Menominee Preverbs as Functional Categories

REBECCA SHIELDS University of Wisconsin, Madison

This paper is one piece of a larger work exploring the proposal that Menominee preverbs occupy head positions in a functional hierarchy like that proposed in Cinque (1999).' The behavior of several preverbs is con­ sidered in light of cross-linguistic properties of similar morphemes, in order to establish their categorial identity among the set of functional heads presumably supplied by Universal Grammar (UG).

PREVERBS AND Menominee has a number of morphemes, called preverbs in Algonquian- ist literature, which encode various meanings including modal, temporal, spatial, and aspectual. A given clause can have several such morphemes, which are sometimes treated as part of the morphological because they occur between a preverbal agreement morpheme and the verb stem (e.g., Bloomfield 1962).2 The position of preverbs in the Menominee ver­ bal complex is shown in (1) below (numerous details about agreement omitted).3 Arguments and other expressions appear both to the left and right of this complex. (2) provides an example of a preverb in what I will call the preverb space.

(1) AGR-PREVERBrPREVERB2-. . .-PREVERBn-V-AGR V J preverb space

1. I am sincerely grateful to elders Marie Floring, Lillian Nelson, Bill Penass, Lavina Shawano, Sarah Skubitz, and Tillie Zhuckkahosee for sharing their time and knowledge of Menominee. Thanks also to the Language and Culture Commission of the Menominee Nation for facilitating research and language preservation, and to the participants of the 36th Algonquian Conference and an anonymous reviewer for very helpful comments. I am much indebted to Monica Macaulay and Marianne Milligan for their extensive support and encouragement. 2. However, there is evidence that preverbs do not form a phonological word with the verb stem in at least some : see for example Bloomfield (1962) for Menominee, Leavitt (1985) for Passamaquoddy-Malecite, and Russell (1999) for Cree.

Papers of the 36th Algonquian Conference, ed. H.C. Wolfart (Winnipeg: University of Manitoba, 2005), pp. 383-406. 384 REBECCA SHIELDS

(2) ne-kes-anohkim. nae- kes- anohkl -m 1- PAST- work.AI -1/2 'I was working.' (11.101) As is well known, preverbs obligatorily occupy this space. For exam­ ple, they cannot appear to the left of the preverbal personal agreement morpheme:

(3) "kes-netdnohkim. kes- nae- anohkl -m PAST 1- work.AI -1/2 (11.102) However, certain elements other than preverbs can also occur in the pre­ verb space: some adverbial expressions, and, according to Bloomfield (1962) and Guile (2001), and certain locative expressions. These other expressions have a much freer distribution than preverbs. For example, adverbials may appear in the preverb space, to the left of the

3. Abbreviations used: 1/2 - firsto r second (local) person agreement, AGR - agree­ ment, AI - animate intransitive, AN - animate, AOR - aorist, CONJ - conjunct order (i.e. embedded clause morphology), CPL - completive aspect, CTR - contrast marker, DIST - distantive aspect ('go off and do X'), DNA - dependent (i.e. obligatorily possessed) animate, DNI - dependent noun inanimate, EPIS - epistemic, HAB - habitual, IC - initial change (morphological ablaut morpheme), II - inanimate intransitive, IMP - imperative, INCEP - inceptive aspect, INF - , INV - inverse theme sign, NEG - negation, NONREF - nonreferential subject, PL - plural, PRED - predicative, PROX - proximate, QUOT - quotative, REP - repetitive aspect ('again, back'), SG - singular, SIMUL - simultaneous, TA - transitive animate, TH - theme sign, Tl - transitive inanimate. Menominee parses and glosses were produced with the aid of a Toolbox parsing database created by Monica Macaulay and Marianne Milligan, based primarily on the morphologi­ cal analysis of Bloomfield (1962). Glosses of some preverbs have been changed to fitth e analysis proposed in this paper. 4. Examples in this paper are given in the modem orthography officially adopted by the Tribe of the Menominee Nation. Note that this is not a phonetic transcription. Data from previously published sources have been modified to conform to the modem orthography. Examples are cited with the source in parentheses following the translation. Data from published sources have the citation along with a page or line number. All other data are taken from elicitations conducted by the University of Wisconsin Menominee Language Project (UWMLP). Unpublished data from elicitation sessions I conducted are marked with elicitation number plus sentence number, e.g., (4.02) - the second elicited sentence from the fourth session. Unpublished data elicited by other linguists are cited by title of text and line or page number (if available). 5. [t] is regularly epenthesized when an agreement morpheme is prefixed to a vowel-ini­ tial stem (Bloomfield 1962 section 4.70), e.g., netdnohkim 'I am working' (UWMLP). MENOMINEE PREVERBS AS FUNCTIONAL CATEGORIES 385

personal agreement morpheme, to the left of a preverbal NP argument, or postverbally:

(4) nekis-ahpdenenew-anohkim. nae- kes- ahpaenenew anohkl -m 1- PAST- always work.AI -1/2 T was always working.' (10.81) (5) ahpaenenew nekis-anohkim. ahpaenenew nae- kes- anohkl -m always 1- PAST- work.AI -1/2 T was always working.' (10.86) (6) ahpaenenew, nohnaeq kes-anohkiw. ahpaenenew nae- ohn kes- anohkl -w always 1- father. DN A PAST- work.AI -3 'My father was always working.' (1.08) (7) nohnaeq kis-anohkiw ahpaenenew. nae-ohn kes- anohkl -w ahpaenenew 1- father.DNA PAST- work.AI -3 always 'My father was always working.' (1.10)

