Institut für Linguistik: Anglistik

Graduiertenkolleg WS 2004/05

21.10.2004 :: Jaklin Kornfilt
Types of subjects and types of clauses in Turkish:
A study of categorial features in argument and adjunct clauses



28.10.2004 :: Nadine Aldinger
Nominalizations and thematic interpretation of genitive attributes

In my dissertation I investigate the properties of German nominalizations and their sortal readings. Sortal readings are illustrated in the following examples, where the deverbal noun "Absperrung" refers to different kinds of entities or sorts:

(1) Um 18.30 Uhr war die Absperrung {des Regierungsviertels} vollzogen. (event)
(2) Die Absperrung {der Lurgiallee} läßt so manchen unverhofften Besucher scheitern. (state)
(3) Hinter den Absperrungen {der Polizei} sammelte sich im Verlauf des frühen Abends eine große Menge Schaulustiger. (physical object)

Moreover, in the sentences (1) and (2), the genitive attributes of "Absperrung" {set in curly brackets} are semantically related to the (internal) Theme argument of the verb "absperren" - denoting the place blocked off -, whereas in sentence (3), the genitive attribute is related to the (external) Agent - denoting the person or group of people responsible for the blocking. In traditional descriptive linguistics, the attribute is called an "object genitive" in (1) and (2) and a "subject genitive" in (3). Note that it seems to be impossible, or at least highly marked, to replace the object genitive in (1) by a subject genitive like in (3):

(1a) ?? Um 18.30 Uhr war die Absperrung {der Polizei} vollzogen.

Is this an effect of the sortal reading of the head noun, "Absperrung"? What (other) parameters decide if a genitive attribute will be more readily interpreted as a subject or object genitive? I collected nominalizations and their genitive attributes from a large corpus of newspaper text and will present the phenomena and some preliminary explanations in my talk; some general problems with sortal readings and their identification will also be discussed.


04.11.2004 :: Tom McFadden
The distribution of subjects in embedded clauses: an account without Case

In recent versions of Minimalism, the role of syntactic Case has been greatly reduced in favor of the EPP. What does remain of the old Case Filter has been retained largely to account for the treatment of subject positions in embedded clauses. In this talk I will argue that the assumption of syntactic Case offers no real insight here either, and that a theory making full use of the EPP and other independently necessary principles of selection provides a better account of the data. The postulation of normal Case, null Case and the lack of Case in specific environments does nothing more than restate the distribution of overt DP subjects, PRO and raising. There is no independent motivation for the distribution of the Case types, least of all from the morphology, given recent demonstrations of the independence of morphological case from positional DP-licensing. Furthermore, there are sets of data that the Case-based theories do not even describe correctly, from the complements of "want" class verbs to clausal gerunds. In particular, I will attempt to show that the environments where overt subjects are disallowed constitute a natural class, whereas the environments where they are allowed does not -- precisely the opposite of what Case theory would lead us to expect.

I will thus propose that we should reverse our take on the distribution of overt DPs. Rather than assuming that they are illicit by default and are required to get into a relationship that can license them (i.e. the assignment or checking of Case), I will hypothesize that they are licit by default in the thematic positions where they are introduced. I will show that the deviations from this default can be accounted for by means of the EPP and its interaction with locality restrictions, selection and constraints on the spell-out of the non-finite complementizer "for". Crucially, I will argue that the assumptions and stipulations necessary for such an account (in particular in dealing with the distribution of PRO) are simpler than those necessary for the standard Case-based account.


