On expletive subjects in Swedish: Another case of syntactic haplology?
In a recent paper, Christer Platzack discusses some interesting cases of backward binding with psychological predicates and tough constructions as in (1) and (2).
(1) Each other’si health worried the studentsi
(2) Pictures of each otheri were not hard for usi [PROi to draw e].
Non V2-languages like English and Italian allow backward binding into the subject, whereas V2 languages like Dutch, Icelandic and Swedish don’t. Platzack assumes, following Cardinaletti (2004), that the non-V2 languages have two subject positions in the T-domain of the sentence, whereas in V2 languages the highest subject position is in the C-domain. On Platzack’s analysis, backward binding is possible just in case the two targets for parallel movement are in the same domain, in which case a form of haplology applies, enabling the required connectivity.
In this talk I will investigate whether Platzack’s use of haplology carries over to some cases in Swedish where expletive subjects, which are otherwise obligatory, tend to be omitted, as in the tough construction in (3) and in the construction in (4), involving finite complements, which seems to be spreading.
(3) Det är (det) svårt att säga.
it is it hard to say
‘It is hard to say.’
(4) Det var dumt att du sa (Engdahl 2010)
it was stupid that you said
‘It was stupid that you said that.’
The syntax of polar questions
Two properties, and only two, are special to direct questions: Q(uestion)-force and a focused variable. Q-force is a variety of illocutionary force, which I take to be a syntactic feature with widest sentential scope (in the spirit of Ross’s 1970 Performative Hypothesis). It means roughly ‘Tell me all values of the focused variable x for which P(x) is true’. In the case of a wh-question, for example (1a), the structure is roughly (1b), where the chain (where, <where>) is the variable, focused by virtue of wh-movement (Q = Q-force).
(1) a. Where did she go?
b. Q [where Foc [IP she go <where> ]]
The range of the variable is typically restricted either by inherent properties of the construction and/or contextually. In (1) it is inherently restricted to places. In addition it may be contextually restricted to, say, Malmö and Lund. The answer picks out one of these as the ‘true value’.
(2) [Lund Foc [she went <where> ]]
The answer copies the LF of the IP of the question, crucially containing a variable, which is bound by the focused place name. The answer is normally spelled out just as Lund – leaving the presupposed IP silent.Direct polar questions have the same components. The variable in this case is inherently restricted to two values: Affirmative or negative, yes or no. This is typically encoded by the Pol(arity) head. Representing variable polarity as [±Pol], the LF of (3a) is (3b), Polarity raised from IP to Spec,Foc (by ‘I-to-C’, in English).
(3) a. Does he speak Swedish?
b. Q [[±Pol] Foc [IP he <±Pol >speaks Swedish]
The affirmative answer ‘Yes’, has the structure (4):
(4) [[+Pol] Foc [IP he <±Pol > speaks Swedish]]
The IP of (4) is copied from the question, with the polarity variable, which in (4) is bound by focused affirmative polarity, spelled out as Yes. The presupposed part is typically silent. Embedded questions don’t have Q-force, but do have a focused variable. The embedded clause in (5) has the structure (3b) minus Q.
(5) I wonder whether he speaks Swedish.
There is no other ‘Q-feature’; the meaning is exhaustively determined by the higher predicate and the focused polarity variable. Focusing of [±Pol] is effected by whether-movement from Pol to spec,Foc. Since there is no Q-force, the embedded question can’t be felicitously ‘answered’ Yes or No. The same is true of speculative questions (i.e. questions not expecting an answer from an addressee) such as Danish (5):
(5) Mon ikke det bliver årets julegave?
‘Will that be this year’s Christmas present, I wonder.’
This approach can provide a characterisation of polar question types (negative questions, confirmation questions, requests, echo-questions), and of the contribution of question markers in different languages/constructions. I will also demonstrate the need for postulating an abstract higher truth-predicate in certain questions, meaning that the basic clause structure is, or can be, [Force [Truth FinP]].
Islands, case and licensing: the neglected role of the attractor
In this talk, I focus on two justly celebrated syntactic proposals that nonetheless fall short of solving the full range of problems that one might expect them to: (1) Case Theory as an explanation for the surface distribution of arguments; and (2) Phase theory as an explanation for island constraints on movement. I argue that the right supplement to Case Theory simultaneously explains certain islands — so that Phase theory, at least, turns out to need neither supplement nor revision once Case Theory is properly supplemented and revised.
Case Theory has accounted successfully for a range of restrictions on nominal arguments not found with non-nominals, but fails to predict an array of restrictions with a similar flavor that make distinctions among the non-nominals. In response to this observation, Pesetsky & Torrego (2006) have argued that case theory interacts with a distinct but closely related requirement that I will call here "Extended Licensing" — which restricts certain possibilities for clausal complementation that would otherwise be allowed by standard Case Theory.
The theory of Phases (Spellout Domains) -- when embedded in a theory in which movement is motivated by the featural properties of an attractor -- accounts for the necessity of successive-cyclicity. and predicts those island effects that can be attributed to the blocking of the phase-peripheral escape hatch by other material. Nonetheless, at least two types of islands, clausal complements to N (one case of Ross's CNPC) and subject position (Chomsky's 1973 "Subject Condition"), have received no explanation in these terms, since normal escape routes through phase-edges appear to be available in both configurations.
I will argue that given the need for successive cyclicity imposed by Phase theory and the hypothesis that movement requires a featurally appropriate attractor, the CNPC and the Subject Condition turn out to reflect independently detectable constraints imposed by Extended Licensing theory on the distribution of the attractor itself. One key argument for this proposal will come from hitherto unnoticed parallels between the distribution of phases whose head hosts successive-cyclic A-bar movement and phases that host A-bar movement that does not proceed further (such as embedded questions).
A Prelude to CP: Argumentstructure from a syntactic point of view
The purpose of this talk is to present an investigation of the theoretical and empirical consequences for argument structure of assuming a minimal, feature-driven implementation in terms of Merge and Agree of the Minimalist program (Chomsky, Pesetsky & Torrego). As my point of department I will take the observation that the maximal number of DP arguments in a single sentence is three. From a syntactic point of view, argument structure is a result of the architecture of syntax and the linking of DPs with three syntactic role families, not a result of information and processes residing within the lexicon. After a brief presentation of the particular tools that I use in my investigation, I will in the main part of my talk present and briefly discuss the theoretically possible argument structures, and show the consequences of these assumptions for cognate objects, unaccusative and ergative verbs, etc.