WorkshopGeometrical Structure in Event Concepts

Thursday/Friday, Oct 7–8, 2004Universitat Konstanz, Germany

SFB 471 / project A5

.organization

Wilhelm GeuderMatthias WeisgerberPeter E. Pause

Irene Wolke (logistics)Melanie Hochstatter (coffee logistics)

.rooms

The conference will take place in Konstanz University in the rooms G 308 (office, coffee)and G 307 (talks). You will find signs in the University. Nevertheless – if you feel you gotlost in the building, please call (inhouse) 2723 (Matthias) or 2424 (Willi). Find more infoand maps on pages 9 ff.

.timetable

.timetable.thu07

9.30 – 9.45 Welcome9.45 – 11.00 Barbara Tversky (Stanford University):

Segmenting Abstract Events

• coffee break •

11.30 – 12.45 Phillip Wolff (Emory University):Causation and Configurations of Force

• lunch break •

14.45 – 16.00 Fritz Hamm (Universitat Tubingen):Plan goal structures

• coffee break •

16.30 – 17.45 Joost Zwarts (University of Utrecht):Spatial structure in semantics: aspect, opposition and modifiers

20.00 – . . . Being together at ‘Weinstube Hinterturle’ (in old town, see map)

2

.timetable.fri08

9.30 – 10.45 Emile van der Zee (University of Lincoln):Motion encoding in language: Reference frame use during the employmentof spatial locatives in a motion context

• coffee break •

11.15 – Carola Eschenbach, Christopher Habel (Universitat Hamburg):Paths, Geometry and Situation Structure

– 13.00 Christopher Habel, Carola Eschenbach:Perspectives in motion events: A geometrical analysis of verbs of branching

• lunch break •

14.45 – 16.00 Seungho Nam (Seoul National University):Locatives in event structure

• coffee break •

16.30 – 17.45 Wilhelm Geuder (Universitt Konstanz):Event structure and spatial structure: the case of open

17.45 – ? Concluding Discussion

.impressumtexts: authors

maps based on: Konstanz Stadtwerke / Tourist Information / UniKNtypeset in LATEX (MW)

Universitat Konstanz, view from south

3

.abstracts

.abstracts.thu07

Bridgette Martin and Barbara Tversky (Stanford University):Segmenting Abstract Events

Although the world is continuously changing, we parse it into discrete events. For familiarevents, like doing the dishes, segmentation is based both on action and on goal achievement;the top-down and bottom-up information is correlated. What underlies segmentation for ab-stract, unfamiliar events? Participants segmented videos of movements of geometric figures,designed to represent two social scenarios: chase and hide-and-seek. Some described whathappened in each segment as they described. Segmentation was either on first viewing oron fifth viewing. Although segmentation was essentially the same for view once and viewfive times, interpretations changed. Familiarized events were more likely to be interpreted asintentional acts; unfamiliarized as movement changes. Detailed analyses of the changes inmovements at segment boundaries suggest that coarse units are distinguished by both startsand turns whereas fine units are distinguished by more refined movement changes, starts,object contacts, stops, and changes in direction. These movement changes are indicative ofdifferent levels of intention.

Phillip Wolff (Emory University):Causation and Configurations of Force

Causal knowledge underlies our ability to explain the past, exert control and predict futureoutcomes. It is a central topic of current research in psychology, linguistics, and philosophy.While significant progress has been made in our understanding of causation, important ques-tions remain. In my talk I will address several classic problems in the causation literaturefrom an entirely new perspective. According to the vector model, causal concepts are definedin terms of the spatial arrangements of physical and/or psychological forces. Support for themodel is provided by experiments in which participants watch 3D animations of realisticallyrendered objects with trajectories that are wholly determined by force vectors entered intoa physics simulator. In these experiments, the very same physical forces used to generatephysical scenes are used as inputs into a computer model to predict how those scenes will bedescribed. In another series of experiments, I account for the difference in meaning betweensingle and multi-clause expressions of causations in terms of the relative pattern of forcesacting on a patient. I also show how a vector approach may contribute to our understandingof the relationship between causality and time.

4

Fritz Hamm (Universitat Tubingen):Plan goal structures

It is well known that facts about tense and aspect in natural languages were one of themain motivations for developing dynamic semantic systems. The talk shows that conceptsfrom logic programming such as “completion” and “integrity constraint” are well suited forformalising the dynamics of temporal reasoning by providing plan-goal structures which arecrucial ingredients of natural language tense systems. In cases of goals which cannot bedeliberately planned the meaning of tense is still derived from path-planning.

Joost Zwarts (University of Utrecht):Spatial structure in semantics: aspect, opposition and modifiers

Space plays an important role in how we conceptualize and organize meaning. Spatial notionslike distance and direction and spatial elements like vectors and paths are not restricted to thecategory of prepositions and the domain of physical space, but they play a central role acrosssyntactic categories (including nouns, verbs and adjectives) and across various conceptualdomains (e.g. kinship, time, force, degree). There is an abstract geometry in the meaningof words. In my talk I will discuss three of the ways in which this spatial structure showsitself in semantics and especially in the semantics of verbs: in aspect (the role of paths),in lexical semantic opposites (antonyms, converses, etc.) and in the interpretation of themodifier ‘close’.

