Optimizing Display Other Formats:      PDF - 350k

Volume 10 Issue 1 (2012)        DOI:10.1349/PS1.1537-0852.A.405

Note: Linguistic Discovery uses Unicode characters to represent phonetic symbols. Please see Optimizing Display for requirements to accurately reproduce this page.

The Seneca Amplification Construction

Wallace Chafe

University of California, Santa Barbara

The polysynthetic morphology of the Northern Iroquoian languages presents a challenge to studies of clause combining. The discussion here focuses on a Seneca construction that may appear within a single clause but may also straddle clause boundaries. It amplifies the information provided by a referent, here called the trigger, that is introduced by the pronominal prefix within a verb or occasionally in some other way. The particle neh signals that further information about that referent will follow. This construction is found at four levels of syntactic complexity. At the first level the trigger and its amplification occur within the same prosodic phrase and the amplification is a noun. At the second level the amplification occurs in a separate prosodic phrase but remains a noun. At the third level the amplification exhibits verb morphology but has been lexicalized with a nominal function. At the fourth level the amplification functions as a full clause and neh serves as a marker of clause combining. Several varieties of amplification are discussed, as are cases in which the speaker judges that no amplification is needed. It is suggested that the typologically similar Caddo language illustrates a situation in which this construction could never arise, simply because Caddo verbs lack the pronominal element that triggers the construction in Seneca.

1. Background

Seneca is a Northern Iroquoian language still spoken by a few dozen people on three reservations in western New York State (Chafe 1996). All the Northern Iroquoian languages are highly polysynthetic, and Seneca is also highly fusional as a result of numerous sound changes, the majority of which probably took place during the eighteenth century, as suggested by a manuscript Jesuit dictionary from about 1700. These languages show an unusually high proportion of words with the morphological structure of verbs. Jean André Cuoq, a 19th century missionary to the Mohawks, wrote the following of the Mohawk language, which is closely related to Seneca:

Ils manquent d’articles, et ils ne sauraient parer à ce défaut d’articles ni par des cas, ni par des prépositions, dont ils sont également dépourvus. Toutefois ils ont d’autres moyens d’y suppléer et de maintenir par là la clarté du discours. [...] Ils ne possèdent que peu d’adverbes et de conjonctions, mais ils sont d’une richesse étonnante en fait de verbes. Dans leur langue, presque tout est verbe ou peut le devenir” (Cuoq 1866:87).

(They don’t have articles, and they wouldn’t know how to compensate for this lack of articles either with case or with prepositions, which they also lack. Nevertheless, they have other ways of establishing and maintaining clarity of discourse. [...] They have only a few adverbs and conjunctions, but in fact they have an astonishing wealth of verbs. In their language almost everything is a verb, or can become one.)

Figure 1 shows the relative proportions of morphologically defined verbs, nouns, and particles in a representative sample of Seneca speech.

When Cuoq mentioned an absence of “articles,” he was probably thinking of the definite and indefinite articles of French. Here we can focus on the Seneca particle neh and the way it functions to link units of discourse. In terms of text frequency it is the most commonly occurring word in the language. In the speech sample examined here it accounts for as much as six percent of all word occurrences. Sometimes it is used in a way that suggests a translation with the English definite article, but such a translation would be misleading. More accurate is the translation ‘namely’, or the colon as a punctuation mark indicating that something relevant is about to follow. Its meaning can be roughly paraphrased as “you are already thinking about a referent and you are about to hear something more about it.” It signals that what follows is a clarification, expansion, or amplification of a referent already introduced. To understand this function in more detail one needs to understand certain properties of the Seneca language itself.

Figure 1: Proportions in Seneca of different verbs, nouns, and particles

Another 19th century writer, Francis Lieber, coined the term “holophrastic” for languages of this type. He described holophrastic words as “words which express the whole thing or idea, undivided, unanalyzed” (Lieber 1837:167). Regardless of one’s language, when one thinks of a specific event or state it necessarily includes a specific participant. ^[1] Events and states are inconceivable without their participants. This general observation offers a satisfying explanation for holophrastic verbs like those of Seneca, which integrate an event or state together with its participant(s) within a single word, thus directly capturing the inseparability of the two. Such words reflect in a direct way what appears to be a universal property of thought.

