"What do you do?" Variation in Interrogative Predicates
Content ('wh') questions have been the subject of numerous studies conducted in a variety of different frameworks, but relatively little of this research has dealt with interrogative predicates, one notable exception being Hagège's (2008) article on the typology of interrogative verbs. Interrogative predicates can be divided into two broad types: i) an interrogative verb is available which may (1) or may not (2) be synchronically analyzable; and ii) a nominal question word combines with a non-interrogative verb (3).
|'What are you doing?'|
|(Wordick 1982: 165)|
|'What have you done to your brother?'|
|(Crowley 1982: 159)|
|(François 2005: 121)|
|(3)||a. What did John do to Mary?|
|b. What happened to the car?|
These strategies, though distinct in terms of their morphosyntax, can perform the same communicative function and express the same meaning. I therefore outline an analysis of their semantics that treats all interrogative predicates as being fundamentally the same. Following Parsons (1990), a sentence which describes an event is taken to distinguish between an event e, the type of that event, and its thematic roles, as in (4).
|(4)||John hit Mary.|
|∃e[hitting(e) ∧ agent(e, John) ∧ patient(e, Mary)]|
I propose that an interrogative predicate presupposes an event involving the specified agent and expresses an enquiry about only part of the verb's meaning, i.e. the type of the event. Adopting a Structured Meaning approach (von Stechow 1990; Krifka 2001), this insight can be captured formally: the restriction of a question containing an interrogative predicate is the type of the event (5).
|(5)||What did John do?|
|<λx.∃e[x(e) ∧ agent(e, John)], EVENT-TYPE >|
This approach to the semantics of interrogative predicates has implications for the analysis of information structure. It is generally assumed that an 'answer constituent', which represents new information, has focus status. However, it is not always the case that an 'answer constituent' is a constituent at c- and/or f-structure (see King 1997). This is true of (6), which is a possible answer to (3b). (The bold represents the VP.)
|(6)||[Charlie put]FOCUS the car [in the garage]FOCUS.|
I contend that the semantics of interrogative predicates are key to understanding and accounting for such data. In short, it is not always the case that 'answer-focus' can be defined in terms of c- or f-structure constituents because the values of i-structure attributes such as Focus are semantic objects. In the case of (6), the new information which has focus status is the verb's event-type (putting) and those entities (Charlie, garage) that bear one of its attendant thematic roles, as specified in the non-interrogative verb's lexical entry, which did not appear in the question, as bolded in (7).
|(7)||Q: What happened to the car?|
|<λx.∃e[x(e) ∧ theme(e, car)], EVENT-TYPE>|
|A: [Charlie put]FOCUS the car [in the garage]FOCUS.|
|∃e[putting(e) ∧ agent(e, Charlie) ∧ theme(e, car) ∧ goal(e, garage)]|
This analysis therefore posits a different view of the nature of information-structure attribute values, as well as a more direct relationship between information-structure and semantic-structure in the LFG parallel architecture than has previously been assumed.
In morphosyntactic terms, the strategies exemplified by (1), (2) and (3) vary with regard to which element introduces the EVENT-TYPE restriction: the verbal predicate, as in (1) and (2), or an 'interrogative chunk', as in (3). The 'interrogative chunk' strategy does not represent a typical content question containing what as in this particular case what does not query an argument subcategorized for by the verb, whether it be a subject (cf. What exploded?), an object (cf. What did you wear?), or a complement with propositional meaning (cf. What do you think?). Only in combination do the two express an enquiry about the type of the event and hence form an 'interrogative chunk'. This construction is thus analysed, on analogy with the treatment of idioms in LFG, as an interrogative predicate which subcategorizes for a nonthematic argument whose FORM value is specified. In this way, the interrogative predicates in examples such as (3a) and (3b) are fundamentally equivalent but not identical to those found in, for example, Yindjibarndi (1), Paamese (2a) and Mwotlap (2b).
Incorporating a new approach to information structure and its place in the architecture of LFG, the proposed analysis is the first to define and account for the features of interrogative predicates and their answers in terms of not only different types of linguistic structure (syntactic, semantic, and information structural), but also the crucial interactions between them.
- Crowley, T. (1982). The Paamese language of Vanuatu. Canberra: ANU.
- François, A. (2005). 'A typological overview of Mwotlap, an Oceanic language of Vanuatu'. Linguistic Typology 9: 115-146.
- Hagège, C. (2008). 'Towards a Typology of Interrogative Verbs'. Linguistic Typology 12: 1-44.
- King, T. H. (1997). 'Focus Domains and Information-Structure'. In M. Butt & T. H. King (eds.) Proceedings of the LFG97 Conference. Online: CSLI Publications.
- Krifka, M. (2001). 'For a Structured Meaning Account of Questions and Answers'. In C. Féry & W. Sternefeld (eds.), Audiatur Vox Sapientiae. A Festschrift for Arnim von Stechow. Berlin: Akademie Verlag. 287-319.
- Parsons, T. (1990). Events in the Semantics of English. Cambridge, MA: MIT Press.
- von Stechow, A. (1990). 'Focusing and background operators'. In W. Abraham (ed.), Discourse Particles. Amsterdam: John Benjamin. 37-84.
- Wordick, F. J. F. (1982). The Yindjibarndi Language. Canberra: ANU