Hokan I: A review of comparative studies

Mikhail Zhivlov Russian State University for the Humanities / HSE University, Moscow; [email protected] Hokan I: A review of comparative studies 1 The article reviews the history of Hokan studies, from the formulation of the Hokan hypothesis by Dixon & Kroeber in 1912–1913 to the present day. Despite more than a hundred years of research, there is yet no consensus on the validity of the Hokan hypothesis. The article argues that one of the reasons for this is the fact that many attempts to compare Hokan languages used a non-standard methodology, in which the study stops at listing the observed sound correspondences, instead of the classical comparative method, which requires a search for complementary distributions and a comprehensive reconstruction of protophonology. Keywords: comparative method; Hokan languages; Native American languages; genealogical classification of languages. 0. Introduction The Hokan hypothesis, combining a number of language isolates and small families in California into a single Hokan stock, was proposed more than a hundred years ago. Despite the long history of the proposal, there is still no consensus on its validity. Although putative Hokan languages remain severely underdocumented, now we have much more data on them than was available to researchers in the 20th century. Time has come for a reassessment of the hypothesis in the light of new data. Such a reassessment must necessarily begin with a survey of what has already been done in the field of comparative Hokan studies. In the words of Lyle Campbell (1997: 290), “[a] thorough understanding of the Hokan hypothesis requires a knowledge of its history”. The present paper is organized as follows: Section 1 is dedicated to the first period of Hokan studies, from the original proposal by Dixon & Kroeber to the work of Sapir. Section 2 surveys an intermediate period from the 1940s to the 1950s, when little has been done on the Hokan hypothesis. Section 3 offers a survey of the period from the 1950s to the 1980s, focusing on the work of Mary R. Haas and her students. Section 4 is dedicated to attempts of general synthesis by such scholars as K.-H. Gursky, D. Leshchiner, T. Kaufman and others in the 1980s. Section 5 surveys the work on Hokan done from the 1990s to the present day. Section 6 discusses general methodological problems relevant to Hokan comparison. 1. The birth of the Hokan hypothesis Although links between, e.g., Yuman, Seri and Tequistlatecan (Brinton 1891: 109–112), Shasta and Palahinihan (Dixon 1905), Chimariko, Shasta and Palahinihan (Dixon 1910: 335–339) had been proposed earlier, the Hokan hypothesis sensu stricto dates from 1912–13, when Dixon & 1 The study was carried out with the assistance of the “Laboratory Of Unnecessary Things” of the Independent University of Moscow. I thank Fernando O. de Carvalho, Albert Davletshin, Yoram Meroz, George Starostin and an anonymous reviewer for their valuable advice. All remaining errors are mine alone. Journal of Language Relationship • Вопросы языкового родства • 22/1–2 (2024) • Pp. 162–188 • © Mikhail Zhivlov, 2024 Hokan I: A review of comparative studies Kroeber (1912, 1913b) announced their findings on the classification of Californian languages in a brief note, published first in “American Anthropologist” and then in “Science”. Their proposed new families were Penutian, consisting of Maidu, Wintun, Miwok, Costanoan and Yokuts; Hokan, consisting of Shasta (including Shastan proper, Achumawi and Atsugewi), Chimariko, Pomo, probably Karuk and possibly Yana; and Ritwan, consisting of Yurok and Wiyot. No evidence was presented in favor of these hypotheses. Then followed a somewhat more substantial paper (Dixon & Kroeber 1913a), where Esselen and Yuman were added to Hokan and a new family was proposed — Iskoman, consisting of Chumash and Salinan. The evidence for Hokan presented in this paper consisted of structural similarities and five sets of lexical cognates — words for ‘tongue’, ‘eye’, ‘water’, ‘stone’ and ‘sleep’, each represented in no less than five out of seven postulated branches of Hokan. The authors also tentatively suggested that Seri may belong to Hokan and that Hokan and Iskoman may be related. Kroeber (1915) connects Seri and Oaxaca Chontal with Yuman and Hokan in general. He lists 35 comparative sets. Kroeber also discusses sound correspondences, but from a decidedly pre-Neogrammarian point of view, e.g.: “This Chontal-Seri correspondence f : x is corroborated by Seri-Mohave f : h in number 15—the fricative character is retained, but the point of articulation changed” (Kroeber 1915: 282). The next researcher to take up work on Hokan was Edward Sapir. Sapir (1917) presents 192 sets of comparisons between Yana and other Hokan languages; some additional Hokan cognate sets without Yana are also included. Sapir’s comparisons are semantically quite straightforward; in most cases the meanings of compared forms are either identical or very close. Some of Sapir’s comparisons must be discarded in the light of new data or reconstruction within accepted small families. Consider his etymology of the word for ‘man’: 60. Yana ʽisi “man, male, husband” Shas. ic “man”; New River Shasta gè-ʼic Chim. itci, itri “man” S. Pomo atcai “man”; N. Pomo tca “person”; C. Pomo tcatc; S. W. Pomo atca Chum. -isǖyix “husband” Chon. acans “person” (Sapir 1917: 9). Checking the Yana, Shasta and Chimariko forms across more modern sources, we can see that these look just as comparable as the forms given by Sapir: Central and Northern Yana hisi ’man’, Yahi hihsi ‘man; husband, husband’s brother’ (Sapir & Swadesh 1960: 98), Shasta ʔís ‘person; Indian’ (Bright & Olmsted 1959: 37), Chimariko ʔiṭi ‘husband, man’ (Conathan 2002: 31). Moreover, we can add Achumawi ís ‘person; Indian’ (Nevin 2020) to the comparison. The Pomoan forms, however, go back to a Proto-Pomo form hardly compatible with the Yana, Shasta, Achumawi and Chimariko forms above: Kashaya ʔaca˚c ‘person, Indian, man’, Southern Pomo ʔač·ay, Northern Pomo čáʔ ‘man, person’, Central Pomo cá·c̓ ‘person’, Northeastern Pomo tá·t-ka·, Southeastern Pomo ca-wi, Eastern Pomo ká·kʰ ‘man, Indian’ < Proto-Pomo *ʔaká·kʔ ‘man’ (McLendon 1973: 81). The Chumash (more precisely, Ineseño) word -isǖyix ‘husband’ was taken from Kroeber (1904: 42). “Samala-English Dictionary”, based on data collected by J. P. Harrington, gives this word as isʰɨ’y ‘husband’, from is- ‘one’s own’ + hɨ’y ‘male’, a shortened form of ɨhɨ’y ‘man, male’ (The Santa Ynez Band of Chumash Indians 2007: 148). Thus, the comparison with Chumash must be rejected. Finally, Lowland Chontal sans ‘person, human being, living creature’ has an irregular plural -sanʸuʔ (O’Connor 2013: 226), pointing to a root san-. The comparison with the forms above does not look promising. Nevertheless, a surprisingly large number of Sapir’s comparisons in (Sapir 1917) stand the test of time. It is interesting to look at some of the recurrent sound correspondences noted by 163 Mikhail Zhivlov Sapir. Yana y corresponds to Chimariko s or c [š] in Yana ʽīya ‘trail’, Chimariko hissa ‘trail’; Yana wêyu ‘horn’, Chimariko h-owec ‘antlers, horn’; Yana -ya ‘female’, Chimariko -sa ‘female’ (Sapir 1917: 9, 16, 23). Yana pʽu corresponds to Chimariko xu in Yana pʽô- ‘to blow’, pʽu-sā- ‘to smoke’, Chimariko -xu-, -xuc- ‘to blow’; Yana pʽū- ‘to swim’, Chimariko -xū- ‘to swim’; Yana pʽuiʽ- ‘to be fat’, Chimariko -xu- ‘fat’ (adj.); Yana pʽun- ‘to paint’, Chimariko -xol- in -po-xolxol ‘to paint’ (Sapir 1917: 13). Despite the obviously inadequate quality of data available at that time, Sapir (1917) remains one of the most persuasive collections of comparanda supporting the Hokan hypothesis. The final statement of Dixon & Kroeber’s views on Hokan was their short monograph “Linguistic families of California” (1919). Here, Hokan includes Karuk, Chimariko, Shastan (including Achumawi and Atsugewi), Pomoan, Yana, Washo, Esselen, Salinan, Chumash, Yuman, Seri and Tequistlatecan (Oaxaca Chontal). No new evidence for Hokan as such is adduced in the 1919 publication. The authors explain this in the following way: “The evidence submitted by the writers as to the unity of these languages is admitted by them to be but slender. Yet they feel themselves absolved from the obligation of presenting further formal proofs through the publication by Dr. Sapir of a recent essay devoted nominally to the determination of the position of Yana in the Hokan stock, but in effect rendering as full a proof of the actuality of the family as could be demanded” (Dixon & Kroeber 1919: 103). However, since the inclusion of Washo in Hokan was a new idea, the authors give two lists of Washo-Hokan parallels, one compiled by themselves and another by Sapir. What is more interesting, the authors provide a “Historical Introduction”, where they explain how they arrived at their classification. In the wake of Dixon & Kroeber’s (1903) typological survey of Californian languages, the authors became interested in lexical similarities between these languages. However, from the start they interpreted these similarities as due to contact between unrelated languages 2: “As evidences of similarities between this and that language accumulated, they were indeed noted, but were consistently interpreted as instances of one unrelated language borrowing either material or machinery from another” (Dixon & Kroeber 1919: 49). A systematic comparison was attempted: “About two hundred and twenty-five English words