Structure of the talk. The semantics of event nominalisation. Event nominalisations and verbal arguments 2



Similar documents
Complex Predications in Argument Structure Alternations

Modeling adjectives in GL: accounting for all adjective classes 1

Chapter 1. Introduction Topic of the dissertation

Support verb constructions

Varieties of specification and underspecification: A view from semantics

Paraphrasing controlled English texts

Outline of today s lecture

CINTIL-PropBank. CINTIL-PropBank Sub-corpus id Sentences Tokens Domain Sentences for regression atsts 779 5,654 Test

Movement and Binding

INF5820 Natural Language Processing - NLP. H2009 Jan Tore Lønning jtl@ifi.uio.no

Overview of the TACITUS Project

Annotation Guidelines for Dutch-English Word Alignment

Non-nominal Which-Relatives

Each vs. Jeweils: A Cover-Based View on Distance Distributivity

Semantics and Generative Grammar. Quantificational DPs, Part 3: Covert Movement vs. Type Shifting 1

A Minimalist View on the Syntax of BECOME *

Syntactic Theory on Swedish

Semantic analysis of text and speech

Double Genitives in English

The syntactic positions of adverbs and the Second Language Acquisition

COMPARATIVES WITHOUT DEGREES: A NEW APPROACH. FRIEDERIKE MOLTMANN IHPST, Paris fmoltmann@univ-paris1.fr

Phase 2 of the D4 Project. Helmut Schmid and Sabine Schulte im Walde

English prepositional passive constructions

How to Make a Domain Model. Tutorial

Title: Chinese Characters and Top Ontology in EuroWordNet

L130: Chapter 5d. Dr. Shannon Bischoff. Dr. Shannon Bischoff () L130: Chapter 5d 1 / 25

[Refer Slide Time: 05:10]

The finite verb and the clause: IP

Do we need Structured Question Meanings? Two Approaches to Questions


English Descriptive Grammar

Special Topics in Computer Science

ONTOLOGY BASED FEEDBACK GENERATION IN DESIGN- ORIENTED E-LEARNING SYSTEMS

PTE Academic Preparation Course Outline

From Logic to Montague Grammar: Some Formal and Conceptual Foundations of Semantic Theory

Chapter 13, Sections Auxiliary Verbs CSLI Publications

The compositional semantics of same

How the Computer Translates. Svetlana Sokolova President and CEO of PROMT, PhD.

ONTOLOGIES A short tutorial with references to YAGO Cosmina CROITORU

IP PATTERNS OF MOVEMENTS IN VSO TYPOLOGY: THE CASE OF ARABIC

NOVEL METONYMY AND NOVEL METAPHOR AS PRIMARY PRAGMATIC PROCESSES 1. Esther Romero González and Belén Soria Clivillés

Introduction. BM1 Advanced Natural Language Processing. Alexander Koller. 17 October 2014

Invited talks (in Hungary unless marked otherwise)

Chapter 7: Functional Programming Languages

Teaching Formal Methods for Computational Linguistics at Uppsala University

Differences in linguistic and discourse features of narrative writing performance. Dr. Bilal Genç 1 Dr. Kağan Büyükkarcı 2 Ali Göksu 3

Clustering Connectionist and Statistical Language Processing

Interactive Dynamic Information Extraction

THE ACQUISITION OF PREPOSITIONAL CONSTRUCTIONS AND THEIR ASSOCIATED CASE-MARKING PROPERTIES IN THE L2 GERMAN OF L1 ENGLISH SPEAKERS

The English Genitive Alternation

Collateral Feature Discharge

Minnesota K-12 Academic Standards in Language Arts Curriculum and Assessment Alignment Form Rewards Intermediate Grades 4-6

Syntactic and Semantic Differences between Nominal Relative Clauses and Dependent wh-interrogative Clauses

Classification of Fuzzy Data in Database Management System

Online multilingual generation of Cultural Heritage content

THE ENTITY- RELATIONSHIP (ER) MODEL CHAPTER 7 (6/E) CHAPTER 3 (5/E)

Type Theory and Lexical Decomposition

Teaching Vocabulary to Young Learners (Linse, 2005, pp )

