143

PAST TENSE, IMPERFECTIVE ASPECT, AND IRREALITY IN

Deborah James Scarborough College, University of Toronto

In the Cree-Montagnais- language complex, and also in at least some other as well, there are two differ­ ent types of morphological device for indicating past tense. One of these is the use of a prefix, which in Cree-Montagnais-Naskapi takes the form /ki:/ or /ci:/.1 The other is the use of what are known in the literature as "preterit" suffixes. In the dialect of Moose Factory, Ontario, with which this paper will be primarily concerned, the pre­ terit suffix takes the form /htay/ when the verb is in the indepen­ dent indicative and when either the subject or the object of the verb is first or second person, and it take a form characterized basically by the element /pan/ in all other cases.2 It is of interest to ask how, if at all, these two devices for indicating past tense differ from each other in meaning and usage. In this paper, I will examine in detail one dialect, the Mosse Cree dialect, and I will show that there are two major types of difference between the /ki:/ prefix and the pre­ terit suffix. The first difference involves an aspectual distinction; the second involves the fact that the preterit—but not /ki:/—can be used to indicate things which are not past tense at all, but are instead hypothetical or irreal. And it will be shown that this latter use of the preterit is semantically linked both with its use to indicate past tense and also more specifically with its use to indicate aspect. In addition, it will be noted that a preterit suffix (unlike the /ki:/ prefix) can ap-

When there is a negative in the clause, then in at least Moose Cree, the /ki:/ prefix is either followed by or replaced by the morpheme /ohci/. In the follow­ ing sentences, both (b) and (c) are equally acceptable: (a) ki:-wa:pame:w 'he saw him' (b) mo:la ki:-ohci-wa:pame:w 'he did not see him' (mo:la 'not') (c) mo.la ohci-wa:pame:w 'he did not see him'

The exact form of the suffix characterized by /pan/ varies somewhat accord­ ing to the order, mode, and stem type of the verb and the subject of the verb. Thus, for example, in the independent indicative, the suffix can be described as having the form /oLpan/ (Ellis 1971), where the /L/ indicates that a preceding short vowel is lengthened. When this suffix is preceded by a vowel, its initial /o/ is dropped. In the conjunct indicative, however, when the subject (and object, if any) is third person and the verb is animate intransitive or transitive animate, then the form of the suffix is /span/. As another example, when the subject is first person, the verb is in the conjunct indicative, and the verb is animate in­ transitive or transitive inanimate, then the form of the suffix is /a:pa:n/. As can be seen, there is a good deal of irregularity in the phonological form of the suffix. 144 Deborah James

pear on a noun or a pronoun, and that in this case it indicates a spe­ cial emphasis on something's being the case in the past but not the present. I will then compare what has been written about these mor­ phemes in other dialects (and something will be said about cognate morphemes in other languages also), and some general conclusions will be drawn about the usage and history of these forms.

I

I will first examine the use of the preterit vs. /ki:/ in those cases where the preterit clearly has past tense meaning. When occurring as a suffix on verbs (noun and pronoun cases will be discussed later), the preterit in Moose Cree is used to indicate the following: activity in progress in the past; habitual activity in the past; and continuous states in the past, where these states are seen as relatively permanent ones holding in the background throughout the course of various other events. These three situations are precisely those which are typically marked by imperfective aspect forms in the world's languages (see, for example, Comrie 1976:25). Accord­ ing to Comrie (1976:3), in perfective aspect, "the whole of the situ­ ation is presented as a single unanalyzable whole, with beginning, middle, and end rolled into one." Imperfective aspect, on the other hand, makes reference to the internal portion of a situation only, with no reference being made to its beginning or end. Similarly, according to Johnson (1981:154), perfective aspect refers to "the time of the whole event itself," while imperfective aspect refers to "times in the developmental phase which are prior to the end of the event." The three classes of situation described above in which the preterit is used in Moose Cree all involve a focus on the portion of the situation prior to its end, and this is why they are normally associated with imperfective aspect forms in languages which have such. The prefix /ki:/, unlike the preterit, is used primarily for past events or states which are perfective in aspect, that is, are treated as single wholes. To illustrate these uses, examples will be given from eight narra­ tives collected in the fall of 1980.1 will look first at examples of the usage of the preterit. Examples (1) and (2) illustrate the use of the preterit to denote activities in progress in the past:

(1) e:kosi ma.ka e:kitohte:ya:hki-pan awa iskwe:w pimohte:-pan and (-) (when)-we-go-out-PRET this woman she-walk-PRET

ne:sta also Irreality in Cree 145

'and when we were going out this woman was walking also (with us)'

