Pronominal Clitics and Indexability Hierarchies in Hanis and Miluk Coosan

Abstract

The subject and object pronominal clitics of the two Coosan languages, Hanis and Miluk, and their associations with inflectional affixes of the verb are examined on the basis of available text corpora, supplementing and correcting Frachtenberg’s original description of the Hanis forms. In Hanis, an additional proclitic position must be recognized in certain imperative and transitive clauses, differing syntactically from the position of the ordinary proclitics; also, the Algonquian-like hierarchy privileging second person over first person that holds for singular enclitics requires modification for nonsingular ones. In Miluk, the pronominal enclitics differ dramatically from the Hanis proclitics in syntactic position and in some cases also in shape, but in most other respects seem comparable in behavior, as do patterns of inflectional suffixation of verbs. Brief comparative remarks are offered on other languages of the southerly Pacific Northwest.

1. Introduction

Hanis and Miluk, the so-called Coosan languages of the southern Oregon coast, are extinct but have been reasonably well documented. Hanis was described in a long grammatical sketch by Leo Frachtenberg (1922a), who had earlier published a collection of Hanis texts collected by himself and by Harry Hull St. Clair (Frachtenberg 1913). Melville Jacobs subsequently published small collections of Hanis texts (Jacobs 1939:19–38, 63–70, 1940:133–36, 227–38) and a much larger corpus of Miluk texts (the remainder of Jacobs 1939, 1940), although he produced no grammatical description of Miluk.¹

The present article draws on these sources to examine subject and object pronominal clitics and their interactions with verb inflection in Hanis and Miluk. One principal purpose is to draw on the text corpus to extend the description of Hanis pronominal clitics beyond what is covered in Frachtenberg’s grammar. Since the corpus is closed, it is often impractical to pursue any given topic in depth; crucial evidence may be scantily attested or absent. Instead, I discuss an assortment of topics ranging over the domain of pronominal clitic grammar. Sections 2–5 focus on Hanis. The sketch of Hanis clause structure in section 2 provides background to the subsequent discussion of both languages. The Hanis pronominal clitic (specifically, proclitic) paradigm, the basic patterns of pronominal clitic usage, and the interaction of proclitics with suffixal inflection of the verb and with negation are presented in section 3. These patterns are reminiscent of Algonquian; notably, in clauses that involve both a first person and a second person participant, pronominal proclitics obey a hierarchy in which expression of second person takes precedence over first person, while special suffixation of various sorts is used in clauses with a first or second person object. Section 4 considers various complications that arise in imperative clauses and in transitive clauses involving nonsingular first and second person arguments: an additional proclitic position needs to be recognized, and simple second person prominence ceases to hold when nonsingular first person acts on second person. Possible approaches to more formal analysis of the Hanis clitics, including the role that indexability hierarchies might play, are examined in section 5.

Section 6 outlines the grammar of pronominal clitics in Miluk as I currently understand it. This will, I hope, facilitate more effective use of the Miluk text corpus for both synchronic and diachronic study—including the troublesome question of just what the nature of the relationship between Hanis and Miluk actually is. The wider areal distribution of Hanis and Miluk person marking patterns is considered programmatically in section 7; this includes the issue of what to make of the Algonquian-like aspects of those patterns.

Presentation of examples is modernized, rather than exactly reproducing the transcriptions of the sources. The representation of Hanis in examples taken from Frachtenberg’s work (examples with sources noted as "FCoosT", "FGr") differs somewhat from that in examples from Jacobs ("JH"); this and other issues of transcription are discussed in an appendix.

2. Basic clause structure in Hanis

The order of constituents of the clause is quite free in Hanis, although it is common for the verb to precede nominals and adpositional phrases. Mithun (1987) has noted some pragmatic effects of different orders—indefinite expressions tend to occur early, as proposed by Dryer (1983), though new topics, contrastive foci, and expressions whose referents are relatively salient or important do so as well. Constituent order freedom is facilitated by an ergative case proclitic = (Frachtenberg’s "discriminative" [1922a:324-25]) for nominal subjects of transitive verbs, as in (1) and (2).² As can be seen in (1), the ergative proclitic follows articles (as does the locative proclitic n= [Frachtenberg 1922a:323]); it also follows possessive pronominal clitics when these are present. Moreover, as shown in (2), it precedes quantificational and adjectival modifiers.

1. (1).

2. See–tr      art= food      art=ERG=  woman

3. ‘The woman saw the food.’ (FCoosT 64.16–17; FGr 324)

4. (2).

5. ERG=two  person  support–freq    adv=back=loc

6. ‘Two men supported him from the back.’ (FCoosT 40.9; FGr 425)

The ergative proclitic is not completely obligatory on subjects of transitives (at least those with human referents), as (3) shows; however, it never appears at all on the subjects of intransitive verbs, nor on objects, as (4) and (5) exemplify.

1. (3).

2. indeed two      person   support–freq

3. ‘Two men were supporting him.’ (FCoosT 40.56; FGr 424)

4. (4).

5. go.home       art=   man

6. ‘The man went home.’ (FCoosT 56.11)

7. (5).

8. old.man   person    see–TR

9. ‘He saw an old man.’ (FCoosT 22.25)

Transitive verbs are often, but not always, marked as such by a transitive, causative, or other suffix, as in examples above. Nominals can act as predicates without a copula, as in (6a) and (6b).

1. (6a).

2. money      art=3p.ps=   stake

3. ‘Their stake (in the game) is money.’ (FCoosT 93.21)

4. (6b). a?yú =g^w∂ éqe, . . .

5. indeed      =CJCTR    corpse

6. ‘That must have been a dead person, . . .’ (JH 23.2–3)

Some clause-level second position enclitics exist. The commonest is the future marker =han (Frachtenberg 1922a:384), seen in (7a) and (7b); also frequent is the customary marker =he, shown in (8). (When these two enclitics cooccur, they take a special contracted form =hanawe.)

1. (7a).

2. run.down–anticausative =fut      art= water

3. ‘That water shall run down.’ (FCoosT 16.9)

4. (7b).

5. Water–all  =fut      1d.incl=  go.down.to.water

6. ‘We two will go down into the water.’ (FCoosT 54.12)

7. (8). in

8. not  =cust    what  find(?)–tr(?)

9. ‘He never found anything.’ (JH 230.4)

"Second position" is a convenient oversimplification. Clause-initial connectives are not counted in determining second position; the connective does not host enclitics, while the word following it does. Connectives of this sort include ta ‘and’, as in (9a), and the demonstrative particle lau ‘then’, as in (9b) (though lau has various other uses); subordinators such as i ‘if, when’ do host enclitics, however.³

1. (9a).

2. and   there   =cust   fetch–freq(?)–impers

3. ‘And someone would go and get it.’ (FCoosT 92.13)

4. (9b).

5.       dem adv=thus   =cust   2S=rdp.inform–inv    art=erg= crow

6. ‘. . . (then) Crow customarily thus informs thee.’ (FCoosT 15.1–2)

It is also clear that second position enclitics can follow a phrase that consists of more than one full phonological word when that phrase is the first constituent of the clause, as in (10a) and (10b). (Phrases consisting of an article or possessive pronominal plus a noun are probably just a single phonological word. The phrases in (10a) and (10b) are more robust than that.)⁴

1. (10a).

2. small   house    =fut     2S=build–tr

3. ‘You shall build a small house.’ (FCoosT 68.28)

4. (10b).

5. all     when   =fut.cust   instr=dem–instr 2S=speak–resultative(?)

6. ‘You shall be talking with it (this language) all the time.’ (FCoosT 19.2)

I am aware of no evidence that enclitics such as =han and =he can be inserted in the middle of such phrases. Presumably, then, no role is played in Hanis enclitic placement by phonologically driven processes analogous to the "prosodic inversion" that Halpern suggests switches the order of an enclitic and a following word in Serbo-Croatian (Halpern 1995:3-5, 13-23). While the fact that Hanis enclitics require some sort of overt host to their left may reflect a prosodic requirement of these enclitics, the processes by which enclitic and host end up in their surface order could well be ordinary syntactic ones, such as movement of a phrasal constituent to a syntactic position to the left of the syntactic position occupied by the enclitic.⁵

3. The Hanis pronominal proclitics: the basic set and its functions.

Hanis has a set of pronominal proclitics used as both subjects and objects, presented in table 1. (The table follows Frachtenberg [1922a:321-22, 350-52], but notes some allomorphy detectable only in Jacobs’s data. Not all types of allomorphy are shown; see below for further discussion.) There is a three-way number opposition for all persons, with dual in addition to plural; exclusive and inclusive are distinguished in the first person dual, but not the first person plural. The third person singular form is zero (not normally shown in examples).

As Doty (2011, 2012) points out, variation in the phonetic forms of some of the proclitics as they appear in Jacobs’s texts, to a considerable extent, reflects systematic allomorphy: the first person singular proclitic appears as ()∂n= before coronal obstruents, while elsewhere it takes the shape ∂= (sometimes n∂=); the second person singular proclitic is normally e= (sometimes eh=) rather than e= before obstruents (this proclitic often displays the vowel quality a before word-initial a, and sometimes before other words with a in the initial syllable). Some less systematic variation appears in the first person dual exclusive, first person plural, and second person plural proclitics, where Jacobs sometimes records the vowel as ∂ rather than i; the first person dual exclusive also appears as x^wun=. None of these variants appear in Frachtenberg’s texts; for instance, he writes the first person proclitic consistently as ∂n= (〈〉, in his orthography) and the second person singular proclitic consistently as e=. I assume (in part from having worked with Frachtenberg’s Alsea material) that Frachtenberg was inclined to normalize the transcription of words and morphemes that he recognized, while Jacobs had a greater tolerance for recording variation-probably including both variation actually existing in speakers’ pronunciations, and variation in his own perceptions.

