Predicting transitivity alternation
Recent developments which revive central elements of Generative semantics assume that no verb is intrinsically causative and causativity enters into a derivation through a functional head (e.g. Pesetsky 1995). This paper proposes, on the other hand, that verbs vary in whether causativity is a lexical property of the verbal root, or verbs are combined with the zero affix "cause" as the result of syntactic incorporation. Given this assumption, this paper develops an analysis of the transitivity alternation, which assumes that verbs can enter freely into different syntactic structures, whereas ill-formed projections are ruled out using tools of compositional semantics and semantic selection.
Cleo Condoravdi & David Beaver
Modification: an Event-free Neo-Davidsonian Semantics
Davidsonian semantics in Davidson\'s original proposal and subsequent work incorporates two main innovations. The first is the use of an ontological category additional to the Carnap/Montague inventory, that of events. The second is an analysis of predication and modification which allows a predicate with a fixed arity to take variable numbers of arguments. Though often conflated, we will demonstrate that these two ideas are independent by reconstructing events in terms of assignment functions. The resulting event-free version of Davidsonian semantics has the following advantages over standard Montague Grammars: (1) it gives a uniformly typed interpretation for all projections of the verb, (2) it gives a uniform type for all modifiers, and (3) it allows a movement-based account of scope without traces.
Scalar complexity and the structure of events
The aspectual structure of change-of-state events often correlates to privileged participants in the event (e.g. incremental themes, paths, cf. Krifka (1989), Tenny (1992)). In this paper I generalize these correlations and argue that the relevant factor for understanding aspectual complexity (primarily durativity) is the complexity of the scale of change, where the two complexities are highly correlated. Intuitively, this explains the difference between achievements, which are punctual and involve simple transitions, and accomplishments, which are durative but involve gradable transitions. This correlation also explains cooccurrence restrictions between certain aspectual classes of verbs and result/goal XPs in resultative constructions (for instance non-gradable AdjPs may only cooccur with punctual verbs, e.g. "John shot/*beat Bill dead" (Wechsler 2002)). I provide a mereological account of the complexity correlation (following Krifka (1998)) by assuming that all change-of-state verbs have scale arguments and that event and scale arguments stand in homomorphic relationships preserving mereological complexity. Different lexemes impose different sortal constraints on the mereological complexity (simplex or complex) of their event or scale arguments, and the homomorphic mapping guarantees the correlation between aspectual and scalar complexity. The generality of this approach allows for a wide-range of complexity generalizations to be stated beyond the change-of-state verbs discussed here.
Aspectual verb construction in Korean
The paper investigates aspectual verb constructions in Korean and their aspectual (Aktionsart/situation aspect (Smith (1991)) properties. It is proposed that the situation aspect is expressed by combining a series of verbs in Korean, and (a)telicity is determined by the second verb in the aspectual verb constructions in Korean. This analysis assumes mereological structures.
Discontinuous reciprocals and inherently symmetric events
Languages commonly have multiple ways to form reciprocal constructions, distinguished not only in form but in their semantics and distribution. (e.g., EINANDER and SICH in German). I show that the "discontinuous reciprocal" construction, shown in (1), is in many languages restricted to predicates denoting what I will call "inherently symmetric events": events expressing a binary relationship whose two participants have necessarily identical participation.
(1) O Giannis filithike me ti Maria (Greek)
the John kissed-Rcp.Sg with the Maria
`John and Maria kissed each other'
Contrary to appearances, discontinuous reciprocals are not semantically reducible to the correspending simple reciprocal. I argue that in this construction the reciprocal verb is semanticaly a two-place verb, and the comitative argument functions as an argument distinct from the subject.
Analogous constructions are found with symmetric predicates that are not grammatically reciprocal, suggesting that symmetry is essential to the discontinuous reciprocal construction. While thematically identical, the two participants remain distinguishable. Their positions differ in syntactic status and discourse prominence, and can be separately targeted by inferences and syntactic constructions.
The concept of infinitesimally small objects
My paper addresses the lower boundaries of event ontology. Natural language reference to events in aspectual semantics, as contrasted with the modelling of negative polarity items (of a certain kind), appears to pose inconsistent requirements on ontology. Specifically, NPIs suggest that even the extensions of homogenous predicates seem to allow for subevents that are outside the extension of the predicate. On basis of an isomorphism to infinitesimally small numbers, as have been developed in model theoretic algebra in the last century, we can resolve the apparent inconsistencies.
Reference to embedded eventualities
In the interfaces between morphology or syntax and semantics, semantic contributions of the involved constituents are opaque, i.e., semantic construction is blind to their inner structure. This principle is violated in many cases where the semantic contribution Cs of a syntactic or morphological constituent C includes embedded eventualities (i.e., eventualities that do not show up as arguments of Cs). Reference to them by a syntactic or morphological sister constituent of C should be impossible - but it is not. E.g., in the preferred reading of (1), 'person characterised by beautiful dancing', the adjective qualifies an eventuality of dancing, even though it is embedded in the semantics of dancer (Larson 1998): (1) beautiful dancer I discuss the derivation of semantic representations for cases of reference to embedded eventualities. The relevant interfaces use underspecified semantic representations.
