Response to Newmeyer's 'Grammar is Grammar and Usage is Usage'

Abstract

In an article in a recent issue of Language (Newmeyer 2003), Frederick Newmeyer argues for a clear separation between what he terms ‘knowledge of language [i.e. grammar] and use of language [i.e. pragmatics]’ (682). In developing his argument, N makes frequent reference to linguistic analyses that are corpus-based, that is, centered on information taken from large datasets of actual speech or writing (e.g. transcriptions of conversations, newspaper articles, novels). In N’s view, grammar is distinct from pragmatics, and because a corpus contains examples of actual language use drawn from a large community of speakers and writers, it can yield only performance data. As a result, N argues, ‘there is no way that one can draw conclusions about the grammar of an individual from usage facts about communities, particularly communities from which the individual receives no speech input’ (696). Implicit in this argument, however, is the view that the traditional way of collecting linguistic data—introspection—somehow gives us insights into the competence of the native speaker that corpus data cannot. The real issue, we argue, is not which kind of data gets us closer to the native speaker’s competence, a goal that we believe is largely unobtainable, but rather how linguists can best collect data relevant to the linguistic analyses they are conducting. There is considerable evidence that a corpus can enrich our understanding of language and, in many cases, provide linguists with examples they would have never considered had they relied only on data obtained through introspection. To support this view, it is worth reviewing N’s critique of Manning’s (2003) use of corpus data to challenge Pollard and Sag’s (1994) analysis of verb subcategorization in English. Manning (2003:299) notes that Pollard and Sag claimthat the verb regard can be followed by as-complements (1) but not by predicative to-complements (2).

Full text

Response to Newmeyer’s ‘Grammar is grammar and
usage is usage’*
C
HARLES
F. M
EYER
H
ONGYIN
T
AO
University of Massachusetts, Boston University of California, Los Angeles
In an article in a recent issue of Language (Newmeyer 2003), Frederick Newmeyer
argues for a clear separation between what he terms ‘knowledge of language [i.e.
grammar] and use of language [i.e. pragmatics]’ (682). In developing his argument, N
makes frequent reference to linguistic analyses that are corpus-based, that is, centered
on information taken from large datasets of actual speech or writing (e.g. transcriptions
of conversations, newspaper articles, novels). In N’s view, grammar is distinct from
pragmatics, and because a corpus contains examples of actual language use drawn from
a large community of speakers and writers, it can yield only performance data. As a
result, N argues, ‘there is no way that one can draw conclusions about the grammar
of an individual from usage facts about communities, particularly communities from
which the individual receives no speech input’ (696).
Implicit in this argument, however, is the view that the traditional way of collecting
linguistic data—introspection—somehow gives us insights into the competence of
the native speaker that corpus data cannot. The real issue, we argue, is not which
kind of data gets us closer to the native speaker’s competence, a goal that we
believe is largely unobtainable, but rather how linguists can best collect data relevant
to the linguistic analyses they are conducting. There is considerable evidence that
a corpus can enrich our understanding of language and, in many cases, provide
linguists with examples they would have never considered had they relied only on
data obtained through introspection.
To support this view, it is worth reviewing N’s critique of Manning’s (2003) use of
corpus data to challenge Pollard and Sag’s (1994) analysis of verb subcategorization
in English. Manning (2003:299) notes that Pollard and Sag claim that the verb regard
can be followed by as-complements (1) but not by predicative to-complements (2).
(1) We regard Kim as an acceptable candidate.
(2) *We regard Kim to be an acceptable candidate.
However, in an analysis of texts in the New York Times, Manning (2003:300) found
examples such as 3, where regard can take a to-complement.
(3) Conservatives argue that the Bible regards homosexuality to be a sin.
Manning notes that counterexamples such as 3 were not anomalous: he found many
additional discrepancies between Pollard and Sag’s intuitions and the data appearing
in his corpus.
In commenting on Manning’s observations, N states:
Perhaps [example 2] is generated by Pollard’s grammar and perhaps it is not. Perhaps [example 2] is
generated by Sag’s grammar and perhaps it is not. But we will never find out by reading the New York
Times. The point is that we do not have ‘group minds’. No input data that an individual did not experience
can be relevant to the nature of his or her grammar. (696)
* We wish to thank Brian Joseph and Richard T. Oehrle for very helpful comments on an earlier version
of this response.
226

