Morphology in the Muskogean languages 1

The indigenous languages of the Americas exemplify a number of uncommon typological patterns, especially in their morphology. Here, that rich morphology is illustrated via the Muskogean languages of the southeastern United States. Muskogean languages are agglutinative, but even more interesting and uncommon patterns emerge in an analysis of their morphology. These include subtractive morphology, suppletion, infixation, ablaut, and the use of suprasegmentals. These morphological patterns present considerable complexity. Inflected verbs in narratives and conversation often reflect more than one of the morphological processes. This morphological complexity also demon-strates characteristics of being nonlinear, of being prosodic yet not aligning with neat prosodic boundaries, of not having direct correspondence between grammatical categories and surface segments or suprasegmentals, or having more than one of those characteristics. Six of the seven Muskogean languages are still currently spoken by fluent first language speakers, and many of the tribal nations who represent these languages are involved in ongoing documentation and revitalization efforts, often in partner-ship with linguists. Thus, despite their highly endangered status, excellent existing documentation and new questions in research create an opportunity to collect even more intricate inflected forms that will enrich models of morphology and morphological theory while having broader impacts, like supporting tribal language revitalization.


| INTRODUCTION
The languages native to the Americas are noteworthy for their morphology, especially the Muskogean languages, leading to significant developments in linguistic theory. Tribes originally located in the southeastern United States spoke Muskogean languages, with some tribes forced westward due to the Indian Removal Act of 1830. Consequently, these communities and languages now extend even further geographically west, to Texas and Oklahoma. Munro and Ulrich (1984) of these facts invokes metrical structure and Kiparsky's theory of Lexical Phonology (Kiparsky, 1982), and Hayes (1995) presents a detailed iambic metrical analysis of Rhythmic Lengthening in Choctaw.
Muskogean agglutinative verbs and their diverse morphological processes provide rich data sets and insights for linguistic theory. In the next sections, I focus on the morphology of the languages in this family, showing the instantiation of typologically varied word-building processes. One goal is to describe those morphological features that are typologically uncommon, while also demonstrating how Muskogean languages instantiate these features. A second goal is to show how these uncommon morphologies can interact. Because of existing analyzed data, uncommon typological patterns, and current research projects with speakers of a number of Muskogean languages, the prospect of newer and richer data sets for investigation is very likely to emerge, fueling new avenues of theoretical investigations.

| AFFIXATION AND AGGLUTINATIVE PROPERTIES
Affixation predominates in Muskogean verbs, whether prefixes, suffixes, infixes, or circumfixes. Overall, Muskogean nouns allow fewer categories or markers in terms of affixation, while verbs are richly inflected, marking grammatical relations, locatives, directionals, instrumentals, tense, aspect, modality, evidentiality, and mood. Verbal morphology can affect valency via causatives or applicatives (see Martin 2000, for example), and there are morphological markers of clause combining like switch reference and complementizers. The various grammatical categories contrast with a language where similar concepts are lexically realized, like English. 6 To illustrate this, (1) shows inflected Chickasaw forms of the verb takchi "to tie it." 7 (1) Chickasaw affixed verbs from paradigms of takchi "to tie it" These examples illustrate aspects of morpheme order in the verb. Munro (2005: 139) notes that Chickasaw verb stems may combine with up to six additional elements (pronominals, modality, and tense/aspect) and even more if the verb stem includes reflexives, reciprocals (like 1c-d), or applicatives (like the comitative suffix in 1b-e). Co-occurrence restrictions between grammatical categories influence how many possible affixal combinations there are. And as Section 4 shows, Muskogean languages also display nonlinear morphophonology.
Switch-reference morphology plays a prominent role in Muskogean clausal morphology by marking whether or not subjects in subordinate clauses co-refer with the subject of the matrix verse (2a) or whether the two have different subjects (2b).
(2) Switch reference in Choctaw (Broadwell, 2005: 193) Broadwell (2005) notes that /-t/ or /-sh/ marks same subject while different subject is marked by nasalizing the verb's final vowel making it homophonous with the case suffixes. McKenzie (2015) surveys switch reference in 69 North American indigenous languages and notes that while homophony with some morpheme is typologically common for switch-reference markers, it is not always with the case markers as seen in Muskogean. More research into switch reference is called for. Broadwell notes that "the interplay between switch-reference marking and topic continuity needs more careful study " (2005: 197), while Munro comments that "Selection of particular connectives and switch-reference markers is an aspect of Chickasaw discourse structure that would reward further study " (2005: 148).