.abstracts.fri08

Emile van der Zee (University of Lincoln):Motion encoding in language: Reference frame use during the employment of spatial locativesin a motion context

Levinson (1996) distinguishes between three different frames of reference to describe a Figurelocation in relation to a Ground when using projective terms like ‘in front of’ and ‘to theleft of’. Locating a Figure in relation to a Ground in a context where both are in motionraises the question of whether three reference frames are sufficient, or whether an additionalreference frame – derived from motion – is necessary to explain the use of projective terms.After making a case for such an additional reference frame on the basis of some examples, Iwill present the results of two experiments suggesting support for this idea. The experimentsshow, e.g., that different projective terms result from employing a relative, intrinsic or a’motion frame’, and also that Levelt’s (1996) Principle of Canonical Orientation (excludingthe use of the intrinsic frame in certain situations) does not apply in motion conditions.

5

Carola Eschenbach, Christopher Habel (Universitat Hamburg):Paths, Geometry and Situation Structure

Natural language descriptions of motion frequently combine verbs of motion and directionaladverbs (Bill ran into the kitchen). This talk presents an analysis of German expressions fromthese two classes and on the spatial structures (called path) that link them. Paths and therelations accessing them are formally specified in a geometric framework and with referenceto the situation structure. The analysis distinguishes verbs of motion (run), for which thepaths represent the trajectory of the moving object, from other kinds of verbs that combinewith directional adverbs (shout, look) and embed paths in a different manner. The approachprovides a basis for explaining certain observable restrictions and preferences in combining(German) verbs of motion with directional prepositional phrases.

Christopher Habel, Carola Eschenbach:Perspectives in motion events: A geometrical analysis of verbs of branching

The critical phase in following a verbal route instruction is choosing between alternativetracks at so-called ‘decision points’, e. g. when the instructed person approaches a crossing.The appropriate choice can be described using verbs that relate a path of motion to aconfiguration of tracks, as in “Cross Main Street at the light and turn left onto 7th St.”, orusing verbs that simply specify a configuration of tracks, as in “The road to Oxford branchesoff three miles from her”. The configurations of tracks (as parts of the spatial environment)are static and neutral regarding orientation or direction. Many verbal description, however,contain directional or temporal expressions imposing a perspective and dynamics onto thescene.

German particle verbs like abbiegen, einbiegen, einmunden and abzweigen [turn/branch-off,turn/branch-into, lead-into, branch-off] express such a directional perspective. Nevertheless,these verbs combine with directional adverbs. Based on a geometrical characterization ofpaths and of non-oriented curves, we can describe the different perspectives imposed ondecision points by natural language expressions and the restrictions in combining the differentverbs with further specifications.

Seungho Nam (Seoul National University):Locatives in event structure

0. This paper focuses on the syntactic and semantic asymmetry between Goal (e. g., into thestore) and Source locatives (e. g., from the store). In the literature, classified as directionallocatives, Goal and Source phrases have not been properly analyzed from a contrastive andformal perspective. (cf. Jackendoff 1990, Koopman 1997) Based on their syntactic andsemantic asymmetry, the paper argues that they have distinct underlying base positionsin extended VP-structure and further that they have different semantic scope/contributionin event structure. The present paper does not deal with non-directional locatives (cf.Maienborn’s (2001)).

6

1.1. The paper argues that Goal PPs (PPG) are generated under the lower VP2 as shownin (1), and they semantically compose a core event (result state: E2). Thus, the sentence(2) denotes an event structure of (3).(1) [VP1 DP1[V 1′V1[VP2[V2’ V2 (DP2) PPG]]]](2) John swam to the boat.(3) Event structure: [E0[E1=P john swim]&[E2=S john BE-AT the-boat]]

1.2. The paper also argues that Source PPs (PPS) are generated under the higher VP1as shown in (4). The sentence (5) denotes a complex event of (6), where the sourcePP modifies the process sub-event (E1). The source phrases do not contribute to thecomposition of core event.(4) [VP1DP1 PPS [V1’V 1[VP2[V2’(DP2) V2 (PP)]]]](5) John swam to the boat from the beach.(6) Event structure:[E0[E1=P[MOD from-the-beach][E1 john swim]] & [E2=S john BE-AT the-boat]]

The paper illustrates various syntactic and semantic asymmetries including the following,which all support the above proposal.

2. Goal-Source Asymmetry in Syntax: Extending Koopman (1997), Travis (2000), andKracht (2002), we further argue for the structural asymmetry between Goal and Source.

2.1. Preposition incorporation: Goal-directional pre/post-positions can be incorporatedinto verb more easily than Source or non-directional ones do (Baker 1990, Koopman 1997).Munro (2000) also illustrates that Chickasaw verbs can occur with more than one applicativeprefixes, and Goal applicative prefixes always come closer to the verb stem than Source ornon-directional applicatives do.