were selected on which material was most likely to be accessible in reasonably accurate and comparable form, and the known native equivalents in sixty-seven dialects of the twenty-one stocks were entered in columns. Comparisons were then instituted to determine all inter-stock similarities that seemed too close or too numerous to be ascribed to coincidence. The purpose of the study was threefold: first, to ascertain the nature and degree of borrowing between unrelated languages; second, to trace through these borrowings any former contacts or movements of language groups not now in contact; third, in the event of any relationship existing between languages then considered unrelated, to determine this fact.” (Dixon & Kroeber 1919: 49). The resulting “lexicostatistical matrix” made no sense in terms of language contact: “Families some distance apart on the map often had more stems in common than those in juxtaposition; if the remote group was regarded as once in contact with the one with which it shared most words, it must have been in contact also with others with which it shared but few words. ... Finally, in a mood rather of baffled impotence, an interpretation of the cases of most abundant resemblance as due to genetic relationship was applied. At once difficulties yielded, and arrangement emerged from the chaos” (Dixon & Kroeber 1919: 50). The number of resemblances in the “lexicostatistical matrix” between branches of what would be called Hokan ranged from 4 (Karok-Yana) to 21 (Shasta-Pomo). The authors themselves admit that their comparisons Fernando O. de Carvalho (p.c.) points out that these interpretations by Dixon & Kroeber make sense within the general Boasian outlook of North American linguistics at the time (see Campbell 1997: 62–66, 72). 2 164 Hokan I: A review of comparative studies “might or might not be considered as including radical words due to a common origin: they certainly included words not due to such common origin but derived by loan. ... Further, the comparisons used being avowedly superficial, that is, not based on analysis, a certain number of false coincidences were bound to have crept in” (Dixon & Kroeber 1919: 52). Still, it is clear that there was more to Dixon & Kroeber’s Hokan data than just five lexical sets published in Dixon & Kroeber (1913a). This is important in view of the following misleading statement by Lyle Campbell: “Hokan had the shakiest of origins. ... This hypothesis was based on inspectional resemblances involving only five words in these languages: 'tongue', 'eye', 'stone', 'water', and 'sleep'” (Campbell 1997: 290). Of course, the unfortunate fact that Dixon and Kroeber did not publish their evidence, thus setting a precedent for later researchers in the field, does not mean that their evidence was confined to five words. Sapir (1920a) demonstrates an interesting paradigmatic morphological similarity between Chimariko and Salinan: verbal subjective pronominal prefixes have the following form in both languages (Table 1; see also Zhivlov 2018). Table 1. Chimariko and Salinan pronominal prefixes, after Sapir (1920a) Sing. Plur. Chimariko Salinan 1 y-, i- e- 3 h- 2 m- m- 1 a-; ya- a- 2 q- 3 h- - k- (subject of 2nd per. plur. imperative) - Sapir’s (1920c) review of J. Alden Mason’s monograph “The Language of the Salinan Indians” contains, apart from many valuable remarks on Salinan synchrony, some diachronic observations. The most interesting is the comparison of plural infixes in Salinan and Yana, worth quoting in full: “One of the most interesting and irregular features of Salinan is the formation of the plural of nouns and of the plural and iterative of verbs. No less than a dozen distinct types and a large number of irregular formations are discussed and illustrated by Mason, the great majority of them involving a suffixed or infixed -t-, -n-, or -l-. Significantly analogous plurals, often of great irregularity though of less frequency, are found in Yana ; e. g. such Salinan plurals as ṭ-eṭeyitinai ARROWS (sing. ṭ-eṭeyini’) and anetem SEVERAL REMAIN (sing. anem) offer more than a cursory parallel to such Yana forms as mut’djaut’i-wi CHIEFS (sing. mu·djaup’a·), k’uru·-wi SHAMANS (-r- < -d-; sing. k’u·wi), sa·dimsi- SEVERAL SLEEP (sing. samsi-, sams-). The Salinan type with infixed -h-, -x- (e. g. mehen- HANDS, sing. men-; kaxau SEVERAL SLEEP, sing. kau) may be analogous to such Yana forms as dja·li- SEVERAL LAUGH (from *djahali- ?), sing. djal-” (Sapir 1920c: 306). Similar infixed plurals are attested in Chontal, Seri and Pomoan (Langdon 1990; Zhivlov 2018). Sapir’s second large paper on Hokan (Sapir 1920b), actually written before “The position of Yana...”, tries to demonstrate the connection between Hokan and Coahuiltecan — a grouping of languages in Texas and northeastern Mexico proposed by Swanton (1915). Swanton’s Coahuiltecan includes Tonkawa, Comecrudo, Cotoname, Coahuilteco, Karankawa and Atakapa. Sapir does not question the coherence of Coahuiltecan as a group, but tries to connect it as a whole to Hokan. Almost half of the comparisons in this article are between just one of the 165 Mikhail Zhivlov Hokan branches and Coahuiltecan and thus are irrelevant for the Hokan hypothesis in the narrow sense. If Coahuiltecan (or some parts of it) was related to Hokan, we would expect Hokan-Coahuiltecan parallels to have a better distribution within Hokan proper. Sapir (1921) gives Salinan additions to comparisons from (Sapir 1920b), as well as 28 new Salinan-Hokan comparisons. The paper also briefly discusses the position of Washo within Hokan. Sapir (1925) presents the evidence for relating the small Mesoamerican family of Tlapanec-Subtiaba to Hokan-Coahuiltecan. We know now that Tlapanec-Subtiaba forms part of the Otomanguean family (Campbell 2017a, 2017b). Even if Hokan is related to Otomanguean as a whole (Kaufman 2016), it is unlikely that many of Sapir’s comparisons will hold. The importance of this paper for Hokan studies lies not in comparisons with Subtiaba, but in the fact that it is here that Sapir laid out his views on Hokan morphology, specifically, nominal, adjectival and verbal prefixes. At least one of Sapir’s findings in this area was confirmed by later research. Sapir (1925: 506–512) reconstructed Proto-Hokan stative prefix *m- on the basis of the rather inadequate data he had at the time. Traces of such a prefix were later found in Eastern Pomo (McLendon 1975: 48), Yuman (Miller 2001: 60) and Seri (Marlett 1981: 70–71). While Dixon & Kroeber did not offer much evidence in support of Hokan, Sapir’s work represented a huge step forward in collecting and organizing such evidence. In his works on Hokan, Sapir used the method now known as “mass comparison”. He understood very well the limitations of this type of work: “Some of these comparisons are doubtful at best and a number of them will, on maturer knowledge, have to be discarded. A certain amount of groping in the dark cannot well be avoided in the pioneer stage of such an attempt as this” (Sapir 1920b: 289). However, a careful application of the comparative method, as in Sapir’s own work on Uto-Aztecan (Sapir 1913, 1919), was impossible here because the only phonologically adequate documented Hokan language at the time was Yana, described by Sapir himself. Still, Sapir does not limit himself to comparisons of lexical lookalikes; he notes recurrent sound correspondences (1917: 6, 7, 8, 9, 10, 13, 16, 21, 23, 27, 1920b: 284, 286, 287), provides preliminary Hokan reconstructions 3 (1917: 33); discusses Hokan root structure (1925: 493-495), derivation (1925), and bits of paradigmatic morphology (1920a). Surprisingly, there were no published attempts to systematically reconsider Sapir’s evidence for Hokan and Hokan-Coahuiltecan in the light of new data on the languages involved. 2. A temporary standstill The following decades saw no new work on Hokan in the proper sense. The only papers published in that period were attempts to add new languages to Hokan. Rivet’s (1942) article is an unsuccessful attempt to show Hokan affiliation of a fragmentarily attested Colombian isolate, Yurumanguí. “The result of the dearth of Yurumangui data combined with the fecundity of Rivet's etymological imagination is that most of the morphological analyses he proposed rest on hypothetical affiliations with Hokan, and so remain undemonstrated” (Poser 1992: 216). If anything, Rivet’s work shows that with the loose application of the “mass comparison” approach it is possible to “demonstrate” Hokan affiliation of just about any language. 3 “Sapir was aware that the available data, except for Yana, was phonologically underspecified if not totally inadequate. I reckon that he intended his pHok starred forms as an approximation to what a full reconstruction would be when phonologically adequate data was available” (Kaufman 1989: 80). 