11.11.2004 :: Kirsten Gengel
Pseudogapping, Focus, and Comparatives

In this talk I discuss the ellipsis phenomenon of Pseudogapping, illustrated in (1) and (2):

(1) Mary invited John, and Tarzan did _ Jane.
    (noncomparative Pseudogapping)

(2) We treat our apprentices better than they do _ their career employees.
    (comparative Psg)

It has been noted that Pseudogapping is more acceptable in a comparative frame (Levin 1978, 1986). This suggests that comparison and contrast may be a factor which needs to be introduced into a theory of Pseudogapping. As I argued elsewhere (Kleinwalsertal, July 2004), Pseudogapping in non-comparative sentences is derived via focus movement. A (strong) focus feature is taken to induce movement of the NP or PP remnant out of the VP in a focus position. Afterwards, the VP is deleted, as illustrated below in (3), the structural representation of (1).

(3) ... [CP Op [TP TarzanSUBJ did] [FocP JANE] [vP tSUBJ [VP invite tOBJ]. This theory may account for the contrastive focus (as indicated with capitals in (3)) on the Pseudogapping remnant.

As ellipsis cases mostly occur in coordinate structures, and as it is possible to view comparative structures as coordinate structures, I propose that these cases should not be treated as distinct structures. In particular, as comparatives inherently possess a strong contrastive structure, the similarity to the focused coordinate structures seems obvious.

The goal of this talk is to illustrate the parallels between the ellipsis cases in comparative and non-comparative structures, and to promote a unified analysis of both types of structure.

Building on the assumption that in non-comparative ellipsis structures, the contrasting elements bear focal stress, I will propose that there exists a syntactic/semantic link between the comparative operator in comparatives and the focus operator in non-comparative structures.

Selected references:
Fox, D. (1999): Focus, Parallelism and Accommodation. SALT 9.
Kennedy, C. (1999): Projecting the Adjective: the Syntax and Semantics of Gradability and Comparison. New York: Garland.
Lasnik, H. (1995) A Note on Pseudogapping. In: R. Pensalfini and H. Ura (eds) Papers on minimalist syntax, MIT Working Papers in Linguistics 27, 143-163.
Lasnik, H. (1999) Pseudogapping Puzzles. In: E. Benmamoun and S. Lappin (eds), Fragments: Studies in ellipsis. Oxford: Oxford University Press.
Lechner, W. (2004): Ellipsis in Comparatives. Berlin: Mouton de Gruyter.
Levin, N. (1978): Some Identity-of-Sense Deletions Puzzle Me. Do They You? CLS 14, 229- 240.
Levin, N. (1986): Main Verb Ellipsis in Spoken English. New York: Garland.
Merchant, J. (2002): Subject-auxiliary Inversion in Comparatives and PF Output Constraints. In: K. Schwabe and S. Winkler (eds.) The Interfaces: Deriving and Interpreting Omitted Structures, 55-77. Amsterdam/Philadelphia: Benjamins.


18.11.2004 :: George Tsoulas
On two recent accounts of coreference and their implications for grammatical architecture

The question of the integration of the theory of anaphoric binding and pronominal coreference in the minimalist architecture has recently given rise to an array of diverging apporaches which, nevertheless, share a certain number of assumptions and goals. Interestingly, due to the nature of the object the different theories proposed for capturing the notions of binding and coreference with minimalist tools and within a fairly disciplined minimalist framework show a remarkable divergence in their implications concerning the architecture of grammatical theory. Elucidating these implications for two such recent approaches will be my purpose in this talk. Although there have been a number of approaches in recent years I will concentrate here on the views expressed by Kayne(2002) and Zwart(02). Both authors are led to the conclusion that a large set of data for which traditional binding theory was designed can and should be derived by movement. This is not a surprising conclusion given the basic assumptions of the MP. What is more interesting is that they are led to diametrically opposed conclusions in what concerns the notion of accidental coreference (see Lasnik 1976). On the one hand Kayne concludes that there is no place for accidental coreference in the theory of grammar, whereas Zwart's investigation leads him to conclude that all pronominal anaphora is a case of accidental coreference. My aim here is not to compare these two theories in order to pronounce one of them correct. I think the two approaches are in fact representative of two deeply different conceptions of the architecture of the grammar, namely Kayne's view requires some sort of meaning which is mapped onto linguistic expressions whereas Zwart's view is consistent with the idea that the output of the syntax should be delivered to the interfaces and one ought to be content with them doing anything they please with it. Having established this I will then turn to some empirical issues regarding both theories. Finally, time and wit permitting, I will turn to an alternative model.