(2) e:yako ma:ka e:kape:siya:hki-pan e:ispisimo:nike:ya:hk once (—) (when)-we-camp-PRET (when)-we-make-a-floor asta:kanak e:a:pacihakihcik, oki:c ma:ka papa:mohte:-pan brush we-use(ing)-it on-top (—) he-walk-around-PRET awa atim this dog

'once when we were camping, when we were making a floor using brush, this dog was walking around on it'3

In the case of each of the above examples, the narrative then goes on to describe something else that happened while these activities were in progress. Sentences (3) and (4) illustrate the use of the preterit to denote habitual activities in the past: (3) e:pe:cipi:htokwe:-span mama anta ka:sime:towa:niwahk (when)-he-come-in-PRET regularly there poolroom api:-pan te:ta:witak mo:citak he-sit-PRET in-middle floor

'when he used to come in regularly there, into the poolroom, he would sit in the middle of the floor'

A comment should be made here on the reason why the verb /e:ispi§imo:- nike:ya:hk/ '(when) we make a floor' does not have a preterit suffix. Usually, embedded clauses in Cree are marked for tense relative to the tense of the higher clause. Thus, if the event described in the embedded clause is contemporaneous with the even described in the higher clause, the embedded clause will have pre­ sent tense marking, regardless of what the tense of the higher clause is. Because of this, /e:ispi§imo:nike:ya:hk/ is marked as present tense. However, a speaker can also choose to emphasize that an embedded clause is past tense relative to the time of the speech act or some other relevant point of time in the discourse, even though it is contemporaneous with its higher clause. In this case the em­ bedded clause does have past tense marking (and this could be either /ki:/ or the preterit, depending on which was semantically appropriate). This often happens, for example, at the beginning of a narrative or at the beginning of a flashback section of a narrative. It is because of this that the embedded clauses in (1) and (3) and the first embedded clause in (2) have past tense marking. In (2) it is evi- dentally sufficient to put the marking on the first embedded clause and not the second. 146 Deborah James

(4) na:spic pikwantaw osihta:-pan ayamiwiniliw e:ohcipa:hpina:- very anything-at-all he-make-up-PRET words (so-that)- niwahk people-laugh-about-it

'he used to make up anything at all to make people laugh'

Sentences (5) and (6) illustrate the use of the preterit to denote general states in the past. Sentence (5) occurs following another sen­ tence in which someone gives an order; it explains why the person in question had the rightt o give the order:

(5) e:kwana e:okima:wiya:hki-pan this-(is)-the-one who-we-have-as-boss-PRET

'he was the one who was our boss'

Sentence (6) is about a dog, and is given as background information in the narrative in which it appears (later on in this narrative, there is some question as to whether the dog is going to bite someone or not, a matter which is crucial to the story):

(6) mo:la wi:ska:t kiske:lihta:kwan-o:pan cima:koma:t not ever it-be-known-PRET for-him-to-bite awe:nihka:na anyone

'he was never known to bite anyone'

In the narratives in my data, there are 90 verbs with preterit suf­ fixes. Of these, 88 clearly fall into one of the above three semantic categories. By contrast, verbs with the prefix /ki:/ most normally occur in sentences such as (7) to (10):

(7) ki:-kipihcika:po:w ka:ni:ka:nohte:t ki:-he-stop-(walking) the-one-who-be-at-the-head

'the man who was at the head (of the line) stopped walking'

(8) e:kosi ma:ka misiwe: n-ki:-mi:la:na:n la:rans ma:nsis e:tas- and-so (—) all we-ki:-give Lawrence a-little we-who-be Irreality in Cree 147

iya:hk a:lahkona:wa so-many bannock

'and so we all gave Lawrence a little bannock'

(9) pe:yako ki:nkwa:niska:k mi:n'a:pihta: mwe:hci n-ki:-moh- one hour again half exactly we-ki:- ta:na:n walk

'we walked for exactly an hour and a half

(10) na:spic ma:ka ki:-milopalina:n e:pe:ciki:we:ya:hk. ki:-milosin very (—) ki:-it-go-well-for-us (when)-we-go-back ki:-it-be-good me:skan. ki:-maskawa:w trail ki:-it-be-hard

'everything went very well on out way back. The trail was good. It was hard'