In this paper I use syntactic distribution criteria to distinguish two distinct syntactic categories: PREVERBS and ADVERBS. The syntactic category PREVERB contains only words which obligatorily appear in the preverb space: between the preverbal agreement morpheme and the verb stem. Examples include kes ',' aweh 'distantive aspect (go off and do X),' and cew 'epistemic modality.' The syntactic category consists of words which may appear in the preverb space, but may also precede or follow the verbal complex. Some examples are ahpaenenew 'always,' ydhpec 'absolutely,' and kanapac 'maybe'.7 This paper is limited to preverbs as defined in this way. Preverbs generally appear in a fixedpositio n and relative order in the clause, they constitute a closed class, and their meanings are those typi­ cally identified as "functional" or grammatical as opposed to lexical. This paper begins an exploration of the hypothesis that the syntactic categories

6. An adverbial may not appear directly to the right of the personal agreement mor­ pheme, however: "netdhpaenenew-anohkim (UWMLP). This gap in distribution may be explained if the agreement morpheme is an affix that selects only preverb or verb hosts for affixation. For varying explanations for the syntactic distributions of preverbs and adverbs, see Dahlstrom (2000) and Shields (2005). 386 REBECCA SHIELDS of Menominee preverbs are drawn from the set of universal functional categories. I attempt to determine which functional category each preverb corresponds to, to the extent that sufficient cross-linguistic description of the categories exists.

SEMANTICS OF SELECTED PREVERBS An extensive hierarchy of functional categories is argued to be universal in Cinque (1999). Cinque claims that UG determines both the inventory of functional categories available to a language and the order of the pro­ jection of these categories in the clause. Before we can ask whether Menominee preverbs corroborate this hierarchy, we first need to establish exactly what the preverbs mean: this is the goal of the present paper. I take as an initial hypothesis the semantics of Menominee preverbs presented in Bloomfield (1962), which is largely extendable to contempo­ rary Menominee. This paper presents argumentation for a subset of pre­ where this semantics is not adequate for contemporary data. To simplify the task, the discussion is limited almost exclusively to the inter­ pretation of preverbs in matrix clauses. The case of embedded clauses is more complex, because interaction with preverbs in dominating clauses must be taken into account. In trying to identify the semantics of preverbs I draw on observations about the behavior of functional categories cross-linguistically. An impor­ tant assumption underlies this method: that while variation in lexical mor-

7. Note that this terminology and architecture of syntactic categories differs from that in Bloomfield (1962). In the present paper the term preverb is used in a narrower sense. Bloomfield assumes a single syntactic category "particle," which he further divides into various subcategories including "preverb particle" and "independent particle." Any parti­ cle is an independent particle when it occurs outside the preverb space. A particle occur­ ring inside the preverb space may be either a preverb particle or an "incorporated" independent particle. Bloomfield further subcategorizes preverb particles into preverbs of the first class and preverbs of the second class, primarily by morphophonological criteria. He states that most preverbs of the first class do not have corresponding independent parti­ cles, while most preverbs of the second class do. 8. Note that the goal of this paper is synchronic description of the language in its current state. The analysis in Bloomfield (1962) was based on fieldwork conducted in the 1920's; the language has certainly changed since this time. I make no general claim about the cor­ rectness of Bloomfield's or other analyses with respect to past states of the language. I do however draw on examples from the past where possible, and point out certain continu­ ities and discrepancies that I am aware of between contemporary and past data. MENOMINEE PREVERBS AS FUNCTIONAL CATEGORIES 387 phemes is limited only by the human experience that makes such morphemes useful, yielding an essentially infinite number of possible lex­ ical denotations, the inventory of possible functional morphemes is defined in UG, and is therefore small and highly constrained. It is thus fruitful to compare unrelated languages and to consider the question of whether two functional items are "the same thing." If this assumption turns out to be false, an entirely different approach is required.

Verbs without preverb Let us first consider how verbs are interpreted in the absence of aspectual or temporal preverbs. Such verbs can receive a variety of interpretations, including habitual/generic, present, and past. A future interpretation does not seem to be possible. The same range of interpretations can be found in Bloomfield (1928, 1962). The following sentences show examples of each of the possible interpretations.

HABITUAL/GENERIC (8) ahpaenenew mekdhtowak. ahpaennew mekahti -w -ak always fight.with.each.other.AI -3 -PROX.PL ' They are always fighting.'(8.15 ) (9) nohnaeq mihekanaehkaew. nae- ohn mlhekanaehkae -w 1- father work.on.road.AI -3 'My father does road work (for a living).' (UWMLP)

PRESENT (10) mekdhtowak. mekahti -w -ak fight.with.each.other.AI -3 -PROX.PL "They're fighting (right now).' (8.21) (11) Q: If you look out the window, and you see someone picking berries, what would you say to describe the situation? A: nawendew. nawenae -w gather.berries.AI -3 'He's picking berries.' (9.47) 388 REBECCA SHIELDS

PAST (12) Mesden netdekwah, kan 's newahkondet. Mesaen nae- aeN -Eko -w -ah kan Michael 1- say.so.to.TA -TA.INV -3 -3.SG NEG as newahkonae -t AOR be.very.hungry.AI -3.CON J 'Michael told me that he's not hungry.' (2.05) (13) nis kiqsehsan meydhkiwaewen. nisw klqsehs-an maeyahkiw -ae -w -en two girl -OBV meet.TA -TA.DIR -3 -QUOT 'He met two girls.' (Bloomfield 1920-1949, The Bead Man, 1. 7)

