2.4 The Theory of Paths: General
2.5.1 Origins and development
One important aspectual innovation of Latin was the introduction of a periphrastic tense that unambiguously denoted perfect (Penny 2002:165). This construction exists in the Romance languages today. In French and most dialects of Italian and Romanian it has, as expected, generalized to perfective. In most dialects of Spanish it still has perfect meaning and contrasts with the preterit, but the perfect is gaining terrain just as in other Romance languages (Bybee et al. 1994:85, Comrie 1976:53). Therefore, the outcome of the process that started in Latin is
exactly as predicted by the perfective path; in certain varieties, the resultative has generalized to perfective, whereas in others, it is at different stages in the parallel development.
Originally, Latin had only a synthetic preterit; cantavi. This form could be used both to denote past actions with and without current relevance, that is, it also conveyed perfect meaning (Comrie 1976:53). Latin had a content verb called habere that had the lexical meaning of possession; “to have, to own”. Originally, habere was typically used in possessive phrases:
(18)
Cultellum habeo cultellu-m habe-o
knife-ACC have-1SG.PRES.IND
‘I have a knife’
However, there existed a construction consisting of habere + past participle (PP) which was used in phrases with resultative meaning, as the following:
(19)
Habeo cultellum comparatum habe-o cultellu-m compar-atum
have 1SG.PRES.IND knife ACC bought PART.ACC.
‘I have the knife which is bought’
In (2), the participle, comparatum, appears in the accusative case to concord with the noun cultellum; there was agreement between participle and object. There also existed a parallel construction that was used with the intransitive verbs, and that used esse (‘be’) instead of habere. Remains of this construction still exist in French and Italian (though not in Spanish), but their use is not determined by intransitivity, rather by unaccusativity;
(20)
Elle est allée à l’école elle est all-é-e á l’école
she be 3SG.PRES.IND go-PAST.PART-FEM to the school
‘She has gone/went to school’
(21)
Lei é andata alla scuola lei é and-at-a alla scuola
she be 3SG.PRES.INF go-PAST.PART-FEM to the school
‘She has gone/went to school’
Notice also how both participles agree in gender with the subject. In the French example, however, the agreement is only reflected orthographically (though this is not always the case). I now return to the habere + PP construction. This construction could be regarded as resultative because it originally denoted a state that was brought about by some action in the past (Bybee et al. 1994:63). This sense is similar to the perfect but different for the following reason: the perfect indicates that the past action has relevance in the present. But where the perfect has current relevance, the resultative consistently signals that the state persists at the reference time. For example, (22) is possible, but (23) is not:
(22) He has gone and come back already (23) *He is gone and come back already
This is precisely because the resultative points to the state resulting from the action, but the perfect points to the action itself. We can be quite sure that the constructions that now have perfect, perfective or simple past meanings in for example Romance and Germanic did arise form a resultative construction, since there do exist good written sources. Bybee et al. (1994:63) point to the fact that the resultative may be lexically restricted, that is, that the same construction does not always have a resultative meaning. This was true of Latin. For example, early in its development, it seems that the resultative construction was used exclusively with typical change of state verbs as persuade, learn and discover. Its grammaticalization can be observed as the construction generalizes to also include other types of dynamic verbs. When this has happened, the perfect sense evolves. This is because the meaning of this construction when used with a non-change of state verb is more an action with lasting relevance rather than a state resulting from an action. In these constructions, the participle indicated earlier action (i.e. relative past tense), while the verb in the present, habeo, indicated that the action belonged to a period of time still current (that is, still relevant), the meaning of such phrases was close to that of a perfect; I have bought the/a knife.
This structure originally only occurred with transitive verbs and explicit agreeing objects (as in example 19). The process of grammaticalization can be observed in sentences like (24) and
(25). Here, the verb has lost parts of it specific lexical meaning, it has undergone semantic bleaching. According to Bybee et al. (1994:69), as we have seen I the previous sections, this kind of generalization comes about in discourse contexts when the resultative is used to set the stage for a subsequent action. In this way, its use is extended not only to denote actions that produce states but also actions that precede other actions. Consequently, the perfect sense evolves.
Instances that indicate grammaticalization are of two types. (24) exemplifies a use with a verb that is logically incompatible with the notion of possession. The sentence in (25) shows no agreement between noun and participle.
(24) Non transitive uses of verbs
Habeo intellectum habe-o intellect-tum
have-1SG.PRES.IND understand-PAST.PART
‘I have understood’
(25) Phrases without agreement between participle and noun (Hopper and Traugott 2003:65)
Haec omnia probatum habemus haec omni-a prob-atum habe-mus
those ACC.PL all-ACC.PL try-PART.PAST have-1.PL.PRES.IND
‘We have tried all those things’