Sam Featherston

Data in generative grammar : The stick and the carrot

Abstract This paper is intended to lay out for broader discussion some arguments for the importance of data in work in generative syntax. These are accepted by many linguists, but a significant number of others still seem reluctant to accept them. The basic claim is that it is no longer tenable for syntactic theories to be constructed on the evidence of a single persons judgements, and that real progress can only be made when syntacticians begin to think more carefully about the empirical basis of their work and apply the minimum standards we propose. We advance two groups of reasons for syntacticians to do this, negative and positive. The negative ‘stick’ group concerns the inadequacy of current practice. We argue that linguists are producing unsatisfactory work with these methods. Data quality is a limiting factor: a theory can only ever be as good as its data base. The positive ‘carrot’ group concerns the descriptive and theoretical advantages which become available with more empirically adequate data. We hope to tempt linguists to adopt new methods by showing them the insights which better data makes available.

New standardised texts for assessing reading performance in four European languages.

Aims: To develop standardised texts for assessing reading speed during repeated measurements and across languages for normal subjects and low vision patients. Methods: 10 texts were designed by linguistic experts in English, Finnish, French, and German. The texts were at the level of a sixth grade reading material (reading ages 10–12 years) and were matched for length (830 (plus or minus 2) characters) and syntactic complexity, according to the syntactic prediction locality theory of Gibson. 100 normally sighted native speaking volunteers aged 18–35 years (25 per language) read each text aloud in randomised order. The newly designed text battery was then applied to test the reading performance of 100 normally sighted native speaking volunteers aged 60–85 years (25 per language). Results: Reading speed was not significantly different with at least seven texts in all four languages. The maximum reading speed difference between texts, in the same language was 6.8% (Finnish). Average reading speeds (SD) in characters per minute are, for the young observer group: English 1234 (147), Finnish 1263 (142), French 1214 (152), German 1126 (105). The group of older readers showed statistically significant lower average reading speeds: English 951 (97), Finnish 1014 (179), French 1131 (160), German 934 (117). Conclusion: The authors have developed a set of standardised, homogeneous, and comparable texts in four European languages (English, Finnish, French, German). These texts will be a valuable tool for measuring reading speed in international studies in the field of reading and low vision research.

Antecedent Priming at Trace Positions: Evidence from German Scrambling

In Chomskys theory of grammar, syntactic representations are said to contain movement traces, i.e., syntactically active but phonetically null copies of displaced constituents. Correspondingly, traces have been claimed to form part of the processing of sentence structure by showing that at trace sites the parser reactivates a moved constituent. This view has been contested, however, by researchers arguing that experimental findings can better be explained in terms of direct associations between subcategorizers and arguments. Against this background, we investigate antecedent reactivation effects in scrambled double-object constructions of German in two cross-modal priming experiments. We found significant priming effects at positions at which a movement analysis of these constructions would postulate an empty category, thus suggesting that the antecedent is indeed reactivated at the gap position. The Direct Association Hypothesis, on the other hand, cannot account for the priming effects we found. Implications for processing and for syntactic analyses of scrambling in German will be discussed.

Brain Potentials in the Processing of Complex Sentences: An ERP Study of Control and Raising Constructions

In the present study we made use of Event-Related Potentials (ERPs) to examine raising and subject control constructions in German. Our most salient result is that the ERPs elicited at the empty subject position of a raising construction are clearly different from those elicited at the corresponding position of an otherwise identical subject control construction, the former producing a stronger P600. We argue that this result provides an electrophysiological correlate of the theoretical distinction between NP trace and PRO.

Universals and grammaticality : wh-constraints in German and English

Abstract In this article we investigate the movement constraints superiority and discourse linking. These have played an important role in the tradition of generative syntax and might therefore be expected to be universal, but they are usually argued to be absent from German. We looked for evidence of them in German data using the methodology of magnitude estimation of wellformedness, and compared this data with parallel results from English. The results showed these effects to be robustly active in the grammar of German, and revealed few differences between the two languages. We suggest that the reason why linguists have denied their existence in German is that they have been assuming a binary and categorical concept of grammaticality, forgetting that this is merely a simplifying abstraction from the primary linguistic data. We demonstrate that the admittedly convenient assumption of categorical grammaticality is obscuring our view of the syntax, and that studies using our own more empirically adequate assumptions of grammaticality can be productive. In particular, we hope that our conceptions of constraint survivability and definition of syntax relevance may permit insights into the size of the grammar, crosslinguistic variation, and syntactic universals.

