The Minimalist Program has witnessed a number of accounts of the simplest Merge, the most recent one forming an unordered set Merge(x,y)={x,y}. However, as mentioned in Chomsky (2013) and shown in Collins, Stabler (to appear), the free set-Merge, as well as the External/Internal Merge distinction, smuggle numerous elements of Narrow Syntax that cannot be irrelevant for economy. I argue that the general architecture can be made simpler by adapting Zwart’s (2009, 2011, 2015; Trotzke, Zwart 2014) approach to derivation. It is shown that Merge can be single-branching (and thus automatically simpler), working directly on the Lexicon, not on SOs previously selected. Consequently, Narrow Syntax can do without such elements Select and Workspace. In this regard I slightly modify the original approach pursued by Zwart showing that the assumed top-down direction of derivation and the obligatory presence of Numeration are not necessarily involved.
This paper examines language-specific differences in long-distance anticipatory vowel-to-vowel coarticulation using two languages with no lexical stress. Native speakers of Standard French (n=6) and Tokyo Japanese (n=5) served as subjects to a production study. Target tokens /ba.bV/ and /ba.ba.bV/ were embedded within a carrier sentence in each language. The word-final trigger vowel is /a/, /i/ or /u/. Acoustic analysis of continuous F1 and F2 trajectories as well as single-point formant measurements revealed opposite patterns in the two languages. Strong anticipatory long distance vowel-to-vowel assimilatory effects were observed in French. Japanese, however, displayed few statistically significant coarticulation in any vowel preceding any trigger. The results contradict existing hypotheses which propose a language-specific level of long-distance V-to-V coarticulation due to the maintenance of contrastive inventory pressure in vowel space. In that specific view, a French system with 14 vowels (compared to Japanese’s 5) should display a lesser degree of variability in V-to-V coarticulation. In addition, we interpret the current results as an indication that there seems to exist different types of contextual variability: not all phonetic (?) assimilatory effects are due to articulatory overlap. Further to this, French displays V-to-V coarticulation extending to all preceding vowels within the carrier sentence ‘Je veux le jeu ____’, not just limited within the target word. The French results run counter to all existing coarticulation models and may perhaps be better be explained by using a vowel harmony account. From an articulatory perspective, phonological rather than phonetic feature-sharing would allow the reset of all the vowel targets within the time domain. Once reset, there will be no further readjustment of the targets based on distance from the trigger vowel. During articulation, the speaker just tries to linearly realize both the consonant and vowel targets in order. As a result, both the vocalic features (assimilated) and consonantal features (unassimilated) are achieved.
I will present evidence from English, and suggest a new generalization about neg-raising predicates. Essentially, neg-raised readings are not available when embedded in modal expressions that future-shift the temporal orientation of their complement. For example, ‘John will not think it’s raining’ does not give rise to a reading in which negation is interpreted in the lower clause (‘John will think it’s not raining’). The semantic analysis is still in progress. I will explore possible explanations that relate to presupposition projection in future-shifted environments, and conditions on the modal bases selected by attitude and modal verbs.
I outline issues for existing approaches to the semantics of multiple questions in English (Dayal 1996, Kotek 2014, a.o.), focusing on how an in-situ *wh*-expression scopes out of the *question nucleus*. I depart from existing approaches, and motivate a new semantics for questions in which *wh*-expressions denote sets of alternatives in the *ordinary semantic dimension*, which are integrated via two type shifters (Charlow, 2014). I sketch an extension of this analysis to *multiply-headed correlatives* in Hindi, which display many of the same semantic and syntactic properties. I will ultimately conjecture that these two constructions share the same underlying semantic combinatorics.
The constructions in question have not received much attention in the literature, so far. It is quite surprising, since they are syntactically very interesting, for at least two reasons. The first reason is the coocurance of the clitic and the referring NP, which is not characteristic for Serbian. The second reason is the syntactic behaviour of the particles evo/eto/eno. With the data gathered from other languages,I hope I will be able to provide the right analysis of these constructions.
On one traditional understanding, cognitive scientists are in the business of positing mental states with representational content. In recent years Chomsky has sometimes appeared hostile to this conception of linguistics, insisting that the object of linguistic enquiry is a purely syntactic object, an array of linguistic properties internal to the mind. I propose to examine the status of this claim in the context of phonological theorising.
The multi-headed correlative construction is a specialized constructions found in languages which have the (single headed) correlative construction. Following is an example from Hindi.
1) jis laRkii ne jis laRke se khela, us- ne us- ko jiit gayi
which girl ERG which boy ASSOC played, that ERG that ACC won
Literally: Which girl played against which boy, that defeated that.
This has at least one reading which is roughly parallel to the English sentence below (2).
2) Each girl who played (a game) against some boy defeated that boy.
The multi-headed correlative involved a preceding relativizing clause which contains two or more pronouns which relate to the main clause through demonstrative correlates.
Relative pronouns in Hindi and other Modern Indo-Aryan languages are morphologically related to interrogative wh-phrases. Further, what seem to be unexpected readings from the multi-headed correlative are parallel to what is found in multi-wh questions such as (3).
3) Which girl played against which boy?
In this paper, we will show that multi-correlative and multi-wh questions have parallel semantic values. Further, unlike some previous analyses, the semantic value of interrogative and relativizing wh-phrases will be the same in both single wh questions and single wh correlatives.
The Japanese root “give” can appear at the PF interface as five different verbs. One of them, kureru (I neglect its honorific form kudasaru) turns out to be less permissive w.r.t. its arguments. It is argued that this verb can be base-generated in two different positions, reaching V or v head. Projecting distinct phrases, each of them provides different argument structure. Unexpectedly, unwanted readings are filtered out by means of a silent argument appearing in vP. The proposed account argues for a non-lexicalist solution, yet it does not force the SEM interface to stipulate its own argument structure.
In Indonesian, a sentence with the surface structure ‘Ali wants to be kissed by Siti’ can also mean ‘Siti wants to kiss Ali’.
This instance of control is unexpected. Reflexes of movement, properties of object voice constructions and other diagnostics reveal that Crossed Control is the result of object movement and reanalysis of fronted objects as subjects.
Based on data and judgments from Haider (1994), I’ll show a Stochastic OT
analysis of data I collected in a survey on the serialization of relative
clauses and argument clauses in the German Nachfeld (see 1).
(1) Es fiel letzte Woche einem Grammatiker auf [R der das untersuchte ] [A dass dieser Satz grammatisch ist ]
it struck a grammarian last week who analyzed it that this clause is grammatical.
(2) Es fiel letzte Woche einem Grammatiker auf [A dass dieser Satz grammatisch ist ] [R der das untersuchte ]
it struck a grammarian last week that this clause is grammatical who analyzed it.
Since Stochastic OT predicts different degrees of acceptability in different languages, I would like to collect some data from the audience.
Nikos Angelopoulos
Brennan (1993) correlates modal flavour with syntax by claiming that root modality corresponds only to a control structure and epistemic modality only to a raising structure. In this presentation, we discuss a Tagalog necessity modal that can have a raising or control structure but nonetheless takes only a root modal base. We take this fact to show that the raising-control contrast is orthogonal to flavour (cf. Wurmbrand 1999). Nonetheless, in the spirit of Brennan, we propose that syntax has an effect on the Tagalog modal’s interpretation because when it enters the control structure, it gives rise to an inference that the controller is aware of the necessity expressed by the modal. We’ll consider a few options for analyzing this inference.