Lars Konieczny

Syntactic Ambiguity Resolution In German

This chapter discusses sentence processing research from differential perspective and from universal perspective. It argues the ways in which the peculiarities of German give the opportunity to provide a new test ground for theories of sentence processing in general. The chapter explains evidence on constructions which are highly comparable, not only in German and English, but also many other languages. It discusses constraints on models of human sentence processing in general which can be derived from experiments on German. The chapter talks about attachment preferences and the principle of parameterized head attachment based on experiments on PP- and NP attachment as well as extensions of the models which have proven to be necessary to explain relative clause attachment preferences. Keywords: clause attachment preferences; English; German; parameterized head attachment; sentence processing research

Linking Syntactic Functions with Thematic Roles: Psych-Verbs and the Resolution of Subject-Object Ambiguity

Subject-object ambiguities like in (1a-d), where the first NP can either be interpreted as the subject or as the direct object of the sentence, are one of the most intensively studied phenomena of German sentence processing, as is reflected by the number of contributions on this topic in the current volume (cf. the chapters of Markus Bader, Paul Gorrell, Lars Konieczny et al., and Matthias Schlesewsky et al.).

Modifier Attachment: Relative Clauses and Coordinations

In this paper, we will discuss accounts of cross-linguistic differences in attachment preferences for relative clauses. In the first section, we will present an overview of data from two-site attachment ambiguities like (1). We will argue for a modular model of sentence processing where a discourse-based preference for a salient antecedent of the relative pronoun and a syntax-based recency preference contribute to empirically observable attachment preferences. (1) the daughter of the teacher who lived in France In the second section, we will extend this account to three-site ambiguities like (13,3) which were first investigated by Gibson et al. (1996a, b), based on three German questionnaire experiments. (2) the lamp near the painting in the house that was damaged in the flood (Gibson et al., 1996a) (3) the customer with the child with the dirty face and the wet diaper the one with the wet diaper the one with the baby with the wet diaper

Modifier Attachment in German: Relative Clauses and Prepositional Phrases

Abstract In this chapter, we present new results from an eye-tracking experiment on NP1–NP2-modifier attachment in German. We will focus on attachment preferences across modifier types, namely relative clauses (RCs) and semantically highly comparable prepositional phrases (PPs), showing that different preferences hold for these two types of constructions. Whereas PP attachment follows recency-based principles, resulting in low attachment, RCs are preferentially attached high. These data provide an on-line confirmation of results from various questionnaire studies presented in Hemforth, Konieczny and Scheepers (in press; see also Hemforth, Konieczny and Scheepers, 1994). Furthermore, the effects for RCs and PPs showed up most clearly either in regression-sensitive (regression path durations) and regression-insensitive (first-pass reading times) measures, respectively. To provide the means for analysing the re-reading process, we established a new measure (load contribution).

Studien zur performanzorientierten Linguistik

SummaryLooking at relative clause extraposition in German as a concrete example, the paper demonstrates how linguistic model building, corpus study and psycholinguistic experiments combine into an integrational research programme that aims at an improved understanding and linguistically as well as cognitively adequate modelling of human language performance. Starting from the word order theory articulated by (Hawkins, 1994), we formulate hypotheses on the positional distribution of relative clauses and evaluate them on the basis of data drawn from corpus analyses and acceptability experiments. Our empirical studies confirm the expectation that grammatical weight has a significant effect on relative clause distribution; at the same time, there is an interesting asymmetry between language production and perception data.ZusammenfassungAm Beispiel der Relativsatzextraposition im Deutschen zeigt der Beitrag, wie Verfahren der sprachwissenschaftlichen Modellbildung, korpuslinguistischen Untersuchung und des psycholinguistischen Experiments in einem integrativen Forschungsansatz zusammenwirken, der auf ein verbessertes Verständnis und die linguistisch wie kognitiv adäquate Modellierung sprachlicher Performanzprobleme zielt. Ausgehend von der von Hawkins (1994) formulierten Theorie zur Wortstellung werden Hypothesen über die positionelle Verteilung von Relativsätzen formuliert und in Bezug auf Korpusdaten und Akzeptabilitätsmessungen überprüft. Alle beschriebenen empirischen Untersuchungen bestätigen den erwarteten Einfluß von Längenfaktoren auf die Relativsatzdistribution, zeigen gleichzeitig aber eine interessante Asymmetrie zwischen Produktionsund Rezeptionsdaten.

