Corpus Linguistics Lecture 1

1
Corpus LinguisticsLecture 1

• Albert Gatt

2
Contact details

• My email albert.gatt_at_um.edu.mt
• Drop me a line with queries etc, and to arrange
  meetings.

3
Course web page

• Course web page http//staff.um.edu.mt/albert.gat
  t/home/teaching/corpusLing.html
• Details of tutorials, lectures etc will always be
  on the web page.
• Readings for the lecture
• Downloadable lecture notes (available after the
  lecture)

4
Suggested text

• T. McEnery and A. Wilson. (2001). Corpus
  Linguistics. Edinburgh University Press
• NB Over the course of these lectures, other
  readings will also be proposed and made
  available, usually online.

5
Lectures and assessment

• Structure of lectures
• all lectures will take place in the lab
• usually, about half the lecture (1hr) will be
  devoted to practical work
• Course assessment assignment
• Final essay (ca. 1500-2000 words)
• Essay topics will involve research on corpora!

6
Questions

• ?

7
What is corpus linguistics?

• A new theory of language?
• No. In principle, any theory of language is
  compatible with corpus-based research.
• A separate branch of linguistics (in addition to
  syntax, semantics)?
• No. Most aspects of language can be studied using
  a corpus (in principle).
• A methodology to study language in all its
  aspects?
• Yes! The most important principle is that aspects
  of language are studied empirically by analysing
  natural data using a corpus.
• A corpus is an electronic, machine-readable
  collection of texts that represent real life
  language use.

8
Goals of this lecture

• To define the terms
• corpus linguistics
• corpus
• To give an overview of the history of corpus
  linguistics
• To contrast the corpus-based approach to other
  methodologies used in the study of language

9
An initial example

• Suppose youre a linguist interested in the
  syntax of verb phrases.
• Some verbs are transitive, some intransitive
• I ate the meat pie (transitive)
• I swam (intransitive)
• What about
• quiver
• quake
• Are these really intransitive?

Most traditional grammars characterise these as
intransitive
10
One possible methodology

• The standard method relies on the linguists
  intuition
• I never use quiver/quake with a direct object.
• I am a native speaker of this language.
• All native speakers have a common mental grammar
  or competence (Chomsky).
• Therefore, my mental grammar is the same as
  everyone elses.
• Therefore, my intuition accurately reflects
  English speakers competence.
• Therefore, quiver/quake are intransitive.
• NB The above is a gross simplification! E.g.
  linguists often rely on judgements elicited from
  other native speakers.

11
Another possible methodology

• This one relies on data
• I may never use quiver/quake with a direct
  object, but
• other people might
• Therefore, Ill get my hands on a large sample of
  written and/or spoken English and check.

12
Quiver/quake the corpus linguists answer

• A study by Atkins and Levin (1995) found that
  quiver and quake do occur in transitive
  constructions
• the insect quivered its wings
• it quaked his bowels (with fear)
• Used a corpus of 50 million words to find
  examples of the verbs.
• With sufficient data, you can find examples that
  your own intuition wont give you

13
Example II lexical semantics

• Quasi-synonymous lexical items exhibit subtle
  differences in context.
• strong
• powerful
• A fine-grained theory of lexical semantics would
  benefit from data about these contextual cues to
  meaning.

14
Example II continued

• Some differences between strong and powerful
  (source British National Corpus)
• strong
• powerful
• The differences are subtle, but examining their
  collocates helps.

15
Some preliminary definitions

• The second approach is typical of the
  corpus-based methodology
• Corpus A large, machine-readable collection of
  texts.
• Often, in addition to the texts themselves, a
  corpus is annotated with relevant linguistic
  information.
• Corpus-based methodology An approach to Natural
  Language analysis that relies on generalisations
  made from data.

16
Example (British National Corpus)

• British National Corpus (BNC)
• 100 million words of English
• 90 written, 10 spoken
• Designed to be representative and balanced.
• Texts from different genres (literature, news,
  academic writing)
• Annotated Every single word is accompanied by
  part-of-speech information.

17
Example (continued)

• A sentence in the BNC
• Explosives found on Hampstead Heath.
• ltsgt
• ltw NN2gtExplosives
• ltw VVDgtfound
• ltw PRPgton
• ltw NP0gtHampstead
• ltw NP0gtHeath
• ltPUNgt.

18
Example (continued)
new sentence

• ltsgt
• ltw NN2gtExplosives
• ltw VVDgtfound
• ltw PRPgton
• ltw NP0gtHampstead
• ltw NP0gtHeath
• ltPUNgt.
• Explosives found on Hampstead Heath

plural noun
past tense verb
preposition
proper noun
proper noun
punctuation
19
Important to note

• This is not raw text.
• Annotation means we can search for particular
  patterns.
• E.g. for the quiver/quake study find all
  occurrences of quiver which are verbs, followed
  by a determiner and a noun
• The collection is very large
• Only in very large collections are we likely to
  find rare occurrences.
• Corpus search is done by computer. You cant
  trawl through 100 million words manually!