| INFIXES AND CIRCUMFIXES
Muskogean languages also display less common morphology such as circumfixes and infixes. The data in (3) show how Chickasaw negation functions in a circumfixal fashion, with an irrealis pronominal prefix occurring in combination with the negative suffix, -o to negate an action. Generally, pronominals are prefixal aside from first person singular agent (3a). Note that the first person singular pronominal is a suffix in declaratives (3a-b) and a prefix in negatives (3c).
(3) Chickasaw circumfixes When the prefixes are used without the negative -o, they indicate hortative forms, as in iktakchi' (iktakchi-') "let him/her tie it," requiring also a glottal stop as suffix. While the Chickasaw irrealis is a prefix, its cognate in Alabama "appears variously as a suffix, infix, or prefix with the shapes -ki, -ikko, -ki-, -kii-, or ik-" (Montler & Hardy, 1991: 1) along with another strikingly familiar element, an -o suffix: 9 (4) Infixing in Alabama negation (Montler & Hardy, 1991: 4-5) 10 Alabama has two additional infixing patterns. Montler and Hardy (1991:2) show an affix, -l-, which alternates between distribution as a prefix (5a) versus as an infix (5b-c); they describe the [i] as epenthetic breaking up illicit consonant clusters: (5) Alabama middle voice and infixation patterns (Montler & Hardy, 1991: 2) When we see infixation, the distribution of the morpheme -lis following the nucleus of the initial syllable. If the resulting form would create an unsyllabifiable string of consonants, the morpheme surfaces with the epenthetic vowel following it, so as -li.
Finally, the second person agent pronominals in Alabama divide into two sets, one, is-(with a post-sonorant variant of -his-), surfacing as prefix (6a) or infix (6b) and the other, -chi, as either a suffix (6c) or infix (6d).
(6) Alabama second person "mobile" affixation The data includes additional complexities in Alabama, including vowel elision and the irregular patterning of nochi with the so-called control series of the active or agent pronominals.
While the phonological aspects of this system are reasonably well documented, the semantics are less so. As noted by Munro and Willmond (1994) for Chickasaw and by Broadwell (2006) for Choctaw, not all the grades are equally productive, and some uses are lexicalized. Grades mark different aspectual meanings, but a verb's lexical aspect, or aktionsart, is not always compatible with the aspectual qualities of a particular grade. The full nature of the restrictions between the morphological aspect of grades and the aktionsart individual verbs is "not well understood, and needs more careful examination" (Broadwell, 2006: 161). Importantly, this and other excellent and insightful analyses go beyond description, providing the foundation for theoretical treatment and the exploration of related implications. Ulrich (1986) analyzes an extensive number of distinct (yet phonologically interconnected) grades and is likely the most comprehensive and complete phonological treatment because it includes the largest number of distinct grade formations (six) and the grades' interactions with syllable structure, rhythmic lengthening, and accent.
(7) Choctaw verb grades (adapted from Ulrich, 1986, Chapter 6) Ulrich (1986) notes that when the glottal grade, seen in various inflected forms in (7f), Ulrich (1986) notes that when it is used, it has a meaning that varies along with its particular affixes, shown above as they express different functional information; the first example includes switch reference suffix, and the second one includes the ireless/negative pattern of circumfix and -o suffix.
The generalizations and example forms of the grade patterns appear in (8), analyzed into morphological components. Other affixation has been stripped off for ease of analysis in highlighting the relevant patterns.
(8) Patterns of Choctaw verb grades As (8) shows, the penultimate vowel is the locus for implementing the grade effects, and the grades make use of one or a combination of the same basic elements, primarily suprasegmental: accent, length (C and/or V), nasalization, laryngeals, and glides. The intricate and interwoven elements of this aspectual marking are not phonologically contained within the boundaries of a single prosodic element such as a syllable or even a foot, making it a challenge to express the descriptive generalizations in terms of some linear segmental string.
A considerable amount of phonological knowledge is necessary to correctly form verb grades: the verbal base; the penultimate syllable; the appropriate manifestation of the nasal; and in some grades, the generation of a copy of the penult vowel (which must bear an accent), as well as the prosodic conditions where the various permutations occur, for example, when the penultimate syllable is open or closed, light or heavy. A host of phonological conditions requires sophisticated knowledge in order to determine the surface shape of an inflected verb grade form, as well as morphological awareness of the base form and affixes.
Referencing the penultimate syllable suggests an active metrical foot. Prosodic morphologyoriented accounts using feet were proposed throughout the 1980s and 1990s, including Ulrich (1986Ulrich ( , 1987Ulrich ( , 1994, Lombardi and McCarthy (1991), Weeda (1992), Hammond (1993), among others. Building even further on the metrical tools noted earlier in this paper's introduction, analyses of the ablaut system treated the internal phonological modifications within the prosodic morphology theoretical framework. Lombardi and McCarthy (1991) drew on prosodic circumscription and authentic units of prosody, for example, an iambic foot and an affixed mora, to analyze surface phonological patterns of ablaut in Choctaw. Authentic units of prosody like the foot are key components of the Prosodic Morphology Hypothesis (McCarthy and Prince 1986), and invoked in competing analyses like Hammond (1993), Ulrich (1994), and, later, Restle (2002).
While these analyses rightfully (in my opinion) draw on insights that moraic and metrical phonology provide the key underpinnings to analyzing the verb grades, they fall short in limiting the scope of the phenomena analyzed. For a counterpoint from another (unrelated) Native American language, Fitzgerald (2012) examines a wide range of rhythmic and prosodic morphological data from Tohono O'odham that bear on the role of quantity, including gemination and vowel lengthening. A similar in-depth analysis of quantitative phenomena in a Muskogean language simply is not out there. The above analyses of Choctaw provide obscured glimpses of the morphophonology of quantity, but what is still lacking is a deeper comprehensive analysis of the many quantitative processes that play a role in Choctaw (or any of the Muskogean languages). I return to this point in subsequent sections.