Introduction. Philipp Koehn. 28 January 2016

ORAKEL: A Portable Natural Language Interface to Knowledge Bases

Information Technology for KM

Constraints in Phrase Structure Grammar

THE SEMANTICS OF ASPECTUALIZERS IN ENGLISH. Tünde Nagy Debrecen University

Information Retrieval, Information Extraction and Social Media Analytics

Compositionality, of, word-formation

Some Methodological Clues for Defining a Unified Enterprise Modelling Language

A Service Modeling Approach with Business-Level Reusability and Extensibility

A Chart Parsing implementation in Answer Set Programming

A Beautiful Four Days in Berlin Takafumi Maekawa (Ryukoku University)

Appendix B Data Quality Dimensions

NON-CORE PARTICIPANT PPS ARE ADJUNCTS. György Rákosi. University of Debrecen. Proceedings of the LFG12 Conference

The Syntax of Predicate Logic

Transcription:

Structure of the talk Sebastian Bücking 1 and Markus Egg 2 1 Universität Tübingen sebastian.buecking@uni-tuebingen.de 2 Rijksuniversiteit Groningen egg@let.rug.nl 12 December 2008 two challenges for a semantic description of nominalisations argument linking anti-iconic readings the framework of analysis: underspecified semantic representation the analysis semantic construction Event nominalisations and verbal arguments 1 Event nominalisations and verbal arguments 2 derivational affixes turn verbs into nouns denoting events or one of their participants (1) die Verarbeitung der Daten the processing of the data (2) beautiful dancer arguments of the verb can become optional during nominalisation others are explicitly bound in the nominalisation process this binding refers to thematic roles of arguments agent for the agentive suffix -er theme for the genitive complement of -ung-nominalisations to handle reference to thematic roles and argument linking, we use a Neo-Davidsonian approach (Parsons 1990) verbs and their arguments are linked in terms of explicit thematic relations (3) λyλxλe.eat (x,y)(e) (4) λyλxλe.eat (e) AGENT(e,x) THEME(e,y) no need to bind off syntactically unrealised arguments of the verb stem in the semantics (5) λxλe. y.eat (x,y)(e) (6) λxλe.eat (e) AGENT(e,x)

Event nominalisations and verbal arguments 3 Anti-iconic readings of event nominalisations 1 nominalisation affixes are semantic functions and their stems are arguments the affixes introduce the relevant thematic relations the agentive -er: (7) λpλx e.p(e) AGENT(e, x) the affix -ung (preliminary): (8) λpλxλe.p(e) THEME(e, x) semantic construction for (1) semantics of Verarbeitung processing : λxλe.process (e) THEME(e,x) semantics of Verarbeitung der Daten processing of the data : λe!x.[data (x)] process (e) THEME(e,x) modified agentive nouns exhibit ambiguity due to several possible ways of integrating the modifier semantics (9) beautiful dancer a beautiful person who dances a person who dances beautifully the second reading can be analysed as an anti-iconic, in that the modifier can pertain to the stem only (Egg 2006) (10) λpλx e.p(e) AGENT(e, x) (λpλx.beautiful (x) P(x) (dance )) = λx e.beautiful (e) dance (e) AGENT(e,x) Anti-iconic readings of event nominalisations 2 Underspecified representation of anti-iconic cases 1 but this approach does not work for the simplistic semantics for -ung in (11) (11) [=(8)] λpλxλe.p(e) THEME(e, x) (12) die schnelle Verarbeitung der Daten processing is over soon (external reading) throughput of the data is high (internal reading) in both readings, schnell fast characterises the speed of an event internal and external reading (Maienborn 2003) are not distinguished syntactically (11) characterises the semantics of -ung as intersective no scope ambiguity for the two readings is possible there is only one event for the modifier to predicate over underspecified representation ( constraint ) for the meaning of (9) (13) [[NP]] : [[NP S ]]: λx e. (e) AGENT(e,x) dance λy. (y) beautiful (y) sets of semantic representations (here, λ-terms) are described on a meta-level ingredients: fragments of λ-terms, holes, and relations between them