Table 1.

Hanis Pronominal Proclitics

Possessive pronominal proclitics within the NP, not examined in this article, are identical to the subject-object forms of table 1, except for second person singular ye()= and third person singular u= (Frachtenberg 1922a:396–99).

Subject and object uses of proclitics are discussed in sections 3.1-3.3. All the proclitics of the basic set in table 1 can express the subject. They can also express the object, but when the object is first or second person, various complications arise: the verb requires various kinds of special suffixation in addition to a proclitic, and there are restrictions on the combinability of first and second person proclitics; in most cases, a first person subject or object is not expressed by a proclitic when the verb also has a second person argument. A further difference between third person proclitics, on the one hand, and first and second person proclitics, on the other, is that the third person proclitics are optional (see section 3.4). Finally, negative clauses present some special features (see section 3.5): the proclitics take the negator as their host, often contracting with it, and it can be separated from the predicate proper.

3.1. Basic position and functions of the proclitics

In general, the Hanis pronominal proclitics are fixed on the predicate of the clause, wherever the predicate occurs; in (11a) and (11b), contrast the placement of the second position enclitic =han with that of first person ∂n=.

1. (11a).

2. 1S=go.home   =fut

3. ‘I shall go home.’ (FCoosT 146.12)

4. (11b).

5. Ocean–DIRECTION   =fut     1S=go.home

6. ‘I am going home to the ocean.’ (FCoosT 88.27–28)

Most often, the predicate is a verb. The proclitics may code the subject of either an intransitive verb, as in (11a) and (11b), or a transitive verb, as in (12a) and (12b).

1. (12a).

2. 1S=open–tr   art=      door

3. ‘I opened the door.’ (FCoosT 74.9)

4. (12b).

5. 2S=want–tr      =cond   =ynq

6. ‘Would you like her?’ (FCoosT 70.10)

They also code the subjects of nonverbal predicates, such as the adjective in (13a), the simple nouns in (13b) and (13c), and the nominal phrases (or compounds?) ‘what sort of person’ and ‘medicine-man’ in (13d).

1. (13a).

2. now   2S=good

3. ‘Now you are all right.’ (FCoosT 146.16)

4. (13b).

5. just   very   all      3P=   women

6. ‘They were all women.’ (FCoosT 50.10–11)

7. (13c).

8. 2S=black.bear      =fut

9. ‘You shall be a bear . . .’ (FCoosT 172.26)

10. (13d).

11. 2S=adv=how–whq   person   —   1S=medicine      person    =surely

12. ‘"What sort of person are you? " — "I am a medicine-man."’ (FCoosT 10.2)

There is reasonably good evidence that adjectives such as ‘good’ constitute a distinct part of speech, rather than being a type of stative verb. As modifiers within noun phrases, adjectives precede the modified noun, as in (14), while verbs normally form relative clauses that follow the noun, as seen in (15a) and (15b).

1. (14).

2. one   good  small     woman      dem   =fut     1p= give–dat    art=    spider

3. u=          temísne• č

4. 3.ps=     grandson

5. ‘We will give a pretty, little woman to Spider’s grandson.’ (FCoosT 70.3–5)

6. (15a).

7. Scent–TR–dat(?)          art=erg=    person   sleep

8. ‘The man who slept scented it.’ (FCoosT 102.8)

9. (15b).

10. just   evid     shake–(?)   in.appearance           art= board     lie

11. ‘It seemed as if the board that lay (there) shook.’ (FCoosT 58.21–22)

Inasmuch as the word that hosts a pronominal proclitic can be of any lexical category, provided that the word is serving as the syntactic predicate or as part of a phrasal predicate, the proclitics show "a relatively low degree of selection with respect to their hosts," a typical property of clitics according to Zwicky and Pullum (1983:503). On the other hand, they display some affix-like behavior in negative clauses (see section 3.5). For purposes of the present article, it is not crucial whether the forms in table 1 are best considered proclitics or prefixes; whichever they are, they are at any rate distinguishable from full-word independent pronouns (for some examples of the latter, see (32) and (33) in section 3.4), and the label "proclitic" is convenient.

As (16a) and (16b) and occasional examples elsewhere show, the pronominal proclitics are not possible hosts of second position enclitics such as =han ‘future’ and =he ‘customary’; if a verb with a pronominal proclitic begins the clause, the enclitic follows the verb rather than the proclitic.

1. (16a).

2. 2S=eat–(?)–detr   =fut

3. ‘You shall eat.’ (FCoosT 108.14–15)

4. (16b). ux^w= címsimt =he

5. 3d=    sleep.pl   =cust

6. ‘They two are sleeping.’ (FCoosT 74.1)

The third person proclitics (dual or plural, since third person singular is zero) can be used as objects, as in (17a)–(17c). As seen in (17b) and (17c), a third person proclitic expressing the object precedes a first or second person proclitic expressing the subject (Frachtenberg 1922a:351-52), although text examples are few.

1. (17a).

2. still   there  3p=watch–tr   all

3. ‘She kept watching them there.’ (JH 37.9)

4. (17b).

5. 3p=   2s=hit–tr

6. ‘you (sg.) hit them’ (FGr 351)

7. (17c). lau qac ux^w= ∂n=k^wináˀ–eiwat

8. dem   nonetheless   3d=     1S=see–FREQ(?)

9. ‘I just looked at them (two things).’ (FCoosT 62.22)

First and second person proclitics can express object as well, but require special suffixal morphology on the verb (see sections 3.2 and 3.3).

The third person proclitics are not obligatory in either subject or object function (see section 3.4).

3.2. Inverse forms

When a third person acts on a first or second person (a "speech act participant"), the verb takes an inverse suffix –u or –i (both written with long vowels by Frachtenberg) and a pronominal proclitic expresses the undergoer (Frachtenberg 1922a:350–52); examples of –u are seen in (18a)–(18c), and examples of –i in (19a) and (19b).

1. (18a).

2. dem dangerous.being   =fut     2S=take–tr–inv

3. ‘Then (if I do that) a dangerous thing will take you.’ (JH 22.3)

4. (18b).

5. 1S=work–tr–inv   =fut     art=1s.ps=   erg=son.in.law

6. ‘My son-in-law shall work on me (to cure me).’ (FCoosT 128.20)

7. (18c). . . . , i x^win= áiw–it–u•

8.           if/when   1d.excl=    kill.pl–tr–inv

9. ‘. . . , when they (pl.?) killed us two.’ (FCoosT 120.23)

10. (19a).

11. and   1s=rdp.beat–inv

12. ‘and he (my husband) beats me’ (JH 230.18)

13. (19b).

14. erg=giantess   1p=     scare(?)–inv

15. ‘The (two) Giant Women scared us.’ (FCoosT 84.9–10)

Just what determines the choice of inverse suffix is unclear. Frachtenberg points out that –u normally follows a transitive suffix such as –(V)t or –(V)c, while –i does not. He further suggests that forms with transitive suffix and –u are used for "actions that have been performed once, or that are completed" (perfective actions, approximately), while forms without transitive suffix and with –i are used for an "action that has been performed more than once, or that has not been completed" (imperfective actions, approximately) (Frachtenberg 1922a:352). This may not account for all examples, however.⁶

When the third person actor of an inverse clause is nonsingular, a third person dual or plural proclitic, as in (20), may appear; as in direct (i.e., noninverse) clauses with a third person object, the third person proclitic precedes the first or second person proclitic (Frachtenberg 1922a:351–52).

1. (20).

2.         dem thus   3p=1s=do–tr–inv

3. ‘. . . , that is what they did to me.’ (JH 22.6)

Only a few textual examples of inverse clauses with an overt third person proclitic can be found, however, and it is clear that it is optional, like third person proclitics in other contexts; compare (19b) above and see the discussion in section 34.

I am not aware of any evidence that actor and undergoer in inverse clauses bear different syntactic relations than they do in direct clauses. Actor nominals in both types of clause are marked by the ergative proclitic =, as is seen for inverse clauses in (18b) and (19b); while ergative marking in inverse clauses, such as (18a), is apparently optional, it is also optional in direct clauses (see section 2). While I do not rule out the possibility that some subtler evidence might emerge that actors in inverse clauses are not subjects, for present purposes I use the terms "subject" and "actor" interchangeably when speaking of transitive clauses, and likewise the terms "object" and "undergoer".⁷

3.3. Interactions of speech act participants

Special suffixes mark action of second person on first person and vice versa: –a(•)is ‘second person subject—first person object’ (2→1), as in (21), and –a•mi ‘first person subject-second person object’ (1→2), as in (22). A suffix –∂m is used instead of –a(•)is when the object bears a dative or benefactive relation to the verb.⁸ Regardless of which is subject/actor and which is object/undergoer, the second person pronominal proclitic appears (Frachtenberg 1922a:350-52). This implies that a hierarchy operates among the pronominal proclitics in which second person outranks first person. (Complications arise when the subject/actor is dual or plural; see section 4.2 and section 5 for fuller discussion of issues relating to person hierarchies.)

1. (21).

2. 2S=take.home–caus–2→1    =fut

3. ‘You shall carry me home.’ (FCoosT 30.12)

4. (22).

5. now    =fut     head–loc       2S=club–tr–1→2

6. ‘Now I will hit you over the head.’ (FCoosT 66.2)

I suspect that –a(•)is and –a•mi originally were simply object suffixes for first person and second person, respectively, and that they became limited to interactions between first and second person because the expression of third person acting on first or second person was taken over by the inverse construction.⁹

3.4. Optionality of the nonsingular third person proclitics

While Frachtenberg does not specifically note the point, it is clear from textual data that the third person proclitics are not obligatory. As subjects, they tend to be omitted when a subject nominal expression (underlined) is present, as is seen in the intransitive clauses in (23a)-(23b) and the transitive clauses in (24a)–(24b).