In each of these cases, the action or state is being viewed as a com­ plete, unanalyzable whole, rather than as something in the process of happening or in the process of being the case. Thus, for example, in (10), the verbs meaning 'go well', 'be good', and 'be hard' denote states, just as do the main verbs in (5) and (6); but the verbs in (10) denote states which hold temporarily, for the period 'on our way back'; and they have a clear beginning and ending. They are in this way quite different from the states in examples (5) and (6), which are being presented as things generally true in the background throughout the course of various other events with which the spea­ ker is more immediately concerned, with no indication of when the states began or ended. Thus /ki:/ clearly indicates perfective aspect in all of (7) to (10). /ki:/, however, is not as uniform in its denotation of aspect as is the preterit. Thus, for example, in my narratives there are 51 cases of verbs which indicate habitual activity in the past. Of these, 46 have a preterit suffix, but five have /ki:/, and the cases with /ki:/ do not seem semantically different from the cases with the preterit. Similarly, there are a few cases where /ki:/ is used for past general states such as those illustrated in (5) and (6). This suggests that /ki:/ is the more unmarked past tense marking device in Moose Cree. There are also two other types of evidence for this position. The first is as follows: Of the 90 verbs in my narratives which have 148 Deborah James preterit suffixes, there are five which also have a /ki:/ prefix. In each case, the verb has imperfective aspect meaning and /ki:/ does not seem to add anything to the meaning or alter it in any way. An example is (11):

(11) ma:rkiy ki:-ihta:-pan-ak ne:stapiko mi:kwa:mak ante: tent ki:-they-are-(located)-PRET or teepees there

'they were living there in tents or teepees'

The second additional piece of evidence for the position that /ki:/ is in fact unmarked for aspect lies in the fact that in cases where the preterit is used to indicate things which are hypothetical or irreal (to be discussed below), then if the verb is meant also to refer to past time, it is /ki:/ which is used to indicate past time, regardless of whether perfective or imperfective aspect is more appropriate (the difference is not marked). Examples are (15) and (22) below. I propose, then, that the preterit specifically indicates imperfec­ tive aspect in the past, while /ki:/ is unmarked for aspect. If this is the case, then we would expect that when imperfective aspect is appropriate, this would normally be marked with a preterit suffix, but that it would not be ungrammatical for /ki:/ to be used; and this is what we find. Moreover, we would expect that it would not be inconsistent or contradictory to have a verb be marked both with /ki:/ and with a preterit suffix, and we find that this is in fact poss­ ible. It is, however, semantically redundant, in that the past tense meaning is expressed twice; this would explain the rarity of the con­ struction. We would moreover expect that in the cases where the preterit suffix is not being used to indicate past tense (i.e., in the "irreal" cases), /M:/ would be used to indicate past tense regardless of aspect, and this is again what we find. So far, we have dealt with /ki:/ and the preterit as verbal affixes. However, it should be noted here that a preterit suffix can also ap­ pear on nouns and on one pronoun, and in these cases, while it still expresses past tense, the preterit suffix does not indicate imperfec­ tive aspect, but rather has a somewhat different function. First, it can appear on nouns denoting human beings (most typically kin­ ship terms), and in this case it indicates that the person in question is no longer alive. An example is /nimosom-pan/ 'my late grandfa­ ther'. A preterit suffix can also appear on the pronoun /awe:na/ 'who'. In this case, it can mean either that some person or creature was present before but is present no longer, as in (12), or that some

4Horden (1881) reports that a verb marked with both /ki:/ and a preterit suffix has pluperfect meaning. I found no evidence for this in my data, however. Irreality in Cree 149

person or creature was formerly expected to do something, but has not done it or is not now going to do it, as in (13). There is a sense ot surprise associated with the use of /awe:ni-pan/.

(12) awe:ni-pan otoho:sima ca:n who-PRET his-horse John

'John's horse was gone!'

(13) awe:ni-pancitakosihk who-PRET he-will-come

'he was going to come but now isn't (I wonder why!)'

This use of the preterit seems closely linked with the "emphatic" (the term is that of Bloomfield 1946) use of the preterit as a suffix on verbs which shows up in Ojibwa and other Algonquian languages, and to a minor extent in the Plains dialect of Cree (see below). In this "emphatic" usage, there is a special stress on the fact that the event or state in question is purely in the past and is not the case in the present. We should note here, to clarify this point, that the use of past tense marking in languages does not in general necessarily imply that the event or state is not also true in the present. Thus, for example, one can say in English "John was living here then" when it is still the case that he is living here now. These cases in Moose Cree illustrate this emphatic usage in that in them, it is be­ ing emphasized that some person or creature lived or was physically present only in the past and not in the present, or it is being empha­ sized that some past expectation of how someone or something was going to behave has not been met in the present.

