The externalization of inflection in a templatic system: Cherokee infinitives
The evolution of derivation into inflection occurs in templatic structures as well, where slots are well defined. In both the Southern and Northern branches of the Iroquoian family, every indicative verb is inflected for aspect. Aspect is indicated by suffixes that occur word-finally or that are followed by just one other suffix. The aspectual categories in the two branches are similar but not identical. Cook (1979) analyzes North Carolina Cherokee as containing five aspects: two imperfectives (often the same in form), a perfective, a stative, and a category he labels an infinitive. The Northern languages show imperfectives, perfectives, and statives, but no infinitive. (Stative suffixes are also used to mark perfects, essentially resultant states such as ‘he has arrived’ or ‘I have sat down = I am seated’.) All of the languages show a variety of forms representing each category, but certain similarities can be identified across the Southern and Northern branches: imperfectives in both contain-s, for example, and the Cherokee stative-%î corresponds in a regular way to the most common Northern stative *-îo. In both branches of the family, imperfectives and perfectives generally appear with agentive pronominal prefixes, while statives indicating resultant states generally appear with patient pronominal prefixes.
Cherokee thus shows an additional category within the aspectual paradigm not found in the Northern languages: an infinitive. It is used in dependent clauses (‘for him to give it to her’) and (with a different modal suffix) for potential events (‘he has to give it to her’).
(1)
Infinitives show an intriguing set of allomorphs, not relatable by regular phono logical rule:-ihst,-hst,-‘st,-o‘t,-‘t,-%îht,-oht, and-ht.
(2)
The diachronic source of the new Cherokee infinitive markers can still be recovered. Among the derivational suffixes in the Iroquoian languages are a set of causative-instrumentals. They show a striking array of shapes, appearing in Cherokee as-ihst,-hst,-‘st,-%îhst,-o‘t,-‘t,-%îht,-oht, and-ht, and in all of the Northern languages as-hst,-‘st,-‘t,-ht, and-hkw. They contribute either causative or instrumental applicative meanings, depending to some extent on the animacy of the introduced participant.
(3)
Because they occur in both branches of the family, we can reconstruct the causative-instrumental suffixes *-ihst, *-ht, *-‘t for Proto-Iroquoian. We can do even more. Languages from both branches of the family retain reflexes of a full verb root that was an apparent lexical source for some of the forms: *-ihst- ‘use’.
(4)
The probable antecedent of the suffix-(∧î)ht- is retained in another Cherokee verb root:- ∧îht- ‘use’.
(5)
The odd array of allomorphs exhibited by the Cherokee causative/instrumentals (and reconstructible for Proto-Iroquoian) is exactly the same as the odd array exhibited by the new Cherokee infinitive. The array is too unusual for the resemblance to be due to chance. It strongly suggests that the older derivational causative/instrumental markers evolved into the new inflectional infinitive markers. It appears unlikely that the original verb roots that served as the sources of the causative/instrumentals could have evolved directly and independently into the infinitive markers, given the perfect correspondence between the arrays of instrumental and infinitive forms. Furthermore, individual stems within Cherokee exhibit the same allomorphs of the instrumental and infinitive suffixes, though the allomorphy is not phonologically conditioned. The verb root-hne:‘- ‘speak’, for example, appears with both the instrumental-ihst- and the infinitive-ihst-.
(6)
It is clear that by now, the instrumental and infinitive are distinct suffixes. They may cooccur within a verb.
(7)
The development of the infinitive category in Cherokee may have been stimulated by language contact. The Southeast constitutes a recognized linguistic area, including not only Cherokee of the Iroquoian family, but also languages of the Muskogean and Caddoan families. Both Creek, a Muskogean language, and Caddo, a Caddoan language, contain infinitives (Jack Martin p.c. and Wallace Chafe p.c.) Whatever the initial stimulus, Cherokee speakers exploited language internal resources for the new infinitive markers. In the Iroquoian languages, morphological verbs are commonly used periphrastically as syntactic nominals, providing descriptive names of objects, such as the Mohawk tehahîuhtané:k%î ’his ears are side by side’ = ‘rabbit’. No overt category shifter is necessary: morpho logical verbs simply function syntactically as nominals, without special marking. The instrumental suffix appears in large numbers of such constructions, because introduced objects have so often been described in terms of their uses.