1. (23a).

2. adv(?)   =below person   Ø=   dwell

3. ‘People were living down below.’ (FCoosT 36.11)

4. (23b).

5. thus   Ø=   speak.pl–prg   art=    men.pl

6. ‘Thus the men were speaking.’ (FCoosT 56.20)

7. (24a).

8. erg=two   person   Ø= support–freq       adv=back=loc

9. ‘Two men supported him from the back.’ (FCoosT 40.9)

10. (24b).

11. then   Ø=  see–freq    art=  erg= people    art=   wonderful(?)   rock

12. ‘(Since that time) the people have seen these wonderful rocks.’ (JH 69.4–5)

(In many of the clear examples of clauses with an overt nonsingular NP as subject/actor, the NP has me• ‘person, people’ as its head noun. However, other head nouns can occasionally be found, showing that neither occurrence nor nonoccurrence of a cross-referencing pronominal proclitic for subject/actor is an idiosyncrasy of me•.)

The third person proclitic may be present, however, "doubling" the subject NP, as is seen for the intransitive clauses in (25a)–(25b) and for the transitive clauses in (26a)–(26b).

1. (25a).

2. there   3d=       arrive   art=   grandsons

3. ‘His grandchildren arrived there.’ (FCoosT 20.12)

4. (25b).

5. one       morning   many   young.man        3p= together–inch     art= person

6. 

7. LOC–village

8. ‘One morning many young men came together, people from the village.’ (FcoosT 76.23)

9. (26a).

10. adv=thus evid(?) 3d=     tell   art=erg=   old.people.pl

11. ‘Thus the two old people said to him.’ (FCoosT 132.7)

12. (26b).

13. many   person   3p=   shoot–freq

14. ‘Many people were shooting at her.’ (FCoosT 160.16)

Pending further investigation, I suspect that a subject nominal expression tends to be doubled by a subject proclitic when the referent is specific or topical, as has been claimed for object clitics in some Slavic languages (e.g., Franks and King 2000:251–58).

In the absence of an overt subject nominal expression, it is evidently normal (perhaps required) for a third person dual or plural proclitic expressing the subject to appear, as is seen for intransitive, direct transitive, and inverse transitive clauses in (27a)–(27d).

1. (27a).

2. now   indeed   outside   3p=   go.out

3. ‘So they went outside.’ (FCoosT 50.11)

4. (27b).

5. indeed   3d=      go.down

6. ‘Surely they two went down (into the water).’ (FCoosT 54.16)

7. (27c).

8. indeed   fire        p=see–tr

9. ‘ Sure enough they saw fire, . . .’ (JH 229.9)

10. (27d).

11.          then thus     3p=1s=do–tr–inv

12. ‘. . . , that is what they did to me.’ (JH 22.6)

For objects of transitive verbs, on the other hand, third person proclitics are normally avoided, even when the clause contains no overt nominal object, as in (28a) and (28b).¹⁰

1. (28a).

2. there   =fut       2s=see–tr

3. (Context: ‘Perhaps they (pl.) play there.) ‘You will see them there, . . .’ (FCoosT 98.13-14)

4. (28b).

5. Want–tr       art=erg=   man

6. (Context: ‘The (two) women were pretty.’) ‘The man liked them.’ (FCoosT 126.17)

Third person object proclitics are not absolutely prohibited, however, as (29a) and (29b) show; since the evidence is limited, it is not clear what factors favor or disfavor the presence of a proclitic.

1. (29a).

2. still   there   3p=watch–tr all

3. ‘She kept watching them there.’ (JH 37.9)

4. (29b).

5. dem nonetheless 2d=     1S=see–freq(?)

6. ‘I just looked at them (two things).’ (FCoosT 62.22)

Two nonsingular third person proclitics evidently never appear on one predicate, even when subject and object of the clause are both nonsingular third person, as in (30a) and (30b).

1. (30a).

2. adv=   thus   =cust   3p=tell art=   children

3. ‘That is the way they tell their children.’ (JH 68.5–6)

4. (30b).

5. now     3d=see–tr        art= youths art=  erg=  girls person

6. ‘Now the (two) girls saw the (two) young men.’ (JH 236.6–7)¹¹

In the few clauses in the corpus, such as (31), in which the subject and object are both nonsingular third person, but of different numbers, a proclitic that appears expresses subject rather than object-which is not surprising, given that third person objects tend not to be expressed by proclitics in any case.

1. (31).

2. even.if two       person go nonetheless(?)    3P=    kill–tr

3. ‘Even if two persons passed by, still they (five Grizzly brothers) would kill them.’ (FCoosT 90.10)

As far as I can determine, first and second person subject proclitics are never omitted, even when an independent pronominal subject is present, as in (32a)–(33). There seem not to be clear examples in the corpus in which an object is represented by a first or second person independent pronoun, so I cannot say for certain that pronominal proclitics for the object are retained in this context—though I expect that they would be.

1. (32a).

2. also   =fut      1s.indp   there    1S=go

3. ‘I too will go there.’ (FCoosT 94.22)

4. (32b).

5. 1S=see–tr      =fut     1s.indp.erg

6. ‘I want to see him.’ (FCoosT 180.6)

7. (33).

8. also   =fut      2.indp.erg  one       2s=shoot–tr

9. ‘You too ought to shoot one (arrow).’ (FCoosT 13.1)

This might, of course, simply reflect the fact that first and second person arguments are intrinsically specific.¹²

3.5. Negative clauses

In negative clauses, pronominal proclitics appear on the negator i•n (Frachtenberg) ~ in (Jacobs) rather than on the verb or other predicate, as is seen in (34a)–(34d).

1. (34a).

2. and  dem  evid  3p=      not kill–tr art=   sea.otter

3. ‘So they did not kill the sea-otter.’ (FCoosT 176.27-28)

4. (34b).

5. 3d=      not   know–tr

6. ‘They (two) did not know.’ (FCoosT 74.23)

7. (34c).

8. dem   ever(?)   1d.excl.not   give–dat

9. ‘. . . , we never give them anything.’ (JH 237.2)

10. (34d).

11.       if=fut   2S.not=hurt–tr–2→1

12. ‘. . . if you do not hurt me.’ (JH 134.2–3)

Several of the proclitics form special contractions with the negator (Frachtenberg 1922a:314); the paradigm of negative forms is shown in table 2, as far as it is known. (First person dual inclusive and second person dual forms do not seem to be attested, although there is little doubt about what they ought to be.) Some of the contractions reflect systematic processes, though not ones that appear in other contexts in the language as far as I am aware. Proclitics ending in n all lose the n before the negator. (The glottal stop that appears to replace the deleted n is presumably simply the word-initial glottal stop of the negator i(•)n, left unwritten in other contexts. Frachtenberg’s spellings 〈xwî^în〉, 〈〉, and 〈〉, may imply reduction or loss of the vowel i(•) of the negator, and my retranscriptions of examples from his data render it as such; however, Jacobs indicates that the negator’s vowel is retained.¹³) The second person singular form, too, looks like a straightforward phonological reduction of the regular second person singular proclitic e= plus the negator i(•)n, though again I am not aware of other contexts in the language in which this particular reduction occurs. The first person singular form ni•, however, is a special portmanteau form, not explicable as the output of any ordinary phonological reduction of first person singular ∂n= plus negative i(•)n.

Table 2.

Hanis Combinations of Pronominal Proclitics with Negator

The fact that the proclitics form morphophonologically idiosyncratic combinations with a negator host is, of course, an affix-like rather than clitic-like trait according to Zwicky and Pullum (1983:504). On the other hand, the fact that the negator rather than the verb hosts the proclitic is a further instance of the proclitics being unselective as regards the lexical category of their host, a clitic-like property. (Recall that, besides the verb and the negator, nonverbal predicates of various categories-including phrases-host proclitics, as is pointed out in the discussion of (13) in section 3.1.) I therefore continue to speak of these pronominals as proclitics.¹⁴

A property of the negator not explicitly discussed by Frachtenberg but robustly exemplified in the text corpus is that, while the negator (along with its associated proclitics) always precedes the verb, it need not immediately precede it. That is, the negator is not simply another proclitic placed between the pronominal proclitics and the verb. Negator and verb can be separated not only by enclitics, as in (35a) and (35b), but also by more substantial material, such as the adverbs asú•‘again’ in (36a) and (36b) or či•č ‘somehow, anyhow’ in (36b).

1. (35a).

2. 2S.not   =fut      hurt–tr–1→2

3. ‘I won’t hurt you.’ (FCoosT 180.27)

4. (35b).

5. dem   3p    1S.not   =fut      see–inv

6. ‘They shall not see me . . .’ (FCoosT 128.23-24)

7. (36a).

8. 2S.not   =fut      again   see–tr–2→1

9. ‘You will not see me again.’ (FCoosT 56.27)

10. (36b).

11. 1S.not  how  again  return

12. ‘I cannot come back.’ (FCoosT 56.24)

Pronominals, nouns, or even short nominal phrases can appear between negator and verb, as in (37a)–(37c).

1. (37a).

2. not   where   what   see–freq(?)

3. ‘Nowhere could he see anything.’ (FCoosT 22.12–13)

4. (37b).

5. at.all   still      3p=   not herring   see–tr

6. ‘Still they (pl.) did not see any herring.’ (FCoosT 32.2-3)

7. (37c).

