Ziegeler, D. Debra.Ziegeler at ARTS.MONASH.EDU.AU
Mon Apr 1 17:20:33 UTC 1996

With reference to the recent discussion on unidirectionality, a possible
question that came to mind was: are bi-directional patterns of change
necessarily so, or are they in some cases representing a unidirectional
pattern which is actually cyclical? Alan King's comments on the
existence of patterns of the nature of A>B, and B>A via C seem to
suggest that this might be happening in some cases.

Bybee et al (1994: 190-1) discuss the sources in some languages of the
grammaticalisation of 'ability' modals as sometimes derived from verbs
with meanings of 'finish', 'reach', 'arrive at' etc., in other words,
attainment of a goal. This is suggested as being implied by the relation
of successful completion of an act with notions of past ability on the
part of an agent.

In English expressions such as: 'He was able to mend the fence', imply,
conversationally, that he did in fact mend it, although the sentence
only expresses past ability (thus, 'He was able to mend the fence, but
he didn't' is quite OK). The direction of inference can be routed as
a) He mended the fence > b) He was able to mend the fence > c) He mended
the fence; i.e. a) accomplishment > b) (past) ability > a)
This would appear to follow either a bi-directional shift of meaning, or
a unidirectional sequence, depending on your viewpoint. However, if we
hypothesised a (c) and a (d) stage, e.g. 'possibility', which is held by
Bybee et al (1994) to be a further stage in the grammaticalisation of
forms expressing ability, then the sequence which should result is: a)
accomplishment > b) (past) ability > c) possibility > d) (past) ability
> a) accomplishment. The question is whether or not such hypothetical
patterns actually do occur in languages. If so, then is there a
possibility that what we are looking at with some cases of
bi-directional semantic change can really be described as a cyclical
pattern of unidirectionality?

Debra Ziegeler
Dept. of Linguistics, Monash University,
Australia  3168

More information about the Funknet mailing list