Sentential subjects in event semantics
This talk is concerned with a particular group of "abstract" verbs which show an alternation between nominal subjects (1) and sentential ones (2). (1) Rebecca erleichterte Jamaal die Küchenarbeit / half Jamaal / gefährdete Jamaals Leben 'Rebecca made the kitchen work easier for Jamaal / helped Jamaal / endangered Jamaal's life' (2) dass Rebecca den Elektroherd repariert hatte, erleichterte Jamaal die Küchenarbeit / half Jamaal / gefährdete Jamaals Leben 'That Rebecca had fixed the electric stove made the kitchen work easier for Jamaal / helped Jamaal / endangered Jamaal's life' The talk will address four issues: (i) It will be shown that the behavior of these verbs with respect to adverbial modification reveals that while in their agentive variant these verbs clearly refer to events, they refer to states in their sentential variant. (ii) The sentential subject in (2) denotes what I will call a "coarse fact", namely a fact that allows a certain amount of redescription without affecting the relation sentences like (2) express. (iii) This relation is not a typical causal one. Instead of causality, the counterfactual concept of supervenience will be employed to show how the subjective nature of the state the object referent gets in depends on facts in the world such as those expressed by the sentential subject. (iv) Finally, it will be argued that the agentive variant of these verbs is derived from the sentential one by a certain kind of type-shifting operation.
Where is the adverbial event horizon?
Adverbials are distributed in “zones” in a sentence; for example, manner adverbs such as tightly normally occur very low, near the main verb, while event-modifying subject-oriented adverbs like 'cleverly' occur somewhat higher in the clause. In this paper I examine the problem of delimiting where the dividing line – the “event horizon” – between these two zones actually is. Specifically, I focus on the contrast between English, in which manner and other “event-internal” adverbs must occur below the lowest auxiliary verb (e.g. progressive or passive be and their equivalents), and Mandarin Chinese, in which these adverbs can occur higher than this point, i.e. above the progressive zai, the passive marker bei, and the preverbal functional head ba. The Chinese event horizon thus appears to be somewhat higher in the clause than it is in English. I first show that these facts cannot be handled easily by current theories of adverbial distribution, and that they cannot be explained by means of standard syntactic moves such as head raising or adverb movement. I suggest instead a line of analysis within the theory of Ernst 2002, in which Chinese zai, bei, and ba can in effect be event-internal operators, while their English counterparts cannot. Thus manner and similar adverbs adjoined above these operators in Chinese are free to continue event-internal modification, while in English they are not. (This is because the theory forbids event-internal modification at any point after the first instance of “event-external” modification, such as would be required by English progressive or passive be.) On this approach we also must consider modifying the most common treatment of restitutive again (Chinese you) and take the latter as similar to measure adverbs (e.g. completely) in its ability to modify a subpart of the event over which it has scope. Finally, if this sort of analysis is right, it shows that we must have a theory that allows both cross-linguistic variation in the location of the event horizon, and a way to group different event-internal modifiers together with respect to the zones they inhabit.
A formalization of Engelberg's progressive
After reviewing various approaches to the semantics of the progressive and the problems they face, Engelberg 2001 outlines an approach centering on a 3-place relation PROG(e,E,Persp) that brings out "the components that have to go into the semantics of the progressive. A strict formalization of this idea is still another matter." The present work paper fleshes out this relation in keeping with its use to overcome a number of problems Engelberg considers. The main intuition behind the proposed formalization is that events are sequences of observations which can be formulated as strings. Event types then come out as sets of strings (indeed regular languages!), and worlds as certain maximal sets of events grounded in time. The lexically projected event structure of an event type can be read off the strings in the language. And the perspectives are recast as functions specifying continuation branches that are, in the extensional case, trivialized, but may, in the intensional case, approximate those of Landman 1992.
Nissim Francez & Michael Kaminski
Unbinding and semantics of preposition-phrase modification
The paper introduces an *unbinding* operator, that converts a bound variable to a free one, applicable to a certain fragment of FOL. Then, it is shown how this operator is applicable to preposition-phrase modification (of PPs like instrument), that act as "arity expanders". The lexical verb meaning is taken as relation among its lambda-bound complements, and existentially quantified role arguments (instrumen,source, destination etc.). If a PP occurs, it has a determiner and noun in addition to the preposition. It *unbinds* the appropriate role argument, and rebinds by the quanntifier induced by the determiner, guarded by the PP noun. Clearly, this operates on representations (and not denotations) and uses *structured meanings*..
Marco Garc?a Garc
Taking a closer look at 'light verb constructions' (to give a kick), 'copula predicative constructions' (to be a doctor), and so called 'cognate objects' (to dream a dream), there are at least two striking characteristics: (i) the predicative NPs of these constructions carry most of the lexical information of the events described; (ii) these NPs generally bear no oblique cases, at least in languages like German. On the basis of an event-semantical analysis that relies on an examination of meaning and argument structure, I would like to give an account for these case restrictions.