Cognitive Parsing in German: An Introduction

Research in human sentence processing is a genuinely cross-linguistic project. Since most theories of human parsing make predictions based on the architecture and the mechanisms of the human sentence processor in general, these predictions are assumed to hold for all human comprehenders irrespective of their mother tongue. Ideally, the only difference between processing phenomena in different languages should be due to different grammatical features of the respective languages. In this view, it should be possible to plug the grammar of a particular language into the sentence processor and have all sorts of syntactic complexity phenomena or preferences in ambiguity resolution fall out (Frazier, 1987). This enterprise can obviously only be pursued if theories of human language processing are applied to and tested in as many languages as possible. Nevertheless, by far the majority of research in human sentence processing has been carried out in English. This picture is now changing, however. More and more cross-linguistic studies on human language processing are being published on a wide variety of languages.

Reanalysen vs. interne Reparaturen beim Sprachverstehen

Quasi-deterministische Modelle der Satzverarbeitung wie das Garden-PathModell (Frazier & Rayner, 1982) und der SOUL-Prozessor (Konieczny, Scheepers, Hemforth & Strube, 1994; Konieczny, Hemforth, Scheepers & Strube, 1995) mussen auser Prinzipien der ersten Analyse ebenso eine Theorie der Reanalyse angeben, um erklaren zu konnen, wie Versuchspersonen von einem eventuellen Holzweg zuruckfinden. Die Begriffe Reanalyse und Reparatur sind im Kontext der Satzverarbeitung oft synonym verwendet worden. In diesem Beitrag werden wir aufzeigen, das Reparaturen von Reanalyseprozessen unterschieden werden sollten. Wir werden Reanalyse und interne Reparatur1 (im folgenden kurz Reparatur genannt) als Mechanismen beschreiben, die auf unterschiedliche Art und Weise mit lokalen Inkonsistenzen zwischen bereits analysierter Struktur und neuem Material umgehen.

Studien zur performanzorientierten LinguistikStudies in Performance-Oriented Linguistics: Aspects of German Relative Clause Extraposition

SummaryLooking at relative clause extraposition in German as a concrete example, the paper demonstrates how linguistic model building, corpus study and psycholinguistic experiments combine into an integrational research programme that aims at an improved understanding and linguistically as well as cognitively adequate modelling of human language performance. Starting from the word order theory articulated by (Hawkins, 1994), we formulate hypotheses on the positional distribution of relative clauses and evaluate them on the basis of data drawn from corpus analyses and acceptability experiments. Our empirical studies confirm the expectation that grammatical weight has a significant effect on relative clause distribution; at the same time, there is an interesting asymmetry between language production and perception data.ZusammenfassungAm Beispiel der Relativsatzextraposition im Deutschen zeigt der Beitrag, wie Verfahren der sprachwissenschaftlichen Modellbildung, korpuslinguistischen Untersuchung und des psycholinguistischen Experiments in einem integrativen Forschungsansatz zusammenwirken, der auf ein verbessertes Verständnis und die linguistisch wie kognitiv adäquate Modellierung sprachlicher Performanzprobleme zielt. Ausgehend von der von Hawkins (1994) formulierten Theorie zur Wortstellung werden Hypothesen über die positionelle Verteilung von Relativsätzen formuliert und in Bezug auf Korpusdaten und Akzeptabilitätsmessungen überprüft. Alle beschriebenen empirischen Untersuchungen bestätigen den erwarteten Einfluß von Längenfaktoren auf die Relativsatzdistribution, zeigen gleichzeitig aber eine interessante Asymmetrie zwischen Produktionsund Rezeptionsdaten.