A Brief Introduction to Corpora and the Lingonet Video Corpus

Dr Mike Nelson

 

1. Corpora: what are they and where did they come from?
 

What is a corpus ?
The term corpus (plural corpora) comes from the Latin word meaning body.  Thus, in the case of language, we are talking about a body of language - a collection, usually computer-stored, of language which can then be used for analysis.

How did corpora begin and develop ?
Corpus linguistics - using corpora to study language, has had a somewhat chequered history in the world of language research. Its origins can be traced back mainly to the 19th century (though analysis of Shakespearean texts was undertaken already 400 years ago) and the image of  silver-haired professors straining over mountains of text and manually counting occurrences of linguistic features, has been a hard one to dispel. Have a look here for more information about the history of corpora.
 http://www.ling.lancs.ac.uk/monkey/ihe/linguistics/corpus1/1fra1.htm
 

The field was, however, held in some regard until an infamous attack on its basic premises by Chomsky.
In a number of articles in the late 1950s and 1960s, Chomsky challenged the whole notion of empiricism on which corpus linguistics was based and suggested instead a rationalist approach. As Chomsky was more interested in competence than performance, corpus linguistics, which was primarily based on actual performance seemed to be invalidated overnight. It led to a situation that Sinclair describes:

Starved of adequate data, linguistics languished - indeed it became almost totally introverted. It became fashionable to look inwards to the mind rather than outwards to society. Intuition was the key, and the similarity of language structure to various formal models was emphasized.
Sinclair (1991:3)

Work on corpora continued despite these criticisms and, with the advent of the computer, really came into their own. The work begun in the early sixties by Randolph Quirk (the SEU Corpus), and Francis and Kucera (The Brown Corpus) was capitalised on by Svartvik in creating the London-Lund Corpus (LLC), creating a machine readable corpus of spoken language for the first time. By the 1980s corpus linguistics had almost found its way back into mainstream applied linguistics. This rise of corpus-based research can be seen in the number of corpora created. Svartvik (1992:8) shows that only ten were created before 1965, but between 1986 and 1992, 320 have been created.  It is safe to say that most recent work on the large scale has been concerned with dictionary creation, and as teachers we are all aware of, for example, the COBUILD project, and perhaps also the British National Corpus (BNC) which has been developed by Oxford University Press, Longman and Chambers. Corpora are back in vogue.
 

2. Key terminology
In this section we will look at two of the key concepts used with corpora. 

2.1 Concordancing
Concordancing is the act of picking out examples of a given word in context. The software used for doing this are often called KWIC concordancers (Key Words In Context). Thus, for example if we are interested in the word “dog” the concordancing program would search out all examples of the word and place them in rows with the word “dog” in the middle. Have a look below to give you some idea of what a concordance line looks like, using the word ‘media’ taken from the Video Corpus.

 

Concordance 

2.2 Collocation
There are many definitions for what collocation is, but basically it refers to how words co-occur. Thus the word blonde co-occurs almost always with the word hair. They are thus said to collocate. Collocations can be strong e.g. vested interest or weak e.g. brown hair. For more on this look now at these collocations taken from a business English corpus of words that collocate with the word  market.


stock market, market share, single market,  mass market
 market leader, market driven, market segment


In order to do concordancing, first of all you need both a corpus, a computer and some kind of software to run it. There are several commercial packages available for looking at corpora and concordancing, but this package uses TextSTAT:
http://www.niederlandistik.fu-berlin.de/textstat/software-en.html

3. Using Corpora : in research and in the classroom
 Murison-Bowie (1996:182) in an article entitled Linguistic Corpora and Language Teaching  notes that over the last few years there have been two opposing views of using corpora in research and language teaching:

 The Strong View says that without a relevant corpus no meaningful work can be done.
 The Weak View states that corpora give us a valuable insight into language and how it works and this is something we can draw on in our teaching.

 Which side do you come down on? Perhaps the safest way is to stick to the weaker side. For example, much corpus work hinges on the concept of frequency. The theory goes that the more frequent a word is, the more important it is to teach it to students. However, the most frequent word in English is “the” and whilst it is no doubt important to make sure our students can master the article system , it doesn’t get them very far in other ways. Remember the complaints of Chomsky discussed earlier? He dismissed frequency by saying it was of no importance at all and he gave this example:


“I live in Dayton, Ohio.”      “I live in New York”


In the English language the first sentence will always be less frequent than the second, thus the data in any corpus would be naturally skewed and therefore useless. Michael Lewis in his talk at the 1998 IATEFL conference when launching LTP’s new collocational dictionary suggested a sensible middle-ground. Use a corpus - but also use your common sense and intuition. Corpora can and have been used to give us great insights into the language. What follows are a few examples I have come across in my own research.
 

