[print]

Institut für Linguistik: Anglistik

Graduiertenkolleg - WS 2005/2006

16.02.06 :: Florian Schäfer
A mute Voice for some Middles

Besides their Generic semantics, Middle sentences as in (1) have two properties that make them look like passives: (i) as in passives (2) the nominative subject in (1) corresponds to the object of the transitive sentence (3); (ii) as in passives the sentence is understood as involving an implicit agent. However, research has shown that the implicit agent in Middles differs from the one in passives in that only the implicit agent in passives shows syntactic effects (by phrase, agentive adverbs, control,...).

(1) This book reads easily (*by Peter) middle
(2) The book was read (by Peter) passive
(3) John reads the book active

This was taken as evidence that passives are formed in the syntax but middles in the lexicon. In the course of the Middle formation the agent is present in the lexicon but blocked from projecting into the syntax (e.g. Ackema & Schoorlemmer 1994). Such theories further argue that the remaining theme argument is merged externally, i.e. that middles are unergatives.
The syntactic inactivity of implicit agents in middles is a huge challenge for non-lexicalist theories. I will develop such a non-lexicalist theory of middle formation. My proposal will be developed on the basis of the German middle (4) and will be transferred to other languages (English, Dutch, French, Greek). I assume that in middles the agent is not merged. The understood implicit agent is the result of the agentive root semantic of the verb involved in middle formation. Syntactically, middles are reflexive unaccusatives as in (5). In these structures the theme is merged internally and the reflexive is merged as an expletive in the Specifier of a non-thematic (mute) Voice projection. The internal merging position of the theme can be diagnosed by a careful application of some unaccusativity tests. Roots like 'lesen' (read) are not allowed to form eventive anticausatives (6); this is a reflex of their encyclopaedic meaning which implies a driving force behind the activity. However, in a Generic context this interface condition can be circumvented (4); 'lesen' can form a generic anticausative (but has to keep its syntactic transitivity via the expletive reflexive pronoun). The implicit agent is the result of our conceptual knowledge about reading-events.
A consequence of this view is that Generic unaccusatives like (7) which neither imply an implicit agent nor have a Voice projection, are Middles, as argued for in Condoravdi 1989.

(4) Das Buch liest sich leicht middle
(5) Die Tür öffnet sich gerade reflexive anticausative
(6) *Das Buch liest sich gerade reflexive anticausative
(7) Eis schmilzt leicht generic unaccusative
19.01.06 :: Agnès Tutin
Collocations in the semantic field of emotions

In this talk, I will deal with collocations in the semantic field of emotions. I will first present an overview of linguistic properties of collocations. I will then examine more specifically the semantic properties of adjectival and verbal collocates of emotions. We will see that these lexical associations show surprisingly little idiosyncrasy. This leads us to advocate an approach where collocational associations can be explained - and partly predicted - according to the combinatorial properties of collocates.
12.01.06 :: Arndt Riester
On the Semantics of Scales and Exhaustivity

First, I will present some proposals concerning the semantics of scales. These have been claimed to play several roles within focus semantics: on the one hand, focus is able to license Gricean quantity implicatures and, on the other hand, the focus-sensitive particle 'only' is assumed to exhibit two readings -- a quantificational and a "scalar" one.
The question will be discussed whether there is indeed reason to assume these two readings or whether they should better be reduced to a single one, while the semantic ambiguity of 'only'-sentences is due to different kinds of scales associated with the focused element in the scope of 'only'.
Against the background of this scenario I want to develop a new perspective on the notion of exhaustivity, the various disguises in which it occurs, and how different elements conveying exhaustivity can interact.
22.12 :: Regine Eckardt
Or: Free Choice and Nominal Quantification

In this talk, I will revisit the so-called "free choice puzzle" that was raised in early writings by Hans Kamp and has received renewed interest in the last years (Zimmermann 2000, Geurts 2005, Simons 2005, Kratzer and Shimoyama 2002). I will argue that the so-called free choice effects are not restricted to the modal domain. This has consequences for the nature of any successful analysis of "free choice" effects. Further explorations will delimit the similarities and differences between nominal quantification and quantification by modals.

 

20.10.2005 :: Jean-Pierre Koenig
(Proto)-instrumentality and the meaning of English verbs

The syntactic and semantic status of instruments is unclear. For some researchers, (Schuetze and Gibson (1998)) they are arguments; for others, (Dowty (1989)), they are adjuncts; for yet others, they are somewhere between arguments and adjuncts (Larson, 1987; Van Valin and La Polla, 1997). There is no agreement either as to the semantic entailments that define the instrument participant category. Nor is there agreement on whether the need to syntactically distinguish different kinds of instrumental phrases corresponds to a semantic distinction or not (Marantz, 1985; Van Valin and LaPolla, 1997). In this talk, I will discuss the results of two comprehensive surveys of 4,000 English verbs, one quantitative and one qualitative. The results of these two surveys show that:

1. Some verbs (about 12%) require the presence of an instrument in situations they felicitously describe; some verbs (about 30%) permit, but do not require the presence of an instrument in situations they felicitously describe;

