Previous research has noted that verbal suppletion for ergative number agreement (i.e. agreement with the subjects of intransitives and the objects of transitives) is widespread throughout the Uto-Aztecan language family and is therefore reconstructable to Proto-Uto-Aztecan (PUA) (Langacker, 1977). However, no previous works have systematically surveyed the attested forms of suppletion in these languages nor posited specific proposals for reconstructions of particular suppletive morphs back to PUA. We redress this lacuna by surveying the suppletive verbs in the various subgroups of Uto-Aztecan and assessing which of those are sufficiently widespread to reconstruct to PUA. We argue for specific PUA reconstructions for two verbal domains: die and kill, arguing that there were three distinct suppletive verb stems for marking these functions: *muku die.sg, *ko(i) die.pl, and *mɨɁa kill.sg. The plural form of kill in PUA was derived by adding a causative suffix *-ya to the plural stem for die, yielding *ko-ya. Other suppletive verbs in the family are not as easily reconstructable to PUA due to variation in attested forms, although some semantic functions seem to be widespread enough to be reconstructable. The PUA forms serving those functions would have been altered in different ways at different times by a lexical replacement process endemic to cases of strong suppletion, i.e. incursion (Juge, 2000). We also consider the issue of potential areal contact involving suppletion patterns in the areas where Uto-Aztecan languages are spoken, finding limited but suggestive evidence for possible areal effects involving suppletion for verbal number agreement.
Booker follows Voegelin and Voegelin (1977) in utilizing a number of “stocks” which are now heavily criticized (see e.g. Campbell 1997) e.g. “Hokan” “Penutian” etc. Obviously splitting these large-scale macro-phyla into a number of smaller families would in turn increase the number of families represented by verbal number suppletion in North America. Booker’s 13 genetic groupings presenting languages demonstrating clear verbal number suppletion are distributed throughout North America. They are: Caddoan; Chinookan; “Gulf” (which includes Muskogean); “Hokan” (which includes Pomoan and Yuman); Kiowa-Tanoan; Na-Dene (including Haida and Athabaskan); “Penutian” (including Zuni); Salishan; Siouan; Tonkawa (a language isolate); Uto-Aztecan; Yuchi (a language isolate); and Yukian.
Haugen and Siddiqi (2013) give arguments from the grammaticalization of functional morphemes from lexical morphemes against the DM view on a discrete split between “suppletable” functional items and “non-suppletable” lexical items. These arguments include data from Hopi which has suppletive suffixes as well as free verbs (with the former being more conducive to a “functional vocabulary” analysis than the latter) as well as Hiaki where there are suppletive “verb-affix hybrids” which seem to be only partway on the grammaticalization cline from free lexical verb root to bound functional suffix. See also Tubino Blanco et al. (2009) on this second point.
In Stubbs (2011) the column for the liquids is headed by *l; we follow Merrill’s (2013) reconstructions in (7).
According to K. Hill (2011) the Sr sg form for die can also be used in the plural interchangably with the pl form.
K. Hill (2001) proposes that the Hopi dual which applies only to animates derived historically from an alternant PUA plural suffix **-tï; this suffix would have contrasted with a regular plural suffix **-mï which was perhaps for inanimates. See also Hill and Hill (2000) on variation across Uto-Aztecan in patterns of marking plurality on different noun classes.
See also J. Hill (2011) for recent discussion problematizing “Taracahitan” as a grouping.
In Marlett’s (2011) orthography the beginning of a root or a stem is marked with the symbol √.
Marlett’s (2011) Seri grammar indicates suppletion for singular and plural subject agreement for both intransitive and transitive verbs which is unlike the ergative pattern seen in Uto-Aztecan and most other languages (which have agreement for the objects of transitives). Steve Marlett (personal communication) confirms that this is indeed the case for Seri while also noting that some such verbs (e.g. ‘travel to (a location)’) would be unlikely to take plural objects in any case.