The past tense preverb kes Bloomfield (1962) gives the meaning of kes as "completed, past." Cook (2003) concludes based on textual data that this morpheme denotes com­ pletive aspect. In fact, as I show here kes does not have any of the proper­ ties of completive or perfective aspect. Completive or perfective aspect denotes a delimited event (the viewpoint includes the endpoints of the event; cf. Smith 1997). Past tense, by contrast, shifts interpretation to a time prior to some other time (typically but not always speech time in a matrix clause). There are three respects in which kes behaves like a past tense, and not like a completive aspect. First, because completive aspect denotes a delimited event, it is ungrammatical to continue a sentence containing a completive verb with an assertion that the event still continues. But if an activity or has a past tense, there is no requirement that the endpoint of the event be in the past, and an assertion of continuation is possible. Compare for example the following sentences: in Russian the continuation is not possi­ ble when the initial clause has a perfective verb, as in (14b). The continu­ ation is possible when the initial clause contains a past tense marker (14a), similarly in the English translation. Menominee kes, like the Rus­ sian and English past tense, allows the continuation (15).11

9. Bloomfield (1962) gives the underlying form of the inverse morpheme as /-Ek/. Monica Macaulay and Marianne Milligan (p.c.) give it as /-Eko/, with the [o] deleting when followed by third person l-v/l. 10. See James (1982) for arguments that the Cree cognate of kes is likewise a past tense rather than a perfective aspect morpheme. MENOMINEE PREVERBS AS FUNCTIONAL CATEGORIES 389

(14a) Russian Oni side-l -i v parkevchera, (i do six por tarn sid-jat). they sit -PAST-PL in park yesterday, and to these times there sit-3PL 'They sat in the park yesterday, (and they're still sitting there).' (14b) Oni po- side-l -i v parkevchera, (*i do six por tarn sid-jat). they PFV-sit -PAST-PL in park yesterday, and to these times there sit-3PL 'They sat for a while in the park yesterday, (*and they're still sitting there).' (15) kes-apdetdew, (meciw new apdetdew). kes- apaetae -w meciw new apaetae -w PAST- be.warm.II-3 still be.warm.II-3 'It was warm, and it's still warm.' (10.52)

Second, cross-linguistically completive and perfective aspect mark­ ers are incompatible with a stative interpretation (Smith 1997). This is because the stative interpretation requires a non-delimited event, whereas a completive verb denotes a delimited event. If a language with a comple­ tive/perfective aspect marker allows it to occur on stative verbs, the result is a change-of-state event verb. The following Russian and Chinese examples illustrate this phenomenon; compare the addition of the past tense morpheme to a stative verb in Russian and in English, which simply yields a past stative interpretation.

(16) Russian: completive + stative = change of state (16a) Onzna -et. he know-3sg 'He knows.' (16b) On u- zna -et. he CPL-know-3sg 'He will find out. (NOT: He has known / he knew.)'

II. I have not yet come across an example of the type 'it was X, and it's still X' from older sources. If kes historically denoted a past tense, the first clause would have been pos­ sible either with or without kes (since bare verbs can also yield past readings). If kes his­ torically denoted a completive aspect and has only recently been reanalyzed as a past, the version with kes (like 15) would have been ungrammatical. This type of test is difficult to perform with only positive textual evidence, since negative evidence is required to falsify one of the hypotheses. 390 REBECCA SHIELDS

(17) Chinese (Comrie 1976): perfective + stative = change of state

(17a) ta gao. 3sg tall. 'He/she is tall.' (17b) ta gao-le. 3sg tall- PFV 'He/she became tall. (NOT: He/she was tall.)'

(18) Russian: past + stative = past state

(18a) Onzna -et. he know-3sg 'He knows'

(18b) Onzna -I. he know-PAST 'He knew.' (19) English: past + stative = past state (19a) He knows / He is tall. (19b) He knew / He was tall. Menominee kes patterns with the past tense morphemes in this respect; when added to a stative verb, the result is a past state, not a change of state.

(20) Menominee: kes + stative = past state kes-apdetdew nek. kes- apaetae -w nae-ek PAST- be.warm.II -3 1- house.DNI 'It was warm in my house, (not: It became warm or it stopped being warm.)' (9.21)

12. This seems to be true of the kes of the 1920s as well: in Bloomfield (1928) stative verbs with kes are consistently translated as past states, not changes of states. For example: anom esekenan kesiapaehkesekandhtekan kes-tandm nekiqs. anom aeseken -an kaesiapaehkesekanahtek -an this/these.OBV that.sort -OBV stovepipe -OBV kes- tan -am nae-kiqs PAST- have.Tll -Tll.TH 1-son.DNA 'My son had this kind of stovepipe.' (Bloomfield 1928, A Bad Neighbor, 1.1) MENOMINEE PREVERBS AS FUNCTIONAL CATEGORIES 391

(21) taq wdehcetaw new apdehnihsaeh pepehsaeh nekes-dwem. taq waehcetaw new apaehnihs paepehs nae- kes- awe -m well really CTRboy baby 1- PAST- be.AI-1/2 'Well, I was really a little boy. (not: I became a little boy.)' (Bill's Life Story, 1. 2)

Finally, completive and perfective forms are typically used to refer to both past and future delimited events (Smith 1997). The Russian sen­ tence (16b), for example, shows a completive aspect marker on a non-past verb that yields a future reading. In Menominee, however, kes is incom­ patible with a future reading in matrix clauses13 - it always yields a past interpretation (22). In embedded clauses, if a dominating clause shifts interpretation to a future time, kes can denote a time anterior to this time, which may in fact be in the future of speech time (23), but this is exactly the behavior we expect from a past tense morpheme (Enc 1987).