166 Hokan I: A review of comparative studies Harrington’s (1943) article is broadly similar: it attempts to show a Hokan affiliation for a South American language, this time Quechua. The results are similarly unconvincing. Unlike Rivet, Harrington tries to set up sound correspondences, but his approach can only be characterized as pre-Neogrammarian. Cf. the following sound correspondence with examples: ’, q, y—’ Q iwa, plant, tree; Subt i·ci, tree; Q qalyu, for *’alyu, tongue; Chim -pen, tongue; Q yawar, blood; Sal a·´kat, blood (Harrington 1943: 337). Greenberg & Swadesh (1953) propose that Jicaque (Tol), spoken in Honduras, is a Hokan language. The actual scope of the comparison is wider, since it includes not just “HokanCoahuiltecan”, but also “Gulf” languages and occasionally other languages from Sapir’s Hokan-Siouan superstock. Some comparisons look plausible, e.g, house: Jicaque wa; Yuma ʔava; Esselen iwa-no; Chimariko ʔawa; Yana wa-wi; Comecrudo wamak; Subtiaba gʷa (Greenberg & Swadesh 1953: 219). More typical, however, are comparisons like the following: mouth: Jicaque lam, lala; Tonkawa kala; Comecrudo xal; Atakapa kat; Yuma axaʔa; Pomo xacita; Chumash ïk; Chitimacha ša; Tunica šohu; Karankawa emi-akoy. Earliest form perhaps xa-la, with development in two directions as lala and axa, ixa (Greenberg & Swadesh 1953: 219). The “reconstruction” *xa-la, which is not based on regular sound correspondences, allows the authors to compare such dissimilar forms as Jicaque lala and Chumash ïk. 3. Mary Haas and the Survey of California Indian Languages The next period of Hokan studies, which lasted from the 1950s to the 1980s, was dominated by the work of Mary R. Haas and her associates at the Survey of California Indian Languages, established in 1953 in the University of California at Berkeley under the leadership of Haas and Murray B. Emeneau. The main purpose of the Survey was to organize fieldwork on endangered languages of California. The results of this fieldwork gave new impetus to comparative studies on Hokan and Penutian. Haas (1954) attempts to reconstruct the Proto-Hokan-Coahuiltecan word for ‘water’. While Haas discusses sound correspondences for the word in question, she does not provide any evidence that these correspondences recur in at least one other word. Thus, the methodology used in this paper only superficially resembles the traditional comparative method. Haas’ reconstructions are actually “pre-reconstructions” in the sense of Peiros (1997). Bright (1954) compares data from Karuk, Chimariko, Shasta, Achumawi and Atsugewi with the aim of establishing regular sound correspondences. He correctly notes that irregularities in sonorant correspondences may be caused by morphophonemic alternations of nasal and non-nasal sonorants, of a type attested in Karuk. Bright (1956) attempts to carry out experimental glottochronological counts for eight southern Hokan-Coahuiltecan (“Hokaltecan”) languages: Seri, Oaxaca Chontal, SubtiabaTlapanec, Jicaque, Comecrudo, Yuma, Salinan, and Tonkawa. The author uses strict objective criteria of what counts as “similar” forms. In addition, dissimilar forms may be counted as matches based on recurrent sound correspondences. Still, it is intuitively clear that the criteria used by Bright result in many spurious matches, such as Seri ítak ~ Chontal ekał ‘bone’, counted as a match on the basis of the “correspondence” Seri k ~ Chontal k, or Seri ʔánoł ~ Salinan ̣népukʰ ‘arm’, with the “correspondence” Seri n ~ Salinan n. The fact that such matches are not distinguished from obvious similarities like Seri ʔaX ~ Chontal axáʔ ‘water’ or Seri ʔapł ~ 167 Mikhail Zhivlov Chontal ípaL ‘tongue’ deprives the resulting counts of any value. See also Marlett’s (2008) criticism of Bright’s Seri-Salinan comparisons. In a series of three short papers, Olmsted (1956, 1957, 1959) tries to show that Achumawi and Atsugewi do not form a subgroup with Shasta within what he calls Northern Hokan. Olmsted uses his own field materials on Achumawi and Atsugewi. Unfortunately, the phonological accuracy of his field records leaves much to be desired (Nevin 1998: 15–18), which makes the work in question less useful than it might have been. Jacobsen (1958) compares Washo — a language he himself worked on — with Karuk, for which a good description (Bright 1957) was available. Some of his comparisons are striking: Washo í·biʔ ~ Karuk ʔípih ‘bone’, Washo -á·gal ‘into or in the mouth’ ~ Karuk -kara ‘into one’s mouth’, Washo -íwl ‘to roll’ ~ K -ívruh in ikrívruh (sg.) ‘to roll (as a ball)’, ʔinívruh- (du.-pl.) ‘to roll (as balls)’, Washo -í·šib ‘to be straight, correct, right’ ~ Karuk ʔíšip ‘to extend, be in a line’, Washo -ámad ‘through a tubular space’ ~ Karuk -vara ‘in through a tubular space’, Washo -íliʔ ‘up, upwards’ ~ Karuk -rih ‘up’. Since these comparisons find no counterparts in other Hokan languages, and even Jacobsen himself does not postulate a specific relationship of Washo and Karuk, a contact explanation seems preferable. It is interesting that many of these parallels are locative-directional suffixes/postpounds — a category whose spread in the Northern California language area must have involved language contact (Mithun 2007). Jacobsen lists sound correspondences between Washo and Karuk, but his approach is rather mechanistic. The comparison of Washo émle ‘heart’ and Karuk iθva·y ‘breast, chest, heart’, according to Jacobsen, illustrates the following correspondences: W e ~ K i, W e ~ K a·, W m ~ K v, W l ~ K zero, W zero ~ K y, W zero ~ K θ. Each of these correspondences, save the last, is supported by at least one additional example. While no segment is left unaccounted for, one may doubt the realism of the whole picture. One of the parts of Morris Swadesh’s introduction to the Yana Dictionary (Sapir & Swadesh 1960) is dedicated to “Yana-Hokan Notes”. Swadesh compares the basic vocabulary of Yana and Chimariko and finds 22 “likely cognates” among 113 semantically equivalent word pairs. Swadesh correctly emphasizes the necessity of recurrent sound correspondences: “The principal fact in this connection is that the compared elements show consistent phonetic relationships, as strikingly illustrated by the correspondence of Chimariko /x/ to Yana /p/ before rounded vowel in four cognate sets, including "swim" and "blow," listed above, and two others ("fat," "smear") given by Sapir” (Sapir & Swadesh 1960: 17). He also notes the importance of lexical sets represented in many daughter languages: “Even more striking is the multilanguage agreement; such words as "water," "two," "drink," "eat," "tongue," "thou" show cognates in six to ten different languages or language families” (Sapir & Swadesh 1960: 18). Haas (1963) starts from an observation that certain cases of long vowels in Shasta go back to earlier sequences *VmV with intervocalic *m. This *m completely disappears in Shasta, but is reflected as /w/ in Okwanuchu. This finding allows Haas to propose a number of ProtoHokan reconstructions with such sequences: PH *išamaruk̓a ~ *išamak̓aru ‘ear’, PH *č-imapasi ~ *imačipasi ‘liver’ (cf. Sapir’s *ipasi), *imarak̓ʷi ~ *imak̓ʷari ‘navel’, *imapaki ~ *amipaki ‘neck (nape)’, *yamari ~ *imari ~ *irama ‘nose’, *amirax̓ʷa ~ *amix̓ʷara ~ *s/kirax̓ʷa ~ *s/kix̓ʷara ‘nails (claws)’, *iš/čarima ~ *iš/čamira ‘arm’, *ipari ~ *ipawari ~ *ipariwa ‘tongue’, *as/cima ~ *is/cama ‘sleep’. Proto-Hokan, as it emerges from these reconstructions, is a language with rather long words, from three to five syllables, with no consonant clusters and with frequent metathesis 4. Fernando O. de Carvalho (p.c.) points out that Haas’ willingness to go on with such “protoform stuffing” and to postulate rampant metathesis likely stems from her prior work on Muskogean (see, e.g., Haas 1969: 40–42). In the Muskogean domain this kind of strategy, yielding three syllable proto-forms for disyllabic, widely 4 168 Hokan I: A review of comparative studies Now, one can hardly object to the comparison of such forms as, e.g., Antoniano Salinan -épal ‘tongue’, Southeastern Pomo bal ‘tongue’, Yuma i·pálʸ ‘tongue’, for which Haas reconstructs PH *ipari — whether or not supported by recurrent correspondences, we at least have here a clear case of phonetic similarity. But the same cannot be said about Barbareño Chumash elʔew ‘tongue’ and Washo álŋ ‘to lick’, which are given under PH variant reconstruction *ipariwa. The problem is that the correspondence of Chumash and Washo zero to a labial stop in other languages is not confirmed by additional examples, and thus, is not recurrent. Comparing dissimilar forms not connected by recurrent correspondences may allow one to call cognates any two forms at will. Another problem is that the length of the reconstructed forms allows comparisons between words which supposedly continue different parts of a longer form, so that “cognates” may share no cognate material at all. Thus, both Karuk ti·v ‘ear’ and Barbareño Chumash tuʔ ‘ear’ are said to go back to Proto-Hokan *išamaruk̓a ‘ear’, but the Karuk word continues the *išam- part, whereas the Chumash word reflects *-ruk̓- (Haas 1963: 46). In the same way, Yana ima ‘liver’ and Chimariko -ši ‘liver’ are given under Proto-Hokan *č-imapasi ‘liver’, but the Yana form reflects *-ima-, while the Chimariko form reflects *-si (Haas 1963: 47). The device of metathesis allows Haas to compare words that are unrelated within one uncontroversial family: Cocopa iya·kal ‘navel’ (supposedly from PY *imakʷáli < PH *imak̓ʷari) is compared to Yuma malʸpú· ‘id.’ (< PY *imalikʷí < PH *imarak̓ʷi), with the following comment: “PYu. *kʷ > p in Maricopa, Yuma, and Diegueno; > k in Kahwan and Cocopa.” (Haas 1963: 48). Actually, Proto-Yuman *kʷ is retained as such in all Yuman languages (Langdon & Munro 1980: 126). Lyle Campbell (1997: 294) was certainly right to conclude that “[i]n the absence of a more fully developed proposal for the historical phonological developments, one might suspect, for example in the case of 'ear', that Karuk t<i·v (< means that the sound is assumed to have undergone a change of assimilation), Chimariko -sam, and Chumash tuʔ may not really be cognate forms in genetically related languages and that they may not derive from Haas's proposed proto form *išamaruk’a/*išamak’aru (1963b:46); similarly, Achomawi owè·> 'liver' is a stretch from the assumed Proto-Hokan *č-imapasi/*imačipasi (1963b:47); and Chumash top’o, Achomawi alu, and Washo í>ʔb ‘navel’ are a far leap from each other and from the proposed Proto-Hokan *imarak’ʷi/*imak’ʷari”. Haas (1964) compares Yana and Karuk — two Hokan languages for which published dictionaries were available at the time. Comparisons with other Hokan languages are also adduced, but sound correspondences are listed only for Yana and Karuk. In what would become a model for further binary comparisons, correspondences, including non-recurrent ones, are simply listed, and no attempt is made to show that they form a coherent system or to assign protophonemes to each correspondence. Haas also points out the problem of “intersection” between Hokan and other Californian families: Penutian, Yukian and Ritwan. She compares her Proto-Hokan reconstructions to words with the same meaning in non-Hokan languages of California. Thus, Proto-Hokan *imarak̓ʷi ~ *imak̓ʷari ‘navel’ is compared to such “Penutian” forms as Chukchansi Yokuts ut-, Central Sierra Miwok pó·ṭi-, Rumsen lop-, Wintu naq and Maidu betéke. What these comparisons demonstrate is not the extent of lexical diffusion in California, but rather the ability of Haas’ method to integrate almost any word from any language under polysyllabic reconstructions unsupported by recurrent sound correspondences. McLendon (1964) presents 149 lexical comparisons between Eastern Pomo and Yana. Data from other Hokan languages are included wherever possible. While comparisons between diverging reflexes (e.g. Choctaw fani ‘squirrel’ : Creek íłu ‘id.’ < Proto-Muskogean *ixʷaNi/u), with different languages retaining different "parts" of the etymon, was apparently successful, and some Muskogean languages like Creek do in fact have very frequent cases of metathesis. 169 Mikhail Zhivlov Eastern Pomo and Yana mostly look reasonable, it is not always clear whether the compared forms from other languages are intended as cognates, and if yes, what exactly is being compared. Consider the set for ‘belly’: Pe wi·ní ‘pregnant’ : Yn, Yy wil(la). ... Cf. Ps wi·ní ‘pregnant’; K višva·n ‘belly’; S ʔíčnit ‘belly’; Yuy i·tó ‘belly’ (McLendon 1964: 129). Southern Pomo wi·ní ‘pregnant’ is straightforwardly related to the Eastern Pomo form. Karuk višva·n ‘belly’ may originally be a compound whose second part may be related to Pomoan and Yana forms. Shasta ʔíčnit ‘belly’ is perhaps adduced because the -ni- part can be compared with -ní in Pomoan words, but nothing in the Shasta form corresponds to wi- of Pomoan and Yana words. It is a complete mystery to me why Yuma i·tó ‘belly’ is given in this set. This case shows how important it is to have a working model of sound correspondences for the family as a whole — without it, there is no clear boundary between plausible and implausible comparisons. Silver (1964) compares Shasta and Karuk, although forms from other Hokan languages are occasionally adduced. Some of the comparisons may represent genuine Hokan cognates: ‘drink’ S -ic·-, K ʔíš; ‘rock’ S ʔíc·aʔ, K ʔasa; ‘sleep’ S -icmas-, K ʔasiṽ; ‘water’ S ʔác·a, K ʔá·s. Some other comparisons, without parallels in other Hokan languages, are likely to be areal loans: ‘gopher’ S xáras·aʔ, K axra·s; ‘sack’ S pú·ʔas, K pu·viš (cf. also Yurok puwi·š ‘sack’); ‘dog salmon’ S ʔičmú·n·a, K ačvu·n. Turner (1967) tries to show that Oaxaca Chontal languages are not related to Seri. He lists eight “cognates” between Highland Chontal and Seri on the 100-word Swadesh list plus twelve additional “cognates” from a bigger 500-word list and comes to the following conclusion: “Evidently all of the above cognates are either accidental or the result of borrowing. Because of the lack of actual cognates, it seems unnecessary to even attempt to describe systematic phonemic correspondences”. Turner’s lack of familiarity with historical linguistics is evident from the following passage: “It would be difficult to explain how two related languages could have different grammatical categories, and such is the case with Seri and Chontal”. A list of supposedly insurmountable differences between the two languages includes the following: “Seri identifies blue and green by different words whereas Chontal does not have separate words for these two colors. ... Seri has separate words for the urine of men and women as well as separate words for the act of urination by men or women; Chontal does not have this feature”. Bright (1970) in his rebuttal remarks “(1) that it is impossible to prove the LACK of relationship between ANY two languages, and (2) that the positive evidence for the relationship of Seri and Tequistlatec is in fact greater than admitted by Turner”. Langdon (1974) is a detailed survey of the history of Hokan studies up to 1970. Her final evaluation is quite optimistic: “We may conclude that Hokan-Coahuiltecan as conceived by SAPIR has withstood the test of time fairly well. Seriously disputed as to membership are only Tonkawa and Karankawa, and a new member, Jicaque, is likely. Less clear is the validity of SAPIR’s subgroupings, which, however, have no serious competitors to replace them, except the very cautious view that no two of the families grouped under the label HokanCoahuiltecan can be shown as yet to be closer to each other than to the others. ... While a full demonstration of the validity of the Hokan-Coahuiltecan hypothesis is not yet a reality, there is a growing sense of excitement as convergent results are reported” (Langdon 1974: 86). Silver (1974) discusses suffixes/postpounds in names of plants, trees and bushes. She draws attention to parallels with non-Hokan Californian languages and suggests, e.g., that the Yurok suffix of plant-tree-bush terms -ep̓ may be borrowed from the Karuk morpheme (ʔ)i·p with the same function. 170 Hokan I: A review of comparative studies Turner (1976) shows that noun plural formation follows different rules in Chontal and Seri; he uses this fact as a further support for his argument that the languages are unrelated. Silver (1976) raises the problem of “morphemization”, i.e. fossilization of morpheme sequences, “which often has the result of obscuring diachronically significant underlying phonological representations of morpheme strings”. According to Silver, aspirated and glottalized consonants in Hokan languages can go back to sequences like *CVx, *CVh and *CVʔC. James M. Crawford (1976) compares Chimariko and Yuman, presenting 134 lexical sets. While many comparisons are quite good, e.g., Chimariko ’awa, Cocopa wá ‘house’, Chimariko ’awu, Cocopa wí ‘mountain’ or Chimariko hoxu, Cocopa i·xú ‘nose’, in some other cases long dissimilar forms are compared on the basis of one matching segment. E.g., the set for ‘smooth’ compares Chimariko luyu’it, luyu’- with Cocopa xlqáy, Yawapai halowári (these two Yuman forms evidently have different roots), with the following comment: “[o]nly laterals are compared”. There are other cases where the Yuman forms within one set are not related to each other, as in the set ‘bird’, where Cocopa šá and Mohave ʔaciyér are unrelated. The author lists sound correspondences between Chimariko and Proto-Yuman. Out of 45 consonant correspondences, 14 are encountered only in one example (i.e., they are non-recurrent), and nine more are encountered in two examples. This is one more case of a binary comparison where analysis stops at the stage of listing the correspondences. No attempt is made to distinguish between chance coincidences and actual correspondences, or to integrate the correspondences into a coherent system. Judith G. Crawford (1976) compares Seri and Yuman languages. The paper contains 227 comparative sets; forms from other Hokan branches are also adduced. Sound correspondences are listed, but not analyzed; no reconstruction is attempted. In some cases, long forms are compared based on one “corresponding” segment only; e.g., Seri kkap ‘to fly’ is compared to Mohave himán ‘fly, get up’, Havasupai mánika ‘fall’ with the comment “Only S p and Yuman m are being compared”. Some other cases involve loose semantics, as in the comparison of Seri ʔáapXa ‘cottontail rabbit’ with Cocopa prxá·w ‘fox’. See also criticism of this work by Marlett (2007). Waterhouse (1976) gives a detailed reconstruction of Proto-Chontal (Tequistlatecan), based on a systematic comparison of Highland and Lowland Chontal. She also provides several lists of comparisons with other languages: “Comparison with Yuman languages”, “Comparison with Seri”, “Comparisons with other Hokan languages”, “Comparisons with Karankawa and Cariban” and “Comparison with Tlappanec (and Otomanguean?)”. Many of her comparisons with Yuman languages look quite promising, e.g.: Paipai yipúk ~ Lowland Chontal -epúh ‘nape’; Cocopa ipáł ~ Lowland Chontal, Highland Chontal -paL ‘tongue’; Tipai tú ‘stomach’ ~ Lowland Chontal, Highland Chontal -etú ‘navel’; Cocopa ixʷáṭ ~ Lowland Chontal, Highland Chontal -Wac’ ‘blood’; Yuma ʔamáṭ ~ Lowland Chontal, Highland Chontal amác’ ‘earth’; Diegueño ʔakʷí ‘cloud’ ~ Lowland Chontal, Highland Chontal akwí ‘rain’; Kiliwa amʔál, Diegueño ʔamal ~ Lowland Chontal, Highland Chontal ʔmaL ‘century plant’; Mohave naʔáy ‘father (w.s.)’ ~ Highland Chontal aʔáyiʔ ‘father’; Yuma ʔaxán ‘good’ ~ Lowland Chontal hanáʔ ‘all right’; Yavapai ʔaháʔ ~ Lowland Chontal, Highland Chontal aháʔ ‘water’; Diegueño ʔíł ‘worm’ ~ Lowland Chontal, Highland Chontal ʔiL ‘flea’; Cocopa išáł, Paipai ṣál ‘hand, arm’ ~ Lowland Chontal asáL, Highland Chontal išáL ‘wing’; Paipai ṣmá ~ Lowland Chontal sma-, Highland Chontal šmay- ‘to sleep’; Paipai xʷíx, Mohave hahʷé·k ~ Lowland Chontal Weloo-, Highland Chontal We- ‘to smell’; Paipai oʔo xʷáy ~ Lowland Chontal We-, Highland Chontal Way- ‘to smoke’; Yuma ʔu·nʸé ~ Lowland Chontal, Highland Chontal áne ‘road’; Yavapai kʸúli ‘long’ ~ Lowland Chontal kulʸíʔ, Highland Chontal akulíʔ ‘far’. Lyle Campbell (1976) supports the Jicaque-Hokan hypothesis with a list of 30 resemblances between Jicaque, Chontal, Comecrudo, Seri, Proto-Pomo and Proto-Palaihnihan. According to 171 Mikhail Zhivlov him, “[t]he Jicaque-Hokan hypothesis looks even better when the full spectrum of Hokan languages is compared”. Oswalt (1977) discusses in detail the Proto-Pomoan word *hibal ‘tongue’ and its development in Pomoan languages. He also briefly lists cognates from other Hokan branches. This type of an in-depth treatment of an individual etymology is unfortunately quite rare in Hokan studies. Especially interesting is Oswalt’s demonstration that Pomoan languages have a fossilized noun prefix *Hi-, frequent in terms for body parts. Jacobsen (1979) is a detailed survey of Hokan inter-branch comparisons. Jacobsen gives some figures to show to what extent errors in badly controlled Washo data recorded by Dixon and Kroeber can vitiate the comparisons: “In my opinion at least 10 (16.7%) in the Dixon and Kroeber list, and 14 (13%) in the Sapir list fall into this category. Most of the mistakes involve misanalysis of the forms, usually failure to recognize the presence of more than one morpheme”. A special critical section deals with Gursky (1974) 5. Analyzing two-member sets from Gursky’s compilation, Jacobsen comes to the conclusion that 89 sets represent probable loanwords: “These 89 dubious sets represent 20.5% of the two-member sets, and 12.6% of the total number of sets”. He also finds “the strong onomatopoetic component in at least 23 of the sets, mostly bird names”. Langdon (1979) compares Pomoan and Yuman — the two branches of Hokan for which solid reconstructions were available. Langdon’s work is unlike other “binary comparisons” produced at the time. Instead of merely listing sound correspondences, she tries to show how comparison with Yuman can shed light on the problems of Pomoan reconstruction. According to Langdon (1979: 594-595), “Hokanists recognize that the concept of "Hokan" or for that matter, of any of the other groupings subsuming such ancient connections, differs somewhat from that of the prototypical "language family" in that the relationships it encompasses, while probably partly genetic in the traditional sense, must have been also affected by early as well as ongoing non-genetic areal pressures”. One might argue that prototypical “language families” in this sense, i.e. cases where genetic relationship is not accompanied by subsequent contact, simply do not exist. Langdon’s advice is quite reasonable: “the rational way to approach the "Hokan problem" is to work within the strong hypothesis that Hokan is in fact a language family in the classical sense, thus keeping alive the eventual aim of fully established sound correspondences and solidly reconstructed meaningful elements in the context of their grammatical structure, while simultaneously welcoming and actively seeking evidence for diffused and universal traits”. Langdon shows that some superficial similarities between Pomoan and Yuman languages are due to parallel development. Thus, voiceless resonants, found in Eastern Pomo and Diegueño, result from independent innovations. Langdon suggests that contrast between /t/ and /ṭ/, known to be secondary in Yuman languages (in Proto-Yuman, [t] occurred pre-stress, [ṭ] — post-stress), may also be secondary in Pomoan. She notes that “among the affixes reconstructed by Oswalt, only *t and *tʰ appear” (Langdon 1979: 603). Langdon also demonstrates the parallelism between the effect of stress in Yuman, where consonants are never lenited in pre-stress (root-initial) position, and Pomoan, where pre-stress (root-initial) consonants are accompanied by a “laryngeal increment” (/h/ or /ʔ/). A similar process is attested in Yana (Langdon: 1979: 614-615). The article also includes a discussion of morphosyntactic parallels between Pomoan and Yuman, and a list of 49 lexical comparisons between the two families. Most of Langdon’s comparisons look much more probable than average comparisons by other authors of that period. Webb (1979) compares terms for ‘tree’ and particular tree species across Hokan family. Each of her comparisons is represented in many daughter branches, thus minimizing the risk 5 172 For this work see Section 4 below. Hokan I: A review of comparative studies of areal loans. Despite the statement that “[t]he cognate lexeme sets are recognized on the basis of recurrent consonant correspondences”, no such correspondences are given in the paper. Some of the compared forms are unreliable. E.g., Webb gives Yana ʔaxa ‘tree’, a form I was unable to find in the primary sources. Webb’s inclusion of some Proto-Uralic forms as external comparanda raises the question of just how easy it is to find accidental similarities in the absence of recurrent sound correspondences. The main part of Campbell & Oltrogge’s (1980) article is a detailed reconstruction of Proto-Tol. The paper also contains a brief discussion of external affiliation of Tol. The authors suggest a specific relationship between Tol and Oaxaca Chontal, listing 57 probable cognates. While some of their comparisons do indeed look probable, e.g., Proto-Tol *lo ‘leaf’ ~ Highland Chontal -la ‘leaf’, some others can be shown to be wrong. Take, for example, the following comparison: “to be sick: WJ pte-vé sick, EJ peʔ, pa¢ die; -tafba sick; -fwana sick” (Campbell & Oltrogge 1980: 221). Highland Chontal -tafba ‘sick’ is actually ł-aštaf-ba ‘sick person’, derived from the verb d-eštaf-ʔma ‘to make sick; to get sick’. This form is unrelated to Highland Chontal ł-af-gwana ‘sick one’, derived from ʔal-gwana ‘sickness’. The authors’ conclusion that “regardless of the ultimate outcome of distant genetic research involving Hokan generally ... the Jicaque-Tequistlatec relationship will hold up” (Campbell & Oltrogge 1980: 222–223) is certainly too optimistic. Webb (1980) compares Esselen with other Hokan languages. Again, some of her data are unreliable: I was unable to find the source of Washo amatu- ‘land/earth’. Webb (1981) lists Hokan comparisons for some kinship terms as well as terms such as ‘man’ and ‘woman’. Here, too the compared material is highly unreliable: the word xmár ‘boy/child’, listed as Chimariko, is actually Yuman. While Hokan comparative studies of that period used new accurate field data, they failed to create a working model of comparative Hokan. This is at least partly due to a nonstandard methodology which is discussed in detail in Section 6 below. 4. Attempts at synthesis The next period in Hokan studies, partly overlapping with the previous period, was characterized by attempts at synthesis. The work of Karl-Heinz Gursky (1964, 1965, 1966a, 1966b, 1968, 1974, 1988, 1989, 1990, 1995) logically continues that of Sapir. Like Sapir, Gursky was initially interested in large-scale classification. In his early work he compares Hokan with two poorly attested isolated languages, Quinigua (1964) and Waikuri (1966b), as well as with “Gulf” and the wider “Algonquian-Gulf” (1965, 1966a, 1968) — hypothetical macrofamilies suggested by Mary Haas. Later, Gursky turned to collecting lexical and grammatical comparative sets for Hokan itself. Especially important is Gursky’s magnum opus “Der Hoka-Sprachstamm” (1974) with 707 comparative sets — the largest published compilation of Hokan lexical comparisons. Subsequently, Gursky published a three-part addenda et corrigenda (1988, 1989, 1990). While Gursky occasionally gives preliminary Proto-Hokan reconstructions (marked with # rather than with the usual asterisk in order to emphasize their tentative nature), he thought that a reconstruction of Proto-Hokan would be premature (1974: 173–175). Although methodologically his work adopts the “mass comparison” approach, it is strikingly different from the most widely known examples of this approach, such as Greenberg (1987). Gursky pays close attention to phonological accuracy of the forms compared, trying to verify all the forms with the help of the most reliable sources. He painstakingly distinguishes between forms which can be transcribed phonologically, and forms known only in pre-phonological notation. His attention to 173 Mikhail Zhivlov details led him to an unexpected discovery that Olmsted’s Achumawi dictionary (Olmsted 1966), based on Jaime de Angulo’s fieldnotes, contains a number of Eastern Pomo words. These words were taken from de Angulo’s manuscript comparison of Achumawi and Eastern Pomo, compiled with the aim to show that these two languages are unrelated (Gursky 1987). In his paper “Some grammatical evidence for the Hokan stock” Gursky (1995) gives 138 comparative sets for grammatical morphemes. This is undoubtedly the most extensive compilation of grammatical evidence for Hokan. One can agree with Terrence Kaufman’s assessment of Gursky’s work: “The contribution of Gursky to comparative Hokan studies is in my view quite valuable and probably not given as much attention as it merits. I was amazed at the number of undoubtedly