2. This distinction between obligatory and optional instrument verbs correlates with the specificity of the instrument role information that verbs encode: Obligatory instrument verbs circumscribe the properties of their instruments much more than optional instrument verbs do;

3. Obligatory instruments are arguments of the verbs that require them; optional instruments are adjuncts of the verbs that permit but do not require them;

4. The semantic space of verbs that require or allow instruments can be subdivided into two dozens or so coherent semantic subclasses. In each of these subclasses, the instrument plays a slightly different role in the causal event denoted by the verb and the additional instrument properties help define the meaning of each semantic subclass;

5. The causal role of instruments differs across these semantic subclasses and exceeds the range of causal notions countenanced in works such as Talmy (1985) or Jackendoff (1990), requiring the introduction of a non-intentional notion of "helping".

Finally, I will present results of on-line sentence comprehension experiments that independently support points 2. and 4.

 

10.11.05 :: Giuseppe Longobardi
On the Form and Functioning of Mapping Parameters

In this presentation, I will explore how we may begin to design a restrictive minimalist theory of parametric variation and will argue that such an approach, based on considerations of simplicity and conceptual necessity, may naturally lead us to also attain a higher degree of the classical explanatory adequacy in one intricate area of the syntax/semantics interface.
In this spirit, it will be suggested that most or all syntactic parameters fall into a very limited set of recurrent formal schemata or formats, which may be instantiated in the same way on a huge number of features or syntactic categories variable from language to language. Each of these formats might in principle be conceptually justifiable and evolutionarily understandable with respect to one of its instantiations and could be regarded as later extended to other categories, so explaining the proliferation of parametric variation through grammars. It will be argued that among the most pervasive schemata of parametric variation are formats like the following: is a certain feature F grammaticalized or not in language L? Is a certain grammaticalized feature strong or not in L? Many syntactic features appear to be crosslinguistically polymorphic precisely along such dimensions, which are actually reminiscent of proposals by Kuroda (1992) and Huang (1982), respectively.
A specific version of this framework will be applied to the controversial domain of nominal mapping parameters (Chierchia 1998). Since Longobardi (1994), it has been argued that in Romance there is a unique and syntactically specifiable position, traditionally labeled D, serving as the locus for reference to individuals (either type of Carlsonian individuals, namely kinds or objects) in the structure of nominal phrases (Topological Mapping Theory, TMT). Thus, only referential constants (essentially lexical nouns) overtly associated with D seem to be able to refer to kinds (common nouns) or objects (proper names). In particular, a D remaining lexically empty cannot be a referential constant and will imply the interpretation of the DP as a variable. The existence of languages exhibiting a way of pairing nominal structures with interpretations which is superficially different from that identified in Romance points toward the need for some parametric theory of nominal mapping. In this paper I want to address two questions:

1) Why is precisely D specialized as the position for reference to individuals?
2) Why are there languages with apparently different mapping strategies?

Under the restrictive theory of phrase structure proposed by Chomsky (1995, ch 4.), requiring each syntactic object to consist of at least one feature interpretable in the A-P or C-I system (i.e. not just of formal, uninterpretable features), the first question will be addressed by identifying D with the Person head (cf. Platzack 2004, Bernstein 2005) and by arguing that the latter feature is a crucial ingredient to allow type-shifting from property- to individual-denotation. Under the announced minimalist theory of parameter formats, the second question will be addressed by showing that all and only the possible polymorphic realizations of the feature Person admitted by such a theory (+grammaticalized Person, + strong Person) are crosslinguistically instantiated and determine exactly the three major types of languages so far observed with respect to nominal mapping phenomena, saliently exemplified, among other varieties, by Japanese, English, and Italian.

 

17.11.05 :: Carmen Dobrovie-Sorin
The Genericity of French Plural Indefinites and the Domain of Adverbial Quantification

Theoreticians of genericity agree that the generic reading of indefinite expressions is dependent on the genericity of the sentence: the indefinite contributes a variable that gets unselectively bound by an overt or covert adverb of quantification. The unselective binding hypothesis is confronted with various empirical and conceptual problems, which have led to the view that Q-adverbs are to be analyzed uniformly as selective quantifiers over events or situations (Rooth (1985, 1995), Schubert & Pelletier (1987, 1988), de Swart (1991, 1996), Krifka & alii (1995), Krifka (1995), etc.). Those configurations in which Q-adverbs seem to bind (one or more) individual variables (or rather an n-tuple of individual variables) would constitute a side-effect of quantification over events: quantifying over events induces quantifying over the tuple of participants to the event. Given this analysis of Q-adverbs, a generically-interpreted indefinite is indirectly bound by a Q-adverb that quantifies over events.

In this talk, I will diverge from the above-mentioned authors in assuming that Q-adverbs may directly quantify not only over events, but also over individuals. The distinction between the two possibilities will prove to be crucial for the analysis of the generic readings of plural des-indefinites in French.

 

24.11.05 :: Michèle Goyens
The impact of syntactic variation: bare nominals vs prepositional phrases from Latin to French.

