Corpus Preparation and Corpus Methods

Querying corpora: Getting started

John Bateman and Kerstin Fischer

17 April, 2002

1. Try answering some of the questions below by means of corpus queries:

• What's wrong with `allow to investigate...'?

• Is `would' in if-clauses always wrong?

• What's the spelling of the Italian vegetable beginning with `bro...'?

• Is it `wholistic/wholism' or `holistic/holism'?

• Is `disinterested' or `uninterested' correct?

• Is it `a book/paper/article/lecture on' or `about' something?

• What is the difference between 'sheer' and 'pure'? Some answers may be available here.

(What you should consider in answering such questions:

1. What does the question involve, what could the problems be?

2. What could be a suitable corpus or suitable corpora for answering this query?

• What kinds of topics should my corpus deal with? For instance, dialogues on appointment scheduling may not be particularly useful for answering questions about vegetable.

• How old should the texts in my corpus be? For instance, can Shakespeare's plays help me determining whether to write `wholism' or `holism'?

• Is written or spoken language more suitable or is it of no relevance? For instance, `oh' may have very different functions in spoken and written language, whereas checking a transcribed corpus for the spelling of a particular word may tell us only something about transcription conventions, not about language use.

• Do I need frequency information? In this case a concordance program that yields only 40 examples may not be useful.

• Do I need syntactic information? In this case I may want to choose a tagged corpus.

3. How should I formulate my query? For instance, in the case of different spelling, it may be useful to ask several queries or to use a query syntax that allows alternatives.

4. How valid are my results, is there a good reason to suppose that querying more corpora would be necessary for a founded answer?)

2. Now try to formulate some questions yourself.