Research 

Holmes (1988) made a study of  how the concepts of doubt and certainty are presented in ESL textbooks. She, herself, advocated the use of  small corpora and used a combination of two corpora to compare findings to an examination of four well known textbooks. Her results showed that whilst in some cases, doubt and certainty - were adequately covered, “some textbooks were positively misleading” (Holmes 1988:40). She noted that other books give information of “variable quality” (1988:40). Interestingly, she also dismisses earlier attempts of analysis of this issue on the grounds that the research had not been corpus based, but along rationalist lines.

In another study, Kennedy (1987) looked at quantification, more specifically the use of approximation, and how it is used by native speakers. Two corpora and the Oxford Concise Dictionary were consulted and compared to input from teachers who were asked to give their own intuitive input on the subject. The results showed that the intuition of the teachers gave the largest range of types of approximation terms, but that on its own it was not enough. The vast amount of intuitive information needed ordering and:

the frequency data which the computer-based examination of these types in large corpora now makes possible were also necessary to give the descriptive information pedagogical value.
Kennedy (1987:282)

Flowerdew (1991) compared biology lecture notes to the COBUILD corpus and to published material and found several differences, for example, the use of the word then.

Ma, (1993) studied the difference between a corpus of 50 business letters and published materials for its teaching and also noted several differences, for example,  use of the PS section of letters was not covered in the materials, but widely used in real life.

Pickard (1992) noted that the language of refutation used in textbooks was not actually used in real life.

Ljung (1990),  in “A Study of TEFL Vocabulary” created a corpus of TEFL textbooks used in Swedish schools and compared it to several corpora, notably COBUILD. He found great discrepancies between “real” vocabulary and the materials used for teaching it. A frequency count showed that 204 words out of the top 1,000 were not shared by the two corpora showing a significant difference in emphasis, especially between concentration on concrete terms in the EFL corpus to more abstract terms in COBUILD. In his conclusion, he notes that:

there is reason to be critical of the TEFL texts on two major counts, i.e. the low general level of lexical sophistication and the absence of a clear increase in vocabulary difficulty as we move from the early to the later school years.
Ljung (1990:44-45)

He goes on to note that the materials do not adequately prepare the students for the tasks of the real world.
 Thus, it can be seen from these studies that corpus-based examination can be a very useful exercise. In each of the examples we find that the text books to some extent had got it wrong. Thus if text books are all flawed what should we do ? This leads us very nicely to the next section where we can look at actual corpora and see what we can do with them in the classroom.

 

 4. The Classroom
 So what can you do in the classroom with a corpus? There are several things that can be done very easily:

4.1 Looking at Frequency This is the most simple and most corpus analysis tools will give you the frequency lists of texts. This is perhaps not very interesting, but it can show - apart from all the grammar and function words - which special words, for example, in a given genre are the most frequent. Thus you could feed in the language found in this Video Corpus and compare it to a general English corpus.

Here we see the frequency list of the Video Corpus:


Video Corpus Frequency


4.2 Looking at Collocations The fact that computers can generate such a large amount of collocations in seconds affords the teacher and student access to very interesting and relevant information. For example, a business English student wants to know more about how the word market is used. The word can be keyed in and the collocates examined. In this way new collocates and even idiomatic phrases can be seen in context and learned. Have a look at Michael Lewis’s book Implementing the Lexical Approach  for a much more detailed account of how this can be done.  

Here we can see collocates of the word 'monarchy' taken from the Video Corpus.

Collocations


4.3 Looking at Words in context: Concordance lines can give you not only the words immediately around a key word but you can also see the word or form you are looking at in larger context.
 

4.4 Looking at Test design - using gap-fill corpus-based
 Butler, 1991, used a concordancer to help design tests. He searched out key words in context using a corpus and the test process then became one of choice for the teacher - choosing from items presented - rather than just sitting down and thinking what would be good examples to test the items.

4.5 Getting students to see grammatical patterns:
 For more advanced students you can use a corpus for them to see how certain grammatical patterns are used in real speech/writing. In 1988, Johns use a corpus to get his students to look at the differences when the word “to” is used as an infinitive and when it is used as a preposition. He also asked them to look at how the words “therefore” and “hence” differentiated in their use.

Why not try stating with something very simple, e.g. examining the difference between “make” and “do” by looking at which words collocate with them.

Concordance line of 'make' taken from the Video Corpus:

Make

Concordance line of 'do' taken from the Video Corpus:

Do

 

5. Exercise types you can easily develop from the Video Corpus
 

This next section will show you a few exercise types that are very easy to make once you have a corpus to work from.

 5.1 Working with collocates - before or after the keyword ?
In this kind of exercise you can use a corpus to generate collocates of given key word. Then, mix them up and ask the students to decide if the words come before or after the key word. This helps with familiarity and meaning can be discussed in the lesson. 