II

I will now turn to cases in which the preterit does not refer to past time at all, but rather is associated with the notion of hypothetical or irreal. In this usage, the preterit usually appears in combination with a future tense prefix on the verb. The use of a past tense morpheme to indicate the hypothetical (sometimes with and sometimes without an associated future tense marker) is in fact not uncommon in lan­ guages: it is very widespread in the Indo-European languages, (occur­ ring in English and French, among others), and also appears in at least the Athabascan language Chipewyan, the Wakashan language Nitinaht, the Bantu languages Tonga and Haya, and the Tibeto- 150 Deborah James

Burman language Garo.5 As has been pointed out by Steele (1975) among others, the reason why past tense marking is used in this way is probably because there is a semantic link between the notion of past tense and the notion of remoteness from reality. This link can be described in the following way. If something is in the past, then it can be seen as remote from the present. If it is remote from the pre­ sent, then it can be seen as remote from present reality; and if it is re­ mote from present reality, then it can be seen as remote from reality in general. The situations where this use of past tense is most likely to occur cross-linguistically are those where the greatest degree of remote­ ness from reality is involved—namely, contrary-to-fact situations. Thus, for example, every language which I have looked at which uses past tense to indicate irreality uses it in contrary-to-fact conditional senten^- ces.6 An example of such a sentence in Moose Cree is (14):

(14) kisa:spin Moose Factory ohciko-pan-e:, katamilowe:lihtam- if Moose Factory she-be-from-PRET future-she-like-it- o:pan o:ta PRET here

'if she were from Moose Factory, she would like it here, (but she isn't, and she doesn't)'

It should be noted that this sentence does not refer to semantic past time. Moose Cree also uses the preterit in one other contrary-to-fact situation: namely, in contrary-to-fact wishes. This is also a very common place cross-linguistically for past tense to be used. (15) is an example:

(15) nipakose:lihte:n Ca:n (ke:)-ki:tako§inoko-pan-e: ota:kosihk7

This use of past tense, as a cross-linguistic phenomenon, is described in de­ tail in James (1981).

6As has been pointed out by Karttunen and Peters (1977), among others, contrary-to-fact conditional sentences in English are not in fact always seman- tically contrary-to-fact. Discussion of this is beyond the scope of this paper, but the facts involved do not in any way invalidate the hypothesis made in the text, that is, the hypothesis that the use of past tense to indicate irreality is most like­ ly to occur in situations where the greatest degree of remoteness from reality is involved. Here the future tense marker and complementizer /ke:/ may be present or [ absent optionally. The verb is marked as dubitative here. The verb in the if- clause of (14) is marked as dubitative also. Irreality in Cree 151

I-wish John (fut.)-ki:-he-come-PRET yesterday

T wish John had come yesterday'

We might also note the use of the prefix /ki:/ here to indicate semantic past time (as mentioned above). Moose Cree also uses the preterit in various situations where the events described are not contrary-to-fact, but are nevertheless hypo­ thetical. For example, it uses the preterit in clauses which denote what would be the case if some hypothetical condition were met in the fu­ ture. Thus, for example, the preterit appears in the main clauses of what are sometimes called "future less vivid" conditional sentences. An example is (16):

(16) kisa:spin itohte:yin mo:sonihk kamilowe:lihte:-htay anta if you-go to-Moosonee future-you-like-it-PRET there

'if you went to Moosonee, you would like it there'

The use of past tense marking in this environment is also common in other languages which use this device to indicate irreality (as a few ex­ amples, English, French, Garo, and Tonga use past tense in this situa­ tion). We might note, however, that the preterit is only used in the main clause and not in the if-clause in this type of conditional in Moose Cree (although past tense may be used in the if-clause in other langua­ ges, e.g. English). The condition (in (14) and (16), the if-clause) does not have to be overtly present in order for the preterit to be used on the verb in the clause which denotes the consequence. This is illustrated in (17):

(17) na:natawa:pahte:ta:w a:patisiwin ke:ki:sia:patisiyahko-pan let's-look-for job future-we-be-able-to-work-PRET

'let's look for a job where we would be able to work'

There is an unstated condition here: "a job where we would be able to work, if we had that job". Also, the condition does not have to be expressed by an if-clause; if can be expressed by some other syntactic means, such as a comple­ ment, an embedded question, or a when-clause. (18) to (20) illus­ trate this: 152 Deborah James

(18) tamilosin-o:pan kiciki:we:yahk ka:ki:otohte:yahk fut.-it-be-good-PRET for-us-to-go-back (the-place)-that-we- came-from

'it would be good for us to go back where we came from'

Note that this could be paraphrased as 'It would be good if we went back where we came from', the condition is formally a complement here, however, rather than an if-clause.