Jean M. Gawron
Motion, scalar path, and lexical aspect
This paper investigates a class of predicates like "cover" which license both event and state readings in examples like (i) [Jackendoff 90]. These are called "spreading motion predicates" (SMPs): (i) Snow covered the valley (from the floor to the east ridge). It is argued that two properties license state readings for SMPs: They have path roles and, in this case, the paths have a scalar structure which "measures" or is bounded by another argument (the "container" argument represented by the valley in (i)). Thus the state readings of spreading motion predicates can be assimilated to those of scalar predicates such as "tall", "weigh", and "cost". The analysis extends naturally to path-shape verbs such as "zigzag" and to load-spray verbs.
Adverbial functions of Slavic prefixes
Under the assumption that in Slavic languages, telicity is grammaticalized by internal prefixes that apply to verbs (i.e. lexical items) and are part of the argument structure whereas external prefixes apply to (telic or atelic) predicates (VPs) without changing the telicity property of the predicate they apply to, I will take a closer look at the functions of external prefixes. Russian external prefixes mark temporal boundaries, only, and fulfil functions of outer aspectuality. In Czech, however, external prefixes primarily mark spatial boundaries and thus modify the sentence in a rather different way. Hence, Slavic external prefixes that are traditionally considered to be aspectual only, can fulfil adverbial functions that are not necessarily aspectual in nature as it is the case in Czech.
Wilhelm Geuder & Seongsook Choi
Deriving states in Korean
We investigate complex predicate constructions in Korean involving the light verbs "cita" (literally: 'fall') and "issta" ('exist'). These light verbs are special in that they do not serve to build up a more complex event structure from an array of subevents but rather appear to have a reductive effect on the event structure: "cita" derives inchoatives and "issta" derives stative predicates from eventive ones (esp. resultant states); the two verbs can also occur in combination. We argue for the existence of a syntactically derived stative form in Korean in which the stativiser "issta" (unlike German or English participle formation) embeds a full Voice phrase (in the sense of Kratzer). Still, stativisation is tied to an argument reduction effect similar to Germanic passive participles. We argue, therefore, that the argument projection of the Korean stative is controlled by semantic and syntactic conditions that filter the output of the stativising operation; it is not the lexical-conceptual content of the state that projects its argument structure. This result confirms Kratzer's analysis of participle formation in which the stativiser is also neutral wrt. argument structure.
Existential readings for bare plurals in object position
In this paper we present an analysis which explains the availability of existential (as opposed to generic) readings for bare plurals in object position. We do this by employing carefully-distinguished notions of situation and event, showing how both are needed for an adequate account. We conclude that an adequate analysis of bare plurals requires the formal notion of situation. This is not identical to the “event argument” employed in some accounts to distinguish between s-level and i-level predicates. We show that an event argument can always fulfil the role of a situation and hence license the existential inference . However, the more general notion of situation is required in cases where no event argument is present and the situation is supplied by other means, e.g. by context.
The conjunction 'weil' and the semantics of verbs: the relation between verb type and causal interpretation
The objective of the paper is to systematically relate the meaning of the causal conjunction 'weil' (because), to the lexical semantics of verbs in German. The issue is twofold: First, it will be shown that the different types of causality like 'causa finalis' vs. 'causa efficiens' of 'weil'-complexes can be predicted by means of the lexico-semantic properties of the sentential arguments of 'weil'. On the other hand, 'weil'-sentences can be used as a reliable diagnostics for what event participants have to surface in the semantic representation of a verb complex and for the thematic status of the event participants. I will formulate the corresponding conditions in the sense of meaning postulates, which serve as directives for the interpretation of a 'weil'-sentence. These are linguistically reflected as selectional restrictions of the expression 'weil', which restrict its combinatorial options by means of a calculation of the lexico-semantic properties of the verbs in question.
Cornelia Endriss & Stefan Hinterwimmer
The influence of tense and number marking on quantificational variability effects
Based on observations regarding the availability of Quantificational Variability Effects (QVEs) in sentences with singular and plural (in)definites modified by a relative clause, we argue that QVEs come about in different ways depending on number marking: In sentences with singular indefinites, QVEs are a consequence of the fact that the denotation of the indefinite is mapped onto the restriction of the Q-adverb. As a side effect of this, the time interval that contains the running times of the eventualities bound by the Q-adverb is preferably determined on the basis of relative clause internal information, so in the default case the tense of the relative clause and the tense of the matrix clause have to agree. But, crucially, this can be overwritten if there is an explicit indication in the matrix clause to choose a different topic time. In sentences with plural (in)definites on the other hand, QVEs are only possible if the relative clause inside the DP introduces a plural eventuality that fulfils the following constraints: It has to be constituted by subeventualities that are regularly distributed over the interval introduced by the relative clause, and the parts of the sum individual have to be distributed regularly over those subeventualities.
Alternating vs. non-alternating 'fill'-verbs: where does the difference come from?