2.2. Prepositional Passive: Couper-Kuhlen (1979) shows that Goal is a typical subjectof prepositional passive, while Source or non-directional one is hardly found in the subjectposition. Thus, we have this contrast: If the boat is jumped into/*from it may capsize.

2.3. Movement and Ordering: Source PPs can be dislocated/topicalized more easily thanGoal PPs can.

2.4. Locative Alternation: Locative alternation in many languages involves Goal-type PPsrather than Source or non-directional ones. (Levin 1993; Hale & Keyser 1993, 2002) Moregeneralized, V-modifiers/complements like Goal-type PPs allow alternations more often thanVP-adjuncts like Source-type or non-directional PPs do.

3. Goal and Source in Event Structure

3.1. Lexical Complex Event Structure: Extending recent proposals on complex lexicalevent structure (Jackendoff 1990; Grimshaw 1990; Pustejovsky 1991, 1995, among others),we identify the distinctive semantic contributions of Source and Goal in terms of semanticscope in event structure.

7

3.2. Event Structures for VPs and Locative PPs: Recent studies on adverbial modifica-tion argue for a finergrained event structures (Eckardt 1998; Ernst 1998; Tenny 2000; Travis2000). We further show that Goal PPs constitute a core event (i.e., result state) whereassource PPs do not. Thus, the goal PP in John sent the book to New York again introducesa result state, so again can give a restitutive reading as well as a repetitive reading. In Johnsent the book from New York again, however, again does not allow restitutive (narrow scope)reading, since the source PP does not form a core event.

3.3. Aspectual Division: Cinque (1999) and Travis (2000) divide at least two aspectualdomains in syntax, i.e., ‘higher/viewpoint’ aspect and ‘middle/situation’ aspect in Tenny’s(2000) terms. We claim that, unlike Goal PPs, Source PPs scope over the situation aspect,so do not shift the aspectual character of the core event. Thus, He ran from the library (forten minutes/*in ten minutes). Goal PPs, however, composing the core event, play a crucialrole in aspectual composition.

4. This paper, based on the extended Event and VP structures, identifies an interfaceprinciple between syntax and semantics of Goal/Source locatives. The paper will show thatour proposal is ready to handle other types of locatives such as symmetric path-type (e. g.,through the tunnel and over the bridge) and non-locative uses of directional PPs (e. g., Hesmashed the glass to pieces; Harry died from AIDS complications.).

Wilhelm Geuder (Universitat Konstanz):Event structure and spatial structure: the case of open

The adjective open and the verb derived from it are notable for the large spectrum of readingsthat they allow (even if purely metaphorical uses are excluded, as will be done in this paper).For instance, Taylor (1992) posits a network of polysemic variants which clusters in threemajor subgroups (but probably still isn’t exhaustive): (a) ‘move apart’: open a book / one’seyes; (b) ‘create aperture’: open the door / the lid; (c) ‘make accessible’: open a room. Inthis talk I argue that most of the polysemy can be explained away if one assumes a uniformunderlying meaning of the adjective open in terms of an abstract situation schema. Thevariability of meaning can then be accounted for by an additional process which regulatesthe subsumption of concrete objects under the schematic properties. The situation schemaexhibits an interweaving of spatial and functional features, which may differ in prominencein different readings. I tentatively conclude, however, that no polysemy needs to be assumedin the spatial-functional meaning. Furthermore, the causal structure of the event to opendoes not seem to interact with the spatial-functional structure, hence the dynamic verb toopen has no polysemic extensions of its own (in the domain under discussion) and inheritsall its spatial-functional features from its resultant state.

8

.local info

.some phone numbers

Taxi Phone: +49 / (0)7531 / 22222

“geoEvent emergency”: +49 / 171 / 209 46 38 (Matthias Weisgerber, mobile)

.taking the bus . . . → See also bus map on p. 12.

• Line 9A / B (not on Saturdays):

The Line 9 bus stops just in front of the University.

Universitat . . . .13/.26/.43/.56 . . .↓

Central Station 8.32 8.47 9.02 9.17 9.32 . . . .32/.47/.02/.17 . . .↓

Universitat 8.46 8.59 9.16 9.29 9.46 . . . .46/.59/.16/.29 . . .

• Line 4: (Saturdays’ option)

Taking the Line 4 bus, you will reach the University within a five minutes walk fromthe bus stop “Egg”.

Central Station 8.08 8.38 9.08 9.38 . . . .08/.38 . . .↓

Egg / Universitat 8.23 8.53 9.23 9.53 . . . .21/.51 . . .

Egg / Universitat . . . .10/.40 . . . 19.40 20.16 21.16 . . . 0.16↓

Marktstatte (C. St.) . . . .26/.56 . . . 19.56 20.32 21.32 . . . 0.32

9

.maps . . .

Map 1: The old town of Konstanz

Maps 2, 3: University of Konstanz

Map 4: Bus lines