(19) ke:kwa:nliw piko ke:natawe:lihtama:n kami:liko-htay whatever only I-will-want fut.-she-give-me-PRET

nka:wiy my-mother

'my mother will give me whatever I want'

Note that this could be paraphrased as 'If I were to want something (no matter what), my mother would give it to me'. The condition is formally an embedded question rather than an if-clause here, how­ ever.

(20) ispi kanawa:pahtahk ki:kiliw, takiske:lihtam-o:pan ispic when he-look-at your-house fut.-he-know-PRET how-much

ke :itakihtahki-pan fut.-he-price-it-at-PRET

'when he has looked at your house, he will know how much he should price it at'

The when-clause here describes the condition from which (if it is met) the state of affairs described in the main clause will follow. It is for this reason that the verb meaning 'know' is marked as preterit. (The last verb in this sentence also has preterit marking; this, how­ ever, is an example of a different situation in which the preterit is used, to be described next.) Another common way in which the preterit is used is to denote the notion of obligation, as expressed in English by words and phra­ ses such as "should", "must", and "ought to". This again is also a Irreality in Cree 153 common way in which other languages use past tense to indicate the hypothetical. For example, in English, the words "should" and "ought" are historically past tense forms. Some examples of this (in addition to the one in sentence (20)) are given in (21) to (23):

(21) kakodihta:-htay kiciwa:pamacik fut.-you-try-PRET for-you-to-see-them

'you ought to try to see them'

(22) me:ri taki:ona:pe:mi:-pan sa:ma Mary fut.-ki: -she-marry-him-PRET Sam

'Mary should have married Sam'

(23) e:kosi ma:ka ka:otihtama:hk ke:sia:patisiya:hki-pan and-so (—) we-got-to (where)-fut.-we-work-PRET

'and so we got to where we were supposed to work'

In addition, Moose Cree uses the preterit in complements which de­ note something general and hypothetical, when the matrix clause gives some kind of value judgement about what is being expressed in the complement. Interestingly, I know of no other language which uses past tense marking in a way quite comparable to this. Some ex­ amples are (24) to (27):

(24) e:isita:pwe:htama:n ni:la we:sa: kise.lihw kiciwi:kima:- (the-thing)-that-I-think I too he-is-old fut.-for-him-to- span-anihi iskwe:wa marry-PRET that woman

'what I think is that he's too old to marry that woman'

(25) ana iskwe:w ka:a:lime:lihtama:n cimilowe:limaki-pan me:ri that woman who-I-find-it-hard fut.-for-me-to-like-PRET Mary

'the woman I find it hard to like is Mary' 154 Deborah James

(26) milohta:kwan anima ka:kitohcika:niwahk kiciki:ni:mina:- it-sound-nice that which-be-being-played fut.-for-people-to- niwahki-pan be-able-to-dance-PRET

'that's nice music for dancing' ('that sounds nice, what's be­ ing played for people to be able to dance')

(27) maska:te:lihta:kwan iskwe:w ni:swa:s na:pe:wa kiciona:- it-be-strange woman seven men fut.-for-her- pe :mi-span to-marry-PRET

'it's strange for a woman to marry seven men'

This completes the list of situations in which the preterit is used to indicate the hypothetical in Moose Cree. It might be noted that this list does not form a natural class, in the sense that while all these sit­ uations involve something which is non-factual and hypothetical, there are also many other situations which also involve something hypothetical but in which the preterit is not used. Two examples are wishes for the future and purpose clauses (and some languages do use past tense marking in these situations; e.g., Chipewyan and Rus­ sian do in the former, and Russian does in the latter). There does not seem to be any semantic feature which is present in only the situa­ tions described in Moose Cree and not others; thus, native speakers must have to memorize exactly which irreal situations the preterit occurs in in Moose Cree. But this is, in fact, the normal situation in languages which use this type of device (see James 1981). At this point, one must ask the following question: why is it the preterit rather than the prefix /ki:/ which is used to indicate the hy­ pothetical? After all, both are past tense markers. However, we should recall here that the preterit is elsewhere specifically a marker of imperfective aspect. And in fact, in a number of languages which use past tense to mark the hypothetical, it is specifically an imper­ fective aspect form which is used for this purpose (French, Old Irish, and modern Greek are examples). I know of no languages, however, which do the opposite: i.e., which distinguish perfective and imper­ fective aspect in the past tense, but use only the perfective aspect form to denote the hypothetical. Thus, there is a cross-linguistic ten- Irreality in Cree 155 dency to associate imperfective aspect with the hypothetical. I sug­ gest that this is because these two notions are linked semantically, and that they are linked by the notion "incomplete", or "not com­ pletely realized". As we have seen, imperfective aspect has as its es­ sential feature the fact that it indicates that an action or state is not being viewed as a complete whole, but rather is being viewed as something going on in time and not yet completed. But likewise, if an action or state is hypothetical, then it is unrealized, and in that sense is not an actual completed action or state. The similarity be­ tween imperfective aspect and the hypothetical is made particu­ larly clear in this quote from Johnson (1981:156) on the subject of imperfective aspect: "imperfective aspect always involves mention­ ing an event whose full realization in actual time, when considered from the position of the reference time, is only a highly probable development, but cannot be called an absolute certainty." Both im­ perfective aspect and the hypothetical, then indicate something which is in some way not fully realized.

