Korpora in der Kognitiven Linguistik
| Wann |
12.10.2008 von 09:30 bis 18:00 |
|---|---|
| Wo | Seminarraum FRIAS |
| Name | Martin Hilpert |
| Kontakttelefon | +0049 761 203-97386 |
| Teilnehmer |
nach Anmeldung |
| Termin übernehmen |
vCal
iCal
|
Organisation: Martin Hilpert (FRIAS)
Cognitive Corpus Linguistics: current issues in theory and methodology
Within cognitive linguistics, there is an increasing awareness that the study of linguistic phenomena needs to be grounded in usage. Ideally, research in cognitive linguistics should be based on authentic language use, its results should be replicable, and its claims falsifiable. Consequently, more and more studies now turn to corpora as a source of data. While corpus-based methodologies have increased in sophistication, the use of corpus data has also generated a number of unresolved questions. In order to promote a critical discussion of these issues, this workshop will take a format that diverges from standard conference proceedings.
In this workshop, participants will discuss a set of critical questions that cognitive corpus linguistics faces at this point in time. Each participant will offer a short 10-minute introduction to a problem or question that will then be discussed for 30 minutes.
Questions to be discussed will include the following:
First, it has become clear that the cognitive category of a prototype is related to raw text frequency, but cannot be equated with it. What frequency measures, if any, are the most reliable predictors of prototypicality, and how can this be determined?
Second, the same problem applies to the notion of entrenchment. How is it that some highly infrequent constructions and collocations still establish themselves as entrenched units?
A third question concerns the level of abstraction that corpus linguists should chose for their respective objects of investigation. Given two or more similar patterns, how can we determine whether these instantiate different cognitive units rather than a single, more general unit?
Fourth, studies increasingly combine corpus analyses with experimental approaches. A question that needs to be addressed is how these approaches can be combined most fruitfully. Specifically, where do experimental results and frequency data make conflicting predictions and how can those be resolved?
Fifth, the expanding methodological toolkit of corpus linguistics has introduced a number of exploratory and inferential statistical techniques into the discipline. How are these tools used most profitably to address questions about cognition? What can be done to communicate their purpose and applicability to the wider community?
The workshop will be advertised as a FRIAS-sponsored satellite event to this year’s ISLE conference. ISLE participants, FRIAS fellows and other guests are welcome to attend.
Program:
Antti Arppe, U Helsinki
"Gains and pains of combining methods and evidence"
Evidence on the same linguistic phenomenon mostly appears to converge, but how neatly does this happen in practice? To illustrate the gains and pains of multimethodological linguistic research, I will walk through the evolution of hypotheses and conclusions in one such study.
Gaëtanelle Gilquin, Centre for English Corpus Linguistics, Louvain
"What corpora can and cannot tell us about cognition: the case of prototypicality"
It has been suggested that language, and in particular language as represented in corpora, can be used to say something about cognition. What exactly it can tell us, however, is still unclear. Using the case of prototypicality, I will show that corpora alone may not be equal to the task of predicting cognitive phenomena.
Dylan Glynn, KU Leuven
"Testing the hypothesis."
Techniques such as Cluster Analysis and Correspondence Analysis offer a means for identifying patterns of usage, but they do not offer a means for testing the significance of their findings. What confirmatory techniques are available for linguists to verify their findings?
Martin Hilpert, FRIAS
"Grammaticality judgment day"
Multifactorial corpus-linguistic techniques allow us to analyze the usage of linguistic forms in terms of several explanatory variables and their interactions. Researchers interested in the psycho-linguistic validation of such results commonly turn to acceptability ratings as their dependent measure. I will argue that any measure calling for meta-linguistic assessments is inherently problematic. Instead, we should develop methods that a) tap into natural language use, b) do not allow participants to identify the task, and c) yield results that can be meaningfully compared to the results of corpus-linguistic analyses.
Anatol Stefanowitsch, Uni Bremen
"Usage and the usage-based model"
In my talk, I will point out the contradiction between the theoretical assumptions of the usage-based model -- that linguistic structure emerges from usage -- and the methodological practice of its main proponents -- who do not look at usage-data systematically, if indeed they look at data at all. I will argue that the investigation of usage data is necessary not only to support the claims of the usage-based model, but also to test its limits.
Daniel Wiechmann, Uni Jena
"More on what corpora can tell us about cognition: the case of entrenchment "
There is a general agreement among researchers working from a usage-based approach to language (cf., e.g., Barlow and Kemmer 2000, Bybee and Hopper 1997) that effects associated with usage frequency are operative in virtually all language related domains: from first and second language acquisition and online processing to diachronic change and cross-linguistic tendencies (for overviews, cf. Diessel 2007, Ellis 2002). However, for at least some notions central to cognitive linguistic theorizing, e.g. prototypicality and entrenchment, the exact nature of frequency effects and how (if at all) they can be estimated via corpus-based methodologies remains controversial (cf., e.g., Gilquin 2003, Nordquist 2004). In my contribution, I shall try and make a case for corpus-based approaches to cognitive dimensions of language in general and, more specifically, suggest some ways of how corpus-based methodologies can help us estimate degrees of cognitive entrenchment of complex linguistic structures.
Arne Zeschel, U of Southern Denmark
"Grammaticality = Familiarity?"