25.11.2004 :: Björn Rothstein
On (imperfect) perfect theories

Post state and extendednow approaches to the perfect seem to be incompatible. According to the former, the present perfect introduces a post state emerging immediately from the eventuality in the perfect (cf. KAMP & REYLE (1993)). According to the latter, the present perfect introduces a perfect time span (=PTS) that starts somewhere in the past and ends - roughly speaking - at the moment of speech (S). The event time (E) is part of PTS (cf. McCOARD (1978)). The GOAL of this paper is to show that a combination of both approaches is necessary to account for German.

THE PROBLEM: It is possible to ask how long with states and activities, but not with achievements, cf. (1). When used in the perfect how long, however, does combine with achievements. Given that the eventuality has culminated, the perfect must denote a lexical post state and a post state analysis such as [PRESENT (POST STATE (EVENTUALITY))] becomes necessary. But in (2), there is no lexical post state. This points toward two independent perfect meanings. A problem for post state analysis is (3) as it requires the eventuality to happen in 1970 which is not necessarily the case. The perfect must therefore introduce a time span, an extendednow, that starts in 1970 and in which (E) and the post state are contained. Hence, only a combined post state & extendednow can account for the present perfect.

THE ARGUMENTATION: It is often claimed that PTS ends - roughly speaking - at (S), but in German, PTS can seperated from the moment of speech (S). Immer requires the eventuality to hold throughout the entire PTS (cf. IATRIDOU et al (2001)). As in (4), immer does not reach up to (S), PTS ends before (S). According to PANCHEVA & STECHOW (2004), the boundaries of PTS are vague. (E) is defined as being a subinterval of PTS. Assuming a vague PTS, however, complicates the analysis of temporal narration as it requires restrictions of PTS by discourse. To give a simpler account, I modify PTS by saying that PTS is a dynamic time span whose length varies due to context. In the default, it is identical with the event time (E) as there should be no reason to assume a PTS distinct from (E) in cases like (5).

KATZ (2003) claims that lexical post states are individual states, but only the former allow positional tense adverbials (cf. (6) vs. (7)). The post state is only licensed if there is an eventuality that has culminated and if there is a point in time distinct from (E) available to license the post state. I claim that this point is the right boundary (RB) of PTS. This allows a uniform meaning of the German present perfect. The availability of stative and non-stative readings depends on the length of PTS. If RB is the final subinterval of (E), the post state is not licensed and the perfect has a non-stative reading. If RB is distinct from (E), the post state is licensed and the perfect is stative.

(1) Wie lange hat Hans die Formel schon entdeckt? / *entdeckte (MUSAN (2002))
How long has Hans the formula already discovered? / discovered
(2) Als ich heimgekommen bin, hat er gekocht.
When I homecome have, has he cooked.
(3) Dieter hat den Präsidenten seit 1970 getroffen (STECHOW (2002))
Dieter has met the president since 1970
(4) Ich habe hier immer gewohnt, ~Ebis vor kurzem (PANCHEVA & STECHOW (2004))
I have here always lived ... until recently
(5) Der Gummiball ist auf dem Boden aufgekommen und wieder hochgesprungen.
The rubber-ball is on the ground hit and again jumped.
(6) Maria ist jetzt bereits abgeflogen
Maria is now already taken-off
(7) *Sie ist jetzt großwüchsig
She is now tall


02.12.2004 :: Martin Everaert
SE-Reflexives and Theta Manipulation

In the discussion about the unaccusative-unergative distinction, verbs of the type zich vergissen 'to be mistaken', zich manifesteren 'to manifest (itself)', zich bukken 'to lean (over)' - intrinsic reflexive verbs/inchoative verbs - have generally been taken to be unaccusatives (Burzio 1981 for Italian, Everaert 1986 for Dutch, Haider 1985 for German). In a recent paper, Reinhart & Siloni (2004) argue against that position. Their approach is couched in a specific version on the lexical representation of unaccusatives (Chierchia 2004), the lexicon and the lexicon syntax mapping (Reinhart 2002, Reinhart & Siloni to appear).

