full independent meaning. The linear sequence of verbs reflects strict temporal order and each VP expresses a complete event distinct form its successor. Syntactically each verb has its full inflectional possibility and the same argument frame as when acting as a single main verb. Informally speaking, the verbs in a CC-SVC share their subject while object sharing is optional. Object and adjunct sharing patterns vary greatly. Likewise, the marking of tense and/or aspect on verbs varies between languages. Below you find an example of a CC-SCV

(1) AKAN (Kwa language spoken in the Volta-Basin area)
- Ama tu-u bayere twitw-a no-a di-i Ama uproot-COMPL yam.OBJ cut-COMPL cook-COMPL eat-COMPL N V N V V V
  
  Ama uprooted (tuber of) yam, cut it in pieces, boiled them (and) ate them.

In Integrated SVCs only two verbs take part to express a clearly identifiable situational profile, in, what one might call, a constructional lexeme. More than one situational profile can be recognized, and the syntactic-semantic composition varies in accordance with the profile, leading to subtypes of ISVCs. It is the intuition of many linguists, working on I-SVCs, that the patterns of composition are restricted, as the semantics of the two verbs in an SVC need to synchronize to derive a compositional meaning. Opposed to CC-SVCs, the first verb in an I-SVC does not always display its full morpho-syntactic potential and its semantics can only be understood in composition with V2.

EXAMPLES from Akan:

(2)

?-de nkrante twa duabasa 3P.SG.MASC-take sword cut branch V N V N

He cut off a branch with a sword

(3)

 -de no fεm-m me 3P.SG.MASC-take 3P.SG.ANIM lend-COMPL 1P.SG. V PRO V PRO

he lent me it (e.g., the horse) (Stewart 1963)

(Glossing adapted to a two-line glossing convention)

CC-SVCs and I-SVCs are attested in a variety of language families, for instance West African Languages of the Volta-Basin area, and most of the South and South-East Asian languages.

Existing Literature

Melanie Owens’s forthcoming dissertation is a typological study of the semantics of SVCs.

Chikara Hashimoto’s dissertation is an in-depth look at SVCs in Japanese, and includes the description of a computational implementation in Jacy.

A Computational Treatment of V-V Compounds in Japanese. Chikara Hashimoto. Kobe Shoin Graduate School Ph.D thesis, 2004 (pdf ps).
A Computational Treatment of V-V Compounds in Japanese. Chikara Hashimoto and Francis Bond. In Proceedings of the 12th International Conference on HPSG, pp.143-156. 2005 8.

http://hpsg2005.di.fc.ul.pt/

Last update: 2011-10-09 by anonymous [edit]