| SUBTRACTIVE MORPHOLOGY
Subtractive morphology, where deletion serves to mark some grammatical function, occurs in five of the Muskogean languages. Subtractive morphology provides another rich example of how morphological processes in Muskogean languages go beyond a straightforward additive, item, and arrangement type of affixation. The most significant theoretical discussions in the literature focus on Koasati, but Broadwell (1993) and Martin (1994) note that five of the Muskogean languages invoke subtractive elements, especially in plural morphology. Broadwell (1993) uses related subtractive processes in plural and repetitive verb formation to argue that Chickasaw, Choctaw, Alabama, Koasati, and Mikasuki constitute a Southern Muskogean branch of the language family, following Munro (1985cMunro ( ,1987aMunro ( , 1993. Mikasuki deletes the rhyme of the penultimate syllable to mark the contrast between singular versus plural forms, shown in (9). While (9a-c) shows that this can target a medial short or long vowel in an open syllable, (9b) shows that it is truly rhyme deletion since the coda consonant in a closed syllable is also deleted.
(9) Mikasuki (Broadwell, 1993, via David West pc) Kimball (1991) outlines a robust pattern in Koasati where subtractive morphology forms plurals. Within a trisyllabic singular verb, all or some of the rhymes of the medial syllable of the singular are deleted in plural formation. The target of the subtractive process applied to this medial syllable represents both open and closed syllables and can be a long vowel (10a), a diphthong (10b), a vowel plus coda consonant (10c), or in some cases, just the coda consonant (10d).
(10) Koasati subtractive plural formation (Kimball, 1991) 14 While the original discussions by Kimball (1983Kimball ( , 1985 handled this via a series of rules doing deletion and replacement of parts of the verb base, following analyses invoked an approach invoking a rime template. Martin (1988) treated this as dissociation of a prosodic constituent, and Lombardi and McCarthy (1991) used prosodic circumscription of the final syllable of the root, applying the delete operator to remove syllabification. The segmental material reassociates into the preceding syllable, which means only the onset consonant surfaces, since it can change its affiliation into a coda with that syllable. A similar approach is found in Weeda (1992).
Some of the recent phonological treatments of this phenomena have drawn on approaches relying on Optimality Theory's faithfulness (or anti-faithfulness) constraints, as in Kurisu, 2001, Horwood, 2002 For example, Kurisu (2001) uses lexically specified constraints in analyzing Koasati, ranking a ban on deletion in singular forms ranked more highly than a ban on deletion in plural forms. A constraint ranked between them forces morphological exponence to have a phonological interpretation; deletion is that exponence. The relational approach contrasts with Trommer and Zimmermann (2014), who invoke mora affixation in their analysis of Alabama subtractive morphology, where they argue that when the affixed moras have defective prosodic structure, subtractive processes are one of the possible outcomes.
While these analyses in different ways draw on moraic phonology in accounting for these phenomena, it is not clear that they sufficiently integrate the pervasive and complex influence played by prosody in Muskogean family. The interrelated moraic and prosodic patterns call for an approach that better exploits an analysis where all the morphological and phonological systems are working together.