In my talk I will return to my original considerations (i) to analyse Dutch inherent reflexive constructions as unaccusatives, (ii) to generalize over unaccusative verbs and reflexive verbs and (iii) to grant SE-elements a special status of accusative case absorbing elements, and will review them in the light of the Theta System. To what extent are these considerations still valid in the Theta System approach? How might the empirical intracacies of these constructions be captured in the Theta System approach? And how does this fit into Reinhart & Siloni's (to appear) approach to parametrization of reflexive SE-verbs (cf. also Papangeli 2004)?

Alexiadou, A, E. Anagnostopoulou and M. Everaert (eds) (2004) The Unaccusativitiy Puzzle: Studies on the syntax-lexicon interface, Oxford University Press; Burzio, L. (1981) Intransitive verbs and Italian Auxiliaries, PhD diss MIT; Chierchia, G. (2004) A Semantics for Unaccusatives and its Syntactic Consequences, in A. Alexiadou et al.; Everaert, M (1986) The Syntax of Reflexivization, Mouton; Everaert, M. (2002) The case of the Theta System, Theoretical Linguistics 28; Haider, H. (1985). Von sein oder nicht sein: Zur Grammatik des Pronomens "sich". In W. Abraham, ed., Erklärende Syntax des Deutschen; .Papangeli, D. The Morphosyntax of Argument Realization: Greek Argument Structure and the Lexicon-Syntax Interface, Doct. Diss Utrect, LOT Diss series 86; Reinhart, T. (2002) The Theta System: an Overview. Theoretical Linguistics 28; Reinhart, T. and T. Siloni (2004) Against an unaccusative analysis of reflexives, in A. Alexiadou, et. al.; Reinhart, T. and T. Siloni (to appear) Thematic Arity Operations and Parametric Variations, Linguistic Inquiry.


09.12.2004 :: Britta Sauereisen
On the intersectivity vs. non-intersectivity of adjectives

In the literature, adjectives have traditionally been divided into intersective and non-intersective ones. However, it would be desirable to have only one category, either treating all adjectives as intersective or as non-intersective. A non-intersective analysis of adjectives has already been proposed in the Montague tradition.
In this talk, I will discuss the possibility of interpreting all adjectives as being intersective in nature. This discussion will be based on the following examples:

  (1)  a red book
  (2)  a red onion
  (3)  a big lion
  (4)  an alleged murderer
  (5)  a former president
  (6)  a fake pistol
  (7)  a beautiful dancer

My analysis will rely on a more complex structure for the adjective itself, and a less restricted meaning for the noun it modifies.

Selected references:
Alexiadou, A. (2004) 'Patterns of adjectival modification'. Talk given at the 30th GGS meeting at the IDS Mannheim.
Kamp, H., and P. Partee (1995) 'Prototype theory and compositionality'. Cognition 57: 129-191.
Larson, R. and G. Segal (1995) Knowledge of Meaning: Cambridge: MIT Press.


16.12.2004 :: Roger Schwarzschild
Association with only, Givenness and Anaphora Resolution

Association with focus is usually viewed as a rule governed phenomenon whose inputs include the focusing operator (eg only) and the scope of that operator along with any focus-features it may contain. On this view, (a) John only ATE lunch differs in meaning from (b) John only ate LUNCH because (a) and (b) differ in the distribution of focus features in the scope of only; a difference that is manifested in the placement of the nuclear accent, here indicated with capital letters.

Anaphora Resolution and Givenness. The following sentences differ with respect to whether the pronoun receives main stress:

(1) Peter spoke to John and then he spoke to George.