Ill

At this point, one might compare these facts about the use of the preterit and the prefix /ki:/ in Moose Cree with what has been writ­ ten about the use of these forms in other dialects of the language. I will look first at the use of these forms to indicate past tense. The only Cree or Montagnais dialect for which this has been des­ cribed in any detail at all, as far as I know, is Plains Cree. In this dia­ lect, according to Bloomfield (1928) and Wolfart (1973), there are three preterit morphemes, and they are not amalgamated into one paradigm, as is the case with the two forms /htay/ and /pan/ in Moose Cree. These three morphemes are described as the /h/, /ht/ (or /htay/), and /p/ (or /pan/) preterits. The /p/ form is said to be rare and is simply said to denote "past occurrences no longer true in the present". The one example given of its use as a verbal suffix is /pima:- tisipan/ 'he was alive (and now is dead)' (Bloomfield 1928:429). In addition, /pan/ appears as a suffix on nouns and on the pronoun /awi:na/ 'who', in usages exactly parallel to the Moose Cree examples given above of the use of /pan/ with nouns and the pronoun 'who' (except that no examples comparable to (13) are given). Thus, re­ gardless of where it is used, the /p/ form in Plains Cree seems to be used only to indicate emphatic past, that is, to indicate a special em­ phasis on the event or state in question's being in the past and not being still true in the present. As for the /h/ and /ht/ preterits, according to Wolfart (1973), it is unclear as to whether or not there is any difference in meaning be- 156 Deborah James tween them.8 Wolfart (1973:45) reports that the /h/ form is often used with imperfective aspect meaning; for example, he says that it may denote "occurrences which persist", as in a sentence such as 'he kept hauling wood'. With regard to the /ki:/ prefix, both Wolfart and Bloomfield gloss it simply as 'past tense'. However, Wolfart (1973:45) does refer to it at one point as "the perfective (?) preverb /ki./." In addition, we might note that the missionary grammars of Lacombe (1874) and Hives (1948) both label the preterit the "im­ perfect tense" and the /ki:/ form the "perfect tense." Thus, it would appear that there is at least a strong tendency for imperfective as­ pect to be associated with the /h/ and /ht/ preterits and perfective aspect with the prefix /ki:/ in Plains Cree. As regards other dialects, MacKenzie and Clarke (1981:140) re­ port that "a preverb /ki:/ or /ci:/ marking past or completed action is used in all dialects." They also note that in at least one dialect (that of North West River, Labrador), the prefix is fairly rare, and the preterit is the primary past tense marker; and that in at least one other dialect (that of Mistassini, ), the converse has happened and the preterit has virtually disappeared, at least in the indicative mode, leaving the /ki:/ form as the primary past tense marker. They do not comment on any semantic differences between /ki:/ and the preterit, however, except for the irreality uses to be mentioned in the next paragraph. Let us now turn to the use of past tense markers to indicate the hypothetical. Wolfart (1973:45) reports that the preterit in Plains Cree can be used to indicate irreality, mentioning in particular that it is used, together with a future prefix, in irreal conditional senten­ ces.9 This is what we have already seen in Moose Cree. In addition, Hunter (1875) and Hives (1948) both mention that a future prefix plus the preterit indicates a meaning of "should" or "would". Uses of the preterit to indicate irreality are also mentioned in connection with four Quebec or Labrador dialects in MacKenzie and Clarke (1981). In each case, MacKenzie and Clarke report that one parti­ cular preterit form—variously the /h/ or the /p/ form— is not used to indicate past tense, but rather only to indicate some type of irreal­ ity. Specifically, in Fort George, Quebec, the /h/ preterit is used only with a modal preverb to indicate the meaning 'should'; in the North West River dialect of Labrador and the Lower North Shore dialect

8 It has been suggested by David Pentland that the /h/ and /ht/ forms are pho­ nological variants of the same morpheme. Marguerite MacKenzie also supports this view.