| SUPPLETION
Suppletion is the process where semantically related words surface with essentially different shapes, lacking predictability in their form with regard to other paradigm members, instead of showing regularity of form and phonological similarity in a given word's paradigm. Classic examples of suppletion in English are go, went (*goed) which contrasts with regular paradigmatic and phonologically similar formal patterns of walk, walked.
Numerous Native American languages exhibit suppletion in verbs when indicating grammatical number, as noted in Booker (1982), and suppletion is well attested in Muskogean languages. As observed by Haas (1948), Heath (1980), and others, it is especially attested in verbs of stance or motion, known as positional or classificatory verbs. Positional verbs in Muskogean have received a fair amount of attention by scholars of Muskogean languages and by linguists interested in suppletion or in how grammars utilize shape. Haas (1948) further notes intriguing characteristics of these verbs with regard to their behavior when the nouns involved with predicates are cloth or liquids. Nouns representing cloth objects, such as handkerchiefs, surface with a dual marked verb, intransitive when they function as subjects and the transitive stem when functioning as objects. 15 Formal properties of suppletion in Muskogean show that it can be partial or full. Heath (1980) describes a set of Choctaw verbs as having partial or full suppletion to mark distinctions in number, aspect (which he characterizes as distinctions between unmarked, distributive, and repetitive), or perhaps also membership in a lexical-semantic class. Martin (2011a: 197) describes Creek suppletion in distinguishing number, with (11) showing singular, dual, and triplural forms. Data from Martin (2011: 199-200) illustrates the suppletive nature of grammatical number for positional verb stems, both intransitive (11a, c-e) and transitive (11b). 16 (11) Creek suppletion (Martin 2011: 199) Examples (11d-e) further show that some of the dual (or plural, where paradigms show only a twoway number contrast) suppletive forms include an internal element, -ho-. Intransitive triplurals can also show internal modification, where -icor -ycreplace the stem-final k found in singular forms. Creek suppletion reinforces the point in the previous section that Muskogean morphology frequently intertwines more than one of these typological patterns. Data from Chickasaw make this point even more nicely.
Positional verbs in Chickasaw also show suppletion and a three-way number contrast in verbs. The paradigms below, such as exemplified by the examples in (12a), show that the dual form shows full suppletion, with no similar segments, and the triplural form is partially similar, showing partial suppletion.
(12) Chickasaw positional verbs and suppletion 17 (Munro, 2006: 4-6) Like Creek, Chickasaw has suppletion intersecting with another typological pattern, but here, it is with ablaut. Munro (2006) notes that the triplurals are typically used in concert with máa, the plural auxiliary, and that the non-triplural forms are grade forms of related verbs. For example, the singular form in (12a), bínni'li, is a g-grade form of biniili, "to sit down (punctual)." In (13), graded examples from (12) are shown with details about which grade is involved and what base the grade operates on. In addition, the number of the suppletive form is indicated. While the data are primarily showing graded forms for non-triplurals, (13f) shows that triplurals do on occasion represent a grade.
(13) Chickasaw verb grades and suppletion (entries in Munro & Willmond, 1994) The interaction of suppletion with the prosodically driven grade process adds a uniquely Muskogean dimension to this process. The patterns in (13) also demonstrate yet another context where prosody plays a role in how the different types of morphology surface.

| CONCLUSION
The Muskogean languages exhibit an extraordinary constellation of typologically interesting features, as shown in this overview of the most uncommon morphology processes they display. Theoretical linguistics have drawn on analyses of Muskogean languages, and the descriptive documentation is detailed for most of the languages, potentially fostering novel insights about how linguistic theories operate.
The patterns here were meant to accomplish two things. First, by trying to give examples that isolated particular characteristics, a clearer exposition of the distinct typological categories was possible, although at the cost of less morphological complexity in individual words. Simpler, more cleaned up types of examples isolate these characteristics in such a way as to highlight them. However, each section also built on the morphological characteristics shown in earlier sections to show how these features combine and surface together in an intricate way.
With the volume of text collection and analyses both complete and ongoing, much more complex linguistic examples of Muskogean verbs exist, particularly in a narrative context. Current documentation and revitalization collaborations provide opportunities to augment what we know of these languages to enrich typology and linguistic theory and to mobilize those findings back into communities to support their efforts to maintain and revitalize their heritage languages.