(22) kes-onikdewak. kes- onikoe -w -ak PAST- build.house.AI -3 -PROX.PL "They built a house. (NOT: they will build a house.)' (9.10) (23) kes-poset, [...] enehpeh aw-pahkdesamah. kes- pose -t PAST- board.vessel.AI -3.CONJ enehpeh aw- pahkEs -am -ahk then IRR- cut.off.TI -TI.TH-1PL.CONJ 'When he has ridden in [boarded] the basket, [...] then we'll cut it off.'15 (Bloomfield 1928, The Red Swan, 11. 304-307) All the evidence therefore points to an analysis of kes as past tense, not as completive or perfective aspect. A past interpretation is possible without kes as well, on verbs that lack a preverb (cf. examples 12, 13 above). Why then would kes be needed at all, and what determines its occurrence? It appears to me that kes is used when the context alone is not

13. I have not encountered a kes example from the 1920s with a future interpretation either, although again negative evidence is required here. 14. When it occurs in the same clause with a future shifting morpheme, the interpretation involves both a shift to the future and a shift to the past; see the discussion of kataew, below. 15. Note that the English translation of this sentence uses a present perfect. Menominee, like Russian, does not have a perfect, and uses a simple past in these cases. 392 REBECCA SHIELDS sufficient to provide a shift to the past. When presented with bare verbs out of context, speakers I consulted sometimes rejected a past interpreta­ tion. These same speakers often produce bare verbs with past interpreta­ tions in context, however. This can be understood in the following way: the reference time16 of the matrix clause is by default identified with speech time, but may be identified with some contextually salient past time instead if such a time is available in the discourse. Therefore there are two ways to achieve a past interpretation: identify the reference time with a past time available in the discourse, or identify the reference time with speech time, and use kes. Only the second option is available if no contextually salient past time is present in the discourse. Note that English also allows a past interpretation without the past tense in the construction known as "historical present:"

(24) So I walk into this bar the other day, and this guy comes up to me and says ... This construction in English is limited not only by discourse factors, but by register and narrative genre. It is nevertheless a grammatical way of i *T referring to past times in pragmatically constrained contexts. On my proposal the difference between English and Menominee lies in variation of these contextual constraints, not in the denotation of past tense itself. Bare verb past interpretations are more widespread in Menominee because Menominee allows the matrix reference time to be identified with a past time in more contexts and speech genres than does English. Finally, the lack of a future reading with kes is predicted given the asymmetry in bare verb interpretations just noted: the reference time of a matrix clause cannot, for some reason, be a future time under any circum­ stances, and therefore a past tense in a matrix clause cannot shift to the past of a future time.

16. i.e., the time in C in the theory of Enc (1987). 17. It remains unclear to me why English allows a bare verb in this context, however, rather than the progressive form, which is generally what English requires for identifica­ tion with the reference time. MENOMINEE PREVERBS AS FUNCTIONAL CATEGORIES 393

The epistemic modal cew

Bloomfield (1962) describes the meaning of this preverb as follows: '"might, ought to, must, is said to', means that the speaker does not cer­ tify actual occurrence, but views it as probable." In fact, the meaning of cew is more restricted than this characterization suggests; it does not have the full range of meanings of English might, must, and ought to, and it is used in only a subset of situations where "the speaker does not certify actual occurrence." Furthermore, cew may indicate that the speaker views the assertion as either probable or necessary. Cew always has an epistemic modality, i.e. it indicates that the asser­ tion contained in the utterance was deduced as true based on some domain of knowledge of the world. It does not have the deontic interpretation of must or ought to, i.e. universal quantification over worlds that conform to a requirement of some societal or religious code. In other words, cew cor­ responds to the must or may in There's a light on, so somebody must/may be home, but not in According to the law, you must / may park here. The use of cew was also rejected by speakers in non-epistemic reported speech contexts, even though here too "actual occurrence" cannot be certified. Cew therefore does not simply indicate that the speaker views the occur­ rence as probable or possible, but gives information about the source of this view - deduction according to the laws of some domain of knowl­ edge. Interestingly, it can have either universal or existential force. In other words, it can indicate that the speaker views the truth of the proposi­ tion as necessary (like English must or ought to) or possible (like English might or may). This can be seen in the following examples:

(25) ce-kes-mdciw. cew- kes- macya -w EPIS- PAST- go.away.AI-3 'He must have left; perhaps he left; he might have left.' (8.01) (26) Q: Suppose Ron is in the room. You leave the room to go make a phone call. When you get back, Ron is not there. How would you say "He must have left?" A: ce-kes-mdciw. cew- kes- macya -w EPIS- PAST- go.away.Al -3 'He must have left' (9.01) 394 REBECCA SHIELDS

(27) Q: Same situation. How would you say "He may have gone home. Or maybe he just went outside."? A: (kanapac) ce-kes-esiw akuaceh, kdeh men new kotaeqnas ce-kes-esiw. kanapac cew- kes- aesya -w akwaceh maybe EPIS- PAST- go.thither.AI -3 outside kaehmennew kotaeqnas cew- kes- aesya -w or somewhere EPIS- PAST- go.thither.AI -3 'Maybe he went outside, or maybe he went somewhere else.' (9.04)

Speakers identified the sentence in (25) as ambiguous: it can mean either that the person in question must have left or might have left. The situations in (26) and (27) are designed to force a universal and an exis­ tential interpretation, respectively. Cew is indeed possible in both situa­ tions.18 For the time being, therefore, I will assume that there are two homophonous preverbs cew in Menominee: one universal epistemic modal and one existential.