This talk addresses differences in argument realization of several subclasses of locative verbs in languages like English/French on one side and languages like Japanese (1), Korean (2), Chinese, Kazakh, etc. on the other, taking fill-type verbs as an illustration. While these verbs currently do not alternate in English/French type languages, they do in the other languages mentioned above. Authors vary as to whether the observed differences are related to typological parametric differences and, assuming that these differences are due to differences in lexical specification, as exactly these lexical differences are (cf. Juffs 1993, 1996 for Chinese, Kim 1999, Kim & Landau 1997, H. Lee 1998 and C. Lee et al. 2000 for Korean, Rosen 1996 for German and Chinese, Iwata 2000 for Japanese). We discuss the insights and shortcomings of these various proposals and argue, taking into account additional data from German, Icelandic, earlier periods of English, Slovenian as well as Vedic (Haudry 1977) and Latin (Brachet 2000), that in some cases differences in the possibility of variable realization of arguments of fill-type verbs reflect differences in the meaning of these verbs and that in other cases it reflects differences in the semantics to syntax mapping principles, the two being in some cases historically related.
Non-selected datives, non-active/reflexive morphology and event (de)composition
This paper focuses on constructions with so-called non-selected dative arguments with non-active/reflexive morphology, as attested in Balkan and other languages. Crucially, I claim that non-active/reflexive morphology is a morphological operation that affects the lexical meaning of a predicate by changing either the aspectual template associated with it or the pairing of a name (a constant) with the aspectual template of a predicate. That is, non-active/reflexive morphology in Balkan languages is not an operation that solely affects the number of arguments in the argument structure of a predicate without affecting its lexical meaning. Relying on the model of lexical meaning proposed in Pustejovsky (1991, 1995), I argue that Balkan non-active/reflexive morphology is an operation on the event structure of the predicate. The various readings of the constructions with non-selected datives can be formally and uniformly derived by analysing non-active morphology as an aspectual operation which suppresses either the initial subevent in the event structure of a predicate or the name (constant) that is associated with this initial subevent. I will in particular discuss the ramifications of such an analysis for the passive construction, the nature of reflexivization as a cross-linguistic phenomenon, and more generally the fundamental nature of lexical complexity.
Lexicalization and eventuality structure of Japanese motion verbs
This study examines the lexical components of Japanese motion verbs and how they are reflected in syntactic structure. Japanese motion verbs display variousverb-particle compatibility requirements, which are attributed to the verbs? lexicalization types. In particular, the lexicalization types: Path, Manner and further subclasses play a major role in forming eventuality structure. Furthermore, these lexicalization types of verbs behave differently in aspectual constructions, and telicity plays a crucial role in both the lexical properties of the verbs and the syntactic structure in which they appear. I conclude that verb meanings characterized by the lexicalization types are essential to explain their syntactic behavior and grammatical constructions also play a role in forming eventuality structure.
Modelling the semantics of alternating verbs
This paper focuses on the key role of semantics in a robust deep analysis of alternating verbs. The aim here is to show that the theoretical framework of HPSG (Pollard and Sag 1994) with semantic representations in Minimal Recursion Semantics (MRS; Copestake et al. 1999) constitutes an appropriate theoretical basis for a robust, linguistically-motivated account of valence alternations, which overcomes the natural limitations of previous syntactic and semantic analyses of such constructions (see among others Rappaport and Levin 1988, Pinker 1989, Markantonatou and Sadler 1996), and provides the necessary formal generalizations for the analysis of alternating verbs in a multilingual context, since MRS structures are easily comparable across languages. Our case study here is valence alternating verbs in German.
More on depictives in Finnish
This talk will focus on the possibility of organizing the identification of variables (`linking') under merge on a purely model-theoretic basis. Specifically, we shall look at depictives, as discussed recently by Vivienne Fong. While with locatives such a purely semantic linking can be defined, taking advantage of such notions as movers, it turns out that with depictives this cannot be done. Instead, linking seems to work on a formal basis, using the notion of theme, which in itself is not truth-conditional. Linking depictives is otherwise quite similar to locatives: there is a fundamental split between static and dynamic depictives, the latter being more strict in their linking behaviour. Unfortunately, the semantic theory by Fong, which extends her earlier work on locatives, suffers from a defect: it claims that the meaning of change is present in a directional even when it is selected by a higher head. Our approach extends work we have done on locatives. It is based on the assumption that selection cancels the semantics. This eliminates the need to provide a suitable semantics in the case of selection because the semantics is operative only when no selection takes place. Thus the reason dynamic depictives imply no change when selected by a higher head is not that the meaning of the cofinal mode is weak in Finnish (as Fong suggests) but rather that it is selected more often over the static mode.
On the plurality of verbs
Over the last 15 years or so, Manfred Krifka has explored cumulativity as an important property of nominal and verbal predicates, and in the course of this work, the possibility emerged that cumulativity might correspond to a significant semantic universal. More recently, Fred Landman has defended initial cumulativity specifically for verbs. On this view, all verb denotations are pluralized from the very start. In my talk, I will discuss some surprising consequences of the Opluralization from the start¹ hypothesis for verbs and defend it against serious criticism.
