Mapping out‐ argument structure

Event-Based Semantics

Representing the base-10 structure of numbers is a challenging cognitive ability, unique to humans, but it is yet unknown how precisely this is done. Here, we examined whether and how literate adults represent a number’s full syntactic structure. In 5 experiments, participants repeated number-word sequences and we systematically varied the order of words within each sequence. Repetition on grammatical sequences (e.g., two hundred ninety-seven) was better than on non-grammatical ones (hundred seven two ninety). We conclude that the participants represented the number’s full syntactic structure and used it to merge number words into chunks in short-term memory. Accuracy monotonously improved for sequences with increasingly longer grammatical segments, up to a limit of ~ 4 words per segment, irrespectively of the number of digits, and worsened thereafter. Namely, short chunks improved memorization, whereas oversized chunks disrupted memorization. This chunk size limit suggests that the chunks are not based on predefined structures, whose size limit is not expected to be so low, but are created ad hoc by a generative process, such as the hierarchical syntactic representation hypothesized in Michael McCloskey’s number-processing model. Chunking occurred even when it disrupted performance, as in the oversized chunks, and even when external cues for chunking were controlled for or were removed. We conclude that the above generative process operates automatically rather than voluntarily. To date, this is the most detailed account of the core representation of the syntactic structure of numbers—a critical aspect of numerical literacy and of the ability to read and write numbers.

ABSTRACT: One of the main objectives of valency lexicons is to identify predicate argument structures evoked by verbs. The analysis used in these resources heavily rests on lexical semantics and as such provides an accurate description of event structure in cases in which a verb is used in its prototypical sense, such as when the verb kick is used as a force verb in the example He kicked the door . However, this approach to event structure representation proves inadequate for examples in which a verb is used in a non-prototypical sense in which additional event participants are evoked constructionally, such as the use of kick in a transfer event: He kicked him the ball . In this paper, we propose a two-tier representation of event structure in which lexical semantics and the semantics of argument structure constructions are dealt with independently. We present a detailed model of constructional analysis that is compatible with lexically based analyses of event structure used in valency lexicons, such as Propbank, and show that a mapping between lexical and constructional representations provides a more comprehensive understanding of the event semantics when compared to a lexical representation alone.

The term ‘discourse’ is used to describe the entangled and contested transactions through which real-world planning and policy issues are addressed. Studies of discourses have become an important way of understanding how power is mediated in planning, but the methods through which discourses are identified and evaluated is as yet unclear in the literature. Here, we describe our attempt—with still only limited success—to map discourses using a method that extends the work of Toulmin and Gasper and George. Our method consists of a tabular representation of argument structure to depict the content and structure of a discourse, and a graphical index to the discourse table to reveal higher order patterns in the discourse. Using discourse pertaining to a real-life design-review case, we demonstrate how our approach allows us to understand the internal structure of that discourse better. We conclude with suggestions for how the method might be further improved.

Topic maps have been developed in order to represent the structures of relationships between subjects, independently of resources documenting them, and to allow standard representation and interoperability of such structures. The ISO 13250 XTM specification [] have provided a robust syntactic XML representation allowing processing and interchange of topic maps. But topic maps have so far suffered from a lack of formal description, or conceptual model. We propose here such a model, based on the mathematical notions of hypergraph and connexity. This model addresses the critical issue of topic map organization in semantic layers, and provides ways to check semantic consistency of topic maps. Moreover, it seems generic enough to be used as a foundation for other semantic standards, like RDF [].

In our approach to event extraction, dependency graphs constitute the fundamental data structure for knowledge capture. Two types of trimming operations pave the way to more effective relation extraction. First, we simplify the syntactic representation structures resulting from parsing by pruning informationally irrelevant lexical material from dependency graphs. Second, we enrich informationally relevant lexical material in the simplified dependency graphs with additional semantic meta data at several layers of conceptual granularity. These two aggregation operations on linguistic representation structures are intended to avoid overfitting of machine learning-based classifiers which we use for event extraction (besides manually curated dictionaries). Given this methodological framework, the corresponding JReX system developed by the JulieLab Team from Friedrich-Schiller-Universitat Jena (Germany) scored on 2nd rank among 24 competing teams for Task 1 in the “BioNLP’09 Shared Task on Event Extraction,” with 45.8% recall, 47.5% precision and 46.7% F1-score on all 3,182 events. In more recent experiments, based on slight modifications of JReX and using the same data sets, we were able to achieve 45.9% recall, 57.7% precision, and 51.1% F1-score.