DISCUSSION NOTES 227
But the issues raised by examples 1– 3 have less to do with notions such as ‘group
minds’ and ‘input data’ and more to do with how we collect data for the linguistic
analyses that we conduct. Linguists arguing against the use of corpus data in favor
of introspection seem to think that they have direct access to the native speaker’s
competence—that by consulting their own intuitions, they will be able to make ‘gram-
maticality judgments’. But does anyone (linguist or otherwise) really have access to
linguistic competence? And does intuition equal competence? Schu
¨tze (1996:26) argues
that ‘in principle, there might someday be an operational criterion for grammaticality,
but it would have to be based on direct study of the brain, not on human behavior, if
it turns out to be possible to discern properties of the mind (e.g., the precise features
of grammar) from physical properties of the brain’.
In other words, data collected introspectively or from a corpus is data based on
different types of human behavior: the experiences of the linguist in the former, the
experiences of many individuals in the latter. And linguistic behavior is notoriously
variable. Snow and Meijer (1977) found that linguists and nonlinguists make very
different judgments about linguistic data. Linguists are more consistent in their judg-
ments of sentences than nonlinguists, and they tend to agree more with one another
than nonlinguists do. Morgan (1972:285) asked a number of native speakers whether
they favored a singular or plural verb in the sentence Either Harry or his parents 兵is/
are其coming and found that ‘some speakers accept is and reject are, some accept are
and reject is, some accept both, some accept neither, and some ‘‘just don’t know’’
which (if either) is more acceptable’.
1
Thus, even though The New York Times Corpus, the Switchboard Corpus, or any
other corpus for that matter may not serve as direct and exclusive input data for any
individual English speaker, it strikes us as imperative that linguists seek wider sources
of data for their analyses, including, yes, building larger corpora representing wider
varieties of language. Introspection certainly has its place in linguistic analysis: after
all, a corpus contains a finite universe of data that needs to be supplemented by other
sources of data. At the same time, it is indisputable that corpora can provide data that
the linguist, or the native speaker for that matter, may not have been aware of. Even
Chomsky (1965:18) acknowledges that so-called ‘performance data’ plays a role in
linguistic analysis: ‘Clearly the actual data of linguistic performance will provide much
evidence for determining the correctness of hypotheses about underlying linguistic
structure, along with introspective reports (by the native speaker, or the linguist who
has learned the language)’.
REFERENCES
C
HOMSKY
,N
OAM
. 1965. Aspects of the theory of syntax. Cambridge, MA: MIT Press.
C
OWART
,W
AYNE
. 1997. Experimental syntax:Applying objective methods to sentence judg-
ments. Thousand Oaks, CA: Sage Publications.
M
ANNING
,C
HRISTOPHER
D. 2003. Probabilistic syntax. Probabilistic linguistics, ed. by Rens
Bod, Jennifer Hay, and Stefanie Jannedy, 289–341. Cambridge, MA: MIT Press.
M
ORGAN
,J
ERRY
L. 1972. Verb agreement as a rule of English. Chicago Linguistic Society
8.278–86.
N
EWMEYER
,F
REDERICK
. 2003. Grammar is grammar and usage is usage. Language
79.682–707.
1
See Cowart 1997 for a discussion of how to devise experiments to elicit linguistic judgments and
minimize intersubject variation.

LANGUAGE VOLUME 81, NUMBER 1 (2005)228
P
OLLARD
,C
ARL
, and I
VAN
A. S
AG
. 1994. Head-driven phrase structure grammar. Chicago:
University of Chicago Press.
S
CHU
¨TZE
,C
ARSON
T. 1996. The empirical base of linguistics. Chicago: University of Chicago
Press.
S
NOW
,C
ATHERINE
, and G
UUS
M
EIJER
. 1977. On the secondary nature of syntactic intuitions.
Acceptability in language, ed. by Sidney Greenbaum, 163–77. The Hague: Mouton.
Meyer [Received 19 July 2004;
Applied Linguistics Program revision invited 27 September 2004;
University of Massachusetts, Boston revision received 1 December 2004;
100 Morrissey Blvd. accepted 1 December 2004]
Boston, MA 02125-3393
[meyer@cs.umb.edu]
Tao
University of California, Los Angeles
Department of Asian Languages and Cultures
290 Royce Hall
Los Angeles, CA 90095-1540
[tao@humnet.ucla.edu]