20
The practical objections

• But were linguists not computer scientists! Do I
  have to write programs?
• No, there are literally dozens of available tools
  to search in a corpus.
• Are all corpora good for all purposes?
• No. Some are general-purpose, like the BNC.
  Others are designed to address specific issues.

21
The theoretical objections

• What guarantee do we have that the texts in our
  corpus are good data, quality texts, written by
  people we can trust?
• How do I know that what I find isnt just a
  small, exceptional case. E.g. quiver in a
  transitive construction could be really a
  one-off!
• Just because there are a few examples of
  something, doesnt mean that all native speakers
  use a certain construction!
• Do we throw intuition out of the window?

22
Part 2

• A brief history of corpus linguistics

23
Language and the cognitive revolution

• Before the 1950s, the linguists task was
• to collect data about a language
• to make generalisations from the data (e.g. In
  Maltese, the verb always agrees in number and
  gender with the subject NP)
• The basic idea language is out there, the sum
  total of things people say and write.
• After the 1950s
• the so-called cognitive revolution
• language treated as a mental phenomenon
• no longer about collecting data, but explaining
  what mental capabilities speakers have

24
The 19th early 20th Century

• Many early studies relied on corpora.
• Language acquisition research was based on
  collections of child data.
• Anthropologists collected samples of unknown
  languages.
• Comparative linguists used large samples from
  different languages.
• A lot of work done on frequencies
• frequency of words
• frequency of grammatical patterns
• frequency of different spellings
• All of this was interrupted around 1955.

25
Chomsky and the cognitive turn

• Chomsky (1957) was primarily responsible for the
  new, cognitive view of language.
• He distinguished (1965)
• Descriptive adequacy describing language, making
  generalisations such as X occurs more often than
  Y
• Explanatory adequacy explaining why some things
  are found in a language, but not others, by
  appealing to speakers competence, their mental
  grammar
• He made several criticisms of corpus-based
  approaches.

26
Criticisms of corpora (I)

• Competence vs. performance
• To explain language, we need to focus on
  competence of an idealised speaker-hearer.
• Competence internalised, tacit knowledge of
  language
• Performance the language we speak/write is
  not a good mirror of our knowledge
• it depends on situations
• it can be degraded
• it can be influenced by other cognitive factors
  beyond linguistic knowledge

27
Criticisms of corpora (II)

• Early work using corpora assumed that
• the number of sentences of a language is finite
  (so we can get to know everything about language
  if the sample is large enough)
• But actually, it is impossible to count the
  number of sentences in a language.
• Syntactic rules make the possibilities literally
  infinite
• the man in the house (NP -gt NP PP)
• the man in the house on the beach (PP -gt PREP
  NP)
• the man in the house on the beach by the lake
• So what use is a corpus? Were never going to
  have an infinite corpus.

28
Criticisms of corpora (III)

• A corpus is always skewed, i.e. biased in favour
  of certain things.
• Certain obvious things are simply never said.
  E.g. We probably wont find a dog is a dog in our
  corpus.
• A corpus is always partial We will only find
  things in a corpus if they are frequent enough.
• A corpus is necessarily only a sample.
• Rare things are likely to be omitted from a
  sample.

29
Criticisms of corpora (IV)

• Why use a corpus if we already know things by
  introspection?
• How can a corpus tell us what is ungrammatical?
• Corpora wont contain disallowed structures,
  because these are by definition not part of the
  language.
• So a corpus contains exclusively positive
  evidence you only get the allowed things
• But if X is not in the corpus, this doesnt mean
  its not allowed.
• It might just be rare, and your corpus isnt big
  enough. (Skewness)

30
Refutations

• Corpora can be better than introspectvie evidence
  because
• They are public other people can verify and
  replicate your results (the essence of scientific
  method).
• Some kinds of data are simply not available to
  introspection. E.g. people arent good at
  estimating the frequency of words or structures.
• Skewness can itself be informative If X occurs
  more frequently than Y in a corpus, that in
  itself is an interesting fact.

31
Refutations (II)

• By the way, nobodys saying throw introspection
  out the window
• There is no reason not to combine the
  corpus-based and the introspection-based method.
• Many other objections can be overcome by using
  large enough corpora.
• Pre-1950, most corpus work was done manually, so
  it was error prone.
• Machine-readable corpora means we have a great
  new tool to analyse language very efficiently!

32
Corpora in the late 20th Century

• Corpus linguistics enjoyed a revival with the
  advent of the digital personal computer.
• Kucera and Francis the Brown Corpus, one of the
  first
• Svartvik the London-Lund Corpus, which built on
  Brown
• These were rapidly followed by others Today,
  corpora are firmly back on the linguistic
  landscape.

33
Summary

• Introduced the notion of corpus and corpus-based
  research
• Gave a quick overview of the history of this
  methodology
• Looked at some possible objections to
  corpus-based methods, and some possible
  counter-arguments

34
Next lecture

• We look more closely at some important properties
  of a corpus
• Machine-readability
• Balance
• Representativeness