The simultaneous interpretation preverb maek

Bloomfield (1962) says of maek that it means "action under way." This suggests that it may be a progressive morpheme, like English -ing. Indeed, when mdek occurs on a verb, it is always translated into English with a progressive:

(28) eneq's mdek-anohkit enoh nekot enew -q as maek-anohkl -t enoh naekotw that.INAN.PRED -INDIC AOR SIMUL-work.AI-3.CONJ that.AN one 'That's where that one was working.' (The Farmer and the Visitor, 1. 8) (29) pdekwasekan maek-micwah. paekwasekan maek- mec -w -ah cooked.squash SIMUL- eat.TI3 -3 -3.SG 'He/she is eating the cooked squash.' (UWMLP)

18. To my knowledge, this situation is unusual: many languages appear to be like English in having one lexical item (like must) which quantifies universally but may have varying modalities (epistemic, deontic, etc.), and another (like may) which quantifies exis- tentially. 19. Alternatively, there may be one epistemic modal in the lexicon with underspecified quantificational force. It is not clear in this case where the specification of quantificational force would come from, if not from the lexicon. MENOMINEE PREVERBS AS FUNCTIONAL CATEGORIES 395

However, the distribution of maek does not match that of the pro­ gressive. Maek is often (but not always) used when the time of interpreta­ tion includes the present, or the time of interpretation of another clause; for this reason I call maek a "simultaneous interpretation" morpheme. But as will be seen below, the conditions of its occurrence have not yet been fully established. I am not aware of a universal functional category that has the same properties as Menominee maek. Let us begin by considering how maek is different from a progres­ sive. According to Smith (1997), progressive is a type of imperfective viewpoint aspect which puts focus on the internal stages of a non-stative event. Therefore, a progressive morpheme may not be used with a stative verb (see also Comrie 1976). Thus English she is thinking but "she is knowing. But Menominee maek occurs on all types of verbs, including 20 statives: (30) mdek-tahkiw. maek- tahki -w SIMUL- be.cool.II -3 'It is cool.' (9.49) (31) nahdw eneq naekotdes ndqs taeh 's mdek-apdehnihseweyan ... nahaw enew -q naekotaes naqs taeh well.then then.PRED -INDIC at.one.time may.it.be and as maek- apaehnlhsewe -an AOR SIMUL- be.a.boy.AI -1/2.SG.CONJ 'Well, then, once when I was a boy...' (Bill's Life Story, 1. 1) Maek is therefore by definition not a progressive. Turkish has an imper­ fective morpheme which focuses internal stages of events or states, known as durative (Murvet Enc, p.c). Menominee maek is like the durative in its ability to apply to both delimited and non-delimited event types. However, there are differences from the durative as well. It has been observed that languages tend to have either a progressive morpheme (which can be added to non-stative verbs to yield a present reading), or a morphologically unmarked imperfective (which yields a present reading for all verb types; cf. Smith 1997). English is a language

20. maek occurs on stative verbs in texts from the 1920s as well: ... 's maek-kaeqc- wesakaeseyen. '... while I was ill.' (Bloomfield 1928, The Childhood of Red Cloud Woman, 1. 10). 396 REBECCA SHIELDS

of the progressive type, while French and Russian do not have a progres­ sive and are of the imperfective type. In other words, if a language has a progressive, it is required for the present reading, and only languages which lack a progressive allow a present reading on an unmarked non- stative verb. The Turkish durative is like the progressive in this respect. As mentioned above, aspectually unmarked verbs can in fact get a present reading in Menominee. Compare the habitual interpretation of the English She picks berries with the Menominee examples (10) and (11) above. In light of this it seems incorrect to identify mdik as either a progressive or durative, given that functional/inflectional categories are not generally known to have optional manifestations. A third property of the progressive and durative is that they are required, in languages that have them, for the overlapping event reading. This can be viewed as essentially the same property as the one just dis­ cussed: if a language has a progressive or durative morpheme, its pres­ ence is required for the interpretation time of the clause to be simultaneous with the reference time. Compare for example the Russian sentence below, where an overlapping event reading is possible with an aspectually bare verb, and the English equivalent, where the overlapping reading is only possible with the progressive -ing.

(32) Vanjae -I, kogdajavo- she-l. Vanja eat-PAST, when I in.PFV-go-PAST 'Vanja was eating when I walked in. (overlapping reading)' (33a) John ate when I walked in. (sequential reading only: first I walked in, then he ate) (33b) John was eating when I walked in. (overlapping reading only: he was in the process of eating when I walked in)

Menominee once again patterns with the non-progressive languages here; as in Russian, no aspectual marking on the verb is required to obtain the overlapping reading:

(34) tahndenohpepon kis-esiqtawakuaq, kes-(mdek-)ahpdenenew-nawendewak. tahnaenoh paeponw kes- aesiqtaw -ak -waw every.time.LOC year PAST- visit.TA -1.SO3.C0NJ-3.PL.C0NJ kes- (maek-) ahpaenenew nawenae -w -ak PAST- (SIMUL-) always gather.berries.AI -3 -PROX.PL MENOMINEE PREVERBS AS FUNCTIONAL CATEGORIES 397

'Whenever I went to visit them, they were always picking berries ' (9.22-23)

(35) nekotdes new mamdceqtawak 's nimihaetituaq, enewen desiat sekemaeh. naekotaes new mamaceqtaw-ak as nimihaeti -t -waw one.time CTR Indian-PROX.PL AOR dance.together.AI -3.CONJ-3.PL enew -en ay- aesya -t sekemaehs there.PRED -QUOT IC- go.thither.AI -3.CONJ diver.duck 'One time while the Indians were dancing, diver duck went there.' (Bloomfield 1920-1949, The Bead Man, 11. 5-6)

Sentence (34) was judged to be grammatical and to have the same mean­ ing with or without maek. The context of the story from which (35) was taken makes it clear that an overlapping reading is intended: diver duck's arrival occurs in the midst of the Indians' dancing, yet maek is not present. Bloomfield's texts (1928) contain other, similar examples as well. The fact that maek is not required for the present reading or the over­ lapping event reading indicates that it is not a progressive or durative. Nevertheless, maek is often used in sentences which have these readings, and when it is used the result is always an undelimited reading with focus on internal stages. I suggest therefore that maek is a type of imperfective morpheme indicating that the interpretation time of the clause includes some other relevant time (either speech time or the time of an embedded or dominating clause). It thus disambiguates between the habitual and present readings that would be possible for a bare matrix verb, and between the sequential and overlapping readings that would be possible for a bare embedded verb. A final observation concerning this morpheme is that, for one elic­ ited sentence, speakers found maek to be required:

