traugott at CSLI.STANFORD.EDU
Wed Nov 1 03:08:28 UTC 1995
This is in response to Spike's boundary-challenging questions
for grammaticalization (hereafter GR).
Susanna Cumming rightly asked what theory of grammar was
being presupposed in the discussions. My view of grammar
arises out of the task I have set myself--to develop a theory of
historical discourse analysis. GR plays a role, but not an
exclusive one, in this task. I see grammar as structuring
cognitive and communicative aspects of language; it
encompasses not only phonology, morphosyntax and semantics
but also inferences that arise out of linguistic form (i.e.
linguistic pragmatics such as topicalization, deixis; other kinds
of pragmatics including encyclopedic knowledge are not part of
grammar, but of course are important in the speaker-addressee
negotiation that gives rise to GR).
In my view, GR is concerned fundamentally with gradient
diachronic processes. Early GR is the result of the interactions
of morphosyntax with pragmatics. In particular, conversational
implicatures arising from the principle of say no more than you
mean and imply more thereby may become salient over time,
leading addressees to interpret certain strings as expressions of
those implicatures. What is important for GR is that a change
can be said to occur when a new construction (morphosyntax-
semantics pairing) has arisen. The classic be going to example is
prototypical because i) the original string was syntactically
constrained (be going to V, not *be going to N), ii) the invited
inference of future and purpose arising from to in the
environment of the non-completive came to be semanticized as
the prime meaning of be going to, iii) a new construction came
into being. In addition, there was bonding and later
phonological reduction became possible.
But I do not regard bonding or attrition as criterial. Nominal
clines (nominal adposition > case) and verbal clines (main verb
> TAM marker) were the staples of GR theory in the eighties.
But there are other clines which are of interest as well, such as
PP/serial V > Connective (e.g. because) and Manner Adverb >
Sentence Adverb > Discourse Marker (e.g. in fact, indeed). These
all show decategorialization and generalization. They also show
subjectification in the sense that they become pragmatically
more situated in the Speaker-Addressee situation.
My current definition of early GR is that it is the process
whereby lexical material in highly constrained pragmatic and
morphosyntactic contexts becomes grammatical. More
specifically, lexical material in a syntactic string comes to
participate in the structural texture of the language, especially
its morphosyntactic constructions. Later GR involves already
grammatical material being generalized.
In most cases of GR, extant lexical items in extant syntactic
strings undergo changes as a result of discourse practices. So GR
does not involve discourse > syntax > morphology (pace Givon
1979), but morphosyntax via pragmatic inferencing in
discourse > more general morphosyntax.
What does this suggest for Spike's questions? Whereas most of
the cases we have studied to date fall into cross-linguistic
construction types (auxiliaries, cases, connectives, etc.), each
instance is unique, but the changes it undergoes occur within
the broad-scale constraints of decategorialization and
generalization. So I would not exclude any of Spike's exs. on
grounds of uniqueness. Go-went involves decategorialization of
the parts of the verb that became suppletive, but there is no
generalization; therefore I think that go-went suppletion is
lexicalization not GR. On the other hand, dem > copula involves
change in function (and in the example given) generalization.
The same holds for the third case, reanalysis of V > Pro.
Much of what I have said above is to be found in more detail
and with data in Paul Hopper and Elizabeth Closs Traugott,
1993, Grammaticalization (CUP), especially Chaps. 3-5, and in a
version of my ICHL paper the final draft of which is still in
progress ("The role of the development of discourse markers in
a theory of GR", ICHL XII, Manchester, August 1995). See also Joan
Bybee, Revere Perkins, and William Pagliuca, 1994, The Evolution
of Grammar, U of Chicago Press, for some similar but by no means
Elizabeth C. Traugott
More information about the Funknet