ELSEVIER Lingua 112 (2002) l-6

www.elsevier.com/locate/lingua

Lingua Franca

Grow your own linguistics?

Robert D. Borsley”,*, Richard Inghamb

Received IS October 2000: revised version 8 November 2000

It is a well known fact that applied linguistics at least in Britain has had little con-
tact with mainstream theoretical linguistics in recent years. This presumably reflects
a widespread belief among applied linguists that theoretical linguistics is largely
irrelevant to applied concerns. Theoretical linguists might wish to dispute this view.
It seems, however, that some applied linguists wish to go further. In recent years, a
number of applied linguists have argued that mainstream theoretical linguistics is
fundamentally unsatisfactory and misguided and have gone on to argue that applied
linguistics can and should develop its own theories of language. Thus, Fairclough
speaks of the need for “the development of a new social theory of language which
may include a new grammatical theory” (1995: 10). Similarly, Kress remarks that
“[i]n very many ways, it is here, in our work in Applied Linguistics that a new the-
ory of language is being produced” (1993 : 5), and adds that “I am certain that by the

Lingua is introducing a series of provocative articles under the general heading of Lingua Franca. We
invite essays of about 2000 words on a favourite topic. voicing strong opinions supported by arguments
in an entertaining way. Obvious examples of the genre include Geoff Pullurn’s Topic/Comment contri-
butions to Natural Language and Linguistic Theory, and Neil Smith’s and Elan Dresher’s columns for
Glot International. We anticipate and welcome replies, also limited to about 2000 words, to which the
original contributor would have the right of reply.
Where someone is an obvious target of a particular contribution, he or she will be sent a copy and
invited to respond, probably in the same issue of the journal. Other responses are welcome, provided
new points are being made, and will be published on an ongoing basis. The editors retain the right to
decide on publication, to edit exchanges, and to close off the discussion.
The first example appears in this issue, accompanied by a reply,
* Corresponding author. E-mail: [email protected]

0378-2166/02/$ - see front matter 0 2002 Elsevier Science B.V. All rights reserved.
PII: SOO24-384 I(01 )00034- 1
2 R.D. Borsley, R. &ham I Lingua 112 (2002) I4

end of this century we will have produced more than the outlines of a theory ad-
equate to our present tasks” (1993: 22). Thus, applied linguistics is seen not as com-
plementing theoretical linguistics but as an alternative, more satisfactory approach to
language. We think this is quite important. It matters for linguistics how it is per-
ceived by various outsiders and if people who one might perhaps expect to be apply-
ing mainstream work argue that it is fundamentally misguided that can’t help. In this
note, we want to take a necessarily brief look at this critique.
What do some applied linguists find unsatisfactory in mainstream linguistics?
And what would they like to replace it with? Mainstream work is mainly concerned
with I-language, the cognitive system underlying the ordinary use of language.
Various applied linguists have suggested that this is unsatisfactory and that texts are
a more appropriate object of study.
What is wrong with the concern with I-language? We are not aware of any careful
discussion of this matter within applied work, but there are various remarks that
embody objections of one kind or another. For example, Graddol (1993) criticizes
‘structuralists’, among whom he includes generative linguists, in the following passage:

“In an attempt to discover a language which is analysable and codifiable, structuralists have imagined
that under the messiness of real life experience lies an ideal form. Such idealizations rest on the assump-
tion that context and social processes are of marginal interest . ..” (1993a: 12)

It appears that Graddol thinks that it is a mistake to look for ideal forms “under the
messiness of real life experience”. But this is normal scientific practice. Thus, the
philosopher of science Roy Bhaskar writes that “the objects of knowledge” are “the
structures and mechanisms that generate phenomena” (Bhaskar, 1975 : 25).
The suggestion that structuralist idealizations “rest on the assumption that context
and social processes are of marginal interest” is also a very strange one. To justify
focusing on certain phenomena and setting others aside all one needs to argue is that
the phenomena one is focusing on form a natural class. There is no need to argue
that the phenomena set aside are of marginal interest. A phonetician does not have to
claim that syntax is of marginal interest, she just has to claim that phonetic phenom-
ena are a natural class. Similarly, a student of English literature does not have to
claim that French, German, Spanish, etc. literature is of marginal interest, and a stu-
dent of 19th century history does not have to claim that all other centuries are of
marginal interest, and so on.
Kress (1993) also appears to think that the standard notion of I-language is prob-
lematic. After considering the quite different ways in which a particular story is pre-
sented in two different newspapers, he remarks that:
“
the hitherto accepted model of the autonomous, homogeneous linguistic system, with its core of syn-
tax and phonology, can no longer be sustained.” (1993: 13)