The Decathlon Model of Empirical Syntax

This article summarizes the findings of some of our studies of the data base of syntactic theory, contrasting the characteristics of frequency data and judgement data. Examination of frequency data reveals that the factors affecting its production interact competitively and probabilistically. This contrasts strongly with the patterns observed in judgement data, which point to a system in which violations of constraints produce negative weightings on form/meaning pairs. Since both data types are the result of human linguistic processing, we present a model of the architecture that such a system might have in order to produce such contrasting data. This Decathlon Model has two modules: Constraint Application and Output Selection. The first is blind, exceptionless and applies violation costs cumulatively (Keller 2000), the second is competitive and probabilistic. This constrains frameworks of syntactic explanation: an empirically adequate grammar must include gradient well-formedness, specify constraint violation costs, and distinguish between the application of rules and the selection of outputs. In this paper reports our investigations into the data base of syntactic theory, specifically addressing the similarities and differences between corpus data and judgements and sketching the implications for the construct of grammaticality and the architecture of the grammar which our findings have. The motivation for these studies was a dissatisfaction with the state of affairs in syntax, when, for example, two syntacticians can look at the same phenomenon and come up with widely differing analyses of what is going on. Another disappointment is the lack of any real forward movement in theory: alternative analyses seem to succeed each more due to fashion than due to falsification. We might say that syntactic description, let alone syntactic explanation, is underdetermined by its data base. In fact most data feeding into syntactic theory has significant flaws: it is fuzzy, it reflects multiple factors, only some of which are relevant to theory, and perhaps worst of all, these factors are poorly understood and not clearly

Empty Categories in Sentence Processing

This book reports a research program into one of the most controversial questions in the syntax — processing interface: The behavior of the parser at gap positions. While the work done is largely experimental, the results are analyzed both for their relevance to sentence processing and for their implications for competing syntactic frameworks. In particular the differing predictions of PPT and HPSG for structures with dislocated constituents are tested for their empirical adequacy. The author addresses a broad range of questions about gap processing and uses a broad range of methodologies to cut through the confounds which prevent previous work providing clear answers. Wh-movement, scrambling, raising, and equi structures are all addressed, and all current accounts of the experimental evidence evaluated. The results move the debate forward significantly, and provide clear confirmation of some non-trivial claims of generative grammar.

Relax, lean back, and be a linguist

In this text I wish to raise some questions about the quantity and quality of data required in grammar research. In previous work I have argued, often emphatically, in favour of using more and better data, so that grammatical hypotheses are more descriptively adequate, and theory building less speculative. My arguments have been just a small part of a wider intellectual trend in this direction, since others have had similar thoughts, and research in the field is increasingly using more evidence-based argumentation. I therefore think that we can assume that the case for greater empirical input into grammatical generalizations and theory building is established. My aim here is to question how far this trend need go. The title of this piece is intended to sum up my suggestion that, while it was necessary and indeed urgent for researchers in grammar to consider data and respect data more, it is neither necessary nor perhaps desirable for them to spend too much time worrying about the finer points of data collection and analysis. Some researchers will take a particular interest in methodology and will innovate and set new standards, but others will be more conservative and rely to a greater extent on traditional methods, while still accepting the basic premise that a theory is an account of data. My basic point is that we need to regard both paths as valid, if we wish the field of research in grammar to remain a single unit, and not drift apart into separate discourses, to the disadvantage of both parts. One of the insights of the SFB 441 Linguistic Data Structures was the extent to which the data types and approaches represented had significant features in common. In particular, it became clear that linguistic data is always complex and requires filtering, interpretation, and location within a wider model to yield its full evidential value. The developing different wings of grammar research therefore need each other; neither all-data nor all-theory can have as much value as a judicious combination of the two.

Bridge verbs and V2 verbs: the same thing in spades?

Abstract In this paper we investigate the quality of ‘bridgeness’, the extent to which a verb permits extraction from its complement clause. In spite of considerable study, linguists have only a limited idea of the nature of the factor involved. We address the question whether the quality of being a bridge verb is the same as the quality which permits a verb to take a verb-second complement clause in German. We report the results of two experiments using the technique of magnitude estimation, a method of eliciting grammaticality judgements with a high degree of differentiation. The data reveals that the ‘bridge’ quality and the ‘V2’ quality co-vary very closely, suggesting that they may well be facets of the same factor. We discuss the implications this finding has for the nature of the bridge quality.

Three types of exceptions – and all of them rule-based

A basic premise of this paper is that a simpler grammar is a more adequate one, and that thus exceptions are undesirable. We present studies concerning three different grammatical structures which contain phenomena standardly regarded as exceptions, and show how, in all three cases, the attribution of the status as an exception was unnecessary. In each case, the collection of better data and the explanatory advantages of firstly, a model of gradient grammaticality, and secondly the distinction between the effects of the grammar and the effects of production processing, reveal the phenomenon to be rule-governed.