Holophrasis blurs the distinction between verb and clause, since the information integrated within a verb often coincides with information that would otherwise constitute a clause. If a clause is the linguistic expression of an event or state with the inclusion of one or more participants, a Seneca clause need be nothing more than a verb, but often the verb is accompanied by one or more particles (words like neh that have little or no morphological structure), which serve to orient the event or state to its context.

2. The Effects of Morphological Fusion

Seneca verbs are not available for conscious analysis by those who say or hear them. A linguist may analyze them into a base (expressing the idea of an event or state) plus a pronominal prefix (expressing the idea of its participant(s)), but such an analysis is not what a Seneca speaker experiences. In fact, the high degree of fusion characteristic of this language often prevents even a linguist from segmenting a word without first reconstructing an earlier stage of the language when segmentation was more transparent.

For example, the word ia:s means ‘he eats’. An earlier stage can be reconstructed as * ihraks, where * hra‑ was a masculine singular agent pronominal prefix, *‑ k‑ was the root meaning ‘eat’, *‑ s was a habitual aspect suffix, and a prothetic * i‑ was added at the beginning because the language could not tolerate a one‑syllable verb. In a series of phonological changes the consonant r was lost entirely, followed by a loss of the resulting intervocalic h and of k before a word‑final s. Vowels in final syllables were lengthened unless they were followed by a final h or glottal stop. In the resulting ia:s there is a relic of the pronominal prefix * hra‑ in the vowel a, but the verb root *‑ k‑ has disappeared entirely. Nevertheless, Seneca speakers use and understand ia:s as a word that means ‘he eats.’

In the examples here the following interlinear format will be used to show first the modern pronunciation, then a reconstructed form with hyphens separating the morphemes, then glosses of the morphemes, ^[2] and finally an English translation of the entire word.

Imagine now a situation in which a woman awoke from sleep. In English we might express our idea of that event with the words she woke up , where the phrase woke up captures the idea of the event and the pronoun she captures the idea of the person who did it. In Seneca the idea of such an event would be expressed with the single word wa’e:yeh, whose unitary nature reflects the inseparability of its constituent elements. A linguist familiar with Seneca morphology and its history might identify the vowel e in the middle of this word (part of an earlier * ye‑) as the part that communicates the idea of a single female participant. Seneca speakers are not aware of such an analysis, whereas for English speakers the separateness of the word she in she woke up is apparent. (2) shows the analysis of this word. Although this analysis is not known to Seneca speakers, the word as a whole does convey the fact that the one who awoke was a single third person female.

3. The Amplification Construction

Through their pronominal prefixes, often hidden within them, Seneca words convey the following types of information regarding the participants in events and states:

Often a speaker will decide that a hearer needs more information concerning a referent than these choices provide. The amplification construction is a favorite way of filling that need. Its components can be summarized as:

The trigger is a word whose information is judged inadequate for the hearer’s needs, usually because the information provided by the pronominal prefix is insufficient. The particle neh anticipates an amplification to follow. Often this amplification is a word whose own pronominal prefix repeats all or part of the information contained within the pronominal prefix of the trigger.

The amplification may be added syntactically in ways that show varying degrees of complexity. In the simplest case, or first degree complexity, the entire construction is confined within a single prosodic phrase or intonation unit expressing a single focus of consciousness (Chafe 1994), as illustrated in example (3) below. Second degree complexity is achieved when the amplification occupies a separate prosodic phrase containing nothing more than a noun, as illustrated in (4). Third degree complexity is similar, except that the amplifying phrase contains a word with verbal morphology that has been lexicalized as a noun, as illustrated in (5) and (6). Finally, with fourth degree complexity the amplifying phrase contains a full-fledged verb. It is only then that the amplifying phrase achieves the status of a separate clause and neh functions as a signal of clause combining, as illustrated in (7), (8), and (9).

