Anders Holmberg

Questions, polarity, and verb movement in Germanic and Finnish

The fact that the English negative auxiliary forms don’t, can’t etc. undergo movement (subject-aux inversion, SAI) shows that they are syntactic constituents.

  1. Don’t you speak French|?

That is to say, they cannot be the result of just a phonological contraction rule. As argued by Zwicky and Pullum 1983 (in Language 59), the English negation n’t in don’t, can’t etc. is an inflection. This implies that the negative auxiliaries are not derived from a structure such as (2) by cliticization of the negation to do, will, can , etc. but instead, given that inflected forms are syntactically derived (following Baker 1988 and Julien 2000), derived from a structure (3) by head movement and incorporation of do, will, can , etc. into n’t.

(2) [TP do [NegP not …]]

  1. [NegP n’t [TP do …]]

This is consistent with the fact that the interpretation of can’t is not CAN (NOT) but NOT(CAN), won’t is NOT(WILL), etc. There are exceptions, including mustn’t in You mustn’t leave now, which is MUST (NOT); I will argue that the exceptions are in fact derived by a phonological contraction rule, operating on the structure [must [not …]].

This entails that SAI in negative sentences is not T-to-C movement , but Neg-to-C movement, where Neg contains the tensed auxiliary. This makes sense especially in the case of yes/no-questions (YNQs). I assume that the negation word is the realization of a functional head Pol(arity) when Pol has the value [Negative]. The other value, [Affirmative], is realized as a null-morpheme. Pol c-commands T. In that case YNQs, negative as well as affirmative, are formed (in English) by movement of Pol to C, Pol incorporating T, which in turn incorporates the auxiliary. That is to say, SAI is really Pol-to-C movement. A characteristic property of YNQs is that they have open polarity. In English the syntactic expression of this is that Pol is a trace. SAI in YNQs is thus not just a matter of linearization or ‘lexicalizing C’ or any other such requirement purely dictated by PF. Instead it is semantically motivated: it derives a structure with a Pol trace, interpreted as a variable ranging over two values, affirmative or negative. The question is a request to the addressee to assign a value to this variable.

The question

  1. Do you not speak French?

is also derived by Pol-movement. In this case the spelt-out negation is not the head n’t but the specifier not. T, containing the auxiliary, moves and incorporates in the empty Pol, which in this case is [+neg] by virtue of agreement with its specifier. Subsequently Pol moves to C, just as in (1).

YNQs in the Scandinavian languages are like (4), except that not just auxiliaries but any finite verb can undergo the movement.

  1. Talar du inte franska?

speak you not French

The negation word in the Scandinavian languages is (uncontroversially) always a specifier. The verb moves to T, which moves to Pol (valued negative in this case), which moves to C. Thus ‘V-to-C’ is actually Pol-to-C.

English and Scandinavian YNQs are thus quite similar to Finnish YNQs. In Finnish a negative YNQ is formed by movement of the negation to C, adjoining it to an overt question particle -ko/-kö, while a nonnegative YNQ is formed (in the unmarked case) by movement of the highest auxiliary or verb to C, adjoining it to the question particle.

(6) a. Sinä et puhu ranskaa.

you not speak French

  1. Et-kö sinä puhu ranskaa?

    2. not-Q you speak French

  2. Puhut-ko sinä ranskaa?

speak-Q you French

The negation is uncontroversially a head in Finnish. I assume it is the spell-out of the negative value of Pol. The affirmative value of Pol is a null morpheme into which T, incorporating the verb, is incorporated. YNQs are thus derived by Pol-movement to C.

A complication is posed by the construction You do not speak French. There must be a head position between C and Pol into which the auxiliary moves. Furthermore, this head must not have any interpretable features (such as a feature ‘Finite’), but must be a ‘pure EPP-head’ with no other function than triggering movement of the subject and the auxiliary, if SAI/V-movement in YNQs is indeed Pol-to-C. Again there are interesting parallels with Finnish. I will also offer some speculations about V2, residual and full V2, where Pol-to-C movement is not semantically motivated.