8. 1S.not   =fut   poor         people(?)–loc      be.among

9. ‘I will not mingle with poor people.’ (JH 65.1-2)

Proclitics that are separated from the verb in negative clauses relate to the suffixal morphology of the verb in the same way as proclitics in nonnegative clauses: when first person acts on second person, the negator takes a second person proclitic and the verb takes the suffix –a•mi, as in (35a); when first person acts on second person, the negator takes a second person proclitic and the verb takes the suffix –a(•)is, as in (36a); when third person acts on first, the negator takes a first person proclitic and the verb takes an inverse suffix, as in (35b); and so on.

Nonenclitic material between the negator and the verb is generally within the scope of negation. The narrative context of (36a), for instance, makes it clear that the speaker-a woman who is ending a visit to her family to return permanently to the seals with whom she is now living-means ‘there will be no occasion on which you see me again’, not ‘there will again be an occasion on which you do not see me’. Nominals, pronominals (e.g., di•() ‘something, anything’), and adverbs of manner and location (e.g., qanč ‘somewhere, anywhere’, či•č ‘somehow, anyhow’) in this position are typically nonspecific or nonreferential, as in the above examples.

Adverbs and indefinite nominals can precede the verb in nonnegative clauses, too, as seen in (38a) and (38b); in some cases, such as (39), they clearly have nonspecific or nonreferential interpretations analogous to those in negative clauses. But in the nonnegative cases, the pronominal proclitics attach directly to the verb rather than precede the indefinite nominal or adverb.

1. (38a).

2. again   =fut   up      2S=run(?)

3. ‘You shall run up again.’ (FCoosT 92.6–7)

4. (38b).

5. person      1S=see–TR

6. ‘I saw a person (spy).’ (JH 36.2)

7. (39).

8. if=fut.cust   person   2S=see–freq(?)

9. ‘When you see a person, . . .’ (no particular person is meant) (FCoosT 19.5)

A usefully close contrast is seen in (40) and (41). In both, the indefinite adverbial ‘somehow, anyhow’ (a fuller variant of occurs before the verb and after an operator-like element—the negator in (40)¹⁵ and the modal particle naci• ‘it is doubtful that . . .’ in (41). And in both examples, the indefinite adverbial is clearly not referential. (Though Frachtenberg did not directly render it in his own free translations, it evidently means something like ‘in any way’ or ‘at all’, within the scope of the operator-like element and emphasizing the impossibility or unlikeliness of the event.) But in (40), the first person plural proclitic attaches to the negator and precedes the indefinite adverbial, while in (41) the same proclitic follows the indefinite adverbial and attaches to the verb.

1. (40).

2. 1P.not   expected   adv=how   seize–tr

3. ‘We cannot seize her at all.’ (FCoosT 56.18)

4. (41).

5. doubt   adv=how   1p=    seize–tr

6. ‘We doubt whether we shall catch her at all.’ (FCoosT 56.19-20)

In other words, separation of the pronominal proclitics from the verb in negative clauses is clearly due to the presence of the negator, not to the presence of nonspecific or nonreferential expressions before the verb. Possibly, then, informally speaking, the syntactic position of the proclitics differs in positive and negative clauses: immediately before the verb in a positive clause, but in some sort of earlier position, preceding the negator and indefinite material, when the negator is present. (In a movement-based syntactic framework, one might wish to explore the possibility that the proclitics are base-generated immediately before the verb, but move up to the negator in negative clauses.)¹⁶

4. An additional proclitic position in Hanis.

Besides the positions occupied by the proclitics of the basic set as discussed in section 3, an additional proclitic or prefixal position must be recognized; it is occupied by a second person morpheme that resembles, but still must be distinguished from, the second person singular proclitic of the basic set. Such a second person morpheme appears in certain imperative clauses (see section 4.1), and in transitive clauses where nonsingular first person acts on second person nonsingular (see section 4.2). In the transitive clauses, moreover, the otherwise valid generalization that second person prevails over first person in proclitic choice partially breaks down.

4.1. Imperatives.

Hanis has a variety of imperative formations. For present purposes, many details are not crucial; a rapid survey suffices.

Some intransitive imperatives, such as those in (42a) and (42b), are marked by suffixes (Frachtenberg 1922a:348).

1. (42a).

2. Go–imv

3. ‘Go on!’ (JH 25.10)

4. (42b).

5. just   eat–imv

6. ‘Just eat!’ (FCoosT 104.13)

For transitive imperatives with third person object, as in (43), a suffix –∂ is used for most transitives (Frachtenberg 1922a:347; the suffix also appears as –a or –e in Jacobs’s material).

1. (43).

2. now     again   shoot–tr–imv

3. ‘Shoot it again!’ (FCoosT 13.3–4)

Some transitive imperatives with first person object (not of a dative or benefactive sort) use a special variant of the second-person-acting-on-first-person suffix, namely, -∂is, as in (44), at least according to Frachtenberg (1922a:349).¹⁷ (This form does not appear in Jacobs’s texts.)

1. (44).

2. art=1s.ps   loc=belt     hold.on–1→2.imv

3. ‘Hold me by this my belt.’ (FCoosT 54.12)

However, sometimes the ordinary form of this suffix, -a(•)is, appears in imperatives, as seen in (45).

1. (45).

2. crossways   club–TR–2→1

3. ‘Hit me crossways.’ (FCoosT 64.28–29)

As seen in the examples above, with all these types of imperative, the second person singular proclitic e= is not used. Transitive imperatives with first person object, as in (44) and (45), have no first person proclitic, either (see section 5 for discussion).

Dual or plural subjects of imperatives of the above types are marked by a second person dual or plural proclitic, in addition to the imperative suffix (if any), as is seen for intransitive verbs in (46a) and (46b) and for a transitive verb in (47).

1. (46a).

2. a.little.bit   towards.fire/river   2d=   dance–imv

3. ‘You two dance somewhat closer (to the fire)!’ (FCoosT 82.19)

4. (46b).

5. now     2p=   go–imv

6. ‘Now go!’ (FCoosT 32.6)

7. (47).

8. dem   2d=   eat–tr–imv

9. ‘You two eat this!’ (FCoosT 120.16; FGr 347)

As far as can be determined, in negative imperative clauses, the second person dual or plural proclitic attaches to the negator, as in (48) (unfortunately, this may be the only relevant example in the corpus).

1. (48).

2. 2P.not   eat–tr–imv   art=   lunch

3. (Then he said to his relatives:) ‘Don’t you eat this lunch.’ (FCoosT 148.15)

For most intransitives, no special imperative suffix is used; a proclitic (or prefix) e= appears instead (Frachtenberg 1922a:347), as in (49a) and (49b).

1. (49a).

2. nearby   imv=play

3. ‘Play close by here!’ (FCoosT 60.21)

4. (49b).

5. now     imv=sleep

6. ‘Now you sleep!’ (FCoosT 102.3)

This imperative proclitic looks like the ordinary second person singular proclitic e=, but must be distinguished from it on at least two grounds. For one thing, dual and plural imperatives of this sort combine imperative e= with the second person dual and plural proclitics (Frachtenberg 1922a:321), as seen in (50) and (51); in nonimperative clauses, the dual and plural second person proclitics do not combine with second person singular e=.¹⁸

1. (50).

2. 2D=imv=   stand

3. ‘You two stand up!’ (FCoosT 120.15)

4. (51).

5. 2p=imv=   get.up

6. ‘You (pl.) get up!’ (FCoosT 68.21, 122.8–9)

For another thing, in negative intransitive imperatives, the element e= consistently attaches to the verb rather than to the negator, as is shown in (52a)–(52c); the ordinary second person singular proclitic e= attaches to (and contracts with) the negator (see section 3.5).¹⁹

1. (52a).

2. not   how   imv.intr=do.rdp(?) grandson

3. ‘Don’t you do anything, grandson.’ (FCoosT 100.25)

4. (52b).

5. not   imv.intr=(be).here

6. ‘Don’t stay here!’ (JH 32.1–2)

7. (52c).

8. not   all    how   imv.intr=speak–rfl

9. ‘do not be talking every which way!’ (JH 238.3)

One would certainly like to know what happens in negative dual and plural imperatives of predicates that form the imperative with e=, but there do not seem to be any relevant examples in the corpus.

As Frachtenberg notes, imperative morphology also cooccurs with the hortative particles, first person dual (inclusive) k^wis (in Jacobs’s transcription, g^w∂s) and first person plural k^win, seen in (53a)–(53b) and (54), respectively;²⁰ these particles historically must represent contractions of some modal element with the corresponding pronominal proclitics (1922a:393). Unlike the ordinary pronominal proclitics, while they precede the verb, they need not do so immediately; various kinds of adverbs, at least, may intervene. (Too few examples are attested to determine what the range of possible intervening elements is. In all of them, the hortative particles are clause-initial.)

1. (53a).

2. hort.1d.incl   now     again   go–imv

3. ‘Now let us two go again.’ (FCoosT 124.22)

4. (53b).

5. hort.1d.incl   chop–tr–imv

6. ‘Let us two chop wood!’ (FCoosT 26.15–16)

7. (54).

8. hort.1p   quickly kill–imv

9. ‘Let us kill him quickly!’ (FCoosT 68.3; FGr 393)

While in most examples, the verb is of a type that marks imperative by a suffix, a single example (55)–probably the only hortative example in Jacobs’s texts-evidently has the imperative proclitic e= (here realized as a=).²¹

1. (55).

2. come!   hort.1d.incl=imv.intr=play

3. ‘Come! let us play a game.’ (JH 29.6)

4.2. Interactions of nonsingular speech act participants.

In a transitive clause where first person acts on second person, or vice versa, and both arguments are singular, the pronominal proclitics follow a person hierarchy such that a second person proclitic appears regardless of whether the second person argument is the subject/actor or the object/undergoer (see section 3.3). However, in some of the cases where one or both of the arguments is dual or plural, a first person proclitic appears.