The second, third, and fourth degrees of complexity all exhibit two prosodic phrases joined by neh. Sometimes this neh appears as the last word in the trigger, anticipating the amplification to follow. Sometimes it occupies a separate position between the two phrases, and sometimes it is the first word in the amplification. These three options have subtly different effects, but no attempt to describe them will be made here.

4. First Degree Complexity

In (3) the speaker had described how a man and his daughter went together into the woods. Then came the following: ^[3]

Up to this point the focus had been on two individuals, the man and his daughter, but (3) described an action performed by only one of them. The fact that it was only the father who went hunting was made clear by amplifying the masculine singular agent prefix in hadówä te’s with neh haöhwö’ ‘namely he himself’ with haöhwö’ distinguishing him from his daughter.

The amplification here might seem to have provided little information beyond what was provided by the masculine singular prefix in hadówäte’s, which had already established that the person who went hunting was a man and did not include the daughter. What, then, is the difference between the communicative effect of a pronominal prefix and the effect of an entire word?

The following analogy may help shed light on the difference. Suppose we compare a Seneca verb with a soup whose various ingredients combine to produce the total gustatory effect, each ingredient no longer easily distinguishable from the others. Suppose one of those ingredients was a tomato, and contrast this soup with a separate tomato that retains its identity as a round red object. For Seneca speakers the word hadówäte’s is like the soup, with parts no longer easily separated. The word haöhwö’ is like the separate tomato, with an independent identity on which one can focus separate attention. Assigning this man to the haöhwö’ category let the hearer individuate him in a way the prefix integrated within hadówäte’s did not.

5. Second Degree Complexity

In (4) the speaker remembered an incident from her childhood, when she went to visit some relatives and took with her a teddy bear. ^[4]

The pronominal prefix in (4a) specified overtly the first person agent of the taking event but not the neuter patient (the bear). In the Northern Iroquoian languages neuter participants have their own overt marking only when no animate participant is present. In this case it is understood that she took something that can be translated ‘it’, and it is that implied neuter referent whose nature was amplified with the noun in (4b).

The only significant difference between (3) and (4) is the division into separate prosodic phrases that reflect the separate foci of attention allotted to the act of taking and the object that was taken. The pitch trace in Figure 2 shows the two phrases and the pause between them. We will see that this kind of prosodic separation offers the possibility of elaborating the amplification phrase in ways that go beyond the simple noun in (4b).

Figure 2: Pitch contours in (4)

6. Third Degree Complexity

In (5) the speaker introduced the idea of a visiting event with the verb wa:ya:jö’s ‘they (du) visited’. Its pronominal prefix, reconstructable as * hy‑ but with loss of the h and compensatory lengthening, carried the information that the agents of this event (the visitors) were two males. Seneca culture attaches special importance to kinship: knowing who is related to whom and what the relationship is. This speaker realized the hearer would want to know more than just the fact that the visitors were two males, and to amplify that information he first extended (5a) with neh and then explained in (5b), a separate prosodic phrase, that the visitors were a father and his son. The y at the beginning of (5b) is a reflex of the same masculine dual agent prefix * hy‑ reconstructed for the verb of (5a). The ‑ atat‑ in (5b) expresses the reciprocal relation of being father and son to each other.

Kinship relations are expressed in Seneca with words like that in (5b), which have the morphological structure of verbs but refer to people, not events or states. (5b) differs from (4b) in exhibiting verb morphology, but its lexicalized status prevents it from being interpreted as a full-fledged verb. The two phrases of (5) thus constitute a single clause, parallel to the two phrases of (4).

