Jan

Svartvik,

Department

of

English,

Lund

University,

Sweden

Lexis

in

English language

corpora

1.

The second

corpus

generation

Many

more years ago than

I

care to remember,

on

the

occasion

of my inaugural lecture

at

Lund University,

I

spoke with some enthusiasm about the bright future of corpus-based

study

of

spoken language, what with tape-recorders getting smaller,

and

computers

getting bigger.

In

1992,

at the

Fifth Euralex Congress

in

Tampere,

the

future

of

corpus

linguistics

seems even brighter than

on

that previous

occasion.

Yet,

while tape-recorders

may indeed

be a bit

smaller

(the

stereo

set,

though, seems colossal compared

to our

gramophone), computers are actually getting smaller

too:

there has been

a

radical devel-

opment from

the

mainframe

to

the micro, personal, desktop, laptop, palmtop

and

note-

book.

But not only are computers getting smaller but also faster and cheaper. This fantas-

tic

technological

hardware

development that

we are

witnessing

is of

course only

one

reason for my

belief

that the future of corpus linguistics is even brighter now than

at the

beginning

of the

seventies.

The

best

part

is

that

the

hardware

is

also becoming well

matchedby software,

and

software development is indeed crucial if the corpus approach

is

going

to

fulfil its promise.

The

meaning of "corpus" as given

in

most dictionaries is rather vague

and

gives little

indication of bright prospects, for example:

• MACQUARIE DICTIONARY:

"a body of data".

•

COLLWS COBUILD DICTIONARY:

"a

large number of

articles,

books, magazines, etc that

have been deliberately collected together for some purpose".

• LONGMAN DICTIONARY OF CONTEMPORARY

ENGLISH:

"a

collection..

of

material or infor-

mation

for

study" (New edition,

1987).

• LONGMAN DICTIONARY OF THE ENGLISH LANGUAGE

0^Jew

edition,

1991) is

more

explicit:

"a

collection

of

spoken and/or written language

for

scientific

study

of

word forma-

tion, sentence structure, sounds, etc".

COBUILD

adds

the warning:

"a

formal, technical word" ft>ut, like

LONGMAN,

also gives the

helpful hint that the plural can be either

corpora

or

corpuses).

AIl of the definitions in these

recent

works

fail

to

specify "machine-readable", which is ofcourse the current norm and

also

the

topic

of

this paper,

in

particular electronic corpora

of

spoken English.1 Only

LONGMAN

gives

a

clear indication that there are,

and

should be, corpora

of

speech

- by

far

the most common use of language and the variety that has too long been neglected

in

both grammatical

and

lexicographical description.

It

is not

often that

we can

date

the

beginning

of a new bud on the

linguistic tree

structure,

but

this

is

indeed possible with corpus linguistics,

at

least English corpus

linguistics.

It is now getting mature, just over

30

years

of

age.

From the humble beginning

engaging only

a

small number

of

linguists, corpora have become

"the

flavour

of the