This experiment was designed to determine whether prepositional phrases are treated as a single undifferentiated type, or whether the parser may recognize different subtypes. In the experiment, participants read temporarily ambiguous prime and target sentences that had either agent or instrument prepositional phrases in the syntactically disambiguating position. Agent and instrument primes both led to significant priming effects for agent targets. Agent primes led to a nonsignificant priming effect for instrument targets, and this priming effect was smaller than the effect that instrument primes had on instrument targets. This pattern can be explained if verb argument structure information is used in structural decisions, and if agent but not instrument roles are obligatory for the class of verbs tested here. The data suggest that readers are likely to activate an implicit agent when they read prime sentences that explicitly mention an instrument, but are not likely to activate an instrument when they read prime sentences that explicitly mention an agent. If the structural representations that incorporate activated arguments persist, or are reactivated more quickly following an appropriate prime sentence, this could lead to facilitated processing of sentences that have the same structural configuration.

Split-intransitive systems of argument marking provide an excellent opportunity to study the structure of the lexical-semantic representations that underlie argument structure alternations and argument linking rules. Yukatek Maya has a typologically rare split-intransitive pattern of argument marking controlled by overt aspect-mood marking. Kramer and Wunderlich (1999) have advanced an analysis according to which the linking of thematic relations to syntactic arguments is governed by lexical aspect as the sole lexical-semantic property linking principles are sensitive to in this language. Critical evidence against this proposal comes from the transitivity alternations of three classes of intransitive verbs: degree achievement verbs, non-internally-caused process verbs, and posture verbs. Transitivity alternations emerge as being governed by the distinction of internally-vs. externally-caused events. The Yukatek facts suggest that argument linking operates on a lexical information structure (event structure) that partially determines (and thus also underspecifies) both lexical aspect and participant structure.

General considerations In this chapter the linking between semantic representations and syntactic representations in complex sentences is explored. An important question to be investigated is the extent to which the linking algorithms proposed in chapter 5 for simple sentences must be modified to deal with complex sentences. Most complex sentences pose no particular difficulties for the linking system developed in chapter 5. Clausal junctures, for example, are composed of clauses, each of which links like an independent clause. Moreover, nuclear junctures act for linking purposes like simple clauses containing a complex predicate, and they basically follow the algorithms for simple sentences. The real challenge comes from non-subordinate core junctures with their obligatory sharing of a core argument, as discussed in section 6.2. Accordingly, linking in clausal and nuclear junctures will be examined first. After the discussion of linking in the different juncture–nexus types and in complex NPs, reflexivization in complex constructions will be investigated, and again the question arises as to the extent to which the principles proposed in section 5.3 will have to be modified to deal with these new phenomena. In the final section an account of the restrictions on so-called ‘long-distance dependencies’ involved in WH-question formation, topicalization and relativization will be proposed. These restrictions, which fall under the principle known as ‘subjacency’ in the generative literature, are significant for linguistic theory, for theories of language acquisition and for related theories of cognitive organization.

Multiword lexical units such as complex predicates, which consist of light verbs and predicative nouns, have long remained on the periphery of traditional syntactic theories. Light verb constructions pose a serious challenge for both theoretical and applied linguistics, as their syntactic structures are not solely determined by verbs alone but also by predicative nouns. There is assumed to be the lack of a comprehensive definition of the concept of light verbs that can account for the cross-linguistic diversity in the types and scope of light verb constructions. In particular, there is still no theoretical framework that adequately describes the complex argument structure formed by the interaction between light verbs and predicative nouns in light verb constructions. This paper investigates the syntactic and semantic representation of Czech light verb constructions. Light verb constructions represent a type of complex predicate where two syntactic elements serve as a single predicate. In this paper I focus to address crucial syntactic-semantic aspects closely related to the fundamental concept of light verbs. Specifically, based on the [+LIGHT] semantic features of desemanticized light verbs, this paper applies reduction and co-reference constraints to analyze the syntactic-semantic structure of light verb constructions. Finally I discuss the question of thematic role assignment between light verbs and predicative nouns.