5.2 Fill in the gaps
 The old standard gap-fill type exercises are much better when you can take them directly from an authentic corpus - just  remove collocates or key words and get the students to fill them back in again.

 5.3 Linking sentences
 Again, another very simple old exercise. Take sentences from the corpus and split the sentences in half. Then ask the students to put them back together again. 

6. Beyond the silent corpus

The big advantage that is to be gained from using the Video Corpus is the fact that the text 'comes to life' and all the texts that are available for concordancing and analysis can be watched and listened to in the precise manner in which the language was originally used. The sound and vision also offer new possibilities for the classroom.

6.1 Listening for linguistic devices

The possibility for listening can help in exercises based around certain linguistic devices, for example, hesitating devices found in the text below. Further use of fillers and false starts can also be analysed.

Right. Um.. Yes, of course, I do think we need a welfare state um.. in order to
accommodate the needs of the weaker members of our society. It's absolutely
essential. Um.. And Britain was really at the forefront in.. in setting up its welfare
state in the middle of the century. Um.. And I think that's something we should be
very proud of. What is rather sad I think is that a lot of people are now feeling as
though they've been conned, um.. particularly elderly people who have erm.. paid their
national insurance stamps for many many years, expecting to be cared for, in their,
when they become older.

6.2 Video for discussion

As the students are able to see and listen to all the texts in the corpus they can be used as preparatory material for group discussions. Individual students can separately watch the opinions given by different members of the panel and then elaborate on their arguments for their own debates on the issues as part of group discussions.

6.3 Video for presentations

In the same manner, students can use the information gained from the corpus and the opinions given to prepare their own presentations on the same topics.

6.4 Video for pronunciation, word stress and inflection

The texts in their audio-visual form can be used to study the pronunciation of words and the word stress and inflections used.

6.5 Video for body language

The video can be used for the study of body language in relation to speech: is there specific body language attached to certain words/ideas?


 7. Conclusions
 In this short article we have seen that despite the arguments against corpora, it can be argued that they are an effective method for both research and for organising exercises for the classroom. The argument over the sufficiency of the concept of frequency over the intuition of native speakers will continue to rage. The middle ground, however, can be taken  - use a corpus to get the data and then use your intuition on how to get the best out of it.
Gathering, collecting and organising a corpus of your own is a time consuming and often boring process. The Video Corpus provides you with a small but useful body of language to utilise in the classroom. You can thus have instant access to authentic, relevant and hopefully motivating language.

 

8. Background reading

There is a marvellous full on-line course on corpus linguistics at Lancaster University http://www.ling.lancs.ac.uk/monkey/ihe/linguistics/contents.htmcreated  by Tony McEnery and Andrew Wilson. Please have a look through- even if you don’t want to do it all.

Note that they have a full section on Corpora in language teaching at: http://www.ling.lancs.ac.uk/monkey/ihe/linguistics/corpus4/4fra1.htm

See David Lee's great site on everything related to corpora Devoted to Corpora

and Bibliography


 Butler, J. (1991) Cloze procedures and concordances: the advantages of discourse level authenticity in testing expectancy grammar  in System 19 (1/2), 29-38
Flowerdew, J. (1991)  Corpus-Based Course Design in Milton, J.C. and Tong, K. (eds) Text Analysis in Computer Assisted Language Learning, 31-43. Hong Kong: The Hong Kong University of Science and Technology and the City Polytechnic of Hong Kong
 Johns, T.  (1988)  Whence and Whither Classroom Concordancing ? in Bongaerts, T. et al. (eds) Computer Applications in Language Learning. Dordrecht: Foris
 Kennedy, G, D.  (1987)  Quantification and Use of English: A Case Study of One Aspect of the Learner’s Task in Applied Linguistics Vol 8 No. 3 Oxford: Oxford University Press
 Lewis, M.  (1993)  The Lexical Approach  Hove: LTP
 Ljung, M. (1990)  A Study of TEFL Vocabulary    Almqvist and Wiksell International: Stockholm, Sweden
 Ma, K.C. (1993)  Text Analysis of Direct Mail Sales Letters in Boswood, T., Hoffman, R. and Tung, P (eds) Perspectives on English for Professional Communication Hong Kong: City Polytechnic of Hong Kong
 McEnery, T. and Wilson, A. (1996)  Corpus Linguistics Edinburgh: Edinburgh University Press
 Murison-Bowie, S. (1996)   Linguistic Corpora and Language Teaching in Annual Review of Applied Linguistics 16, 182-99: Cambridge: Cambridge University Press
 Pickard, V. (1991)  Should We Be teaching Refutation ? Concordanced Evidence From the Field of Applied Linguistics  Paper presented at the Eighth ILE International Conference 15-18 December, 1992
Sinclair, J. (1991)  Corpus, Concordance, Collocation  Oxford: Oxford University Press