9The four examples of this which Wolfart gives contain /h/ or /ht/ preterits. He does not say whether or not the /p/ preterit can also be used in this way. Irreality in Cree 157 of Quebec, the /p/ preterit must either combine with a modal preverb to indicate "a non-factual event", or occur in the suffix /-shkpan/, which denotes "an event the occurrence of which is based on hearsay evidence rather than the speaker's actual observation." And in Mis­ tassini, the dubitative preterit (involving a /p/ form) is likewise used to indicate events known from hearsay evidence (MacKenzie and Clarke 1981:141-143). It is of interest that this last is a usage which does not occur in Moose Cree. It is also relevant to note here that, according to Bloomfield (1958), Ojibwa also has both a /ki: / prefix and a preterit suffix. The /ki:/ prefix is glossed simply as 'completed action'. As regards the preterit, Ojibwa has only the /p/ (/pan/) form and it is clear from Bloomfield's description that this is used both to indicate emphatic past (regardless of aspect) and to indicate specifically imperfective aspect in the past (past habitual activities and past general states are given as examples). Moreover, the preterit in Ojibwa can also be used in irreal conditional sentences. Thus, we see in Ojibwa a semantic pattern basically very similar to that of Cree. As regards other Algonquian languages, as least Menomini (Bloom­ field 1961), Delaware (Goddard 1979), and Malecite (Szabo 1981) also have a preterit in /pan/, and in all three languages this seems to have an emphatic function like that described above for Cree. All three languages also have a past tense prefix (apparently not related to /ki:/) which is not emphatic. The preterit does not appear to have an imperfective aspect use in any of these languages. It does, how­ ever, appear to be used in irreal conditional sentences in at least Dela­ ware. We can see, then, that the /ki:/ prefix seems to be basically a neu­ tral past tense marker, with a tendency towards indicating perfec­ tive aspect, in Cree and probably also in Ojibwa. As regards the pre­ terit, I will look first at the /p/ form. This is associated with three specific types of notion besides the past tense: emphatic past, im­ perfective aspect, and irreality. The meaning most widely distributed across languages and dialects is that of emphatic past, and therefore it is reasonable to conclude that this was the original meaning of the morpheme historically. If this was the case, how would the other two meanings have arisen? It should be noted that just as there is a se­ mantic link between irreality and imperfective aspect as opposed to perfective aspect, there is also a semantic link between irreality and emphatic past as opposed to non-emphatic past. After all, if some­ thing is emphatically in the past, then it is emphatically remote from the present, and thus emphatically remote from present reality. By comparison, a non-emphatic past form does not imply as much dis­ tance from reality. This would explain why the /p/ preterit but not 158 Deborah James

the /ki:/ form would have developed an irreality sense in addition to its past tense sense. Further, once the /p/ preterit had an irreal­ ity sense in addition to its past tense sense, the presence of the im­ perfective aspect meaning, where it occurs (e.g., in Moose Cree and Ojibwa), can be explained as resulting from the close semantic con­ nection between the notion of irreality and the notion of imperfec' tive aspect in the past. As regards the /h/ and /ht/ forms of the preterit, these seem to occur only in Cree-Montagnais-Naskapi, and in addition to the past tense notion, they have associated with them the two notions of im­ perfective aspect and irreality. These two forms may be different forms of the same original morpheme. If we assume this to be the case, what would the meaning of this morpheme have been origin­ ally? If we assume that it was simply general past tenseunmarke d for aspect, then we have no way of explaining why it is the /h/ or /ht/ form rather than the prefix /ki:/ that is associated with imperfective aspect and irreality. It could, of course, have originally denoted im­ perfective aspect in the past, and the irreality meaning could have developed from that as a result of the semantic connection between imperfective aspect and irreality. Another possibility, however, is that the morpheme was originally purely a marker of irreality. In this case, it would subsequently have developed the meaning of past tense, and, further, specifically imperfective aspect in the past, because of the semantic links between irreality and these other no­ tions. A case very similar to this has been argued for by Seiler (1971). Seiler argues that in Latin the morpheme /-ba-/, which marks past tense with imperfective aspect, is historically derived from a form of the verb 'be' in the present optative; the optative was a type of ir­ reality marker in early Indo-European. It is of interest, then to note that the /p/ preterit and the /h/ or /ht/ preterits may originally have denoted quite different meanings in Cree-Montagnais-Naskapi—the former, emphatic past, and the latter, irreality. However, they have now fallen together in meaning to such an extent that in some dialects, such as Moose Cree, they now form a single syncretized paradigm; and in others, one form (for example, in Plains Cree, the /p/ preterit) has virtually disappeared, having been replaced by the other form(s). It should be emphasized, however, that much data is needed on the occurrence of these morphemes and how they are used in the various dialects, and also in other Algonquian languages before we can speak with more assurance about their history.10