Why Kress thinks his conclusion follows from his observations is unclear. We can
only speculate that he is under the impression that mainstream linguists believe that
all members of a speech community have the same I-language and use it in the same
way, but of course no one believes this. The reference to a ‘homogeneous’ system is
 R.D. Borsley, R. lngham I Lingua 112 (2002) I4i 3

also puzzling since mainstream linguists generally see the linguistic system as mod-
ular, hence non-homogeneous.
Instead of a focus on I-language a number of applied linguists advocate a concern
with ‘texts’. Stubbs, for example, asserts that “[tlhe unit of study must be whole
texts” (1996: 32). Similarly, Graddol advocates a “postmodem model of language”
and remarks that “[plostmodem theories are concerned pre-eminently with texts,
with how texts are produced, and how texts are used and situated within other cul-
tural practices” (1993: 18). In similar vein, Kress suggests that “a set of questions
central in any approach to language” are “How do texts arise? How do they come
into being? How and why do they get produced? Why in short do people talk and
write?” (1989: 4), and criticizes mainstream work because it cannot provide “any
revealing account of the production of text” (1993: 6).
Thus, the suggestion is that linguistics should study texts and especially how they
are produced. Stubbs defines a text as “an instance of language in use, either spoken
or written: a piece of language behaviour which has occurred naturally, without the
intervention of the linguist” (1996: 4). On this definition, texts will include the
American Declaration of Independence, Hamlet, an episode of Eastenders, the
words on a tombstone or a beer mat, an advert for soap powder, what we said to each
other yesterday, etc., etc. The idea that some discipline should have as its object of
study the production of such diverse things is very odd. There are, of course, people
who are concerned with some of these things. Thus, there are political historians
who are interested in the production of the American Declaration of Independence
and similar documents and literary historians who are interested in the production of
Hamlet and other literary works. Are we to believe that some kind of linguist could
do a better job? Surely not.
It would be much more plausible to propose that linguistics is concerned with the
linguistic properties of texts. This would mean properties like involving a simple
vocabulary, containing a large number of passives, or containing many discourse
connectives, but not properties like being interesting, being libellous, or being avail-
able from all good bookshops. There are in fact various reasons for studying the lin-
guistic properties of texts. In the case of spoken texts, one might learn something
about the production mechanisms. If some sentence type occurs very rarely in natu-
rally occurring speech that may be significant. A study of texts might also establish
a number of text types with distinctive properties. This would be of interest because
it raises a question about speakers’ knowledge of text types. Is this a part of the I-
language (Hudson, 1996; Paolillo, 2000) or part of a separate ‘user’s manual’ (Culy,
1996). Thus, the idea that linguistics should study the linguistic properties of texts is
a reasonable one. Clearly, however, it presupposes some understanding of the lin-
guistic properties, but this is what the study of I-language seeks to provide. Thus, far
from being an alternative to the study of I-language this kind of research presup-
poses its existence.
A second important feature of mainstream linguistics is a reliance on native
speakers’ intuitions as a major source of data. Various applied linguists think this is
unsatisfactory, and suggest that corpora of naturally occurring speech and writing are
a more satisfactory source of data.
4 R.D. Borsley, R. Ingham I Lingrta 112 (2002) 14

Stubbs is particularly outspoken here, but what he has to say has been said before
by various people. He remarks that “a linguist’s introspections are private and not
open to criticism by others” (1996: 45). However, it is simply not true that intro-
spections are not open to criticism by others. As Chomsky emphasized a long time
ago :

“It is not that these introspective judgements are sacrosanct and beyond any conceivable doubt. On the
contrary, their correctness can be challenged in various ways, some quite indirect. Consistency among
speakers of similar backgrounds, and consistency for a particular speaker on different occasions is
relevant information. The possibility of constructing a systematic and general theory to account for these
observations is also a factor to be considered in evaluating the probable correctness of particular obser-
vations.” (I 964: 79-80)

Stubbs also criticizes the use of intuitions on the grounds that “[nlormally one
expects a scientist to develop theories to describe and explain some phenomena
which already exist, independently of the scientist” (1996: 29). Stubbs has appar-
ently forgotten the role of experiments in science. Scientists conduct experiments
precisely because the “phenomena which already exist, independently of the scien-
tist” are uninformative and do not allow them to evaluate their theories. As has often
been pointed out (e.g. Chomsky, 1986: 36; Jackendoff, 1993: 46-49), investigations
of intuitions are informal experiments. It would be as odd for linguists to avoid such
informal experiments as it would be for physicists to avoid their much more sophis-
ticated experiments. One might argue here that more use should be made of more
formal experiments. In fact a number of researchers have argued for this position,
e.g. Bard et al. (1996), Cowart (1997). However, this is quite different from reject-
ing any use of intuitions.
A further criticism comes from Graddol et al., who remark that “[tlhe embarrass-
ing fact is that native speaker intuitions about ‘acceptability’ or ‘grammaticality’ are
not infallible” (1994: 19). It is not clear why they think there is something embar-
rassing here. Would they say that it is embarrassing that experiments in psychology
or physics are not infallible? Presumably not. None of these things are embarrassing.
?‘hey are simply a reflection of the fact that scientific research is hard and that there
is no infallible method that one can employ. As Bhaskar (1975: 168) remarks,
“[slcience is work that requires creative intelligence, and there can be no mechani-
cal surrogate for that”.
As we have indicated, Stubbs, Graddol and others see a corpus of speech or
writing as a better source of data than intuitions. There are various problems here.
One is that a corpus is only as useful as one’s ability to search it. There is no prob-
lem if one is just interested in lexical statistics, but mainstream linguists are gener-
ally interested in much more abstract matters. One might, for example, want to know
whether some language allows parasitic gaps within subjects, in other words,
whether it allows sentences like She is the kind of person that friends of would do
anything for, where there is a parasitic gap following of and an ordinary gap follow-
ing for. There is no mechanical way to search for sentences like this. Hence, one
would have to do it by hand. This looks like a good way to waste time. It would
clearly be more sensible to ask native speakers.
 R.D. Borsley, R. Ingham I Lingua 112 (2002j l-4 5

