Corpora in lexical studies Corpus Linguistics Richard Xiao lancsxiaoz@googlemail.com
Aims of this session Lecture Lab session Corpus-based lexicography Collocation and colligation Lab session Collocation using WST Collocation using AntConc Collocation and colligation in Xaira Using the BNCweb to study collocation
Corpus revolution in lexicographic and lexical studies Lexicographic and lexical studies are the greatest beneficiaries of corpora Corpora have “revolutionised” dictionary making and reference publishing It is now nearly unheard of for new dictionaries and new editions of old dictionaries published from the 1990s onwards not to claim to be based on corpus data
Why use corpora in dictionary making? Machine-readable corpora allow dictionary makers to extract all authentic, typical examples of the usage of a lexical item from a large body of text in a few seconds Corpora allow dictionary makers to select entries based on frequency information Corpora can readily provide frequency information and collocation information for readers Textual (e.g. register, genre and domain) and sociolinguistic (e.g. user gender and age) information encoded in corpora allows lexicographers to give a more accurate description of the usage of a lexical item
Why use corpora in dictionary making? Corpus annotations such as part-of-speech tagging and word sense disambiguation also enable a more sensible grouping of words which are polysemous and homographs A “monitor corpus” allows lexicographers to track subtle change in the meaning and usage of a lexical item so as to keep their dictionaries up-to-date Corpus evidence can complement or refute the intuitions of individual lexicographers, which are not always reliable because of potential biases in intuitions
Five emphases Changes brought about by corpora to dictionaries and other reference books - five “emphases” (Hunston 2002) an emphasis on frequency an emphasis on collocation and phraseology an emphasis on variation an emphasis on lexis in grammar an emphasis on authenticity
Top 1000 written / spoken words Authentic examples
Corpus-based learner dictionaries First ‘fully corpus-based’ dictionary Collins Cobuild English Dictionary (1987) Some corpus-based learner dictionaries Longman Dictionary of Contemporary English (3rd edition) Oxford Advanced Learner’s Dictionary (OALD, 5th edition) Cambridge International Dictionary of English (1st edition)
Frequency dictionaries
Collocation Collocation is among the linguistic concepts which have benefited most from advances in corpus linguistics What is collocation? strong tea, powerful car (Halliday 1976) “collocations of a given word are statements of the habitual or customary places of that word…the company that words keep” (Firth 1968:181-2) “One of the meanings of night is its collocability with dark” (Firth 1957:196) “a frequent co-occurrence of two lexical items in the language” (Greenbaum 1974:82) expel a school child vs. cashier an army officer “I propose to bring forward as a technical term, meaning by collocation, and apply the test of collocability” (Firth 1957: 194)
Meaning by collocation “There is frequently so high a degree of interdependence between lexemes which tend to occur in texts in collocation with one another that their potentiality for collocation is reasonably described as being part of their meaning” (Lyons 1977: 613) Complete description of the meaning of a word would have to include the other word or words that collocate with it “You shall know a word by the company it keeps!” (Firth 1968:179) Collocation is part of the word meaning
Two types of collocation Coherence collocation vs. neighbourhood (horizontal) collocation (Scott 1998) Coherence collocation Collocates associated with a word (e.g. letter – stamp, post office) Neighbourhood collocation Words which do actually co-occur with the word (letter - my, this, a, etc)
Coherence collocation “A cover term for the cohesion that results from the co-occurrence of lexical items that are in some way or other typically associated with one another, because they tend to occur in similar environments.” (Halliday & Hasan 1976:287) candle – flame – flicker hair – comb – curl – wave sky – sunshine – cloud – rain Difficult to measure using a statistical formula
Neighbourhood collocation Collocation in corpus linguistics Structure of collocation – collocation window “We may use the term node to refer to an item whose collocations we are studying, and we may then define a span as the number of lexical items on each side of a node that we consider relevant to that node. Items in the environment set by the span we will call collocates.” (Sinclair 1966:415) Casual vs. significant collocation Significant collocation: collocation that occurs more frequently than would be expected (in a statistical sense) on the basis of the individual items n.b. Neighbourhood (horizontal) collocations can include some coherence collocations
Intuition vs. collocation Greenbaum (1974): “people disagree on collocations” in introspection-based elicitation experiments Although “collocation can be observed informally” on the basis of intuitions, “it is more reliable to measure it statistically, and for this a corpus is essential” (Hunston 2002: 68) Intuition is often a poor guide to collocation “because each of us has only a partial knowledge of the language, we have prejudices and preferences, our memory is weak, our imagination is powerful (so we can conceive of possible contexts for the most implausible utterances), and we tend to notice unusual words or structures but often overlook ordinary ones” (Krishnamurthy 2000: 32-33) Collocation can be measured on the basis of co-occurrence statistics (MI, z, t, LL etc) – more discussion to follow
Collocation is syntagmatic Langue (Language system) paradigmatic famous boots. On the stroke of full time the Stoke the lead on the stroke of half-time with a goal Smith sin-binned on the stroke of half-time, added a clinched their win on the stroke of lunch after resuming chase by declaring on the stroke of lunch. <p> With a lead expectant crowd, on the stroke of midday. The bird hour began not upon the stroke of midnight but upon the of midnight but upon the stroke of noon. There was, booked in advance. On the stroke of seven, a gong summons Promptly on the stroke of six 'clock, the chooks from Edinburgh on the stroke of the Millennium. Parole (Utterance) syntagmatic
Collocation vs. colligation Relationship between a lexical item and other lexical items Relationship between words at the lexical level E.g. very collocates with good Colligation Relationship between a lexical item and a grammatical category Relationship between words at the grammatical level E.g. very colligates with ADJ
WST Collocate settings Concord tab
WST collocates Strength of relationship is displayed as 0.000 if it hasn't yet been computed
Strength of collocation relationship A wordlist is required
Highlight and double click…
…to see the selected collocate
Collocates in AntConc
Collocation in Xaira
Colligation in Xaira
Exploring collocation with BNCweb http://bncweb.lancs.ac.uk/bncwebSignup/user/login.php
Search for “sweet”
Concordances of “sweet” KWIC view
KWIC view
Dropdown menu: collocations
Collocation setting
Collocation database (default settings)
Adjusting settings
Noun collocates of “sweet” Click on a word to see its collocation info
Collocation info of “sweet” + “smell” Click on a number to see concordances of collocates at that position
Concordances of “smell” at R2
Collocation statistics
Rank by frequency “Sweet Maxwell” is a personal name. Frequent words crowd into the top of the collocate list: Are they genuine collocates?
Rank by the t test Also focusing on frequent words?
Rank by MI Infrequent words at the top of the list n.b. - “Sweet Afton” is a phrase from the lyrics expressing the beauty of the River Afton; “sweet nothings” means romantic and loving talks between sweethearts; “sweet marjoram” is the name of a herb for cooking. Infrequent words at the top of the list How useful are they (especially to English learners)?
Rank by the z score Like MI, the z score also over-estimates infrequent items (e.g. nothings, afton, marjoram)
Log-likelihood test
Rank by MI3
Rank by dice coefficient