The distribution of French prepositional constructions is threefold: some prepositions are chosen by the speaker according to semantic values and can thus be replaced by others (parler ` vs avec vs de vs pour quelqu'un), some are obligatory and may not be replaced by others nor left out (compter sur quelqu'un, s'intiresser ` quelque chose vs *compter ` quelqu'un, *s'intiresser sur quelque chose, *s'intiresser quelque chose), others finally can be used next to others, without real apparent semantic implications (habiter `/dans/x Paris).
From an historical point of view, the distribution of French prepositional constructions has been subject to important modifications. A great deal of French verbs which used to be construed with a direct object have developed a prepositional construction (aspirer quelque chose > aspirer ` quelque chose), or vice versa (contredire ` quelqu'un > contredire quelqu'un). Lots of verbs changed prepositions (ichapper de quelque chose [vx] > ichapper ` quelque chose), while others develop several prepositional constructions (s'excuser ` quelqu'un > s'excuser auprhs de, envers quelqu'un).

In this seminar, I would like to focus on the development of Latin verbs with dative complement, in order to show the lack of regularity in the evolution of these constructions: some dative complements develop into prepositional constructions with `, which seems the most "standard" evolution. Others however evolve in a more "irregular" way: the most crucial period in this matter seems to be Middle French and the 17th century. What is to be examined is the semantic implication of these changes: is there a modification in the signification of the verb, or are these constructional changes mere syntactical phenomena?

 

01.02.06 :: Hubert Haider
What it means to be VO

There is a syndrome of grammatical properties that characterizes VO and OV languages, respectively:

1. Complex head-initial projections involve shell structures, complex head-final ones do not. 2. A VO-language may have quirky subject constructions, an OV language does not allow this. 3. Subject expletives are mandatory in VO languages, but crucially not in OV. 4. Head-initial structures are /compact/, head final ones are not. Corollary: Scrambling applies only in head-final structures. 5. VO languages with particle stranding may strand a particle between objects, OV languages do not permit this. *6*. Complex head final verbal projections cluster (with clause union effects), with verb order variation in the cluster. There is no verb oder variation in head-initial verbal projections.

These syndromes follow from a single parametric difference (directionality) in an otherwise universal system of phrase structure. The universal properties and the parameter option becomes understandable once universal grammar is put into a usage perspective of grammar, namely in mental language processing.

 

01.12.05 :: Kirsten Gengel
Contrastivity and Deletion: A Feature-Based Account of Ellipsis

In this paper I put forward a uniform account for ellipsis (specifically, VP ellipsis (1), Pseudogapping (2), Gapping (3), and Sluicing (4)), based on Merchant's (2001, 2004) E-feature, and implemented in terms of contrastivity.
(1) Claire read a book, and Heather did too. (VP ellipsis)
(2) Claire read more books than Heather did magazines. (Pseudogapping)
(3) Claire read a book and Heather a magazine. (Gapping)
(4) Claire read a book but I don't know which. (Sluicing)
The E-feature is based on the notion of E-Givenness, modelled on Schwarzschild's (1999) Givenness, thereby introducing a relation to focus in ellipsis structures, since a phrase can be deleted only if it is e-GIVEN (cf. Merchant 2001), i.e. if it has a proper antecedent.
Building on his analysis for Sluicing, I propose that it be slightly modified to account for other ellipsis types as well, in its interaction with contrastivity. I suggest first (following Merchant) that the E-feature is placed on the head that is redundant with respect to the antecedent, and subsequently to be deleted. I assume further (contra Merchant) that the E-feature triggers the deletion of the maximal projection of the head in question. That this implementation is possible is shown in the Pseudogapping example in (5), which employs an additional feature, the EPP feature (in Lasnik's 1995, 1999 terms).
(5) ... than [TP Heather [T did [AgrOP[EPP] magazinesi [VP [V[E] read [NP ti]]]]]] I show that it is not the EPP which is involved in Pseudogapping, but that the E-feature may interact with another feature, namely a contrastive feature on the element contrasted with its antecedent. This feature (a) captures the impossibility of deleting focused or new material, and (b) triggers movement of contrasted elements out of the phrase marked for deletion. This is not only applicable for the Pseudogapping structure in (5), but also holds for Sluicing, if we consider the moved wh-element to be somewhat contrastive to (or, at least, less specified than) its counterpart in the antecedent. Moreover, the combination of the E-feature with a contrastivity feature allows us to treat gapping as ellipsis, more specifically, TP deletion, and to account for the need of a contrastive subject in the Gapping cases in (6).
(6) a. Claire read a book, and Heather a magazine.
b. *Claire read a book, and she a magazine.
c. Clairei read a book, and SHEk a magazine.
(7) ... and [TopP Heatheri [FocP a magazinek [TP ti [T[E] [AgrOP tk [VP [V[E] read [NP tk]]]]]]]]
Hence, the modification of the mechanism invoked by the E-feature, in combination with a contrastivity feature on contrastive elements, permits a uniform picture of various types of ellipsis.