A similar structure is presented by (6), the beginning of a Seneca story. In (6b) the verb wá:hdë:di’ ‘he set out’ introduced a masculine singular agent. Without amplification the story would be heard as beginning in medias res, a pretense that the hearer was thrown into the middle of events with no background context (Chafe 1994:228). That device, however, is more at home in written literature, and in this case the narrator fulfilled his obligation to the hearer by ending (6b) with neh and adding an amplification in (6c) with the word for ‘man’.

As in (3), the amplification here might seem to have provided little information beyond what was provided by the masculine singular prefix in wá:hdë:di’, which already established that the person who set out was a man. Here again we can contrast the communicative effect of a pronominal prefix with that of a separate word. Assigning this man to the hö:gweh category activated a complex set of associations including his role in Seneca society, his relation to his family, and his expected behavior, properties that went beyond his status as nothing more than masculine singular.

Although it may be less obvious, hö:gweh shares with yadátawak in (5) certain verbal properties. For example, when its root, * ‑ökweh, is incorporated with a verb root as in högwé’ di:yo:h ‘he’s a nice man’, it appears in the nominalized form *‑ ögwe’t‑, suggesting an origin as a defective verb. However, hö:gweh refers to a person, not a state, and (6), like (5), can be regarded as a single clause rather than two.

It is also worth noting that the inclusion of masculine singular in wá:hdë:di’ was not enough to make this referent identifiable or “definite” (Chafe 1994:93‑107). (6c) is better translated ‘ a man’ than ‘ the man’, thus removing whatever temptation there might have been to regard neh as a definite article. To be identifiable a referent evidently needs prior assignment to a lexical category, like that supplied here by hö:gweh. Characterization in terms of person, gender, number, and case alone is not enough.

7. Fourth Degree Complexity

We are finally ready to look at examples in which the amplification construction functions as a device for clause combining. In (7) the speaker was talking about a man who used to walk through the woods near her house, inspecting the gas pipelines that ran through her property. She said:

In (7a) the word ha:gëh ‘he sees it’ contains the masculine singular agent prefix ha- (from * hra). The speaker then decided that the hearer needed to know more about the man who kept seeing many things, not just the fact that he was masculine singular. In the amplification she explained that he was a man who was moving about. She had introduced this man earlier but had talked about something else in the meantime, so (7b) served to bring him back into the picture. To describe people as moving about, using a verb root whose earliest meaning was ‘stir’, is very common in Seneca discourse. The verb do:dawë:nye:h ‘he is moving about’ contains the masculine singular patient prefix ho- (from * hro-), whose patient role is dictated by the stative aspect ending, a regular feature of Northern Iroquoian verb morphology.

An alternative view of this example might interpret neh as a relative pronoun, ‘the one who was moving about’, and it does lend itself to that translation. However, neh is not referential but, as we have seen, a signal of amplification. Still another alternative would be to regard neh as a nominalizer, so that do:dawë:nye:h would be in apposition with the pronominal prefix of ha:gëh. That alternative, however, misses the underlying motivation of this construction as a device for amplifying the insufficient information provided by a preceding pronominal prefix.

(8) is a statement made by the original “False Face,” represented by the kind of wooden mask used for curing that is a distinctive Iroquois art form (Fenton 1987):

The pronominal prefix *khe‑ in (8a) combines a first person singular agent with a third person plural patient. The prefix *yök‑ in (8b) reverses these roles by combining a third person plural agent with a first person singular patient. The False Face amplified the information in (8a) by explaining who he would help.

In (9) the amplification in (9b) provides further information about the implied neuter patient of (9a), which is not plural, as the English translation suggests, but “distributive”: distributed in this case over a variety of things the man told about (Mithun 1999:88‑91). The neuter singular prefix of nö’ö:wëh in (9b) shares that referent.

8. Nested Amplifications

It is not unusual for one amplification to appear inside another. (10) quotes a statement that was made by a man who had constructed a large chair as a gift for his grandchildren. The pronominal prefix * hrën‑ of ë:nöjë:’ ‘they will sit’ in (10a) introduced a masculine plural agent of the sitting, and the immediately following neh anticipated the amplification in (10b) with the noun hadíksa’shö́’öh ‘children’, which shared the same masculine plural agent prefix. But the speaker decided that that was not enough and she added a second neh that led to a second amplification with the verb hë:ne’s ‘they are around’ (in the sense of being actively present), modified by the adverb jáwë’öh ‘always’.