Another problem with a corpus is that it is difficult to conclude anything from the
fact that some type of example does not appear in it. This might mean that such
examples are ungrammatical but it might not. This is especially true when the kinds
of example one is interested in involve complex syntactic properties. Suppose, for
example, that one wants to know whether some language allows the topicalization of
a reflexive pronoun, in other words whether it allows examples like Himself, Kim
most admires. If one does not find a relevant example in a million words of text, can
one conclude anything? One might conclude something if there was some reason to
think that such examples would appear in a million words of text if they were gram-
matical, but as far as we know, there is no reason to think this.
A final problem concerns meaning. A corpus is just a collection of sounds or
marks on a surface or an electronically stored version of these. In themselves the
sounds or marks have no meanings. They are endowed with meanings by native
speakers. Hence the only way to find out about meanings is to ask native speakers.
Anyone who doubts this should try analyzing a corpus of speech or writing from a
language that they do not know without the help of a native speaker. Surprisingly,
those who object to the use of intuitions are quite happy to make statements about
meaning which could only derive from intuitions. Thus, Stubbs, for example, says a
great deal about meaning, discussing inter alia presupposition, modality, and illocu-
tionary force. Perhaps, the critics assume that intuitions about meaning are funda-
mentally different from intuitions about grammaticality. As far as we are aware,
however, no one has actually argued for this position, and it seems to us that it would
be very difficult to do so.
It seems to us, then, that applied linguists who advocate reliance on a corpus have
failed to appreciate the limitations of such a methodology. Of course, none of this
implies that corpora cannot be useful. In fact there is increasing use of corpus data
in various areas of mainstream linguistics. Moreover, there is obviously no alterna-
tive to corpus data if the speakers are dead. Hence, mainstream work on dead lan-
guages is inevitably based on a corpus (supplemented by intelligent guesses about
meanings). Corpus data is also central in work on the early stages of child language
acquisition. However, when intuitions are available they are an invaluable source of
data.
The issues that we have been concerned here deserve a more extended discussion.
It seems to us, however, that the applied linguists that we have looked at have
neither a serious critique of mainstream linguistics nor a coherent alternative. In the
circumstances, the idea that applied linguistics represents an alternative, more satis-
factory approach to language cannot be taken very seriously.

References

Bard, E.G., D. Robertson and A. Sorace, 1996. Magnitude estimation of linguistic acceptability. Lan-
guage 72( l), 32-68.
Bhaskar, R., 1975. A realist theory of science. Leeds, Leeds Books.
Chomsky, N.A., 1964. Current issues in linguistic theory. Mouton, The Hague.
Chomsky, N.A., 1986. Knowledge of language. New York: Praeger.
6 R.D. Borsley, R. @ham I Lingua 112 (2002) 14

Cowart, W., 1997. Experimental syntax: Applying objective methods to sentence judgements. London:
Sage Publications.
Culy, C., 1996. Null objects in English recipes. Language Variation and Change 8, 91-124.
Fairclough, N., 1995. Critical discourse analysis. Second edition. London: Routledge.
Graddol, D., 1993. Three models of language description. In: D. Graddol and Boyd-Barrett (eds.), Media
texts: Authors and readers, 1-21. Buckingham: Open University Press.
Graddol, D., J. Cheshire and J. Swann, 1994. Describing language. Second edition. Buckingham: Open
University Press.
Hudson, R.A., 1996. Sociolinguistics. Second edition. Cambridge: Cambridge University Press.
Jackendoff, R., 1993. Patterns in the mind: London: Harvester Wheatsheaf.
Kress, G., 1989. Linguistic processes in sociocultural practice. Oxford: Oxford University Press.
Kress, G., 1993. Cultural considerations in linguistic description. In: D. Graddol and D.L. Thompson
(eds.), Language and culture, l-22. Clevedon: Multilingual Matters.
Paolillo, J.C., 2000. Formalizing formality: An analysis of register variation in Sinhala. Journal of Lin-
guistics 36, 215-259.
Stubbs, M., 1996. Text and corpus analysis: Computer-assisted studies of language and culture. Oxford:
Blackwell.