valid grammatical comparisons found in Gursky 1966b. Although his scope of comparison was broader that just Hokan, the Hokan data can be examined by itself. ... The work of K-H Gursky in assembling likely Hokan lexical cognate sets (especially Gursky 1974) deserves the highest praise also. Even though he used all the previously published work of other scholars, he has found much that is new, he has judiciously split apart sets that were unlikely in the first place, and has brought all this together in the compass of two or three major articles” (Kaufman 1989: 111). In his well-known monograph “Language in the Americas”, Joseph Greenberg (1987) gives a list of 168 exclusively Hokan comparisons plus 100 Hokan comparisons with parallels in other “Amerind” languages. Greenberg’s Hokan is maximally inclusive: it consists of “Achomawi (including Atsugewi), Chimariko, Chumash, Coahuilteco, Comecrudo, Cotoname, Esselen, Jicaque, Karankawa, Karok, Maratino, Pomo, Quinigua, Salinan, Seri, Shasta (including Konomihu, probably a distinct language), Subtiaba (including Tlappanec), Tequistlatec (sometimes called Chontal of Oaxaca), Tonkawa, Waicuri, Washo, Yana, Yuman, and Yurumangui” (1987: 132). Thus, Greenberg calls Hokan what others called “Hokan-Coahuiltecan”. Greenberg was an advocate of the method of “mass comparison” and his Hokan comparisons are assembled based on superficial similarity rather than sound correspondences. Greenberg’s treatment of Salinan and Yurumangui data was criticized by Poser (1992). The years 1988–1990 saw two attempts at reconstructing Proto-Hokan based on regular sound correspondences. One of these was made by Dmitry Leshchiner with the assistance of Sergei Nikolaev (Leščiner 1989, 1990; Leshchiner & Nikolaev 1992). Leshchiner reconstructs ProtoHokan consonants, vowels and tones, giving their reflexes in Achumawi, Karuk, Yana, ProtoPomoan, Proto-Yuman, Seri and Tequistlatecan. The reconstructions are accompanied by 295 cognate sets, which makes Leshchiner’s work the only one with both cognate sets and explicit ProtoHokan reconstructions. Unfortunately, Leshchiner’s reconstruction has a number of serious drawbacks. First, his Proto-Hokan consonant and vowel systems (Tables 2 and 3) are much larger than any system in any of the attested daughter languages. He reconstructs 58 consonants and 11 vowels (not counting vowel length). Typologically, this system looks rather unexpected when compared to attested phonological systems of Hokan and other Californian languages. Table 2. Proto-Hokan consonants, after Leshchiner & Nikolaev (1992). ᵐb ᵐp’ p’ d t’ ᵐpʰ pʰ (tʰ) ᵐf f s m n ᵐp ᵐw 174 b p w t r ḍ ṭ’ ṭʰ ṭ ṛ ṇ ƛ’ č’ čʷ’ k’ (kʷ’) k kʷ ƛʰ čʰ čʷʰ kʰ ł š šʷ x ƛ l č y čʷ ŋ q’ qʷ’ kʷʰ qʰ qʷʰ xʷ x̣ x̣ʷ q qʷ ʔ Hokan I: A review of comparative studies Table 3. Proto-Hokan vowels, after Leshchiner & Nikolaev (1992). i e ɛ ü ö ɨ u a ɔ ə o Second, probable borrowings were not systematically excluded: “When we weren’t sure that the words compared are not true cognate, then this item remains in the dictionary. We note the cases which strongly seem to be borrowings” (Leshchiner & Nikolaev 1992: 366). This strategy can only lead to proliferation of spurious sound correspondences, which in turn leads to overbloated protophonology. Third, many comparisons exhibit loose semantics, leading one to suspect that the forms compared are neither true cognates nor loans. Thus, the list of Hokan body parts (Leshchiner & Nikolaev 1992: 373–380) contains the following comparisons: ‘cheek ~ forehead’; ‘nose ~ forehead’; ‘brain, Adam’s apple, head ~ eye’; ‘spleen ~ milt ~ body, meat’; ‘back ~ shoulder blade ~ elbow’; ‘calf of leg ~ elbow’; ‘shoulder blade ~ arm ~ finger, hand ~ thigh, inside of leg’; ‘palm of hand ~ arm above elbow ~ thigh’; ‘ankle ~ nail’. Finally, the regularity of correspondences is also in doubt. As an example, we can look at the reflexes of the alleged Proto-Hokan phoneme *ᵐb (see Table 4). This phoneme should give /v/ in Karuk, at least word-initially, but out of four examples with Karuk reflexes, two have /v/, one /p/ and one /ʔ/. Yana should have /m/, but out of five cases with Yana reflexes, three have /m/ and two /p/ without any complementary distribution. Moreover, none of Leshchiner’s papers give any comments on complementary distribution of reflexes. Overall, Leshchiner’s work cannot be used as a working model in further attempts to reconstruct Proto-Hokan. Table 4. Reflexes of Proto-Hokan *ᵐb in Leshchiner & Nikolaev’s (1992) reconstruction. PHo* Ač K Ya PPo* PYu* Se Ts *ᵐb m/p- v- m- b m- ?-/-p- ø- pōk vásih *bak’ṓ *mák *ᵐbāk’ṍ ‘back’ *siᵐb[ü] ‘moss’ simi?túl (PR sap̄’a) *xäᵐbu ‘carrot’ hamuč (PR h’ām’ač) ‘wild carrot’ *ᵐb˳xʌ ‘bear’ *ᵐbHéšö ‘meat’ *ᵐbaka ‘woodpecker sp.’ *ᵐbūk’ǻ ‘worm’ mísuč (PR mīsuč’) *max˳á ‘badger’, *max˳áta ‘bear’ mūxa (N) ‘porcupine’ pū́fič ‘deer, venison’ ?ī́š pa(na) ‘deer, deer meat’ *bo?ó ‘to hunt; elk’ pasi (N), pahsi (Y) *bixe makmáka amuq (PR āmoq’ ‘caterpillar, grub, maggot’) (Pe) [xālḗ] sībú ‘tree moss’ (?) la-picúlâ? ‘buttocks’ (Pe) šībú wah’ (PR woh’) *ᵐb[ō]wʌ́ páw (PR puwwāwi ‘cow’) (~-- ?) ‘elk’ mak’i ‘back, backbone’ *ma?úl ‘antelope’ ?ap ‘venado buro’ ipḫási l-išík’ (Pe) bākáka vákaj cāmuk’u (N) ‘acorn worm’ ł-âmúšâkʰ ‘large white worm sp.’ 175 Mikhail Zhivlov In 1989, Terrence Kaufman, a leading authority on Mesoamerican languages, published his work “A Research Program for Reconstructing Proto-Hokan: First Gropings”. The 119-page paper contains a survey of Hokan phonological systems, tables of sound correspondences between Hokan branches and a list of 1118 Proto-Hokan reconstructions. Kaufman’s Hokan includes the following languages: Pomoan, Chimariko, Yana, Karuk, Shastan, Achumawi-Atsugewi 6, Washo, Esselen, Salinan, Yuman, Cochimí, Seri, Coahuilteco, Comecrudoan, Chontal, Tol. Kaufman reconstructs the following phonological system for Proto-Hokan (see Tables 5, 6). Table 5. Proto-Hokan consonants, after Kaufman (1989) p p’ pʰ f m ṭ ṭ’ ṭʰ r l n t t’ tʰ θ ¢ č ¢’ č’ ¢ʰ čʰ s š kʸ kʸ’ kʸʰ xʸ q q’ qʰ x̣ kʷ kʷ’ kʷʰ xʷ 7 h lʸ nʸ y w Table 6. Proto-Hokan vowels, after Kaufman (1989) i e u a o length /:/ stress /ˊ/ This work might have put an end to the debate on the validity of Hokan, were it not for one thing: Kaufman lists his reconstructions but omits the data underlying those reconstructions. The absence of reflexes in Kaufman’s work makes its evaluation a hard, although not impossible, task. Kaufman’s Proto-Hokan phonological system looks typologically and areally plausible for a Californian language, and his reconstructions look more realistic compared to those by other authors. Compare Kaufman’s #(a-)lʸafu ‘navel’ (Kaufman 1989: 133) with Haas’ *imarak̓ʷi ~ *imak̓ʷari ‘navel’ or Leshchiner’s *ʔä́Hlúp’ʌ ‘navel’. Kaufman’s protoform looks more similar to such forms as Chimariko -onapu ‘navel’, Karuk ʔárup ‘navel, navel cord’, Shasta ʔéˑraw ‘navel’ and Proto-Yuman *mlʸpu(:) ‘navel’ (Miller 2018: A61). Nevertheless, some of Kaufman’s reconstructions must be rejected in the light of new evidence. For instance: (a) #saHmaH [s ~ š] ‘heart’ (Kaufman 1989: 133) is based on the following comparison by Gursky (1974: 193): Eastern Pomo sa·ma·i ‘heart’, Highland Chontal -unšahmaʔ ‘heart’. It is widely believed that Achumawi and Atsugewi constitute a language family (or a subfamily of Hokan), sometimes called Palaihnihan. There is a published reconstruction of Proto-Palaihnihan phonology (Olmsted 1964). However, Nevin (2019) has persuasively shown that Olmsted’s reconstruction is based on faulty methodology and inaccurate data. According to Nevin, “a proto-language ancestor common to Achumawi and Atsugewi has not been reconstructed, and they cannot be claimed to be more closely related to each other than either is to the Shastan languages or perhaps even to Yana” (Nevin 2019: 48). 6 176 Hokan I: A review of comparative studies In fact, the Highland Chontal form is derived from the verbal stem -unšá- ‘to breathe’ with the instrumental suffix -hmaʔ, therefore, the comparison should be rejected. (b) #i-7ip ‘navel’ (Kaufman 1989: 133) is based on the following comparison by Judith G. Crawford (1976: 316): Seri ʔatóošx iʔíip ‘navel’, Washo íʔib ‘navel’ The Seri word literally means ‘where the umbilical cord is’, from ʔatoošx ‘umbilical cord’ and iʔiip ‘where it is’, derived from the verb kaap ‘to be’ (Moser & Marlett 2010: 76, 379, 435). While the similarity of Washo and Seri forms is striking, it is undoubtedly due to chance resemblance. (c) #panA [a ~ o] ‘river’ (Kaufman 1989: 160) is based on a comparison by Campbell & Oltrogge (1980: 222): Tol sɨ-pones ‘river’, Highland Chontal -banaʔ ‘river’ The Tol word is most likely a compound of ʔɨsɨ́ ‘water’ (Dennis & Royce de Dennis 1983: 55) and pɨné ‘big’ (Dennis & Royce de Dennis 1983: 30). In some other cases, Kaufman’s reconstructions can be amended based on new data. (d) #x̣owK ‘coal’ (Kaufman 1989: 142) is probably based on the following entry in Gursky (1974: 183): “COAL Ach //həwk// coal; Sh //xúkʷ// coal (ember), charcoal; Chi kōwa coals”. Judging by Kaufman’s reconstruction and correspondences, the Chimariko word does not belong here. The diphthong in #x̣owK is apparently based on the Achumawi word, for which the online “Achumawi Dictionary” gives variants h̓úk and h̓uq̓ (Nevin 2020). Thus, we can safely assume that *ow should be replaced with *u in the reconstruction of this word. The cases above are, however, exceptional, and on the whole Kaufman’s comparisons hold water remarkably well. For