Spike Gildea spikeg at OWLNET.RICE.EDU
Fri Oct 27 19:27:46 UTC 1995

Hey all,

We've been having a discussion here involving the term GRAMMATICALIZATION.
Our discussion has evoked memories of the Workshop on Diachronic Syntax at
the ICHL in Manchester last August.  Harris and Campbell's 1995 book on
diachronic syntax gives a pretty precise defintion of REANALYSIS, and
everybody seemed pretty happy with it, but then when the word
grammaticalization came up, arguments began flying around about which cases
of reanalysis could also insightfully be called GRAMMATICALIZATION.  It
made me wonder if there is really a precise notion of GRAMMATICALIZATION
that we all share, or if the notion is not perhaps more meant to evoke a
concept with fuzzy edges, that everyone sees just a little bit differently.

Anyway, to explore this question a little bit, I wonder if we could get
some discussion here on FUNKNET about examples like the following:

We all agree that the term GRAMMATICALIZATION is appropriate when a noun
becomes an adposition and then continues on to become a case inflection, or
when a complement-taking verb becomes an auxiliary and then continues on to
become a tense-aspect inflection.

Is it appropriate to use grammaticalization to refer to the kind of change
which leads to suppletive paradigms for the surviving word? -- e.g. the
modern paradigm for English _go_ was created in a process whereby the past
tense form _went_  of the separate verb _wend_ became the past tense of
_go_, and wend developed new, regular past tense form _wended_.  Although I
argued against the use of grammaticalization for this case, (since the
change is limited to a single lexical item rather than occurring in a broad
system), it can be argued that grammatical change is also taking place,
since a new "allomorph" is being created to mark past tense.

Another potentially controversial case is that where a demonstrative
pronoun gets reanalyzed as a the third person present tense copula, perhaps
later being extended to other persons (cf. Chinese) or tenses (cf. Panare),
or perhaps becoming a suppletive form in a fuller inflectional paradigm
(cf. Modern Hebrew and a number of Cariban languages).  Given that a
pronoun is already a "function word", and belongs to a closed class, and
given that copulas are also essentially function words (although they may
formally belong to the open class of verbs), do we have grammaticalization
in any of these cases?  Is it simply lexical change (especially in the case
where the new forms end up as part of a suppletive paradigm)?

A third case is that of the reanalysis of a verb to a pronoun.  How's that?
you ask.  Well, some verbs get reanalyzed as adpositions which inflect for
person, leading to (in essence) a new series of case-inflected pronouns.
If a certain case is lost (i.e. becomes unmarked) on nouns, then the
pronominal forms become unanalyzable as morphologically complex.  The
etymological verb form is now a series of pronouns.  Is this
grammaticalization carried out to its logical conclusion, or is this
something else (either a 'reversal' of grammaticalization, or a case of

There must be a bunch more cases out there that would lead to disagreement...


More information about the Funknet mailing list