Underspecified representation of anti-iconic cases 2 s 1 the described semantic representations ( solutions ) are derived by identifying fragments and holes formally, a bijective mapping (the plugging of Bos 2004) the solutions of (13) (14) λx e.dance (e) AGENT(e,x) beautiful (x) (15) λx e.dance (e) AGENT(e,x) beautiful (e) possible relations between the meanings of verbal predicates and their nominalizations verb nominalisations have the same meaning as their verb stem (Parsons 1990; Cocchiarella 1996) verb nominalisations introduce concrete events that are characterised (relation ) by abstract event concepts in terms of verb stem meanings (Cocchiarella 1996) we follow position 2 and distinguish types for concrete events and for event concepts (cp. Asher 1993) variables for concrete events are abbreviated as e variables for event concepts are abbreviated as E s 2 s 3 the more elaborate semantics of the affix -ung is split into a main and a secondary fragment (16) main fragment: λpλe.e λe.p(e) (17) secondary fragment: λpλxλe.theme(e, x) P(e) the main fragment introduces a bipartite event structure -ung does not bind the event argument of the verbal stem it introduces a new event argument, i.e. the concrete event e, that is related to the verbally given event concept argument E via the secondary fragment introduces the theme argument the modifier can pertain to two different entities the underlying verbal E (resulting in the internal reading) the nominal e (resulting in the external reading) the verbal theme argument is made available for binding by a subsequent genitive DP argument semantic representation of Verarbeitung processing (18) [[N]] : λe.e λe. (E) [[N S ]] : λxλe.theme(e,x) process (e)

Underspecified representation of event nominalisations 1 Underspecified representation of event nominalisations 2 semantic representation of Verarbeitung der Daten processing of the data (19) λe!x.[data (x)] (e) [[NP]] : [[NP S ]] : λe.theme(e,x) process (e) λe.e λe. (E) constraints like (19) have one solution only unwanted scope ambiguities between top fragment of nominalisations and DP argument fragment are blocked this is due to the typing of event (concept) variables the DP semantics introduces concrete events e the inner event variable in nominalisations is an event type E verb (and adverbial) semantics introduce a general event type e T (e T = e E) the only solution for (19): (20) λe!x.[data (x)] e λe.theme(e,x) process (E) in the following, reference to these types is mostly dropped Underspecified representation of event nominalisations 3 Underspecified representation of event nominalisations 4 representation for schnelle Verarbeitung der Daten fast processing of the data (21) [[NP]] : solutions by starting (22)... from the left: the internal reading λe!x.[data (x)] e λe.theme(e,x) process (E) fast (E) λe!x.[data (x)] (e) λe.e λe. (E) [[NP S ]] : λy. λe.theme(e,x) process (e) (y) fast (y) (23)... from the right: the external reading λe!x.[data (x)] e λe.theme(e,x) process (E) fast (e)

Locatives 1 Locatives 2 locatives are another instance of external vs. internal reading (24) die Zubereitung des Huhns in der Pfeffersauce the preparation the chicken GEN in a pepper-sauce (25) *die Zubereitung in der Pfeffersauce des Huhns the preparation in the pepper-sause the chicken GEN (26) die Zubereitung des Huhns in der Küche the preparation the chicken GEN in the kitchen diamond for Zubereitung des Huhns in der Pfeffersauce preparation of the chicken in the pepper sauce ; simplified w.r.t. the DP argument semantics (27) λe.e λe. (E) λy. (y) in (y,p) λe.theme(e,ιx.chicken (x)) prepare (e) Locatives 3 Locatives 4 solutions of (27) (28) (a) internal reading λe.e λe.in (E,P) THEME(E,ιx.chicken (x)) prepare (E) (b) external reading λe.in (e,p) e λe.theme(e,ιx.chicken (x)) prepare (E) the second reading is ruled out pragmatically: the preparation event as a whole is localised in the pepper-sauce the first reading raises the question of what it means to localise the lexical base E, i.e., how does this apply to the chicken? Maienborn s (2003) proposal for adverbial locatives (29) MOD*: λqλpλx[p(x)&r(x, v)&q(v)] Condition: if MOD* applies to categorial type X, R = part-of, otherwise (in an XP-environment) R is identity. (30) [ VP [ PP in der Küche] [ VP das Huhn zubereiten]] [ VP [ PP in the kitchen] [ VP the chicken prepare]] (31) [ VP das Huhn [ V [ PP in einer Pfeffersauce] [ V zubereiten]]] [ VP the chicken [ V [ PP in a pepper-sauce] [ V prepare]]] (32) external: λe... prepare (e) in (e,k)... (33) internal: λe... prepare (e) part-of (e,v) in (v,p)... in the internal case, the free variable v is conceptually specified