(2) Peter spoke to John and then HE spoke to George.

We tend to understand the pronoun as referring to Peter in the first case and to John in the second case. This can be explained by assuming that in each case, the second clause is meant to contrast with the first and that focusing marks just the points of contrast. In effect, rules for focusing conspire with a free assignment of values to pronouns to achieve the result. There is no particular rule that needs to tie the two together.

Being a quantifier, only is context sensitive though in a more complicated way than a personal pronoun. The context constrains the domain of the quantifier, not it s value. Nevertheless, if different focus configurations correlate with different choices of an antecedent for a pronoun, it should also correlate with different choices of domain for a quantifier. For the pronoun, there is no rule of grammar that enforces these correlations, but in many cases, it seems as if the rule of Association with Focus is doing that for the quantifier only. What does that tell us about the status of such a rule? Do we actually need the rule in the first place? Various authors have considered an alternative idea whereby focus reflects the nature of the discourse the sentence is uttered in and it is coherence within that discourse that produces the effect of Association with Focus.

To test this hypothesis we need a discourse in which coherence would require one domain for the quantifier while the rule of Association with Focus requires another. I offer two such discourses with conflicting results. In one case, incoherence ensues, in the other, the rule of Association with Focus appears to be violated.

A conservative conclusion can be drawn: our proposed rule of Association with Focus has to be weakened to cover the case where the stronger variant of the rule is violated. I explore a different solution, which starts by treating only as a scalar particle (eliminating alternatives that are highly ranked on a relevant scale). Under that view, I show that in the examples discussed, Association with Focus becomes a direct result of the fact that the entire clause containing the only contrasts with previous utterances. This still leaves open the question of the status of a rule of Association with Focus, especially in cases where only does not appear to be scalar.


27.01.2005 :: Florian Schäfer
The dative causer construction in German

In this talk I deal with the different interpretations that a free dative can have in German change of state contexts. As is well known, German has two classes of decausative verbs, one that is morphologically unmarked (cf. (1)) and one that comes with a reflexive pronoun in the intransitive use (cf. (2)).

(1) a. Hans zerbrach die Vaseb. Die Vase zerbrach
(2) a. Die Menschheit veränderte das Klima   b. Das Klima veränderte sich

A free dative with verbs of class (1) can be interpreted either as affected by the change of state or as the unintentional causer of the event of change (cf. (3)).

(3) Dem Hans zerbrach die Vase
   'The vase broke and Hans was affected by this'
   'Hans unintentionally caused the vase to break'

But with class (2) verbs only the former interpretation is available. The unintentional causer interpretation is missing (cf. (4)). In this respect the German reflexive construction differs from its counterpart in e.g. the Romance languages.

(4) Der Menschheit veränderte sich das Klima
   'The climate changed and the mankind was affected by this'
   'The mankind unintentionally caused the climate to change'

In my last Kleinwalstertal talk I showed that with respect to all tests for causation the two types of decausative verbs behave alike. Hence the blocking of the unintentional causer reading cannot be due to the presence of a causative meaning component with class 2 verbs (where it could be assumed that the reflexive pronoun is the external/causer argument).

Here I give a syntactic account for both readings. I assume that in both cases the Dative DP is located in the specifier of an applicative head where it is assigned inherent case. The affected dative is introduced by an applicative head within the VP that relates the Dative DP with the theme's result state (cf. Cuervo 2003). Following Travis (2004), I assume that Agents and Causers are base generated in different positions. (Intentionally acting) Agents are introduced by agentive Voice, while Causers originate in a lower position, namely within the VP that constitutes the change of state event. Causers have to leave their base position for case licensing reasons; they can either move to the specifier of a Voice head, if this head is part of the numeration (transitive structures (1)), or otherwise to the specifier of an applicative head which is located above the change of state VP (dative causer reading of (3)). I take class (2) decausatives - although semantically intransitive - to be syntactically transitive because they project two DPs that are marked with nominative (theme argument) and accusative (sich) respectively. This means that besides the theme position within the VP these verbs project a further DP position in the specifier of a non-thematic Voice head. This non-thematic specifier (which either hosts the expletive sich or alternatively attracts the theme whose trace gets spelled out as sich; both options will be discussed) intervenes between the Causer's base position within the change of state VP and Spec, ApplP on top of vP. As a result, the movement of the Causer is blocked by minimality.