When first person singular acts on second person dual or plural, the hierarchy is followed, as in (56) and (57); a second person proclitic of the appropriate number appears, but no first person proclitic.

1. (56).

2. just   2D=   tease(?)–tr–1→2

3. ‘I was just playing with you (two).’ (JH 228.8)

4. (57).

5. this.here   behind always     =fut meat   2p=    give.food–1→2

6. ‘After this I will always give you (pl.) meat.’ (FCoosT 166.18–19)

The hierarchy is also followed when second person dual or plural acts on first person singular, as in (58).²²

1. (58).

2. 2p.not   =fut    again   see–TR–2→1

3. ‘You (pl.) will not see me again.’ (FCoosT 130.21)

It is also followed when nonsingular second person acts on nonsingular first person-or at any rate that is what (59) indicates, though unfortunately it seems to be the sole clear attestation of such clauses in the corpus.²³

1. (59).

2. if=fut      2d=want–tr–2→1 2d=inform–2→1 =fut      adv=how 2d=do.rdp(?)

3. ‘If you (two) want us (two), you must explain to us what you do.’ (JH 236.12)

Clauses in which singular second person acts on dual or plural first person seem to be unattested in the corpus.

When first person dual or plural acts on second person, however, the hierarchy of section 3 is not followed. A first person dual or plural proclitic appears, along with a special second person object proclitic (or prefix) e=, as seen in (60) and (61).

1. (60).

2. now =fut    1.indp.erg 1d.excl= 2.ob=tell–1→2

3. ‘When we two tell you (sg.), . . .’ (FCoosT 126.21-22)

4. (61).

5. food    =fut    1p=2.ob=give.food–1→2

6. ‘We will give you (sg.) food, . . .’ (JH 229.20)

The e= in these forms, like the imperative e=, at first glance looks like the ordinary second person singular proclitic e= (and in Jacobs’s material it shows similar allomorphy, with final lost before obstruents). In fact, however, second person object e= must be distinguished from the ordinary second person singular proclitic. For one thing, the e= that appears with nonsingular first person subject/actor is indifferent to the number of the object/undergoer; e=, rather than a second person dual or plural proclitic, appears even when the object is dual or plural, as in (62) and (63).²⁴

1. (62).

2. 1d.excl=2.ob=want–tr–1→2

3. ‘We (two) want (as wives) you (two).’ (JH 236.11)

4. (63).

5. just   1p=   2.ob=visit–tr–1→2

6. ‘We just came to see you (pl.).’ (FCoosT 130.19-20)

Moreover, it also seems that in negative clauses this e= is fixed on the verb, like imperative e=, rather than being attached to the negator like ordinary pronominal proclitics, while the nonsingular first person proclitic appears in the normal proclitic position before the negator. Unfortunately, the evidence is limited to the single example in (64).

1. (64).

2. 1p.not  2.ob=want–TR–1→2

3. ‘We do not want you (sg.?)!’ (JH 32.1–2)

In short, the e= that appears in the context of a nonsingular first person subject/actor is a special element comparable to the e= in imperatives. (Presumably, though, it is not a coincidence, at least diachronically, that these various distinguishable second person forms share the shape e=.)²⁵

5. Indexability hierarchies and related issues in Hanis.

I do not attempt a formal analysis of the Hanis proclitics and the correlated verb morphology here. Since there are gaps in the data, especially for the phenomena surveyed in section 4, there may well not be enough evidence to conclusively support one analysis over others. Instead, I identify some things that a formal analysis would need to account for—especially, but not exclusively, the question of what role indexability hierarchies might play among the principles governing proclitic occurrence in Hanis.²⁶

The paradigmatic positions of pronominal proclitics and other person-marking morphology are summarized in the informal template in figure 1. Third person proclitics (dual and plural only, since third person singular is zero), in slot A, precede the ordinary first and second person proclitics, in slot B; the negator and material within its scope intervene in slot C between these elements and the special imperative marker of section 4.1 or the special second person object marker of section 4.2 (slot D), the verb and its derivational morphology (slot E), and the verb’s inflectional suffixes (slot F).²⁷ The issue of an indexability hierarchy arises for several slots within this template, and these need to be considered separately; as Macaulay (2009:358) points out for Algonquian, it is more useful to think in terms of local hierarchies relevant to particular parts of the system, rather than a global hierarchy valid for the whole language.

Figure 1.

Informal template for Hanis proclitics and person marking.

For the inflectional suffixes, a simple hierarchy where first and second person outrank third person is relevant: an inverse suffix –u or -i is used when third person acts on first or second, but not otherwise (see section 3.2). This hierarchy plays no role for the proclitics; third person proclitics (slot A of the template) can freely cooccur with other proclitics. Although third person proclitics are sometimes omitted, this has nothing to do with what the other arguments in the clause are: third person proclitics tend to be omitted when they are objects, regardless of what the subject is, and also when they would double an independent NP in the same argument role, regardless of what the person of the other argument in the clause is (see section 34).

The suffixes –u and -i are not used for combinations of first and second person. There is no strong reason to think of either the second-person-acting-on-first suffixes -a(•)is and –m (dative) or the first-person-acting-on-second suffix –a•mi as inverse. True, in many clauses with –a•mi, the proclitic occupying slot B represents the object/undergoer, as it does in clauses with –u or -i, and this might suggest that –a•mi is a sort of inverse marker, as is sometimes suggested for first-person-acting-on-second-person "thematic" suffixes in Algonquian languages. However, the correlation of–a•mi with an object proclitic in slot B does not hold in all cases; in some –a•mi clauses the slot B proclitic is the subject/actor—namely, when the subject/actor is dual or plural first person (see examples (60)–(64) in section 4.2). Hence, I reject the idea that –a•mi is an inverse.²⁸ More generally, the choice of proclitic is best seen as independent of the choice of suffix.²⁹

Turning to the first and second person proclitics, there is a sense in which second person is uniformly prominent over first person in Hanis: in any nonimperative clause containing a second person argument and a first person argument, some sort of second person marker will occur among the proclitics-either one of the ordinary second person proclitics in slot B of the template, or the special second person object marker in slot D-whereas a first person argument is in many cases left unrepresented among the proclitics in the presence of a second person argument. (Slot D is in part a place where second person is morphologically represented when it is excluded from slot B.)

More narrowly, there is the matter of determining whether it is a first person argument or a second person argument that is represented in slot B, where the two are mutually exclusive. An indexability hierarchy that references only person, and in which second person outranks first (2 > 1)-meaning that slot B is always occupied by a second person proclitic if the clause contains a second person argument-accounts for many clauses in Hanis, but fails when a nonsingular first person acts on second person (again, see examples (60)–(64)). One might try to account for these clauses by revising the indexability hierarchy so that dual or plural first person outranks second person, while second person continues to outrank first person singular: 1D/1P > 2 > 1S. However, this reformulation of the hierarchy is not adequate either, if we can trust the evidence of (59). There, in a clause in which the nonsingular first person is the object rather than the subject, a second person rather than a first person proclitic appears in slot B. Example (59), especially its first clause, forms a near-minimal contrast with (62). The clauses are identical in terms of the person and number features of their arguments-all are transitive clauses with a first person dual argument and a second person dual argument-but which of these arguments is represented by a slot B proclitic depends on which one is the subject. While it is unfortunate that conclusions have to rest on just two crucial examples, still, as far as the evidence goes, it seems that the principles that determine whether slot B is occupied by a first person or a second person proclitic must refer to the grammatical relations of the arguments, not just to an indexability hierarchy that takes account only of their referential properties such as person and number. Informally, the rules for slot B in nonimperative clauses are as follows: it is occupied by a nonsingular first person proclitic expressing the subject; if there is no such subject, then it is occupied by a second person proclitic if any argument is second person; and if there is no second person argument, it is occupied by a first person proclitic if any argument is first person.

Transitive imperatives with singular subject and a first person singular object (see examples (44) and (45), and discussion in section 4.1) clarify further the nature of the principles that determine occupancy of slot B. In nonimperative clauses, one might think that second person prominence–the failure of singular first person arguments to be represented by a proclitic in a transitive clause in which the other participant is second person–results from competition between first and second person proclitics to occupy slot B, with a first person proclitic being blocked when a second person proclitic has preempted the slot. But imperative clauses such as (44) and (45) rule out this analysis: no second person proclitic appears, but neither does a first person proclitic. Rather, insertion of a first person singular proclitic in slot B is evidently prohibited when the subject of the clause is second person, regardless of whether that second person subject is coded by a proclitic or not. This implication is clear even though no information is available as to what happens when the first person object of an imperative is dual or plural.³⁰

Another sort of complication that a more formal account would need to deal with is the fact that the pronominal proclitics (slots A and B) can be syntactically separated from the verb, at least in negative clauses (see section 3.5). The principles of proclitic occurrence, therefore, are not simply rules for the phonological exponence of morphological features of the verb. (As is seen in section 6, this problem is even more acute in Miluk, where the pronominal clitics are always second position enclitics, and thus can never be analyzed as part of the verb morphology.)³¹

It is clear, then, that while prominence of second person over first person figures among the factors governing proclitic choice and occurrence, it cannot be the whole story. Those who have carefully studied person and number marking in the Algonquian family are, of course, well aware that what is sometimes alleged to be a straightforward system of second person prominence in those languages is, in fact, considerably more complicated than that,³² but complications arise, too, in the morphologically simpler system of Hanis.

Algonquian patterns of person marking are an obvious typological point of reference in the discussion above. Section 7 raises the question of whether these similarities of pattern, some of which extend to Miluk as well (see section 6), might have diachronic significance.