The periphrastic expression of pluractionality
Traditional in the morphological description of some West African and North American languages, the notion of pluractionality has recently been extended to the semantics of some verbal prefixes in the Slavic languages (Filip & Carlson 2001) and to the overall treatment of atelicity (Van Geenhoven, to appear). This paper confronts the issue of the overt expression of pluractionality by monoclausal verb combinations. It will be shown that a class of Romance „aspectual“ periphrases can be semantically analyzed as pluractional markers. The description focusses on Spanish ir + Gerund and andar + Gerund, which are not, as often assumed, variants of a „progressive viewpoint“, but „eventuality modification“ periphrases. In giving a formal characterization of the derived temporal structures brought about by periphrastic pluractionals, we will propose answers to the following questions: (i) to what extent distributive pluractionals instantiate standard distributivity, and what event „ingredients“ (subevent times, locations, participants) act as „distributor“ and „distributed share“; (ii) in how far the incremental structure contributed by ir + Gerund differs from those associated with „basic“ telic predicates (Krifka 1998, Hay & al. 1999) or with modifiers such as gradually (Piñón 2000); (iii) whether the differences in behavior between ir+Gerund and andar+Gerund are at all compatible with the strong association between pluractionality and atelicity assumed by Van Geenhoven (to appear).
Event decomposition and the syntax and semantics of durative phrase in Chinese
This paper investigates the interaction between lexical structures of verbs and adverbial modification with a special focus on Chinese durative phrases.Depending upon the (a)telicity of the situation modified, Chinese durative phrases may follow or precede the object NP with a different interpretation. I show that the syntactic distribution of durative phrases follows from decomposition of verb meanings in overt syntactic structures and the homogeneity requirement of durative phrases. More precisely, durative phrases can be adjoined to every possible (maximal) projection, provided we can interpret them there without violating the homogeneity requirement. In addition, I will argue that a now somewhat popular analysis that incremental theme verbs are not inherently telic as in Kratzer (2003) is not able to account for the distribution of Chinese durative phrases. Instead, I return to the more traditional assumption that Incremental Theme verbs are inherently telic, arguing that the atelic reading of an incremental sentence is the consequence of a partitivity operator associated only with incremental theme verbs. It follows from this assumption that on the telic reading, incremental sentences only allow the object-durative order, but not the durative-object order, parallel to sentences with an achievement verb or a resultative compound verb.
Event based phase structure and the Karttunen-Peters problem
According to Karttunen and Peters (1979), any many-dimensional theory of presupposition fails to account for the intuition that in 'Someone managed to succeed George V on the throne of England.' it is one and the same person, which is making true the presupposition 'Someone had a hard time to succeed George V' and the assertion 'Someone succeeded George V'. A similar problem, which is not confined anymore to many-dimensional frameworks, can be observed with the interplay between existential quantification and presuppositional phase structures, expressed e.g. by the German phase particles 'schon' and 'noch'. In this paper, the second problem is solved within classical first order logic by using Kratzers event-semantical perfect aspect and relating the arguments of the phase particles to resulting states or everlasting preceding states. Along this line of reasoning, a solution for the Karttunen-Peters problem is offered by taking into account the internal event based phase structure of 'manage'.
Distributive predication and collective event
In this paper we provide a criterion for collective event and group formation for purely distributive predicates (e.g. “walk”: “John and Mary are walking along the beach”). Intensional theories consider groups (i) to be sums seen under a certain perspective, and (ii) to be able to fill a thematic role of a singular predicate as a singular individual. Nevertheless even under the collective interpretation, purely distributive predicates cannot be shifted to a predication of a singular event. We propose a criterion for collective event formation that rests on three bases: (i) phases or descriptions of events, (ii) constraints or non-accidental linking, (iii) lumping. We claim that a collective event results from the application of a constraint on the descriptions of the plural events and all the events that they lump.
Eric McCready & Chiyo Nishida
Intransitive reflexives in Spanish and event semantics
This paper considers the semantics of the Spanish reflexive clitic 'se' in intransitive constructions. It is well known that 'se' with transitive verbs must appear in telic environments; we show that the same is true for intransitive 'se'. We provide a generalization of Nishida's (1994) Krifka-style analysis of transitive 'se', based on the unaccusative nature of the verbs with which 'se' can appear. We then point out two unique characteristics of intransitive constructions with 'se'---the possibility of using the 'dative of interest' and the presence of path-like interpretations---and provide semantic analyses of them.
On the interaction between temporal adverbials and event structure
This paper deals with a set of constructions that include temporal adverbials and non-atomic event descriptions. In particular, it will focus on cases where the temporal adverbial acts as a true event modifier, in the sense that inclusion in the time frame set by it is a defining property of the events comprised in a sum represented by a matrix clause. Examples: “John crossed the Atlantic three times in July 2002”; “John wrote three essays in ten days” (and analogously, though with some idiosyncrasies: “John goes to Paris three times a year / “John is writing three essays per year). The purpose is to elaborate on the semantics of these constructions – where temporally bounded quantification over events emerges – and argue for a separate linguistic analysis (distinguished from e.g. simple temporal location or temporal measure structures).
Interpreting adverbial modification of adjectives
Among the principal problems in the syntax and semantics of adverbial modification is how the position and interpretation of adverbs should be related. Efforts toward addressing aspects of this question (from Jackendoff 1972 and McConnell-Ginet 1984 to Cinque 1999 and Ernst 2002) have focused primarily on adverbial modification in the verbal and sentential domain. There are, however, less prototypical uses of adverbs in the adjectival extended projection as well, and interestingly, the interpretation adverbs receive there varies predictably from the one they receive elsewhere. In this respect, adverbial modification in AP offers another perspective on the larger problem. This paper addresses the syntax and semantics of certain such 'ad-adjectival' adverbs, presenting an analysis of how their interpretation arises and of how it relates to the interpretation of their counterparts outside the adjectival domain. Along the way, it touches on broader questions about how degree semantics relates to event(uality) semantics, and about how the structure of the extended AP relates to that of the extended VP with respect to adverb licensing.
