Why suppletion may be seen as grammaticalization

[Oesten Dahl]

Elizabeth Traugott says that she thinks that "go-went" is lexicalization,
not grammaticalization.
However:
Suppletion may play a role in processes that also involve what is
undoubtedly grammaticalization on anyone's definition of the term and
together with the latter contribute to the creation of new grammatical
distinction. Consider for instance the rise of perfective-imperfective
distinctions. In the well-known case of Russian, verbs come in
imperfective-perfective pairs, where the relation between the two members
may be of a rather different character:

1. unprefixed imperfective vs. prefixed perfective, e.g. fotografirovat' -
sfotografirovat' 'photograph'
2  suffixed imperfective vs. plain perfective, e.g. davat' - dat' 'give'
3. different suffixal formations from the same root, e.g. zamykat' -
zamknut' 'close'
4. imperfective and perfective from totally different stems, e.g. brat' -
vzjat' 'take'
and various other more complex cases

The general process that seems to be going on is that erstwhile lexical
distinctions - whether derivational or of a more idiosyncratic character -
have come to be exploited in grammar to create aspectual paradigms. In
Russian this process has not been completed: the members of the pairs are
usually seen as separate lexemes. A closer look at the tense-aspect systems
of IE languages such as Latin, Greek, and Sanskrit shows, however, that the
distinctions between different stems (present, aorist, perfect) in the verb
paradigms have similar origins and that suppletion plays a role there too.

In my opinion, there is motivation for using the term "grammaticalization"
in such a way that it covers all processes that contribute to the rise of
grammatical constructions or grammatical paradigms. Elizabeth's view is of
course also motivated if it is assumed that the fusion of two earlier
lexemes into one should be called lexicalization. But then not only
suppletion but also the inflectionalization of derivational morphemes falls
under this definition of lexicalization. So the conclusion would probably be
that grammaticalization and lexicalization are not necessarily distinct
processes.

Oesten Dahl