6. Miluk

The only published sources from which substantial information on Miluk grammar or lexicon can be derived are the text collections published by Melville Jacobs (1939, 1940); discussion here is based on these texts. It should be borne in mind that Annie Miner Peterson, from whom Jacobs obtained this corpus in 1933 and 1934, was a last or near-last speaker of Miluk whose primary language in her adult life had been Hanis (e.g., Jacobs 1939:4). (She was also the source of Jacobs’s smaller corpus of Hanis texts.) Grant briefly discusses how reliable Mrs. Peterson’s picture of Miluk may have been, concluding that it seems consistent with that seen in other sources, to the extent that the very limited nature of the latter make it possible to judge (1997:149). My own assessment is that, as far as the phenomena focused on in the present article go, Mrs. Peterson’s Miluk on the whole seems internally consistent and also displays regular patterns of similarities to and differences from the corresponding Hanis phenomena; it seems safe enough to take it as a reliable source of information about Miluk in these respects. How far that holds for other aspects of her Miluk remains to be determined.³³

It is problematic just how Miluk is related to Hanis. Grant observes that "the extant data indicate that Miluk Coos was different in many structural respects (for instance, in clitic placement) from Hanis. It also differed greatly in lexicon. As a rule, words in Miluk are identical, or almost so, to their Hanis counterparts, or else are completely different" (1997:149); "Hanis and Miluk Coos . . . were unambiguously separate, and mutually unintelligible, languages, despite sharing a majority of their free and bound morphemes" (1997:150). Exactly how far apart the languages were remains difficult to assess. Grant’s estimate that "the time depth between Hanis and Miluk was probably on the order of that between Upper and Lower Chinookan or between the three branches of Kalapuyan" (1997:150) seems too shallow. Pierce, on the other hand, had earlier argued that the divergences between Hanis and Miluk were so great as to imply that the the languages were not actually related to each other, but had converged due to their geographical proximity (1965:325).³⁴ Recently, Doty (2011, 2012) has expressed similar doubts about whether Miluk is related to Hanis, arguing further that Miluk is related to the Salish family. My working assumption remains that Miluk is related to Hanis, given that a fair amount of the grammatical machinery of the two languages–some of the suffixes of verbs, in particular–looks quite similar. The relationship, however, may well not be particularly close (contra Grant, and contra Thompson and Kinkade [1990:42]), given that some Miluk grammatical morphology, such as the subject-object clitics discussed below, diverges sharply from the functionally equivalent elements of Hanis; contact is likely to be responsible for many obvious resemblances between the languages, such as shared vocabulary and their largely identical syntactic patterns. (And, while Doty’s work to date has not convinced me that there is an actual genetic relationship between Miluk and Salish, he certainly adduces much evidence that suggests some sort of direct or indirect historical connection.) A better understanding of the historical position of Miluk will require a thorough synchronic investigation of Miluk morphology (especially its verb morphology) and lexicon. (See Doty [2012] for some initial steps in this direction.)

While my understanding of Miluk morphology remains provisional, especially as regards the segmentation and identification of suffixes in the verb other than those explicitly discussed below, this should not seriously affect the analysis of pronominal clitics. A short and oversimplified summary of the rest of this section is that, although the syntactic position of the Miluk pronominal clitics, and the forms of most of them, are quite different from those of the corresponding Hanis proclitics, in many other respects the Miluk pronominal clitics pattern quite similarly to the Hanis ones.

6.1. Basic clause structure.

The most obvious aspects of Miluk syntax are quite similar to Hanis. Constituent order is again rather free, and nouns expressing the subject/actor of transitive verbs are frequently marked with ergative =, as in (65).

1. (65).

2. Fight–tr(?)   art= dog      art=erg= young.man

3. ‘The young man kept fighting the dog.’ (JH 180.1–2)

Nominal predicates occur without copula, as in (66a) and (66b).

1. (66a).

2. girl    wealthy.headman   3.ps=child

3. ‘A girl was the child of a wealthy headman.’ (JM 159.1)

4. (66b).

5. not   dog      art= 1s.ps=husband

6. ‘My husband was not a dog.’ (JM 162.9)

My impression-based in part on texts for which Jacobs collected versions in both languages-is that overall the order of nonclitic constituents in Miluk, including the pragmatic effects of placing constituents in particular positions, is comparable to Hanis.

As in Hanis, the future enclitic is =han, exemplified in (67), though its initial h is often deleted in Miluk, as is seen in various examples later on-regularly after singular pronominal enclitics and the subordinator i ‘if, when’, and optionally after other enclitics. The customary or habitual enclitic is =du, quite different from Hanis =he.

1. (67).

2. now   =fut      return art=   young.man

3. ‘Now the young man was going to return home.’ (JM 164.11)

Also as in Hanis, enclitics follow the first phrase of the clause, rather than the first word, as seen in (68), and certain clause-initial discourse connectives, notably wi ‘then’, are not counted in determining clausal second position-the enclitic appears after the word or phrase that follows the connective, not on the connective itself, as in (69). Subordinators such as i ‘if, when’ and naim ‘because’ (both identical to Hanis forms) do host enclitics, however.³⁵

1. (68).

2. later      people   fut      dem(?)   eat

3. ‘The next people (the Indians to come) will eat these.’ (JM 195.14)

4. (69).

5.       then   recover(?)   fut

6. ‘. . . then she will get well.’ (JM 197.17)

6.2. The pronominal enclitics.

Miluk has pronominal enclitics for subject and object roles, rather than proclitics; the inventory is shown in table 3. Third person plural =()i is essentially the same as the corresponding Hanis proclitic, second person dual =()is is very similar to Hanis =()iš=, and the first person dual and plural enclitics might be interpreted as reductions of the corresponding Hanis proclitic forms (or else the Hanis forms might involve the accretion of additional material). But the other Miluk enclitics (in bold type in the table) do not resemble the Hanis proclitics at all. The singular enclitics are particularly striking: Miluk second person =n(V) resembles Hanis first person ∂n=, while the Miluk first person singular form resembles nothing in Hanis.³⁶

Table 3.

Miluk Subject-Object Pronominal Enclitics

The Miluk pronominals occur in the same position in the clause as other enclitics. In (70a), the verb is the initial constituent in the clause and hosts the second person subject enclitic; in (70b), the subordinator náim ‘because’ is the initial constituent in the clause and hosts the same enclitic.

1. (70a).

2. Falsehood–prg(?)=2S

3. ‘You lie!’ (JM 214.4)

4. (70b).

5. because   =2S=falsehood–prg(?)

6. ‘because you are lying’ (JM 195.8–9)

Pronominal enclitics do not need to be adjacent to the verb, as (71a)-(71b) and various examples later on show. They precede the tense-aspect enclitics =han ‘future’, shown in (71a), and =du ‘customary’.

1. (71a).

2. indeed   =1s=fut   there   shoot–tr

3. ‘Indeed I will shoot him right there.’ (JM 218.10)

4. (71b).

5. then if/when   =1s     from= Empire(?) depart

6. ‘When I had left Empire . . .’ (JM 110.11)

Morphosyntactically, the pronominal forms of table 3 are clearly enclitic; they must be preceded by an overt element, while, as (70a) shows, they need not be followed by one. Jacobs’s transcription implies that they may be phonologically proclitic to a following word, however; he often writes the pronominal enclitics close with the following word, without space or dash, and my retranscription reflects this by connecting the enclitic to the following word without a space, as for =n∂ in (70b).³⁷

The Miluk negator an (also aN, with devoiced final consonant) has no special effects on the pronominal enclitics. Like other syntactic elements, it hosts enclitics when it is the initial constituent in the clause, as in (72a), but not if it is later in the clause, as in (72b). (The position of the negator itself seems comparable to what it is in Hanis.)

1. (72a).

2. not =1S   how  return

3. ‘I cannot get back.’ (JM 219.15)

4. (72b).

5. but =1S   not  falsehood–prg(?)

6. ‘But I am not lying.’ (JM 214.4-5)

While their position in the clause is different, in most other respects Miluk pronominal enclitic usage parallels that of the Hanis proclitics. The Miluk enclitics code the subjects of intransitive verbs, as in examples above; they can also code the subjects (actors) of transitive verbs, as in (73a) and (73b).

1. (73a).

2. not =1S   money      want–tr

3. ‘I do not want money.’ (JM 89.8)

4. (73b).

5.       dem(?)   =2S=fut   take

6. ‘. . . , you ought to be able to catch it.’ (JM 213.19)

The enclitics are also attested as subjects of nonverbal predicates, as in (74a)–(74c).

1. (74a).

2. not   3p   sea.otter

3. ‘They are not sea-otters.’ (JM 51.8)

4. (74b).

5. two=1d.excl

6. ‘We are two.’ (JM 54.7)

7. (74c).

8. not   =1S   have–wing–have something person   =1S

9. ‘I am not a winged thing, I am a person.’ (JM 163.16–17)

Like the other enclitics, the pronominal enclitics follow the initial phrase rather than the initial word of the clause. This is seen with phrasal nonverbal predicates, such as the modifier-plus-noun phrase in (75a) and the possessive phrase in (75b).

1. (75a).

2. poor            person =1p

3. ‘We are poor people.’ (JM 85.18)

4. (75b).

5. trickster      3.ps=child=2s

6. ‘You are the trickster’s child.’ (JM 202.11)

Various examples below—namely, (85), (88c), and (100b)—show that the pronominal enclitics can follow phrases that are not predicates, as well.

As in Hanis, third person dual and plural enclitics and are not normally used as objects, as (76a) and (76b) exemplify.³⁸

1. (76a).

2. women      =1S=fut make–tr thus    =1S=fut =cust call(?)      1s.ps=voc=

3. hí•me"

4. children

5. ‘I will make some women, and I will call them "my children".’ (JM 186.13–14)

6. (76b).