Finally, I argue that the German vs. Romance difference with respect to the unintentional causer reading is due to a difference in the status of the respective reflexive elements. German reflexives are full pronouns occupying DP positions, while their Romance counterparts are clitics occupying head positions (hence no minimality problem arises).


03.02.2005 :: Matthew Stone
Interpreting Vague Utterances in Context

We use the interpretation of vague scalar predicates like "small" as an illustration of the ability of systematic semantic models of dialogue context to derive useful, fine-grained utterance interpretations from radically underspecified semantic forms. Our account involves two principles. First, we model pragmatic reasoning as a general process that infers consistent collaborative intentions to explain agents' contributions to joint activity. Second, we interpret vague predicates by recovering salient scales and relevant distinctions along them from the dialogue context. Given this framework, we can infer implicit standards of comparison for vague scalar predicates through completely general pragmatics, yet closely constrain the intended meaning to within a natural range. Our account connects closely with dynamic models from formal semantics, but we have implemented it exactly in a natural language interface for describing spatial actions.


10.02.2005 :: Christian Hying
Qualitative Spatial Models for the Map Task

The talk is about the application of qualitative spatial models in the semantic analysis of the HCRC Map Task corpus.

The HCRC Map Task corpus contains a series of dialogues where one participant describes a route on map to another participant. In order to interpret route directions and localisation phrases against the maps of the instruction giver and the follower, respectively, we need to bridge the gap between symbolic formalisms of natural language semantics and two-dimensional, geometrical representations of the maps. Qualitative spatial models serve this purpose. By means of such models we can represent geometrical representations as logically transparent representations which are compatible with the methods of formal semantics.

I will present the generation of a qualitative spatial model from a geometrical representation of a map and I will discuss the mapping of natural language predicates onto the predicates of the model. The discussion of examples from the corpus shows that there are different situations which can only be explained satisfyingly, if the analyses are based on models of different granularity. Due to this observation I will propose pragmatic constraints on the interpretation process controlling the selection of a particular model.


17.02.2005 :: Giuseppina Rota
Cognitive processing of German "garden path" sentences and emotional prosody: insights from two brain imaging studies

The biofeedback based on functional Magnetic Resonance is a technique which allows regional brain activity self regulation. Two linguistic tasks have been selected to assess the improvement in linguistic performance, which is expected as a result of brain regional up-regulation. The first task is a "speeded grammaticality judgement" (Meng, 2000) test with 58 different declarative sentences presented visually. Some of these sentences elicit garden path ambiguities, some are grammatical incorrect, and others are correct. I will present the fMRI results of that study conducted on ten subjects. During the second task subjects had to identify emotional prosody of 58 declarative sentences presented auditory. The semantics and the prosody of these sentences (Breitenberg, 1996) describe happy, sad, angry or neutral scenarios. In my talk I will present fMRI data on the emotional processing which takes place in the brain when people identify prosodic intonation.

Meng, M., & Bader, M., (2000). Mode of disambiguation and garden path strength: An investigation of subject-object ambiguities in German. Language and Speech 43, 43-74.

Breitenstein, C., Daum, I., Ackermann, H., Lütgehetmann, R., & Müller, E., (1996). Erfassung der Emotionswahrnehmung bei zentralnervösen Läsionen und Erkrankungen: Psychometrische Gütekriterien der "Tübinger Affekt Batterie". Neurologie & Rehabilitation 2, 93-101.