The so-called cartographic approach to discourse-related word-order variation is based on the idea that particular interpretations—say, contrastive focus—are licensed in the specifier of particular functional projections—say, a focus phrase. In this paper we present arguments against this view based on scrambling in Dutch. We discuss a range of implementations of the cartographic approach and show that they are either too weak, in that they cannot generate all the word orders found in Dutch, or too strong, in that they fail to capture restrictions on scrambling. The alternative we present dispenses with discourse-related functional projections and instead relies on mapping rules that associate syntactic representations with representations in information structure. On this view, scrambling operations derive a syntactic configuration that matches the structural description of a mapping rule that could otherwise not apply. We suggest that it is this interface effect that licenses the marked structures created by scrambling.

The evolution of phonological theory over the past fifteen years or so mirrors in important respects that of syntactic theory. The common evolutionary characteristic is the emergence of modularity. As is frequently noted following Chomsky (1986: ch. 3), in syntax, rich systems of rules have been supplanted by a relatively small number of discrete sub-theories, such as the theories of thematic relations, Case, Binding, Government, and the theory of empty categories. Characteristically, each sub-theory regulates one specific aspect of syntactic structure, at one or more levels of representation, for example the distribution of overt noun phrases at S-structure. The emergence of the sub-theories reflects a natural shift in investigative focus. Just as studying the facts of language from a systematic and formal perspective led to the discovery of generalizations of fact, originally expressed as ‘rules’, so the study of the rules themselves led to the discovery of higher-order generalizations, expressed by the various conditions or principles that make up the contemporary sub-theories. Although Chomsky (1986:70ff.) lists several important contributors to the development of the new perspective, in the mind of most syntacticians, a watershed event in this evolution was Chomsky's own ‘Conditions on transformations’ (1973). To the extent that this development of syntactic theory is a natural one towards deeper understanding, a comparable one is expected in phonological theory. Although the ‘modularity’ of phonology is less frequently noted and identification of a single watershed event is perhaps more difficult, there are clearly several ‘modules’ or sub-theories that have emerged in post-SPEhistory – three in particular.

In this paper, I propose an analysis of Persian complex predicates, based on the First Phase Verbal syntax developed by Ramchand (2008). I suggest that the light verbs lexicalize the subevent heads into which the verbal phrase is decomposed, while the preverbal element occupies the Rheme position and semantically unifies with the light verb to build one joint predication. Further, I propose a feature specification for some of the most productive light verbs. I argue that the light verb is responsible for the argument structure of the entire predicate (in line with Megerdoomian 2002b, Folli et al. 2005), while the aspectual properties of the complex predicate depend on the interaction between the preverb and the light verb.

Although the organization of first language (L1) and second language (L2) lexicosemantic information has been extensively studied in the bilingual literature, little evidence exists concerning how syntactic information associated with words is represented across languages. The present study examines the shared or independent nature of the representation of verb argument structure in the bilingual mental lexicon and the contribution of constituent order and thematic role information in these representations. In three production tasks, Greek (L1) advanced learners of English (L2) generated an L1 prime structure (Experiment 1: prepositional object [PO] and double object [DO] structures; Experiment 2: PO, DO, and intransitive structures; Experiment 3: PO, DO, locative, and “provide(someone)with(something)” structures) before completing an L2 target structure (PO or DO only). Experiment 1 showed L1-to-L2 syntactic priming; participants tended to reuse L1 structure when producing L2 utterances. Experiments 2 and 3 showed that this tendency was contingent on the combination of both syntactic structure and thematic roles up to the first postverbal argument. Based on these findings, we outline a model of shared representations of syntactic and thematic information for L1 and L2 verbs in the bilingual lexicon.

We propose a new kind of event structure representation for computational linguistics, based on the theoretical framework of FirstPhase Syntax (Ramchand, 2008). We show that the approach not only gives a theoretically well-motivated set of subevents and related semantic roles, it also posits the levels of representation needed for analyzing a linguistic phenomenon that has repeatedly caused problems in computational systems, namely the treatment of complex predication. In particular, we look at V+V complex predicates in Urdu/Hindi and show that Ramchand’s subevent decomposition implemented in a VerbNet-style resource allows for a consistent semantic analysis of these complex events. We also show how the proposed event representation can be added to existing resources in the language, in particular the Hindi-Urdu Treebank and Hindi PropBank.