7. see=2s=ynq

8. ‘Did you see them (two people whom the speaker is pursuing)?’ (JM 155.8)

The presence of a first or second person enclitic in (76a) and (76b) is probably not responsible for the absence of a third person enclitic. Combinations of a third person enclitic plus a first or second person enclitic are possible in inverse constructions, at least (see section 6.3).³⁹

Even in subject function, the nonsingular third person enclitics are not obligatory when there is an independent noun subject, as in (77a) and (77b); omitting the third person enclitics certainly seems to be the most common option in this context.

1. (77a).

2. thus speak–prg(?)   art= people

3. ‘That is what the people said.’ (JM 197.6–7)

4. (77b).

5. adv= thus tell–(?)      art= erg= old.people

6. ‘That is what the (two) old people said to him.’ (JM 215.1–2)

However, doubling of an independent nominal subject is not absolutely ruled out, as is shown by (78a)–(78d) (the subject nominal is underlined).

1. (78a).

2. then help=3d      art=   3.ps=pain art= erg= women      shaman

3. ‘Then those two women shamans helped him with his pain (power), . . .’ (JM 94.16–17)

4. (78b).

5. adv= thus =3p =cust tell–(?)      art= erg= people art= snow

6. ‘That is how the people spoke about the snow.’ (JM 100.10–11)

7. (78c).

8. then art= children wives      then fruits.and.berries =3p=pick/dig

9. ‘The girl (younger) wives picked berries and fruits.’ (JM 844–5)

10. (78d).

11. art 3.ps=erg= parent      art= wealthy 3.ps=child not =3p =cust

12. 

13. something make(?)

14. ‘The parents of a very rich person’s child did not concern themselves (about her?) . . .’ (JM 88.12)

First and second person enclitics are evidently obligatory even when they double an independent first or second pronoun subject, or so (79a)–(79c) indicate. (At least this is true of the singular enclitics; I have not discovered any evidence in the corpus as to whether this is true for dual and plural first and second person.)

1. (79a).

2. in.turn   =1S=fut   1s.indp   rdp.gamble

3. ‘Now I myself will gamble.’ (JM 182.3–4)

4. (79b).

5. also   =1s=fut   1s.indp go

6. ‘I am going to go too.’ (JM 181.12)

7. (79c).

8. also   =2s=fut   2s.indp adv= thus

9. ‘You too will be like that, . . .’ (JM 201.6)

As in Hanis (see section 3.4, n. 12), it is difficult to determine whether doubling by a pronominal enclitic makes any difference to the syntactic positions of nominals, especially third person nominals. The doubled nominals in (78c) and (78d) are evidently left-dislocated, since they precede the hosts of enclitics, but the doubled pronouns of (79a)-(79c) are evidently clause-internal, since they intervene between enclitics and the verbal or nonverbal predicate of the clause; it is not clear just what sort of position the doubled postverbal nominals in (78a) and (78b) occupy. I leave this question open.⁴⁰

6.3. Inverse forms and interaction of first and second person.

Miluk transitive clauses in which the object/undergoer is first or second person closely parallel Hanis, at least when first or second person arguments are singular. (Doty [2012: chap. 6] has independently reached the same conclusions; besides the inverse and the first and second person object suffixes, he discusses various other aspects of transitive verb morphology.) When third person acts on first or second person, an inverse construction is used. The verb is marked with a suffix –u(•)n, or sometimes –en or -i(•)n, and the first or second person object (singular, in the clearest examples) is coded by a pronominal enclitic; first person singular object is seen in (80a) and (80b), and second person singular object in (81a) and (81b).

1. (80a).

2. adv=   thus =1s   tell–inv   dem= 1s.ps=erg= dream

3. ‘That is what my dream has told me.’ (JM 91.10)

4. (80b).

5. indeed   =1s=give=inv   art=   paper      3.ps=payment(?)

6. ‘So then he gave me the payment for the (divorce) papers.’ (JM 113.11–12)

7. (81a).

8. then when=2s=fut thus tell–inv

9. ‘Then when he says (lit., "will say") to you . . .’ (JM 216.6)

10. (81b).

11.       ten   =2s=fut   give–inv   sea.otter 3.ps=hide

12. ‘. . ., he is to give you ten sea otter hides.’ (JM 134.9)

When the subject of an inverse verb is third person dual or plural, the appropriate third person enclitic may appear as well, preceding the first or second person enclitic, as in (82)–(83b); such clauses are fairly frequent in the texts.

1. (82).

2. not   =3d=1S=kill–inv

3. ‘They (two, ‘your parents’) did not kill me then.’ (JM 220.8)

4. (83a).

5. thus   =3p=1S tell–inv

6. ‘That is what they say to me.’ (JM 203.19)

7. (83b).

8. Follow–tr–inv=3p=2s

9. ‘They are following you.’ (JM 156.9)

As with noninverse verbs, a third person enclitic may double an overt noun subject, as in (84), although again the enclitic is not obligatory, as in (85).

1. (84).

2. adv=   thus   =3p=1s=mock(?)–inv      art= erg= cottontail 3.ps=children

3. ‘This is the way those children of cottontail mock me.’ (JM 203.18–19)

4. (85).

5. erg= afterwards(?)   person   =2s=fut take.aboard–tr(?)–inv

6. ‘The next persons⁴¹ will take you.’ (JM 219.15–16)

The inverse suffix –u•n looks like the Hanis inverse suffix –u (whether by inheritance or by borrowing) plus an accreted n, which it is tempting to see as a reanalyzed pronominal enclitic. A similar source might be proposed for the other Miluk inverse suffixes –i(•)n and –en. However, it is even less clear what governs the choice between the inverse suffixes than it is in Hanis, which makes it difficult to securely connect either –i(•)n or –en with Hanis -i.

When second person acts on first person, or vice versa, the verb takes a special suffix, –ai for second person acting on first person, as in (86), or –m when the first person is a dative recipient or a beneficiary, as in (87), and –a•mi for first person acting on second person, as in (88a)–(88c). As in Hanis, a second person clitic but not a first person clitic appears with all these forms—or at least that is the case when both arguments are singular; see section 6.5 for clauses in which one or both arguments are nonsingular.

1. (86).

2. oh   shoot–tr–2→1=2S

3. ‘Oh you (sg.) shot me!’ (JM 218.11)

4. (87).

5. if/when=2s=fut not give–dat.2→1   dem= 1S.ps=wives

6. ‘if you (sg.) do not give me my wives’ (JM 146.12–13)

7. (88a).

8.       then(?) evening =2s=fut get–TR–1→2

9. ‘. . . , (In five days,) on that evening I will come to get you, . . .’ (JM 204.2)

10. (88b).

11. Help–tr–1→2   =2s=fut

12. ‘I will help you (sg.) . . .’ (JM 193.13)

13. (88c).

14. crow myth   =2s=fut narrate–1→2

15. ‘I will tell you a crow myth.’ (JM 170.10 [title])

The Miluk suffixes certainly resemble the corresponding Hanis ones closely enough to imply a historical connection, whether by inheritance or by borrowing (and, like the Hanis suffixes, could well have originated as simply first or second person object suffixes). The fact that –a•mi and –m are identical to the corresponding Hanis forms suggests that these two suffixes might have been relatively recently borrowed in one direction or the other. The second-person-acting-on-first suffix –ai seems more likely to be cognate with its Hanis equivalent -a(•)is (or at least to reflect old borrowing), given that it is similar but not identical to the latter.

6.4. Imperatives.

While a good many clauses with imperative translations occur in the corpus, their structure is difficult to interpret, though a suffix (or , etc.) appears on many intransitive imperatives. Still, it is clear that, as in Hanis, a pronominal clitic is not used for the subject of a singular imperative, whether intransitive, as in (89), or transitive, as in (90), and that a first person (at least, first person singular) clitic is not used for the object of an imperative, even when there is no second person clitic, as in (91a) and (91b).

1. (89).

2. quickly   go.out–intr.imv(?)

3. ‘Get outside quickly!’ (JM 25.10)

4. (90).

5. Water–instr(?)   wash–tr   dem= 2s.ps=mother

6. ‘Go wash your mother with water.’ (JM 217.3)

7. (91a).

8. Give–dat.2→1 dem=   2s.ps=dentalia

9. ‘Give me your money (large dentalia).’ (JM 225.5-6)

10. (91b). ⁴²

11. Name–TR–2→1

12. ‘Name me (i.e., call me by a kin term)!’ (JM 193.13)

When the addressee is nonsingular, it seems that a nonsingular second person pronominal enclitic is used at least some of the time, as in (92)–(93b).⁴³

1. (92).

2. come   follow–tr–2→1=2d

3. ‘Come (dual) follow me!’ (JM 140.14–15)

4. (93a). ⁴⁴

5. cross(?)–tr–2→1=2p

6. ‘Ferry me across!’ (JM 149.2)

7. (93b).

8. not =2P=weep–intr.imv(?)

9. ‘Do not weep (for me).’ (JM 149.14–150.1)

The form ‘come here!’ is fairly frequent (e.g., JM 221.6) in Miluk as in Hanis. In Hanis it is transparently composed of e()= ‘second person imperative’ plus ‘(come) hither’, but there is no evidence that it has any internal structure in Miluk; the form was probably borrowed from Hanis as an unanalyzed unit. At least some of the time, Miluk is evidently a quasi-interjectional particle outside the clause proper; this is seen in (92) above, where is not the host of the enclitic .⁴⁵

I have not identified any Miluk equivalents of the Hanis dual and plural hortative particles k^wis (g^w∂s) and k^win.