The amplifying sequence neh jáwë’öh hë:ne’s suggests a translation with the relative clause ‘who are always around’. Some of the examples above could be translated with headless relative clauses, as (8), for example, could be translated ‘I will help them, the ones who depend on me’. Although such translations are adequate as translations, they reinterpret the Seneca construction with a pattern that is imposed by the grammar of English.

(10b) ended with the verb waë’ ‘he said’ with no prosodic phrase boundary and without functioning as a separate clause. This word is frequently used, as here, to attribute a preceding quote to a masculine singular speaker. It has been grammaticalized to function as an evidential that serves to identify the source of preceding language.

Another example of nested amplifications is provided by (11). The pronominal prefix *‑ shako‑ in (11a) combines a masculine singular agent (who has already been identified as the conductor on a train) with a feminine singular patient (a passenger). With the demonstrative në:gë:h ‘this’, (11b) focuses attention on the passenger and neh anticipates her categorization as an old lady. Knowing that much about her was evidently still insufficient, so in (11c) a second neh anticipated her further characterization as the grandmother of a certain male (‘his grandmother’). Information regarding that male was then amplified with a third neh to make it clear that he was a boy who had been introduced earlier in the story.

All three of the characterizations that were anticipated here by neh— yegë́hjih ‘old lady’, hohso:t ‘his grandmother’, and haksá’a:h ‘boy’—have the morphology of stative verbs, but they were used here to refer to people and not to states as such. Cuoq did not mention it, but one reason for the “astonishing richness of verbs” in these languages is their frequent use to convey ideas that other languages would express with nouns. In some contexts yegë́hjih means ‘she’s old’ (a state), but just as often ‘one who is old’, an old lady. (The German expression die Alte comes closer to the Seneca usage.) Similarly, hohso:t can mean literally ‘she is grandparent to him’ (a state), but more often ‘his grandmother’. And haksá’a:h can mean ‘he’s a small child’ (a state), but more often ‘one who is a small child’, a boy. If we limited “clause” to a sequence containing a morphological verb that expresses an event or state and has not been lexicalized as a noun, then (11) would be a single clause that contained three amplifications.

9. The ne:’ neh Construction

There is an extremely common Seneca usage in which an assertion is introduced with the two‑particle sequence ne:’ neh , which is often the first element in a sentence. The second of the two particles is the neh with which we have been concerned. The first, ne:’, does not easily correspond to anything in English, but its meaning can be approximated with the translation ‘it is’. (Seneca has no copula as such.) Occasionally a Seneca speaker may express agreement with something by saying nothing more than ne:’, roughly ‘that’s right’, although more often ne:’ is supplemented with another particle, such as waih ‘indeed’ in ne:’ waih ‘indeed so’, or nö:h ‘I guess’ in ne:’ nö:h ‘I guess so’.

(12) shows a typical use of the ne:’ neh sequence. Does neh in such a case anticipate an amplification and, if it does, what exactly does it amplify? If one regards ne:’ as implying the maximally vague, neutral referent captured by ‘it’ in the translation ‘it is’, that referent can be thought of as triggering the need for the further information that is supplied by the amplification that follows.

Should, then, the sequence ne:’ neh be considered a clause, as this analysis might suggest? Does ne:’ function as a verb, despite the absence of any verb morphology? From the point of view of the amplification construction, it can at least be said that (12) shows some of the symptoms of a two‑clause sequence, despite the absence of the full range of criteria one might otherwise associate with such a sequence.