Event quantification, distributivity and aspect
In my earlier work (Nakanishi 2003), I proposed that quantifiers in split quantifier constructions (Japanese floating quantifier construction and German split topicalization) measure the events in the extension of the VP and that such a measurement in events is possible when the relevant VP has a part-whole structure. I further proposed that there is homomorphism (or a structure-preserving mapping) from a lattice of events denoted by the VP to a lattice of individuals denoted by the host NP. This analysis successfully accounts for the observation that split quantifier constructions lack a collective reading. In this paper, I argue that events can be measured not only when the relevant VP is pluralized, yielding a distributive reading, but also when the VP has inherent 'subevents' without pluralization, yielding a collective reading. In particular, I show that a collective reading is allowed with atelic VPs and telic VPs in a progressive form, since these VPs have inherent 'subevents'.
Goal and source in event structure:asymmetry in their syntax and semantics
This paper focuses on the syntactic and semantic asymmetry between Goal (e.g., into the store) and Source locatives (e.g., from the store). Based on their syntactic and semantic asymmetry, the paper argues that they have distinct underlying base positions in extended VP-structure and further that they have different semantic scope/contribution in event structure. Thus, we claim (i) Goal PPs are generated under the lower VP2, and they semantically compose a core event (result state: E2) denoted by the lower VP2. And (ii) Source PPs are generated under the higher VP1, and semantically modify the process sub-event (E1). Source locatives do not compose a core event. The paper illustrates various syntactic and semantic asymmetries including the following: preposition incorporation, prepositional passive, movement and ordering, locative alternation, ambiguous adverbial modification, and aspectual division. This paper identifies an interface principle between syntax and semantics of Goal/Source locatives, and further argues that the interface principle should account for the non-directional readings of Goal/Source phrases in natural language.
Verbs of creation
In this paper, an analysis of verbs of creation (e.g., 'build') is proposed that improves on analyses by Krifka and von Stechow. The leading idea is to bring together a treatment of incrementality with a predicate of existence so as to be able to capture the intuition that an object gradually comes into existence. The analysis is then extended to cover examples involving partial creation of a type of object (e.g., 'Rebecca partly built a house'). For these cases an account of partial creation is offered that makes use of quantification over types of objects.
Morphological Conversion in German: Was den Besuch zum Ereignis macht
Nouns like Schlag “hit”, Kauf “buy”, Besuch “visit” are usually analyzed as deverbal conversions. Schlag “hit”, Kauf “buy”correspond to monomorphematic verbs whereas the base verb of examples like Besuch “visit” is morphologically complex. Interestingly, with the latter group it is always possible to realize the verb’s accusative object as a postnominal genitive: (1) der Besuch der Universität “the visit the universityGEN” If the corresponding verb is simple, we have to consider its Aktionsart. A genitivus objectivus is only possible with telic events: (2) der Kauf des Hauses (genitivus objectivus) “the buy the houseGEN” (3) der Schlag des Kindes (genitivus subjectivus) “the hit the childGEN” These diffences can be explained if we assume that a genitivus objectivus is restricted to nouns which are – morphologically and semantically – based on verbs. Besuch provides independent morphological evidence for such a derivation (verbal prefixes do not occur with underived nouns!), Kauf provides independent semantic evidenc (underived nouns cannot denote telic events!). There is, however, no need to assume a verbal base for a noun like Schlag. I will argue that the impossibity of a genitivus objectivus is actually due to the fact that these examples are underived nouns.
This paper presents a unified treatment of the different uses of modifiers, as exemplified by "illegally" and "illegal". It analyzes the manner, clausal and pre-adjectival forms of this adverb, as well as the adjectival form, and extracts a core lexical meaning common to all uses. This core meaning consists mainly of a restriction on context which picks out the relevant law, and the negation derivationally provided by "il". The differences between the uses, or rather the meaning-parts that are specific to a use, are treated as type-shifts. Proposals where they are introduced by local functional structure are also considered. These use-specific parts are general across a class of modifiers (including pysch-adverbs and the like), explaining systematic relationships throughout the class. This analysis can be used as a building block for giving unified accounts of individual modifiers, as well as adverb classes.
Hilke Reckman & Crit Cremers
Implementing event structures for nominalization
We implemented event structure for nominalizations in Delilah, a parser and generator for Dutch, driven by a categorial grammar. We focus on those nominalizations that are ambiguous between an event and an object/result reading, such as "Henks beschrijving van een auto" (Henk's description of a car). For both readings we use a semantic representation in which both the event and the result are present. The nominalization gives rise to an event reading if the event is abstracted over, and therefore available for predication, and the result is existentially closed. If things are the other way around, so the result is abstracted over and the event is existentially closed, then we get a result reading. The main motivation for implementing event structure in the first place, and for implementing it in this particular way for these nominalizations, is to facilitate automated inferencing in the future. The talk will include a demonstration of the working system.
