The externalization of inflection in a layered system: Yupik mood
It is well known that derivational affixes generally occur inside of inflectional affixes, that is, closer to the root, yielding maximal structures like that in Figure 1.
Figure 1. Inflection outside of derivation
Yet we know that all derivational affixes in a language are not necessarily older than all inflectional affixes. Haspelmath (1993) brings to our attention a set of diachronic shifts in morpheme order whose result was to restore the external position of inflection after grammaticalization had added new derivational affixes to the outside of words, trapping inflectional affixes inside. Speakers restored the preferred structure in two stages. First the inflection was doubled, with a copy of the inflectional suffix repeated outside of the new derivational suffix. Next the original internal inflection faded, perhaps by analogy to other inflected words that contained no internal inflectional affixes. Only the external copy of the inflectional suffix remained. Haspelmath termed this process the ‘externalization of inflection’. Here we will examine another kind of externalization of inflection.
A disturbance in the peripheral position of inflection can result from another diachronic process as well. Existing derivational affixes may evolve into inflectional markers, at first in their original internal positions. At a certain point, they may be reanalyzed as part of an inflectional paradigm and take their place among the other members, at the edge of the word. The evolution of their function (a kind of grammaticalization) and subsequent formal reanalysis can thus stimulate a shift in position. Such a situation can be seen in Yup’ik. It will
be recalled that Yup’ik, like other Eskimoan languages, exhibits the morphological structure in Figure 2, repeated here.
Figure 2. Yup’ik Morphological Order
The same structure is reconstructed for Proto-Eskimoan by Fortescue, Jacobson, and Kaplan (1994). Both verbs and nouns show this structure, but the two are morphologically distinct. The inflectional suffix complex in nouns specifies the referent and its possessor (if present), as well as case. (Absolutive case is formally unmarked.) The inflectional suffix complex in verbs specifies mood and the core arguments of the clause. The pronominal suffixes on unpossessed nouns resemble those on third person intransitive verbs. The pronominal suffixes on possessed nouns resemble those on transitive verbs. All nouns and verbs contain one and only one inflectional suffix complex.
Figure 3. Inflectional Suffix Complexes
On the basis of comparative evidence, Fortescue et al. reconstruct four moods for Proto-Eskimoan: indicative, subordinative, interrogative, and optative/imperative. All four persist in modern Yup’ik. A number of additional moods appear in Yup’ik that are not reconstructed for the parent. Many of the new inflectional mood markers are transparently descended from derivational suffixes that are reconstructed for Proto-Eskimoan. One new mood suffix is the past contemporative-ller-, usually translated ‘when (in the past)’. Examples of its use can be seen in (1).
(1)
The past contemporative is transparently descended from a nominalizer that persists in modern Yup’ik with the same form-ller- ‘former (noun)’ or ‘the one that …-ed or was …-ed’.
(2)
The ancestor of the nominalizer is reconstructed by Fortescue et al. with the same shape: *-łət. (Yup’ik orthographic ll is a voiceless fricative [ł], e is a schwa [ə], and r is a voiced uvular fricative [ʁ]. Velar and uvular fricatives regularly alternate with stops in final position, like the r and q here.)
(3)
Traces of the diachronic source of the past contemporative mood are still visible in the shape of the pronominal suffixes that appear with intransitive verbs in this mood. An example of a nominal inflectional suffix complex can be seen in (4). The pronominal suffix is transitive, specifying the possessor and possessed. The case is locative.
(4)
The same nominal endings, complete with locative case suffix, appear with intransitive past contemporative verbs.
(5)
It is clear, however, that contemporatives like ayallratni ‘as they were leaving’ are no longer nominals. They have crossed the line and become verbs. In nominal possessive constructions like ‘the man’s boat’, a noun identifying the possessor (‘the man’) appears in the ergative case. With past contemporatives like ‘as the man left’, a noun identifying the leaver (‘the man’) appears in the absolutive case, consistent with the intransitive status of ‘leave’. The man is no longer classified grammatically as a possessor.
(6)
In transitive verbs, the traces of the historical nominal source of the past contemporative are disappearing. The past contemporative mood suffix is now usually followed by the verbal transitive pronominal suffixes that appear with other connective moods.
(7)
These endings differ from the locative possessive counterparts that would appear with nouns: angya-atni ‘in his boats’.
Other modern Yup’ik mood suffixes can easily be seen to be developing from derivational suffixes as well, derivational suffixes that can be traced back to Proto-Eskimoan and in some cases to Proto-Eskimo-Aleut. The simultaneous contemporative mood marker-(ng)inaner- ‘while’ can be linked to the derivational suffix-nginar- ‘to just (verb)’ plus the nominalizer-neq- ‘activity of …ing’. The consequential mood-(ng)a- ‘because’, has the same form as a derivational suffix-(ng)a- translated ‘to be in the state of having …-ed’, or ‘having been …-ed’, essentially portraying a result. The conditional mood-ku ‘when, if’ resembles a derivational suffix-ku- that derives adverbs pertaining to the future: wani ‘now’, wani-ku ‘soon’; unuk ‘night, last night’, unu-ku ‘tonight’. The participial mood suffix-lria has the same form as a derivational nominalizer-lria. Comparative evidence indicates that the new moods were not all added at the same time, but have been developing during different periods in the history of the family. In fact alternations between nominal and verbal endings on them in the modern language show that some are still in the process of evolving. Yup’ik speakers accept both tangallr-atni (nominal ending) and tangallr-aki (verbal ending) for ‘when he watched them’. Jacobsen 1982 lays out the range of nominalizing constructions in modern Yup’ik along a continuum, from those which are still primarily nominal to those which are primarily verbal.
The evolution of the derivational nominalizers into inflectional mood suffixes has resulted in a shift in morpheme order, consistent with the preference for derivation close to the root and inflection at the periphery of the word. In (8) the derivational nominalizer has a somewhat internal position, followed by a derivational verbalizer-u- ‘be’, then a suffix-yaq- ‘indeed’, then the inflectional suffix complex consisting of the intransitive indicative mood-u- and the pronominal suffix-q. In (9) the inflectional mood suffix appears immediately before the pronominal suffix-ani.
(8)
(9)
It appears that the marker-ller- has moved some distance, over several suffixes (here-llru- ,-u- ‘be’, and-yaq- ‘indeed’), before landing in the inflectional mood slot. But Yup’ik shows a layered structure rather than a templatic one. Derivational suffixes do not occupy fixed positions, so there are no special derivational slots for the nominalizer to hop over. A somewhat different kind of mechanism led to the apparent change in order.
Nominalizers may occur deeply embedded within the derivational morpholo gy, as in (8), but they appear most often at the end of the derivational string, immediately before the case and pronominal endings. Now the pronominal suffix complexes that indicate possession are similar in form to those that specify the core arguments of indicative verbs. Nouns contain no mood markers: the pronominal suffixes directly follow the noun stem. In many deverbal forms, it would be easy to reanalyze a nominalizer immediately preceding the pronominal complex as a mood marker. The important initial shift was not one of position, but rather from derivational to inflectional status. Only as later derivations were constructed would the shift in status lead to a visible shift in position.