10. Amplification Triggered By an Entire Word

The examples so far have illustrated amplification of the information provided by a pronominal prefix or, in (12), by a particle. Sometimes, however, a speaker may decide that an entire complex word and not its pronominal prefix alone needs amplification. At the beginning of (13) the ne:’ neh sequence followed the pattern illustrated in (12) by introducing the idea of a big crowd. But it was crucial to the speaker’s narrative for the hearer to understand that this crowd was composed of white men and not Indians. To make that clear, he went on to amplify gëjóhgowa:nëh ‘big crowd’ with neh hadí:nyö’öh ‘namely white men’. In English and many other languages such information would be expressed with a prepositional phrase such as ‘of white men’. Seneca performs a similar function with the amplification construction.

11. How a Lack of Interest May Preclude the Need for Amplification

Not every referent introduced with a pronominal prefix requires amplification. During a discussion of a local election in which people were paid for voting a certain way, someone said:

This word referred to a paying event that involved two kinds of participants, as captured by the transitive pronominal prefix. The people who did the paying (the payers) were specified as third person plural, while the beneficiary of the paying (the payee) was masculine singular. The identity of the payee was of considerable interest, and in fact he had already been named. In contrast, the identity of the payers was of little interest, and nothing further was said about them. The pronoun ‘they’ in the English translation can be understood as the nonspecific ‘they’ that is found, for example, in statements like ‘they charge too much for these apples.’ A lack of interest in a referent is often expressed in English with the passive voice, and the referent is often said to have been “demoted.” In (14) the passive translation ‘he was paid’ is thus an accurate alternative to ‘they paid him.’ The pronominal prefix *‑ höwa‑ is often used in this way.

12. How a Complex Verb May Also Preclude the Need for Amplification

Quite a different situation was illustrated by a speaker who decided that a word as a whole provided so much information that no amplification was necessary. She was telling about a man who was introduced with the verb in (15). Its pronominal prefix identified him only as one male, and without further information an amplification might have been needed. However, the same verb also carried the information that this man had lost his wife, and with that further knowledge there was no need for more.

13. Comparison with another Polysynthetic Language

The Caddo language of Oklahoma, a member of the Caddoan language family, shares with Seneca a highly polysynthetic morphology, a high degree of fusion, and a morphological structure that has much in common with Seneca (Melnar 2004, Chafe 2005). Although Caddo verbs often include pronominal prefixes, they are not always present. In particular, there is no overt marking of a third person realis agent, an especially common but covert category in Caddo speech. In such verbs the prefix that would trigger the Seneca amplification construction is simply not there, and its absence is associated with a favorite way of introducing a new referent that is rare in Seneca. In (16), the beginning of a Caddo story (Chafe 1977:29), the verb in (16b) has no pronominal prefix and thus no trigger for amplification. Instead, Tsah Wadu’ ‘Mr. Wildcat’ is introduced at the beginning and there is no need to expand the information that has already been provided by his name.

Nevertheless, Caddo sometimes follows a different pattern when introducing a new referent, as illustrated in (17) (Chafe 1977:33):

With its placement of the word for ‘wren’ after the verb, this sequence more closely resembles the Seneca pattern. However, tsínda:kístsi’ in (17c) cannot be considered an amplification of a pronominal prefix within the verb háh’í:’a’ ‘she is present’ because there was no pronominal prefix. The absence of such a prefix might thus be regarded as precluding the development of a construction that is such a conspicuous feature of Seneca. The suggestion is not that the presence of overt third person marking, as in Seneca, predictably leads to this construction, but rather that the absence of such marking, as in Caddo, creates a situation where the construction would not arise. Of course this hypothesis would be disconfirmed if a language like Caddo were found to exhibit the amplification construction. Whether other languages of the Seneca type do have it is for now an open question, but it does appear to be present in Cayuga, Seneca’s closest relative within the Iroquoian family.

14. Summary

The Seneca amplification construction that is introduced with the particle neh provides additional information about a preceding referent, usually one that was introduced with a pronominal prefix. The amplification may occur in the same prosodic phrase as the word that triggered it or it may occupy a separate phrase, in which case it may be a noun, a morphological verb lexicalized as a noun, or a full-fledged verb. Only in the last case does this construction qualify realistically as a type of clause combining.