Two puzzles for a theory of lexical aspect: semelfactives and degree achievements
This paper presents an analysis of semelfactives such as kick and jump and degree achievments such as cool in the framework of the analysis of lexical classes proposed in Rothstein 2004. I argue that the verbs from the four Vendler classes denote events which differ with respect to (a) whether minimal events in each class have or do not have duration and (b) whether or not they denote events of change. States and achievements hold at instants and are thus [-duration] and activities and accomplishments hold at intervals and are [+duration]. Achievement and accomplishment predicates denote events of change, while states and activities do not. I propose an operation of concatenation which concatenates minimal events, and which because of the terms in which it is defined cannot apply to events denote a change from a situation ¬φ to a situation φ. This operation applies to activities and states but not to achievements and accomplishments. The relation between semelfactives and activities is then straightforward: a semelfactive use of jump denotes the set of minimal events of jumping, while the activity use denoted the set of jumping events closed under concatenation. Concantenation applies to states also, but assuming that time is dense and no minimal instants are indentifiable, the minimal set of events in love is not accessible, and the predicate love denotes unambiguously the set of loving events closed under concatenation. Concatenation cannot apply to events of change, since two changes from ¬φ to φ involving the same participant cannot be immediately adacent, but must be separated by an event of change from φ back to ¬φ. (Kamp 1979) This predicts that if there are events of change in P where two events of P can be immediately adjacent, then they ought to be concatenatable. I argue that degree achievements are events of this kind, since they are not events in which a change from ¬φ to φ occurs, but events in which there is a change in the value assigned to x on a scale. Thus a verb such as cool is ambiguous between an achievement reading denoting a single (near) instantaneous change, which parallels the semelfactive reading of kick, and the extended activity–like reading of cool which includes in its denotation events constructed out of concatenations of changes.
Kjell Johan S?b?
The structure of criterion predicate
My aim is twofold: To give an analysis of criterion predicates, and to give an analysis of the "by" locution in English. Criterion predicates, like "break a promise", must meet second-order criteria to do with conventions and intentions. Like Kearns (2003), I group them together with manner-neutral causatives. Unlike Kearns, I analyse such "meta predicates" as, essentially, relations between "basic predicates" and events. The use of DRT style unification allows a compositional analysis of the "by" locution in the spirit of Bennett (1994) and an account of the cases where meta predicates are not modified syntactically.
State and event denotations of German copular verbs
The talk deals with the meaning of the German copular verbs sein ("to be"), werden ("to become") and bleiben ("to remain") and the way in which they are related to each other. According to general knowledge, sein denotes a state, werden denotes a change of state and bleiben the continuation of a state. However, there are data which show that bleiben is not limited to such a durative meaning but that it may denote a change of state, too.The aim is to examine whether bleiben in this second reading does denote a state or an event ? the latter would mean that bleiben and werden have partly identical functions. Apart from this, it can be shown that the main difference between bleiben and sein isn't that between a state on the one hand and some longer (or continuing) state on the other. It rather lies in the fact that bleiben typically appears in a context where one would expect the counterstate to become true in place of the state actually asserted.
Meijia Gao & Hooi Ling Soh
Two '?les' in Mandarin Chinese
Mandarin Chinese particle –le may (sometimes) appear immediately after the verb (verbal –le) or sentence finally (sentential –le), or in both positions at once (double –le): (1a) Women daoda le shan-ding. (1b) Women daoda shan-ding le. we reach LE mountain-top we reach mountain-top LE ‘We reached the top of the mountain.’ ‘We have reached the top of the mountain.’ (1c) Women daoda le shan-ding le. we reach LE mountain-top LE ‘We have reached the top of the mountain.’ We propose that the verbal –le is a perfective marker (Li and Thompson 1981), while the sentential –le is a transition marker, which triggers a presupposition about an immediate past event that is in opposition to the one described by the sentence (see Dowty 1979, Pustejovsky 1990). Our proposal accounts for the meaning(s) of the verbal –le and the sentential –le, the interpretations of double –le sentences and a restriction in the appearance of zhi ‘only’ and budao ‘less than’ in some sentences with the sentential –le.
Minjeong Son & Peter Cole
Semantic and Syntactic Decomposition of Events in Standard Indonesian
A widely held position in the literature on verbal meaning is that the lexical semantic representation of verbs involve complex event structure with semantic primitives like CAUSE and BECOME (e.g., Dowty 1979). A growing number of recent works on predicate decomposition have shown that there is a close correlation between the semantics of event structure and the syntax (e.g., Travis 2000; van Hout 2000). This paper presents a new empirical argument for the view that there is a direct mapping between semantic decomposition of predicates and the (morpho-)syntax by developing an explicit analysis of the semantics and syntax of the verbal suffix -kan in Standard Indonesian. We argue that the verbal suffix -kan is an overt instantiation of the RESULT head, the semantics of which gives rise to a causative interpretation. By treating -kan as the RESULT head, we argue that the current analysis not only provides important empirical support for the syntactic decomposition of predicates but also leads to a unified semantic and syntactic account of -kan, which straightforwardly captures distributional properties of the suffix.
