We now come to the claim that grammaticalization is unidirectional or – by terms which are used synonymously – irreversible or oriented. So we have to define what is meant by the unidirectionality of a process. Let P
be a process leading from an initial stage S1
to a final stage S2
; then for P
to be unidirectional means that there is no process P'
leading from S2
to S1
. Given this definition, our topic is the proposition that grammaticalization is unidirectional.
There is another issue that needs to be settled before we look at the empirical evidence. What is the methodological status of the said proposition? Is unidirectionality a definitory feature of grammaticalization, so that a definition of grammaticalization will contain a sentence such as “Grammaticalization is an irreversible process such that ...”? Or is the concept defined without recourse to this condition, and the unidirectionality of grammaticalization is instead an empirical generalization over observed grammaticalization paths?
This distinction is crucial. Suppose unidirectionality is a definitory feature of grammaticalization. Next suppose that empirical evidence is adduced for the above P'
, i.e. a historical change that involves the same stages as are constitutive of a process of grammaticalization, but in reverse chronological order. Then this would simply not come under the concept of grammaticalization by definition; it would be something else. If, however, unidirectionality is an inductive generalization advanced as an empirical hypothesis about a property of grammaticalization, then it is falsifiable. So if the same P'
is raised, it will be a counterexample to the unidirectionality hypothesis, and the generalization will be no longer tenable.
As a matter of fact, the proposition asserting the unidirectionality of grammaticalization is an empirical generalization arrived at inductively by analyzing a large set of grammaticalization phenomena, constructing their reverse in theory and looking out largely in vain for empirical manifestations of this construct. We are now in a position to formulate the unidirectionality hypothesis. The formulation already takes into account that bit of evidence that has been established as genuine cases of the reverse process. Grammaticalization is unidirectional in the sense that changes that go in the opposite direction of grammaticalization are very rare.
Before we can assess the evidence for this reverse process, we should characterize it in more detail. That is, we define degrammaticalization as the reverse of grammaticalization. Degrammaticalization is a process in which a linguistic sign gains in autonomy, i.e. it becomes relatively free from constraints of the linguistic system.
The methodological standards for the proof of a case of degrammaticalization are exactly the same as for grammaticalization:
For a set of data to count as evidence for degrammaticalization, it must fulfill three conditions:
- There are two historical stages of language
L
, earlierL1
and laterL2
. L1
has formF1
andL2
has formF2
, such thatF2
is diachronically identical withF1
.F1
is more grammatical thanF2
.
Thus, a convincing data set would be one where F1
was an inflected form at stage L1
while F2
was a periphrastic form or a syntactic construction at L2
, so that an inflectional affix becomes an auxiliary or another type of word.