[Corpora-List] intuitions about co-occurrence

[Oakey, David [ENGL]]

Hello Marco,

Going a bit further back, Fox (1987: 146) found that experimental subjects, when asked to think of collocates of 'feet', gave other parts of the body - 'legs', 'toes', 'head', etc - while others gave 'shoe', 'sandals', 'sock'; very few thought of 'tall' or 'high' or 'long', which are frequent collocates in the Bank of English - I don't think she specified the probabilities of intuition/corpora occurrences exactly though. Fox also noted that once respondents were told which were the most frequent lexical collocates, they were not able to say why they had not thought of these themselves - they thought "more of semantic sets than of words which are actually likely to occur in the near vicinity of each other." These sound like the kind of things a flexible distribution model you describe would do well at identifying. 

Individual words - while obvious units of analysis - are not so valuable as test items: "The mental lexicon clearly holds more abstract entities than are identified by computational searches, and neither native speakers nor learners produce word combinations on the basis of their frequency and probability of co-occurrence" (Howarth 1998), hence the current focus - as other posts indicate (thanks Ute!) - on cognition/intuition and fixed multi-word items like academic formulas and lexical bundles, or strings of abstract form/function items like constructions/collostructions and semantic sequences. Research design is an interesting point - whether items are checked with intuition post-hoc as in Ellis & Simpson-Vlach or elicited a priori as in Fox.

Best wishes,

David

Fox, G. (1987). The case for examples. In J. McH. Sinclair (Ed.), Looking Up: an account of the COBUILD project in lexical computing (pp. 137-149). London:
Collins ELT.
Howarth, P. A. (1998). Phraseology and Second Language Proficiency. Applied Linguistics, 19(1), 24-44


-----Original Message-----
From: corpora-bounces at uib.no [mailto:corpora-bounces at uib.no] On Behalf Of marco baroni
Sent: Wednesday, April 24, 2013 6:43 AM
To: CORPORA at uib.no
Subject: [Corpora-List] intuitions about co-occurrence

Dear Corpora-ers,

Is anybody aware of experimental studies where researchers have looked at whether subjects' explicit intuitions about the probability of co-occurrence of two terms correlate with (functions of) the frequency of co-occurrence of two terms in a corpus?

I am aware of studies correlating other psychological variables, such as the degree of association of words in free association norms, with corpus co-occurrence, but I was not able to find anything relevant to the specific question I'm asking above.

Any advice greatly appreciated.

Ciao,

Marco

--
Marco Baroni
Center for Mind/Brain Sciences
University of Trento
http://clic.cimec.unitn.it/marco


_______________________________________________
UNSUBSCRIBE from this page: http://mailman.uib.no/options/corpora
Corpora mailing list
Corpora at uib.no
http://mailman.uib.no/listinfo/corpora

------------------------------
David J. Oakey, BA, MEd, PhD
Assistant Professor
Applied Linguistics Program
Iowa State University
317 Ross Hall
Ames, Iowa 50011-1201, USA
Email: djoakey at iastate.edu
Office: 515-294-7521
http://engl.iastate.edu/directory/doakey
------------------------------

_______________________________________________
UNSUBSCRIBE from this page: http://mailman.uib.no/options/corpora
Corpora mailing list
Corpora at uib.no
http://mailman.uib.no/listinfo/corpora