Event Decomposition and the Ambiguity of tasi 'again' in Korean
The theory of semantic and syntactic decomposition of predicates follows from a number of syntactic and semantic facts with regard to adverbial modification. Von Stechow (1995), for instance, has proposed that the different readings of wieder again in German provide a test for the syntactic and semantic composition of predicates and for a syntactic constituent denoting result state encoded in the verb meaning. Building upon von Stechow s insight, Son (2003) provides a unified syntactic treatment of two types of morphological causatives (MCs) in Korean. By demonstrating the scope ambiguity of tasi again , she argues that all MC formation in Korean takes place in the syntax and that MCs that have been considered to be non-decomposable involve more complex underlying structure than was previously believed. In this paper, I further demonstrate that the again test shows ambiguity for lexical causatives as well in which the meaning of causation is embedded in a single word. Both morphological and lexical causatives are thus shown to contain an abstract CAUSE head and different types of causatives result from the differences in the size of the argument for which CAUSE selects, regardless of whether CAUSE is visibly realized in the morphology.
The notion of path in aspectual composition: evidence from Japanese
In this paper, I would like to argue that the notion of 'path' plays a crucial role in aspectual composition, by eliciting data mainly from Japanese. Specifically, I would like to claim that incremental theme and motion verbs are distinguished from change-of-location/state verbs with respect to the telicity formation. This is evidenced by the complementary distribution of two postpositions in Japanese, ni 'in/at/to' and made 'as far as'. I will show that the notion of path is crucial to the telicity formation for the former verb classes. I also would like to claim that the differenciation of the origins of telicity in these verbs is extended to so-called resultative constructions in Japanese (and English). The difference in telicty formation leads to the answer of the question why Japanese lacks so-called 'strong' resultatives.
Alice ter Meulen
Cohesion in context: the role of aspectual adverbs
Aspectual adverbs (e.g. still, not yet, already, and finally) serve to create temporally coherent contexts, contributing factual information as well as subjective, speaker dependent information. Assuming the DRT account of their dynamic semantics in , this paper addresses (i) how presuppositions of aspectual adverbs, when prosodically marked by high pitch accents in English, are shared or rejected in answers to polarity questions, cf. (1), and (ii) analyses their use in dialogue requests to share the subjective speaker’s perception regarding the timing of the described events, cf. (2). In accommodating presuppositions in a non-empty context, it is argued that conflicts between the presupposition and the conditions asserted to be true at the given reference time cannot be resolved by interpolating another reference time at which the presupposed information is true. Accommodation is, like the notion of logical consequence, static, i.e. unable to adjust the context by interpolating new reference times to resolve information conflicts. This paper contributes to our understanding how asserted, presupposed and entailed information each contribute differently to the constitution of the common ground, representing shared information in a multi-agent system.
Patients in Igbo and Mandarin
I argue from the grammar of resultative verb compounds in Igbo and Mandarin that, in these languages, the patient relation between a verb and its object is introduced by structure extrinsic to the verb. Systematically, Igbo and Mandarin verbs that require a patient when in simple clauses are subject to no such requirement when in a resultative. This is explained, I argue, only if the patient relation is introduced externally to the verbal predicate, whether simple or complex. The facts thus challenge the idea recently explored by Kratzer, that the patient relation should be excluded from our ontology of natural thematic predicates.
Momentary 'until' in Mongolian
Mongolian -tal is puzzling in that it means both 'until' and 'while', which are incompatible with each other in temporal relations. For a sentence with a form P-tal Q, 'until' and 'while' are "until P happens Q holds" and "while P holds Q happens", respectively. In this paper, it is shown that there indeed is a common meaning to these two, and the two readings are derived from the eventualities on the predicates. -Tal's core meaning is to indicate the point at which P and Q maximally approach to each other. The 'until' reading is when P(t)-tal Q(i), in which the duration in the matrix clause stems from the predicate, but not from -tal, unlike English 'until'. When it is P(i)-tal Q(t), we will get the 'while' reading, which requires duration in the subordinate clause. This is a successful description of Mongolian -tal. It eliminates ambiguity in the lexicon, by separating eventuality information from the definition of the suffix. Notably, the meaning of -tal maintains similarity to English 'until', namely, the boundary between until-expression and the matrix. It suggests a possible expansion of 'until'-like items in its clausal constructions.
Prepositional aspect and the algebra of paths
The notion of path plays an important role both in the study of verbal aspect and in prepositional semantics, but these two empirical domains are not often systematically connected. This paper intends to relate the two domains in the most direct way: by treating the denotation of a directional PP as an algebraically structured set of paths that is thematically mapped to a mereology of events. In this way, there will be two-way traffic between event structure and spatial structure. The mereological approach will deepen our insight in the semantics of prepositions because we can exploit the parallelisms with the nominal and verbal domain (and find sums, cumulativity, plurality and a mass-count distinction in the path domain). On the other hand, the lexical semantic definitions of prepositions can be made relevant to event structure in a more principled and more compositional way than often done. The intuitive idea that paths with an end point lead to telicity is shown to be untenable when we take more directional prepositions into account.