6.5. Interactions of nonsingular speech act participants

Clauses in which first person acts on second person or vice versa and at least one of the arguments is dual or plural are difficult to locate in the texts, and can also be difficult to interpret when found. Variation occurs for which there is no obvious principled explanation. Possibly this is a domain in which Mrs. Peterson’s command of the language was less secure, or one that was undergoing change at the time when she acquired Miluk.

Scrappy evidence suggests that when nonsingular second person acts on singular first person, as in (94)–(95b), or vice versa, as in (96), a second person nonsingular enclitic is used, but no first person enclitic; this would parallel Hanis.

1. (94).

2. if=2d =fut      want–tr–2→1

3. ‘ If you (dual) want me, . . .’ (JM 140.14)

4. (95a).

5.       then that.kind(?) =2p   =fut    =cust give–2→1 tobacco(?)

6. ‘. . . , then you (pl.) are to give that tobacco to me, . . .’ (JM 97.13)

7. (95b). ⁴⁶

8.       thus 3s.ps=length=2p   =fut   see–2→1

9. ‘. . . for that long a time you (pl.) will see me, . . .’ (JM 51.11)

10. (96).

11. now =2p=fut   thus tell–1→2

12. ‘When I tell you (pl.), . . .’ (JM 147.4–5)

In (97), however, the second person enclitic unexpectedly takes the form = čn (contrast = čil in (95b)).

1. (97).

2. again   =2p(?)=fut   see–TR–2→1

3. ‘You (pl.) will see me again.’ (JM 48.2)

If it is not a mishearing for =čil, the form =čn might possibly represent conflation of expected =čil with the second person singular enclitic (not the first person singular enclitic, since that is =u or =wu in Miluk), or nonce influence of the Hanis first person singular proclitic ∂n=.⁴⁷

I have not located any clear instances of second person acting on dual or plural first person.

Evidence about clauses in which a nonsingular first person acts on second person is both limited and contradictory. In (98), no second person enclitic appears.

1. (98).

2. not   =1p=want–tr–1→2

3. ‘We do not want you (sg.)!’ (JM 32.2)

(It does not seem likely that a second person singular enclitic =n(V) has been phonetically absorbed by the negator an in this example. That would imply that the second person enclitic precedes the first person enclitic, which is contradicted by examples below.) In (99), it is impossible to tell whether =ni is the first person dual exclusive enclitic or the second person singular enclitic, the two being homophononous in Miluk; I tentatively gloss it as the former.

1. (99).

2. mother’s.sister erg=   both      =1d.excl(?)=like–1→2

3. ‘Mother’s sister! we both like you (sg.).’ (JM 49.7)

Other examples appear to have both a first person plural enclitic and a second person singular enclitic, in that order (with a transitional vowel i before or after the first person plural form), as in (100a) and (100b). (In available examples, the object appears to be singular semantically as well as formally.)

1. (100a).

2. Compensate–tr(?)–1→2=1P=2S

3. ‘We are compensating you (sg.).’ (JM 97.10)

4. (100b).

5. five      day      =1p=2s=fut also give–1→2

6. ‘We will give you (probably sg.)⁴⁸ five days . . .’ (JM 60.14)

It is not attractive to suppose that in these two examples is a nonce borrowing of the Hanis first person plural form . One would then expect to find occasional instances of in the Miluk corpus for first person plural also where there is no second person argument, and I am not aware of any such instances. If (100a) and (100b) are valid Miluk, the construction is reminiscent of that in corresponding Hanis clauses, where both a nonsingular first person proclitic and a second person proclitic appear (see section 4.2); however, the second person form in the Miluk examples appears to be the ordinary second person singular enclitic, rather than a special clitic or affix fixed on the verb as in Hanis.

Yet another instance of nonsingular first person acting on second person, seen in (101), shows essentially the same surprising form of the second person plural enclitic, = č(i)n, as in (97) above. The discourse context fairly clearly implies that the first person subject of (101) is plural, not dual, so it is not possible to explain the n of the enclitic as exclusive dual =n.⁴⁹

1. (101).

2.       then fight–TR(?)–1→2=2P(?) fut

3. ‘. . . then we (pl.) will fight you (pl.)’ (JM 60.8)

The fact that the second person argument in (101) is evidently understood as plural constitutes a contrast with (100a)-(100b), where it is singular. Conceivably = č(i)n might be a special portmanteau form for plural (nonsingular?) first person acting on plural (nonsingular?) second person, though that will not explain the occurrence of = čn in (97).

7. Areal perspectives

Many aspects of the marking of pronominal categories in Hanis and Miluk are reminiscent of patterns found in other languages of the Northwest. I briefly note here a few such resemblances (mostly of grammatical pattern, not of the actual form of morphemes) to suggest directions for typologically or areally oriented research.⁵⁰

Pronominal clitics for subjects appear widely in Northwestern languages, including Siuslawan and Alsea on the Oregon coast (Frachtenberg 1920, 1922b: 467–72, 479–80, n.d.:127–34), Sahaptin in the southern Plateau area (Rigsby and Rude 1996; Jacobs 1931), in much of the Salish family, and so on. Mostly these are second position enclitics of some sort; Hanis is unusual in having proclitics that are more or less fixed on the verb or other predicate. (The Chinookan languages on the lower Columbia River inflect verbs with an elaborate system of pronominal prefixes rather than using clitics [Boas 1911:580–84; Dyk 1933:30–39; Silverstein 1976]; within the Salish family, some languages such as Tillamook [Edel 1939; Egesdal and Thompson 1998; Thompson 1979: 739–40; Kroeber 1999:54–57] have fixed pronominal clitics on the predicate. But none of these are adjacent to Hanis.)

Pronominal clitics in Salish languages are typically limited to the subject role, but Sahaptin, Alsea, and Siuslawan to varying degrees and in different ways allow at least some of their clitics to express objects (or undergoers) as well. These languages differ as to how they use affixal verb morphology to reduce or eliminate ambiguity as to the grammatical roles of the clitics. Inverse morphology comparable to that of Hanis and Miluk is not found in Alsea (where only the nonsingular clitics can be used as objects) and Siuslawan, for instance; on the other hand, transitive verbs in both these languages take object suffixes that distinguish person but not number. (I suspect that such object inflection was the historical source of the Coosan suffixes for first person acting on second and vice versa; see section 3.3.)

Some of the Hanis and Miluk patterns reviewed in this article are strikingly reminiscent of Algonquian—or more precisely, of the inflection of Algonquian verbs in the independent order.⁵¹ The Algonquian pronominal prefixes can represent either subject (actor) or object (undergoer), like the Hanis and Miluk pronominal clitics, and an inverse suffix is used when third person acts on first or second. The Algonquian "thematic" suffixes that mark action of second person on first or vice versa parallel Hanis -a(•)is and –a•mi (see section 3.3) and Miluk –ai and –a•mi (see section 6.3), and like these can equally well be construed as first or second person object suffixes (see section 3.3 and n. 9). Second person prominence in the pronominal prefix paradigm is, of course, a well-known trait of Algonquian—even if not quite as pervasive in Algonquian inflection overall as is sometimes assumed—and this parallels the rather strong second person prominence seen in the Hanis and Miluk pronominal clitics (see sections 5 and 6.3). The idea that there might be some sort of historical connection between Hanis and Miluk, on the one hand, and Algonquian, on the other, is not as absurd as it might initially seem, either. Yurok and Wiyot, which along with Algonquian constitute the Algic family, are located on the northwestern California coast, not far from the location of Hanis and Miluk—the Athabaskan languages that now intervene are certainly recent arrivals—while there is some reason to think that Algic, including Algonquian, originated somewhere in the Plateau region.⁵²

The Coosan patterns most closely match Algonquian proper, rather than the attested states of the nearby Yurok and Wiyot: inverse constructions, for instance, are not employed as extensively in Yurok as in Algonquian or the Coosan languages, and Wiyot has no inverse construction. Second person prominence is not characteristic of Yurok (Robins 1958:69-71); in Wiyot, it appears solely in the negative construction, the only remaining construction in which Wiyot verb inflection retains relics of the pronominal prefixes (Teeter 1964:37, 89). (Still, the very fact that second person prominence is vestigial in Wiyot suggests that it is an old trait of Algic, not something that developed specifically in the Algonquian branch.) And, on the other hand, Hanis shows a fuller array of resemblances to Algic than does Miluk. The Hanis pronominal clitics are proclitic to the verb, a closer resemblance to the position of the Algic personal prefixes than the clausal enclitic positioning of the Miluk pronominals. The negator is the only element that attracts proclitics away from the verb in Hanis, and this is reminiscent of Wiyot, where the negator is the only predicate element on which pronominal prefixes ever appear. And, finally, the third person singular possessive proclitic of Hanis (except when preceded by an article with which it contracts) is u= (Frachtenberg 1922a:396–98)—strikingly similar to the third person (especially possessive) prefixes of Algic: Algonquian *we- (Bloomfield 1970:451), Yurok we- (Robins 1958:25), Wiyot huh- (phonetically [hu-]) ~ Ø - (Teeter 1964:79),⁵³ and quite different from the Miluk third person possessive proclitic d∂=.⁵⁴

There are certainly significant differences between Algic and the Coosan languages as well. The pronominal prefixes of Algic, for example, mark only person and not number, unlike the elaborate clitic paradigms of Hanis and Miluk; nor do the Algic prefixes ever cooccur, while the Oregon languages allow at least limited possibilities for two clitics to appear together. And, on the other hand, there is no real Coosan equivalent for Algonquian’s complex suffixation marking subject and object number. (Wiyot totally lacks number suffixation, however.) Still, the possibility of diffusional connections between the Coosan languages and Algic, or particular branches of Algic, seems worth investigating further.