Chapter two deals with WH in situ in French single WH constructions. In that chapter, I show that WH-in-situ questions in French do not behave exactly like their displaced counterparts. The scope of the WH phrase in situ is fixed, while the scope of the dislocated WH phrase is not. In that chapter, all the technical machinery that will serve as the basic theory for the remaining chapters is introduced. The Scopal ECP is presented in detail, as well as the formal background theory of Skolem functions, predicative indefinites and the distinction between focus and topic.
Chapter three concentrates on partial-WH-movement constructions in German and Hungarian. I argue that while Hungarian scope markers involve movement, German ones do not. I account for the island effects exhibited by partial-WH-movement constructions in terms of scope. I account for the differences between German and Hungarian with regard to intervention effects in syntactic terms: while the relation between the scope marker and the intermediate WH phrase is an A’-relation in German, it is an A-relation in Hungarian.
Chapter four reopens the debate as to whether so-called N-words in French are negative quantifiers or NPIs. The evidence provided suggests that French N-words are negative quantifiers rather than NPIs. The dependency between the French Neg ne and the N-word in situ is sensitive to intervening scopal elements which suggest that these constructions are further instances of split constructions.
Chapter five deals with focus particle constructions. I provide evidence for the view that attributive focus particles are not base-generated where they take scope, but involve base-generation of a dislocated focus null operator. The island effects that such structures exhibit are accounted for in terms of scope.
Finally, chapter six summarises the findings and points to some interesting issues for fiirther research.
^ cf. Higginbotliam and May (1981), Banvise and Cooper (1981), Dowty, Wall and Peters (1981), Partee (1995).
" As pointed out by Honcoop (1998) in his analysis of split constructions, not everyone agrees that the indefinite NP in such constructions functions semantically as a restriction on the range of the WH operator. For example, in the case of (5 (a and b)), many researchers would argue that the restriction over which tlie WH operator ranges is a number. However, like Honcoop (1998), I assume that an indefinite NP functions as a restriction on the range of some operator Op when tlie indefinite NP is interpreted as a property that restricts tlie range of possible valuations of the variable quantified over by Op. What this means is that in examples like (5), the operator ranges, not simply over a number, but over a number of books. Tliis is important because it fits with the interpretation one gets for such sentences. (5a) and (5b) are asking how many books, not, say, how many magazines or cats.
^ I will use the term noun restrictor and semantic restriction interchangeably.
^ Split constructions raise a number of theoretical issues, e.g. for structure dependency and compositionality. However, it is customary in the literature to simply acknowledge tliese constructions (cf. Rizzi 1990). I will follow this trend here. I have, however, more to say about tliese issues in fortlicoming work (cf. Mathieu and Sitaridou 2002). Following ideas by Devine and Stephens (2000), it will be assumed that splitting can be traced back to a paratactic DP consisting of two DPs in apposition, DPi being a null head nominal and DPo the stranded nominal. A modifier restricts the null head nominal in DPi and the null head N is coindexed with the stranded nominal. The idea is tlius tliat split-DP syntax involves discontiguity, rather tlian discontinuity. Discontiguity or split-DP syntax is possible is possible when a language has null head nominals. In other words, what is split is not a sub-constituent o f a DP, but a DP in its ovm right. So in fact, “splitting” is a misnomer.
“ Note that long DP movement is not possible in all dialects of German.
' Note that I am interested in movement of the bare operator and movement of the nominal \\ith wliich it is associated. If there is a stjlistic rule involved in the cases under discussion then it must be the cases like those where the interrogative is split (i.e. when the nominal remains in situ). These split variants are not only used in spoken language, but also in literary genre and poetry (tliis is true for French, Russian, Polish, Greek, etc.). If we can call a construction 'stylistic', then these cases are primal candidates. The nominal may in fact raise (I w ill later argue in detail that stranded indefinites are incorporated nouns). This movement is not feature-driven and may thus be called a stylistic rule. However, raising of the nominal after the bare operator has moved to Spec-CP cannot be considered a stylistic rule, since this option is the default option in terms o f interpretation (it does not carry any presuppositions). In otlier words, tliere is nothing ‘stylistic’ about it. It nevertheless raises witliout being attracting by a feature, since the feature of C has already been checked when the bare operator has moved to Spec-CP. One could argue that the nominal is attracted by a focus feature and ends up in the specifier of a focus phrase. But, tliis is not going to work, since it is generally assumed that questions and focus items compete for the same position (i.e. Spec-Foc, cf Rizzi 1997).