(36) Mesden sewas mdek-apew 'spihtikiyan. Mesaen sewas maek- ape -w as pehtikae -an Michael already SIMUL-sit.AI-3 AOR go.into.a.place.AI-l/2.SG.CONJ 'Michael was already sitting there when I came in.' (9.45) (37) "Mesden sewas apiw 's pihtikian. Mesaen sewas ape -w as Michael already sit.AI -3 AOR pehtikae -an go.into.a.place.AI -1/2.SG.CONJ ""Michael already sat there when I came in.' (9.46) 398 REBECCA SHIELDS

This is true even though the verb apew, like other verbs in Menominee, can receive a present interpretation without maek: (38) ndh! apiw enoh kahkik. nah! ape -w enoh kahkik look! sit.AI-3.SG that.AN chipmunk 'Look! The chipmunk is sitting there.' (UWMLP) It is not clear to me why maek should be obligatory in (36) but optional in sentences like (34) and (35). It is possible that the presence of sewas 'already' has something to do with it, but further research is clearly needed to understand the distribution of this morpheme.

The irrealis preverb aw and the future preverb kataew In this section I will consider two morphemes together, because it is use­ ful to compare and contrast them. These are morphemes that have both been claimed to entail a future interpretation. Bloomfield (1962) says of aw that it means '"will, shall, going to' denotes a future event, without regard to an inner impulse." His definition of kataew is '"intend to, want to, going to' [...] denotes that the actor [...] has an inner tendency to the action which he will presumably carry out." Both morphemes indeed often shift interpretation to the future, and kataew is often used when an actor's desire or "inner tendency" is involved in determining the future:

(39) aw-mdciw. aw- macya -w IRR- go.away.AI -3 'He will leave.' (8.9) (40) katdew-miandceqtawak. kataew- myanaceqta -w -ak going.to- play.AI -3 -PROX.PL 'They're gonna play.' (8.30) (41) [...] katdew-ndeweyan, kena-ndewem. kataew- naew -e -an going.to- see.TA -TA.l.ACC -1/2.SG.CONJ kae- aw- naew -e -m 2- IRR- see.TA -TA.l.ACC -1/2 '[if ever] you want to see me, you will see me.' (Bloomfield 1928, Boy Blessed by Mosquitoes, 1. 105) MENOMINEE PREVERBS AS FUNCTIONAL CATEGORIES 399

But two refinements to these observations are necessary. First, aw does not always shift to the future, as Bloomfield himself notes. When combined with past tense, aw yields "uncertain past events," as Bloom­ field puts it:

(42) netowak 's aw-kes-pitok kdekoh. netowak as I.wonder.PRED AOR aw- kes- pit -6 -k kaekoh IRR- PAST- bring.Tl2-TI2.TH -3.CONJ something T wonder if she brought me something.' (9.32) (43) nekis-nohtdn Mdriih 's aw-kes-anihewaet. nae-kes- noht -a -n manih as 1- PAST- hear.REP -Tll.TH -LCL.TI Marie AOR aw- kes- anehewae -t IRR- PAST- win.from.people.AI -3.CONJ ' I heard that Marie might have won.' (9.37)

If aw were the equivalent of English will, we would expect (42) to mean something like T wonder if she will have brought me something' - a shift to the future followed by a shift to the past. However, there is no shift to the future in (42-43) at all. Rather, native speakers explain that these sen­ tences differ from their counterparts without aw only in that with aw "you're making it doubtful." This morpheme is therefore probably best described as an irrealis marker, as in Cook (2003). Unfortunately the term "irrealis" is used in a variety of ways, and I am not aware of a rigorous semantic definition of it or of specific behaviors that distinguish it from evidentials and other types of modals. The intuitive idea is that an irrealis morpheme indicates a low degree of certainty (or a lack of commitment to any degree of certainty) on the part of the speaker with respect to the truth of the assertion (in any world). Statements about future events, since their truth is in principle unknowable, therefore often include an irrealis marker. Irrealis markers do not entail a shift to the future, however. The important conclusion for the current discussion is that it is not necessary to posit lexical ambiguity for aw: one 'future' and one 'doubt.' Aw is in fact neither a Tense morpheme nor a future-shifting modal like English will. Aw can be viewed as making a consistent "irrealis" contribution to the semantics of sentences like (39) and (42).21 400 REBECCA SHIELDS

The use of aw together with kes appears to have limited distribution. All textual and elicited examples I have found with this combination are in clauses embedded under intensional verbs, such as 'wonder' and 'hear' in the examples above. Furthermore, speakers rejected the combination in the absence of the embedding verb, as in the following sentence:

(44) kdekoh (*aw-)kes-pitoq? kaekoh aw- kes- pit -6 -q something IRR- PAST- bring.TI2 -TI2.TH -INT 'Did she bring me something?' (8.33-34) Kataew, unlike aw, does always shift to the future. This can be seen in the following examples, where it co-occurs with past tense:

(45) kis-katdew-ndtamowdew. kes- kataew- natamow -ae -w PAST- going.to- help.TA -TA.DIR -3 'He had meant to help him.' (Bloomfield 1962, p. 217) (46) kespen kat kis-katdew-mamatdweqtah. kespen kat kes- kafw- mamataweqta -aehk if NEG PAST-going.to-perform.marvelous.action.AI-AI.NONREF.CONJ 'If there had not been a theatrical performance planned.' (Bloomfield 1962, p. 217) Both these examples show a shift to the past, followed by a shift to the future, just as in English He was going to .... This is unlike the case of aw just dis­ cussed, which does not shift to the future when combined with past tense. Unlike aw, kataew is therefore a consistently future-shifting morpheme. However, kataew does not always involve intention, desire, or the "inner tendency" of an actor. First, as Bloomfield noted, it can be used to talk about things like the weather, which have no volitional component:

21. Costa (2002) arrives at a similar conclusion for the Shawnee preverb -ih-. 22. Only older textual examples are given in support of this claim. Native speakers I con­ sulted rejected sentences with combinations of kataew and kes in either order. I do not currently have an explanation for why these morphemes should be incompatible in con­ temporary Menominee. But although the argument that kataew always shifts to the future relies crucially on older examples, the claim is not contradicted by contemporary elicited examples. MENOMINEE PREVERBS AS FUNCTIONAL CATEGORIES 401

(47) kanapac katdew-tahkiw wapah. kanapac kataew- tahk! -w wapan -k maybe going.to-be.cool.II-3 be.day.II -3.CONJ 'Maybe it's going to be cool tomorrow.' (10.3) (48) katdew-kemewan. kataew- kemewan -w going.to- rain.II -3 'It's going to rain.' (Bloomfield 1962, p. 217)

It can also be used with other impersonal verbs that take non-referen­ tial subjects, such as 'happen' in the following example.

(49) eneq saqyaeh kdekoh maceq kdekoh 's katdew-esekemakah. enew-q saqyaeh kaekoh maceq kaekoh then.PRED-INDIC now something bad something as kataew- aesekemakat -k AOR going.to- happen.II -3.CON J 'Something bad is going to happen.' (The Farmer and the Visitor, 1. 26)

Clearly, kataew does not require any particular properties on the part of an actor, since it can be used with verbs which have no actor argument. These apparently different meanings of kataew ('intend to, want to' and 'going to') are in fact quite similar to the range of meanings for English going to. Note that going to, in contrast to will, often has something of a volitional component, and can be used when declaring intentions that are at present certain. In a context where someone is communicating their previous decision to attend a party, for example, I'm going to go is much more natural than I'll go. I'll go, on the other hand, would be more natural if the person in question was not previously planning to go, but has just been talked into going by a friend, or has just at this moment decided to attend. Going to, like kataew, can also be used with impersonal predicates where no intention is involved, as in it's going to rain. In contexts where we have some evidence that it is in fact going to rain (for example a weather report predicting rain, or dark storm clouds), it's going to rain tomorrow is much more natural than it will rain tomorrow. In Menomi­ nee, kataew is preferred in similar contexts:

(50) Q: Suppose you hear the weather report on TV. It says it will be cool tomorrow. You report this news to your friend. 402 REBECCA SHIELDS

A- Eneq detamemakah eneh mahkcth 's kataew-tahkik wdpah. enew -q ay- aet -am -makat'-k that.INAN.PRED -INDIC IC- say.X.so.TIl -TILTH -INAN-3.CONJ

eneh mahkahkw as kataew- tahkl -k that.INAN box AOR going.to- be.cool.II -3.CONJ

wapan -k be.day.II -3.CON J "That's what the box said, that it's gonna be cool tomorrow.' (10.1) Speakers judged the equivalent of (50) with aw instead of kataew to be ungrammatical. The following minimal pair also brings out this differ­ ence between aw and kataew:

(51) Wdpah naeqniw tepahekan nenaw-awih-monahekaem wapan -k naeqniw taepahekan be.day.II -3.CON J three hour nae- aw- aweh- monahekae -m 1- IRR- DIST- work.in.garden.AI-1/2 'Tomorrow at 3:001 may be going to work in the garden.' (10.13) (52) Wdpah naeqniw tepahekan nekataw-aweh-monahekdem. wapan -k naeqniw taepahekan be.day.II -3.CONJ three hour nae- kataew- aweh- monahekae -m 1- going.to- DIST- work.in.garden.AI -1/2 'Tomorrow at 3:00 I'm going to work in the garden (definitely).' (10.16)

I therefore propose that kataew is a future-shifting morpheme that requires that the future event has a sufficient degree of certainty, similar to English going to. The judgment of certainty may be based on various types of information, including but not limited to an actor's known inten­ tions or desires. Much more work needs to be done to understand the semantics of aw and kataew (as well as will and going to). For now, we have established that kataew is a future-shifting modal like going to, while aw is an irrealis marker which is compatible with but does not necessitate a shift to the future.

23. nae- '1st person' + aw TRR' surface as nenaw, due to epenthesis of [n] which Bloomfield (1962) calls an "irregularity;" see his sections 4.85 and 13.70. MENOMINEE PREVERBS AS FUNCTIONAL CATEGORIES 403

Other preverbs

For the remaining preverbs presented here, I give a brief restatement - without further argument - of the characterization of their semantics according to Bloomfield (1962), along with some representative exam­ ples. For some of these, notably aweh, wdep, and yah, there is little cross- linguistic information: kew: HABITUAL ASPECT (a pattern that holds consistently over an interval; cf. Smith 1997)

(53) kew-yah-piwak. kew- yah- pya -w -ak HAB- back- come.AI -3 -PROX.PL "They keep coming back.' (8.47) (54) kew-mecehsowak. kew- mecehsi -w -ak HAB- eat.AI -3 -PROX.PL 'They eat (every time, e.g., after a funeral).' (8.71) (55) kew-mdek-micehsowak. kew- maek- mecehsi -w -ak HAB- engaged.in- eat.AI -3 -PROX.PL 'They would be eating (e.g., every time I went to see them).' (8.73)

(56) kew-awih-andmehdw. kew- aweh- anameha -w HAB- DIST- go.to.church.AI -3 'She goes to church regularly/often.' (8.103) anih: COMPLETIVE ASPECT (a delimited event with focus on the endpoint, 'finish')

(57) ariih-andmehdwak. anih- anameha -w -ak CPL- pray.AI -3 -PROX.PL 'They're finished praying.' (8.130) (58) nekes-anih-awih-anohkim. nae- kes- anih- aweh- anohkl -m 1- PAST- CPL- DIST- work.AI -1/2 'I finished going to work, stopped.' (8.148) 404 REBECCA SHIELDS