(8)
In all of the Northern languages, the imperfective aspect suffix-ha‘ in particularly common constructions of this type has been replaced with the noun suffix a‘. In some of the languages, the forms have been shortened, omitting the instrumental and imperfective. The Mohawk term for ‘car’, for example, probably first coined for ‘wagon’, originated as an instrumental verb based on the root-‘sre- ‘drag’. The instrumental suffix is now omitted in Mohawk, though it is retained in some of the other Northern Iroquoian languages. When the noun root for ‘car’ is incorporated into another verb, however, the instrumental suffix must be present. There is a general grammatical requirement that only formal morphological noun stems may be incorporated. The instrumental suffix can now apparently function on its own as a marker of nominalization.
(9)
A number of other modern nominals in the languages show a similar pattern, though no verbal origin is now recoverable. When incorporated, they include an additional sequence identical in form to one of the instrumental suffixes.
(10)
Why should such forms evolve into infinitives? Infinitives in many languages serve to reify or objectify events and states, a kind of nominalizing function. This nominalizing function may have served as a semantic link between Cherokee instrumentals, frequently used in nominals, and infinitives. Describing Oklahoma Cherokee, Pulte (1975) points to the match in form between certain nominals and infinitives, with only a change in pronominal prefix.
(11)
Of particular interest here is the fact that the evolution from instrumental to infinitive appears to have been accompanied by a drastic shift in morpheme order. The instrumental is an inner derivational suffix in all of the Iroquoian languages, while the aspect markers are outer inflectional suffixes. The suffix morphology of the two branches of the family is compared in Figure 1.
Figure 1. Suffixes
Both the derivational suffixes and the inflectional aspect markers appear to have been in place in the parent language. The inner derivational suffixes generally appear to be cognate, though precise reconstructions are not possible given the brevity of the forms. The Cherokee reversive is-ki:-/-y- and the Northern Iroquoian counterparts *-hkwi-/-hsy-. The Cherokee dative-benefactive ‘for’ has Cherokee forms-e:-/-hsi- and Northern Iroquoian counterparts *-‘se-/ ‘s-/-ati-/-ahs-/-hse-/-7î-. The andative ‘go and’ has forms-hi-/-ihl- in Cherokee and *-h-/-hr-/-hn-/-hsr- in the Northern languages. The purposive ‘in order to’ has the form-e:- in Cherokee and-e- in the Northern languages. (This suffixis probably descended from the root-e- ‘go’ which retains reflexes in all of the modern languages.) The last derivational suffixes are comparable in function between the two branches of the family and show some similarity in shape, but it is difficult to determine whether or not they are cognate. Cook lists the reiterative as-hi‘s-, somewhat reminiscent of the Northern Iroquoian distributives-‘s and-‘sh˛ó. The progressive (termed ‘ambulative’ by Cook for Cherokee) is attached to stative verbs in all languages and shows similar but not identical shapes,-i:to:-/-i:ta in Cherokee and *-atye‘ in the Northern languages, with similar meanings ‘go along …-ing’, ‘go about …-ing’.
In their development from instrumentals to infinitives, the suffixes appear at first to have jumped over the dative, the andative, the purposive, the reiterative, the progressive, the repetitive, and the completive, all of which precede the aspect suffixes. The actual mechanism was probably not a long-distance hop, however, but rather a multi-step process of reanalysis, parallel to that observed in Yup’ik. All of the derivational suffixes that follow the instrumental are optional; most often the instrumental is the last suffix in the stem. The marker probably first began to acquire an infinitival sense in its original position within the derivational morphology. As its function was generalized from that of a derivational instrumental suffix through that of a nominalizer to that of an infinitive, it showed increasing semantic affinity to the set of existing aspectual suffixes. At a certain point it was reanalyzed as part of the aspectual paradigm, to which it was usually adjacent. The shift in status from derivational to inflectional did not necessarily entail a shift in position of the marker itself. If anything, it was the boundary between derivation and inflection that shifted in these constructions, from the right to the left of the suffix. It was only when new complex forms were derived, and other derivational suffixes were positioned to the left of the new infinitive, that the reanalysis became visible.