The amplifying information may vary in its specificity, ranging from assignment to a very general category such as ‘man’ to a specific kinship relation such as ‘father and son’. Sometimes more specific amplifications are nested within more general ones, such as ‘grandmother’ within ‘old lady’. The very common sequence ne:’ neh can be understood as triggering an amplification of an abstract ‘it’ that is implied by the particle ne:’. Occasionally it is the meaning of an entire word, such as the word meaning ‘big crowd’, that triggers the amplification. Sometimes a newly introduced referent fails to trigger amplification, as with a pronominal prefix whose referent is judged to be unimportant, or a word whose meaning is rich enough in itself that no amplification is needed.

There was finally a brief mention of the Caddo language, whose very similar morphological structure fails, nevertheless, to include in its pronominal prefixes any overt marking of a third person realis agent. It was hypothesized that this absence of third person marking creates a situation where the amplification construction would not arise because, in effect, there is nothing to amplify.

Acknowledgments

I am grateful to all three editors of this issue—Jeannette Sakel, Pier-Marco Bertinetto, and Marianne Mithun—for a variety of helpful comments on an earlier version of this paper. I am also grateful, as always, to many Seneca and Caddo speakers over the years for sharing with me their uniquely fascinating languages.

Abbreviations

AGT agent; BEN beneficiary; DIS distributive; DU dual; DUP duplicative; EMPH emphasis; FAC factual; F feminine; FUT future; HAB habitual; M masculine; N neuter; NSF simple noun suffix; PAT patient; PFV perfective; PROTH prothetic; RECIP reciprocal; SG singular; STA stative; STA.CON stative continuative; STA.DIS STATIVE DISTRIBUTIVE; TRANS‑TRANSLOCATIVE; 1 first person; 3 third person

References

Chafe, Wallace. 1977. Caddo Texts. In: Caddoan Texts. International Journal of American Linguistics, Native American Texts Series, Vol. 2, No. 1, ed. by Douglas R. Parksm, 27‑43. Chicago: The University of Chicago Press,

-----. 1994. Discourse, consciousness, and time: the flow and displacement of conscious experience in speaking and writing. Chicago: The University of Chicago Press.

-----. 1996. Sketch of Seneca, an Iroquoian language. In: Handbook of North American Indians. Vol. 17, Languages, ed. by Ives Goddard, 225‑253. Washington, DC: Smithsonian Institution.

-----. 2005. Caddo. In: The native languages of the Southeastern United States, ed. by Heather K. Hardy & Janine Scancarelli, 323‑350. Lincoln, NE: University of Nebraska Press.

Cuoq, Jean André. 1866. Etudes philologiques sur quelques langues sauvage de l’Amerique. Montréal: Dawson Brothers. Reprinted 1966 by Johnson Reprint Corporation.

Fenton, William N. 1987. The False Faces of the Iroquois. Norman, OK: University of Oklahoma Press.

Lieber, Francis. 1837. Remarks on some subjects of comparative philology and the importance of the study of foreign languages. Southern Literary Messenger 3.161‑172.

Melnar, Lynette R. 2004. Caddo verb morphology. Lincoln, NE: University of Nebraska Press.

Mithun, Marianne. 1999. The languages of Native North America. Cambridge, UK: Cambridge University Press.

Author’s Contact Information:
Wallace Chafe
1010 Mission Canyon Rd.
Santa Barbara, CA 93105
USA
chafe@linguistics.ucsb.edu

^[1] Exceptions may be ambient events like rain and ambient states like hot.

^[2] See abbreviations at the end.

^[3] In this orthography vowels with umlauts are nasalized except that ä is a low front oral vowel. The acute accent shows a higher pitch, the colon shows vowel lengthening, and the apostrophe represents a glottal stop.

^[4] The comma at the end of (4b) indicates the rising pitch that is visible in Figure 2.