(59) aw-anih-andmehdwak. aw- anih- anameha -w -ak will- CPL- pray.AI -3 -PROX.PL 'They will be finished praying.' (8.133)

(60) taq, kayes-anih-ackit... taq ay- kes- anih- aki -t well IC-PAST- CPL- be.tired.AI-3.CONJ 'When his weariness had gone, ...' (Bloomfield 1928, Sweet Root 1. 77) aweh: DISTANTIVE ASPECT ('go off and do X')

(61) awih-nipowenon enes wihkikaen. aweh- nepowe -n -on enes wihkikaen DIST- stand.AI -2SG.IMP -MP there make.a.comer.II 'Go stand in that corner.' (8.93) (62) aweh-miandceqtawak. aweh- myanaceqta-w -ak DIST- play.AI -3 -PROX.PL "They're going out to play.' (8.35) wdep: INCEPTIVE ASPECT (a delimited event with focus on the entry point, 'start, begin')

(63) wdep-anohkiw. waep- anohkl -w INCEP- work.AI -3 ' He's starting to work.' (8.112)

(64) katdew-wdep-andmehdwak. kataew- waep- anameha -w -ak going.to-INCEP- pray.AI -3 -PROX.PL "They're going to start to pray.' (8.121) yah: REPETITIVE ASPECT (repeated or reverse traversal of a temporal or spatial path, 'again, back')

(65) yah-etah.

yah- aet -am -h REP- say.so.Tll.FO -TILTH -2.SG.IMP 'Say it again.' (8.44) (66) yah-micehsowak. yah- mecehsi -w -ak MENOMINEE PREVERBS AS FUNCTIONAL CATEGORIES 405

REP- eat.AI -3 -PROX.PL 'They're eating again.' (8.79) (67) ydh-mdciw. yah- macya -w REP- go.away.AI -3 'He went back.' (8.10) (68) ydh-aqtoh. yah- aqt -6 -h REP- place.Tl2 -TI2.TH -2.SG.IMP 'Put it back.' (8.45)

CONCLUSION

In this paper I have investigated the meaning and behavior of several Menominee preverbs in a range of environments. Where possible, I have identified the cross-linguistic functional category that each preverb may belong to. In some cases this was straightforward; in others there appears to be no known universal category that corresponds to the Menominee word, or there is insufficient description of a universal category to make any assignment possible. The following table gives possible correspon­ dences to categories in Cinque's (1999) functional hierarchy for the pre­ verbs discussed in this paper:24

PREVERB LABEL USED HERE CATEGORY FROM CINQUE (1999) kes past tense past tense cew epistemic modality epistemic modality maek simultaneous ? (durative aspect appears closest) aw irrealis mood irrealis mood, evidential mood? kataew 'going to' future future tense? volition? kew habitual aspect habitual aspect anih completive aspect completive aspect aweh distantive aspect distantive aspect wdep inceptive aspect inceptive aspect yah repetitive aspect repetitive aspect

24. Distantive and inceptive aspect are not assigned a position in the hierarchy by Cinque, due to insufficient evidence, but are included in his inventory of functional cate­ gories. 406 REBECCA SHIELDS

REFERENCES Bloomfield, Leonard. 1920-1949. Leonard Bloomfield Papers, mss, National Anthropo­ logical Archives, Smithsonian Institution, Washington. Bloomfield, Leonard. 1928. Menomini texts. Publications of the American Ethnological Society 12. New York. Bloomfield, Leonard. 1962. The Menomini language, ed. by Charles F. Hockett. New Haven: Yale University Press. Cinque, Guglielmo. 1999. Adverbs and functional heads. Oxford: Oxford University Press. Comrie, Bernard. 1976. Aspect: An introduction to the study of verbal aspect and related problems. Cambridge: Cambridge University Press. Cook, Clare. 2003. A semantic classification of Menominee preverbs. Papers of the 34th Algonquian Conference, ed. by H.C. Wolfart, pp. 35-56. Winnipeg: University of Manitoba. Costa, David J. 2002. Preverb usage in Shawnee-narratives. Papers of the 33rd Algonquian Conference, ed. by H.C. Wolfart, pp. 120-161. Winnipeg: University of Manitoba. Dahlstrom, Amy. 2000. Morphosyntactic mismatches in Algonquian: Affixal predicates and discontinuous verbs. Chicago Linguistics Society 36, pt. 2: The panels, pp. 63- 87. Enc, Miirvet. 1987. Anchoring conditions for tense. Linguistic Inquiry 18:633-657. Guile, Timothy. 2001. Sketch of Menominee grammar. An anthology of Menominee say­ ings, with translations, annotations and grammatical sketch, pp. 452-501. Miinchen- LINCOM Europa. James, Deborah. 1982. Past tense, imperfective aspect, and irreality in Cree. Papers of the 13th Algonquian Conference, ed. by William Cowan, pp. 143-160. Ottawa: Carleton University. Leavitt, Robert M. 1985. Passamaquoddy-Malecite preverbs. Papers of the 16th Algon­ quian Conference, ed. by William Cowan, pp. 73-90. Ottawa: Carleton University. Russell, Kevin. 1999. The "word" in two polysynthetic languages. Studies on the phono­ logical word, ed. by T. Alan Hall & Ursula Kleinhenz, pp. 203-221. Amsterdam: Benjamins. Shields, Rebecca. 2005. The functional hierarchy in Menominee: Preverbs and adverbs Paper read at the 41st meeting of the Chicago Linguistics Society. Smith Carlota. 1997. The parameter of aspect. Dordrecht: Kluwer. 1997. The parameter oj aspect. 2nd ed. Dordrecht: Kluwer.