Case marking and disjunction type alternations

In this paper I present an analysis of the semantics of two kinds of disjunction in Finnish and their interactions with case. I claim that a seemingly optional variation between accusative (ACC) and partitive (PART) case in questions actually depends on the presence/absence of focus, and that this explanation can also be used to account for the semantic consequences of ACC and PART case interacting with two types of disjunction in yes/no and alternative questions.

In Finnish declaratives, irresultative verbs select PART objects, and resultative verbs (e.g. ‘to notice’ in ex. 1a) select ACC objects (Kiparsky 1997, inter alia). PART case is also used with mass nouns and indefinite plural count nouns, irrespective of resultativity (Kiparsky 1997, Karlsson 1999, inter alia). However, in yes/no questions with resultative verbs (ex. 1b, 1c), both cases are grammatical – even though the PART case is ungrammatical in the declarative form of the same sentence (ex. 1a). This alternation between ACC and PART in questions is not random, and the two cases are most felicitously used in different kinds of contexts. ACC case is used to mark the object in a question where (i) the NP is in focus (in the sense of Rooth 1985, 1992) or (ii) the polarity of the event itself is in focus. An example of (i) is a context where we are talking about a list of people/things that Pekka saw – so the proposition that ‘Pekka saw X’ is presupposed. Here, the NP (in 1b) is in focus. An example of (ii) is a setting where we know that Pekka and a man both went to the same store and so they might–or might not–have run into each other. On the other hand, (1c), with a PART object, is used when there is no relevant presupposition generated by focus. Further data illustrating that PART objects are incompatible with focus comes from the fact that wh-questions (2a) and questions with the focus-marker ‘only’ (2b) are ungrammatical with PART, but grammatical with ACC case.

This correlation of ACC vs PART case with the presence or absence of focus can be used to explain an otherwise surprising paradigm that arises with yes/no and alternative questions in Finnish. Finnish has two lexical items for ‘or’, ‘tai’ and ‘vai.’ As illustrated in (3a), a question with ACC-marked objects (with a resultative verb) and ‘tai’ can be interpreted either as a yes/no question (possible answers: "Yes, Pekka saw a woman or a man," or "No, Pekka didn’t see a woman or a man") or as an alternative question (possible answers: "Pekka saw a woman" or "Pekka saw a man"). A question with the same verb, but now with PART-marked objects and ‘tai’ (ex. 3b), can also receive both answers, though it seems to show a preference for the yes/no answer. In contrast, ACC-marked objects and ‘vai’ (ex. 3c) can only be interpreted as an alternative question. A question with PART-marked objects and ‘vai’ (ex. 3d) is ungrammatical (unless, of course, the verb is irresultative, in which case its object is always in PART case – even when it co-occurs with ‘vai’ – and doesn’t display the alternation shown in (1b, 1c)). Thus, ‘tai’ is the ‘default or’ and permits both interpretations, whereas ‘vai’ can only occur in alternative questions (see Hakulinen&Karlsson 1988).

Let us first consider my analysis of the examples with ‘vai’. This disjunction can only occur in alternative questions, and thus (3c), with ACC case and ‘vai’, only has the alternative reading. Interestingly, question (3d), with PART case and ‘vai’, is ungrammatical. This is especially striking, since (1c) – which consists of the first part of (3d) – is grammatical. To account for the ungrammaticality of (3d), I follow Han & Romero (2001), who show that alternative questions necessarily involve focus on the disjunctive XP's. Thus, in Finnish, questions with ‘vai’ involve focus. The ungrammaticality of (3d) can now be explained as a consequence of the incompatibility of PART with focus.

We now turn to the examples with the other disjunction, ‘tai’. According to my analysis, the reason why (3a) can receive either a yes/no interpretation or an alternative interpretation is because (i) the default disjunction ‘tai’ is compatible with both interpretations, and (ii) as discussed above, ACC case is used when there is focus on the NP or on the polarity of the event, and so both the alternative and yes/no interpretations are possible. However, ex. (3b), with PART case and ‘tai’, is predicted to permit only a yes/no answer – since PART case is incompatible with focus and alternative readings require focus on the disjuncts. How, then, can we explain the observation that (3b) actually permits both a yes/no answer and an alternative answer? I claim that semantically, (3b) can only be interpreted as a yes/no question. However, due to the pragmatics of questions and answers – more specifically, the fact that if a person happens to know a more specific answer than a question asks for, s/he can provide that more detailed answer – (3b) can also receive an alternative answer. This also explains why, for (3b), the semantically-generated yes/no interpretation is stronger than the pragmatically-derived alternative question interpretation.

In sum, in this paper I claim that – in contexts where both ACC and PART case are grammatical, e.g. (1b, 1c) – the choice of one over the other depends on the presence/absence of focus. Moreover, I argue that the availability of yes/no and alternative question interpretations can be derived from the interactions between case marking (ACC vs. PART) case and disjunction type (‘tai’ vs. ‘vai’.)

Selected references:

Hakulinen, A. & Karlsson, F. (1988). Nykysuomen lauseoppia. Helsinki: SKS

Han, C. & Romero, M. (2001). Negation, Focus and Alternative Questions. Talk given at WCCFL 20.

Karlsson, F. (1999). Finnish – An Essential Grammar. New York: Routledge.

Kiparsky, P. (1997). Partitive Case and Aspect. Stanford University, Ms.

Rooth, M. (1985). Association with Focus. Doctoral dissertation, UMass at Amherst.

Rooth, M. (1992). A theory of focus interpretation. NLS 1:75-116.

Data

(1a) Pekka huomasi miehen/*miestä.

Pekka-NOM noticed man-ACC/man-PART

‘Pekka noticed a/the man.’

(1b) Huomasiko Pekka miehen?

Noticed-? Pekka-NOM man-ACC

‘Did Pekka notice a/the man?’

(1c) Huomasiko Pekka miestä?

Noticed-? Pekka-NOM man-PART

‘Did Pekka notice a/the man?’

(2a) Kuka huomasi Liisan/*Liisaa?

who-NOM noticed Liisa-ACC/Liisa-PART

‘Who noticed Liisa?’

(2b) Huomasiko Pekka vain miehen/*miestä?

Noticed-? P.-NOM only man-ACC/man-PART

‘Did Pekka notice only a/the man?’

(3a) [ACC objects and ‘tai’]

Huomasiko Pekka miehen tai naisen?

Noticed-? P.-NOM man-ACC tai woman-ACC

‘Did Pekka notice man or woman?’

[ok y/n interpretation]

[ok alternative interpretation]

(3b) [PART objects and ‘tai’]

Huomasiko Pekka miestä tai naista?

Noticed-? P.-NOM man-PART tai woman-PART

‘Did Pekka notice man or woman?’

[ok y/n interpretation (preferred)]

[ok alternative interpretation]

(3c) [ACC objects and ‘vai’]

Huomasiko Pekka miehen vai naisen?

Noticed-? P.-NOM man-ACC vai woman-ACC

‘Did Pekka notice man or woman?’

[ * yes/no interpretation]

[ok alternative interpretation]

(3d) [PART objects and ‘vai’]

* Huomasiko Pekka miestä vai naista?

Noticed-? P.-NOM man-PART vai woman-PART

‘Did Pekka notice man or woman?’

[ ungrammatical ]

In the examples, ‘?’ is used to gloss the question particle [–ko]