10Since writing this paper, I have learned from David Pentland that in Nor­ thern Plains Cree, not only a /h(tay)/ paradigm, but also a /pan/ paradigm is pro­ ductive and common; the /pan/ forms are used only with inanimate intransitive Irreality in Cree 159

REFERENCES Bloomfield, Leonard 1928 The Plains . International Congress of Americanists Proceedings 22:427-431. 1946 Algonquian. Pp. 85-129 in Linguistic Structures of Native America. Harry Hoijer, ed. Viking Fund Publications in Anthropology 6. New York. 1958 Eastern Ojibwa: Grammatical Sketch, Texts, and Word List. Ann Arbor: University of Michigan Press. 1962 The Menomini Language. New Haven: Yale University Press. Comrie, Bernard 1976 Aspect. Cambridge: Cambridge University Press. Ellis, C. Douglas 1971 Cree Verb Paradigms. International Journal of American Linguistics 37:76-95. Goddard, Ives 1979 Delaware Verbal Morphology: A Descriptive and Comparative Study. New York: Garland Publishing. Hives, Harry E. 1948 A Cree Grammar. Toronto: Missionary Society of the Church of England in . Horden, John 1881 A Grammar of the Cree Language. London: Society for the Propa­ gation of Christian Knowledge.

verbs, however. Both forms seem to be primarily used with the meaning of Eng­ lish 'would have'. This appears to combine the meanings of past tense and irreal­ ity. Pentland gives in addition one example in which a /pan/ form means 'used to' (/ayimano:pan/ 'it used to be hard'); this suggests either an emphatic past or an imperfective aspect meaning. As regards the /h/ and /htay/ forms, Pent- land suggests that the /h/ form is historically derived from the /htay/ form (as noted in footnote 8); specifically, he proposes that /htay/ became /h/ when not followed by further endings (i.e., pluralizers) by the same change that deleted final /n/ from /toke:n/, which is the dubitive morpheme. The full form /htay/ was then optionally restored to the singular in Plains Cree, he suggests. In Moose, /htay/ would have completely replaced the /h/ forms, in this view. As regards the etymology of /htay/, Pentland suggests that two forms which may be cognate to it are a morpheme /t/ meaning 'future' in Micmac and a type of interrogative mode in Fox with /t/. It is of interest that the notion of the hypothetical is involved in both of these meanings. If these are truly cognate formsoriginall, thiy denotes lendsd somsuppore typt eto o fth irrealitye suggestio. n that the morpheme in question 160 Deborah James

Hunter, James 1875 A Lecture on the Grammatical Construction of the Cree Language. London: Society for the Propagation of Christian Knowledge.

James, Deborah 1981 Past Tense as a Device for Indicating the Hypothetical. Paper read at the Canadian Linguistic Association. Halifax.

Johnson, Marion 1981 A Unified Temporal Theory of Tense and Aspect. In Syntax and Semantics 14: Tense and Aspect. P. Tedeschi and A. Zaenen, eds. New York: Academic Press.

Karttunen, Lauri, and Stanley Peters 1977 Requiem for Presupposition. Pp. 360-371 in Proceedings of the Third Annual Meeting of the Berkeley Linguistics Society. K. Whist­ ler et. al., eds. Berkeley.

Lacombe, Albert 1874 Grammaire de la langue des Cris. Montreal: Beauchemin et Valois.

MacKenzie, Marguerite, and Sandra Clarke 1981 Dialect Relations in Cree/Montagnais/Naskapi: Verb Paradigms. Pp. 135-191 in Linguistique amerindienne II: Etudes algonquiennes. Lynn Drapeau, ed.Recherches linguistiques a Montreal 16. Montreal.

Seiler, Hansjakob 1971 Abstract Structures and Moods in Greek. Language 47:79-90.

Steele, Susan 1975 Past and Irrealis: Just What Does it All Mean? International Journal of American Linguistics 41:200-217.

Szabo, Laszlo 1981 The Use of Verbal Enclitics in Malecite. Pp. 106-113 in Papers of the Twelfth Algonquian Conference. William Cowan, ed. Ottawa: Carle- ton University. Wolfart, H. Christoph 1973 Plains Cree: A Grammatical Study. Transactions of the American Philosophical Society 63, part 5. Philadelphia.