Locatives 5 of anti-iconic cases 1 transfer to internal locatives in ung-nominalizations? here the choice between part-of and identity cannot be conditioned structurally i.e., MOD* can vary freely w.r.t. the two assignments for R external vs. internal locative revisited: (34) λe.in (e,p) e λe.theme(e,ιx.chicken (x)) prepare (E) (35) (a) λe.e λe.prepare (E) part-of (E,v) in (v,p) THEME(E,ιx.chicken (x)) (b) λe.e λe.prepare (E) in (ιx.chicken (x),p) THEME(E,ιx.chicken (x)) somewhere the gap between syntactic and semantic structure must be bridged several possibilities have been suggested syntactic preprocessing (Generative Grammar) involved syntactic construction (L-TAG, Kallmeyer and Romero 2008; LFG, Çetinoǧlu and Oflazer 2006) powerful syntax-semantics interface (Egg 2006) semantic representations are constraints a main and a secondary fragment are distinguished interface rules refer to these fragments and determine them for emerging constituents 2 3 rule for affixation (36) [ X Bs Aff] (morph) [[X]] : [[Aff]]() [[X S ]] : [[Aff S ]]([[Bs]]) semantics of Verarbeitung processing (38) [[N]] : λpλe.e λe.p(e)() affixes like -er that do not influence the secondary fragment of the emerging word have the identity mapping as secondary fragment semantics of -ung (37) [[Aff]]: λpλe.e λe.p(e) [[Aff S ]]: λpλyλe.p(e) THEME(e,y) [[N S ]] : [λpλyλe.p(e) THEME(e,y)](process ) after β-reduction (39) [[N]] : λe.e λe. (E) [[N S ]] : λyλe.process (e) THEME(e,y)

4 5 rule for DP arguments (40) [ X X DP] (SSI) [[ X]]: [[DP]]; [[ X S ]]: [[X S ]]([[DP S ]]) rule for projection to XP (42) [ XP X] (SSI) [[XP]] : [[XP S ]]:[[ X]] [[ X S ]] after the integration of the DP-argument der Daten of the data (41) λe.e λe.(e) [[ N]] : λe!x.[data (x)] (e) [[ N S ]]:λe.process (e) THEME(e,x) Verarbeitung der Daten processing of the data as NP: (43) [[NP]] : λe.e λe. (E) [[NP S ]] : λe!x.[data (x)] λe.process (e) THEME(e,x) (e) Asher, N. (1993). Reference to abstract objects in discourse. Dordrecht: Kluwer. Bos, J. (2004). Computational semantics in discourse: Underspecification, resolution, and inference. Journal of Logic, Language and Information 13, 139 157. Çetinoǧlu, Ö. and K. Oflazer (2006). Morphology-syntax interface for Turkish LFG. In Proceedings of the 21st COLING and the 44th ACL, 153 160. Cocchiarella, N. (1996). Conceptual realism as a formal ontology. In R. Poli and P. Simons (eds), Formal Ontology, 27 60. Dordrecht: Kluwer. Egg, M. (2006). Anti-Ikonizität an der Syntax-Semantik-Schnittstelle. Zeitschrift für Sprachwissenschaft 25, 1 38. Kallmeyer, L. and M. Romero (2008). Scope and situation binding in LTAG using semantic unification. Research on language and computation 6, 3 52. Maienborn, C. (2003). Event-internal modifiers: Semantic underspecification and contextual interpretation. In E. Lang, C. Maienborn, and C. Fabricius-Hansen (eds), Modifying adjuncts, 475 509. Berlin: Mouton de Gruyter. Parsons, T. (1990). Events in the semantics of English. Cambridge: MIT Press.