example, Kaufman reconstructs the following pair of words for Proto-Hokan: #a-px̣a ‘shit’ and #i-px̣a ‘guts’ (Kaufman 1989: 133). While some daughter languages preserve only one of the two words, the opposition between the two is preserved in Southern Pomo ʔahpʰa ‘excrement’, ʔihpʰa ‘intestines, guts’ (Oswalt 1981: 17, 22) and Chimariko ápx̣a ‘excrement’, ípx̣a ‘intestines’ (Berman 2001: 1053). Since there is no evidence of contact between Chimariko and Pomoan, this remarkable parallel is best viewed as inherited from the common ancestor. The two words share the same root but have different fossilized prefixes. These prefixes are not synchronically attested in Pomoan and Chimariko, but they can be identified with Kaufman’s Proto-Hokan #7a:- 1. ‘absolutive of intimately possessed noun’; 2. ‘substance or mass noun prefix’ and #Hi:- 1. ‘body-part prefix’; 2. ‘possessed state of intimately possessed noun’ (Kaufman 1989: 117), thus confirming Kaufman’s morphological reconstruction. Another pair of reconstructions having the same prefixes is #a:-xʸá7 ‘water’ and #i-xʸa7 ‘juice’ (Kaufman 1989: 131). Here, the opposition of the two prefixed forms is preserved in Seri ʔaχ ‘water’ vs. iχ ‘liquid, sap, juice’ and Highland Chontal l-ahaʔ ‘water’ vs. l-iha in l-iha-ʔmał ‘mescal (lit. its-water the maguey)’ (Turner & Turner 1971: 179, 194). Unlike the case with Chimariko and Pomoan, Seri words contain productive prefixes ʔa- ‘absolute (without explicit possessor)’ (Moser & Marlett 2010: 298–299) and i- ‘third person possessor’ (Moser & Marlett 2010: 386). In this case we deal not with isolated lexical parallels, but with reconstructed items that can be morphologically segmented on the Proto-Hokan level. Kaufman (2015) 7 provides a model of Proto-Hokan grammar and a list of approximately 150 reconstructions of grammatical morphemes. Once again, the reflexes in daughter lan7 This paper is a slightly updated version of a manuscript written in 1989. 177 Mikhail Zhivlov guages are omitted, but, unlike Kaufman 1989, this study lists the languages on which a particular reconstruction is based. Langdon (1990) offers a new approach to Hokan morphological comparison: “The alternative then suggested itself to focus on elements of the stem without prejudging whether they function as prefixes, roots, or suffixes. Crucial here is Jacobsen's (1980) notion of bipartite verb stems in Washo, where two basic parts of a stem are postulated but where each part can consist of a root. So instead of thinking of a Hokan verb stem as consisting of one (or more) prefixes plus a root, it seemed more appropriate to focus simply on how many distinct parts a verb stem consists of, and the semantic patterns in which they combine. The consequence is a view of Hokan verb forms as consisting of a sequence of elements (possibly all basically lexical) whose various combinations and recombinations in the daughter languages reshape the identifying criteria of the distinction between root and affix so that the issue of what is a root vs. what is an affix, while important for the synchronic structure of each language, is perhaps less crucial historically” (p. 66). Langdon reconstructs the following Hokan kernel stem structure (Table 7): Table 7. Hokan stem structure, after Langdon (1990) I Bipartite I II Plural infixed III Bipartite II IV Direction/Motion V Plural “The basic insight — if that is what it is — is that in a particular language and from the perspective of comparative Hokan, one cannot assume that prefixes must be compared to prefixes (and only prefixes) and roots only to roots” (1990: 67). For example, “Bipartite I” yields so-called instrumental prefixes in, e.g., Pomoan, but so-called primary stems in Yana. 5. Hokan studies on the wane The period from the 1990s up to the present is characterized by a decline of interest in the Hokan hypothesis. Symbolically, the last of the Hokan-Penutian conferences, which were regularly held since the 1970s, took place in the year 2000 (Golla 2011: 304–305). Poser (1995) argues against Greenberg’s (1987: 132) assertion that binary (rather than multilateral) comparisons have done much harm to Hokan studies. Poser points out that even socalled “binary comparisons” often included forms from all branches of Hokan, being thus essentially multilateral in Greenberg’s sense. “Ironically, the real problems with Hokan classification are due to the very causes that make Greenberg's methodology questionable. What is controversial about Greenberg's methodology is not his comparison of many languages. It is his failure to establish the systematic correspondences between languages that remove the possibility of chance similarity, his lack of concern for loans, and his willingness to postulate relationship on the basis of minute amounts of evidence, precisely the weaknesses that affect parts of Hokan. ... If we are to learn more about which languages really belong to Hokan and how the family is to be subgrouped, it will be via the program laid out by Haas (1963), namely, the acquisition of more and better data, the establishment of phonological correspondences, and the reconstruction of Proto-Hokan and intermediate proto-languages. It is not from the assemblage of lists of vaguely resemblant words that we can expect to improve our knowledge, but from the fieldwork that continues to be done, the study of the vast store of Harrington material, and such initiatives as Kaufman's (1988) reconstruction of Proto-Hokan” (Poser 1995: 142). 178 Hokan I: A review of comparative studies Lyle Campbell’s (1997) monograph “American Indian Languages: The Historical Linguistics of Native America” contains a section with a detailed critique of the Hokan hypothesis. His justified criticism of Haas’ methodology was already cited above. Nevertheless, some of Campbell’s criteria seem overly strict. Thus, by his count, twenty-six comparisons in Haas (1964) are to be rejected as “short forms” (1997: 294). The problem is that many Hokan languages have short roots. Thus, Jamul Tiipay roots have the canonical structure (C)V(C) (Miller 2001: 11). Excluding short forms from comparison amounts to declaring that the comparative method is inapplicable to languages with short canonical roots (e.g., many languages of Southeast Asia). Campbell’s verdict on Hokan is rather pessimistic: “Given the reservations expressed here concerning the many Hokan studies, but also taking into account Langdon's more promising comparisons (and Kaufman's optimism—see the next subsection), I conclude that it is by no means clear or even likely that there was a proto language from which some or most of the putative Hokan languages diverged long ago, but that this hypothesis is fully worthy of continued research”. Regarding Kaufman’s (1989) work, Campbell says: “While Kaufman's proposals have stimulated some other linguists to accept more positive attitudes toward Hokan, they can be evaluated appropriately only after he presents the lexical evidence upon which they are based. Therefore, for the present, we are left with essentially the same uncertainty that has always attended the Hokan hypothesis — there certainly is enough there to make one sympathetic to the possibility of genetic relationship, and yet the evidence presented to date is not sufficient to confirm the hypothesis, regardless of which languages are included”. Mixco (1997) reviews the work of Mary R. Haas on Hokan. According to him, “Haas revealed herself to be an exemplar of a truly "neogrammarian" stamina in the face of the most persistently intractable problems that tax the student of the linguistic prehistory of the Americas (or of the world, for that matter). Time and time again, she treats Hokan as a hypothesis worthy of continuing intellectual devotion. At no point does the knotty tangle of obstacles in the path of a definitive reconstruction discourage her from further efforts”. Mixco defends Haas’ use of metathesis in Hokan comparisons, pointing out that metathesis frequently occurs within the uncontroversial Yuman family. In his publication of Sapir’s Chimariko linguistic material, Berman (2001) lists several comparisons between Chimariko, Proto-Pomo and Proto-Yuman (see Table 8). Berman also notes that Chimariko suffixes -kʿ ‘motion towards here’ and -m ‘motion towards there’ resemble Yuma suffixes -k ‘towards here’ and -m ‘towards there’. The pair *-k ‘hither’ / *-m ‘thither’ also occurs in Achumawi, Atsugewi and Shasta (Zhivlov 2018: 154–157). Table 8. Chimariko-Pomo-Yuman comparisons, after Berman (2001) Chimariko earth eat excrement ám·a hám·a to eat ápx̣a Proto-Pomo Oswalt Proto-Yuman *ʔa(h)ma· Proto-Pomo McLendon *ʔahmáṭ / ʔamáṭ *maʔa- *maʔá food *ʔahpʰa *ʔahpʰá *ma· eat mush *cihmé/ci·me *ʔ-mat hair (body) / fur hím·ı hair (of any kind) *zi(h)me intestines ípx̣a *ʔihpʰa moon ála, allaʳ moon, sun *ʔala·ša q’alá·(xa) / ʔal·á·(xa) *xlʸʔa· *ahqʰa *ahqhá *ʔ-xa water äqʿ *-mi(y)(s) *pxa (Kiliwa pʰaʔ) 179 Mikhail Zhivlov Good (2002), using material from Californian Hokan languages, attempts to show that Proto-Hokan had a three-vowel system (*i *a *u) rather than a five-vowel system (*i *e *a *o *u), reconstructed by Kaufman (1989). However, Good’s comparisons are purely structural, in the sense that no actual cognate sets are discussed. While this is understandable given that Kaufman (1989) does not list the reflexes of his Hokan protoforms, Good’s conclusions can be verified or falsified only by looking at actual correspondences as manifested in cognate sets. Marlett (2007) reviews the evidence for the relationship between Seri, Yuman and Oaxaca Chontal and comes to the conclusion that the relationship has not been demonstrated. Appendix 1 includes a revision of the resemblances between these languages proposed by earlier researchers, with detailed comments on the morphological structure of Seri forms. Marlett (2008) gives a very useful critique of Bright’s (1956) Seri-Salinan comparisons. He concludes that “[w]hen the least likely candidates are eliminated, the percentage of “correlates” between Seri and Salinan is lower than between Seri and the Otomanguean data presented as “Supanec.” This would suggest that the points of comparison may be coincidences— not evidence of a genetic connection. ... Seri and Salinan should be listed as isolates until adequate evidence of a relationship with some language is published and evaluated”. In his book “California Indian Languages”, Victor Golla gives a brief survey of “The Hokan Phylum” (Golla 2011: 82–84) as well as a sketch of the linguistic prehistory of Hokan languages (Golla 2011: 242–248). Golla summarizes his position on the Hokan and Penutian hypotheses in the following way: “My position on the historical validity of these hypothetical relationships is somewhat more positive than that of such methodological conservatives as Lyle Campbell (1997), although considerably more cautious than that of such enthusiasts as Paul Radin (1919), Morris Swadesh (1959), and Joseph Greenberg (1987), or even Edward Sapir in his less restrained moments (as, for example, in Sapir 1921c). While the categories "Hokan" and "Penutian" are undoubtedly meaningful—no one would ever classify Shastan as Penutian or Klamath-Modoc as Hokan, and the consensus is now clear that Chumash and Yukian belong to neither (nor to a phylum of their own)—we continue to debate the nature of the historical relationships these categories imply. … I hope, however, that I make clear the highly speculative nature of such hypotheses, and I try to indulge my predilection for them only when, in my view, they appear to cast some useful light on the past” (Golla 2011: 239). Jany (2013) is a comprehensive bibliography on Hokan languages. The author takes a neutral stance on the question of their genealogical relationship. Haynie (2014) is a publication of one of the chapters of the author’s PhD thesis (Haynie 2012). The work applies a sophisticated statistical methodology to the question of deep relationship between California languages, comparing the initial segments of words in search for recurrent sound correspondences. The results are rather unfavorable for the Hokan hypothesis: “Both in its measure of relationship strength and in its evaluation of statistical significance, this test of linguistic relationships among California languages offers no support for Hokan or Penutian” (Haynie 2014: 439). It must be noted, however, that the limitation of comparison to initial segments does not suit well languages with developed prefixation, where prefixes can easily get fossilized. Consider the following words for ‘tongue’ from Haynie’s wordlists: Achumawi iplē·, Cocopa mpałʸ and Southeastern Pomo bal (Haynie 2012: 283). It is clear that the historically relevant comparison in this case should include Achumawi p, Cocopa p and Southeastern Pomo b rather than Achumawi i, Cocopa m and Southeastern Pomo b. Jany (2017) reviews features of polysynthesis in the so-called Northern Hokan languages and comes to the conclusion that “[s]tructural similarities stem from language contact rather than from genetic affiliation”. 180 Hokan I: A review of comparative studies Zhivlov (2018) gives a systematic survey of some aspects of reconstructible Hokan morphology, namely person and number markers, lexical prefixes, plural infixes, and directional suffixes. The article ends with the following conclusion: “While it has been frequently pointed out that areal diffusion in Northern California makes it hard to distinguish between genetic relationship and borrowing (Jany 2016), some results of the present study suggest a more optimistic outlook. Thus, if the hypotheses offered in this paper are correct, some features of Proto-Hokan (plural infixes and a simple directional system opposing ‘hither’ and ‘thither’ forms) are better preserved by languages outside of the Northern California linguistic area.” (Zhivlov 2018: 158). In his monograph on Esselen, Shaul (2019) provides evidence for the Hokan affiliation of this language. In Shaul’s own words, “When I began the research for this reworking of Esselen, I was skeptical about the Hokan hypothesis. I am no longer: Kaufman's comparative treatment of Hokan (2015) gave me ways of explaining things in Esselen that I could not otherwise explain”. Shaul gives Esselen reflexes for Kaufman’s Hokan reconstructions and adds a number of Hokan reconstructions of his own. Shaul (2020a) provides similar evidence for the Hokan affiliation of Salinan. He defines a “Hokan common core”, consisting of 279 reconstructions with wide distribution in Hokan. Out of these, 126 have reflexes in Salinan. These reflexes are listed in Appendix A together with Proto-Hokan reconstructions, but without reflexes in other Hokan branches. This makes it difficult to verify proposed Hokan etymologies of Salinan words. One more problem is that Shaul’s works on Salinan (2020a, 2020b) completely ignore one of the two phonologically reliable sources of Salinan data — J.P. Harrington’s fieldnotes (the other reliable source — William Jacobsen’s fieldnotes — is less extensive). With data that are largely unreliable in respect to their phonology, it is much harder to establish regular sound correspondences. Lyle Campbell’s (2024) new monograph “The Indigenous Languages of the Americas: History and Classification” has a section on the Hokan hypothesis. Compared to the similar section in Campbell (1997), it contains a detailed discussion of Terrence Kaufman’s work on Hokan. Among Kaufman’s 1118 Hokan comparisons, Campbell finds 242 which occur only in a single linguistic area, 15 with considerable semantic difference in the languages compared, 58 onomatopoetic forms, 17 nursery forms, 23 possible borrowings, as well as a number of short forms and words found also in languages beyond Hokan (Campbell 2024: 348–352). Campbell does not deny, however, that there is a core of very stable words found in most of Hokan languages. These words “must include at least forms with the following meanings: arm/shoulder, ashes, body / meat / deer, bone, dog, earth / ground, eye, grease / fat, hand / give, knee, many, skin, sleep, stone, tail, tongue, and water” (Campbell 2024: 386). He entertains the idea that “the thirty or so words mentioned above that seem like possible cognates that turn up across many of the “Hokan” languages might be inherited not from some Proto-Hokan, but from some Ur-American language of long ago, maybe even one spoken by some of the first inhabitants in the New World”. Another possible option is “to suspect that these thirty or so words are not evidence of a Hokan genealogical connection among those particular languages but rather are instead pan-Americanisms”, i.e. “those seemingly similar forms that appear to pop up repeatedly in broad comparisons of American Indian languages not known to be related to one another, probably a collection of accidentally similar forms, loanwords (maybe including some Wanderwörter), unrecognized onomatopoeia, sound symbolism, and the like” (Campbell 2024: 353). However, Campbell makes no attempt to demonstrate that these “core” Hokan comparisons actually occur in other Native American language families, and without such a demonstration the idea that they go back to a deeper level than Hokan remains speculative. Campbell’s overall assessment of Hokan remains pessimistic: “I, however, remain very 181 Mikhail Zhivlov skeptical about Hokan in general. Nevertheless, there are those tantalizing recurrent similarities that need explanation, and it would be difficult to reject Kaufman’s work entirely, and so research on possible Hokan connections should continue. However, my strong belief is that if there ever was a Hokan genetic unity, it was so far in the past and so much has changed in the languages since then that it will never be possible to confirm a genetic relationship among those languages on the basis of the information that remains today” (Campbell 2024: 353). 6. Conclusion There are two possible answers to the question of why there is still no convincing demonstration of Hokan relationship — or, vice versa, no convincing attempt to explain away all similarities as due to contact. One possibility is that the problem lies in the nature of evidence — it is so meager that the comparative method cannot be applied (either because the languages are not related or because of the great time depth of the family). This is essentially the view of Poser (1995: 140), when he says that “[m]any of the links were originally posited on the basis of very slim evidence, consisting of unsystematic similarities in a very small number of words”. Cf., however, the opposite point of view voiced by Kaufman (1989: 64): “It is a common observation that there is a very small number of etymologies (usually characterized as being "about thirty") that have reflexes in most of the Hokan families and isolates, and that there do not seem to be any more new and good sets forthcoming. This is not a totally appropriate characterization nor does it show a very constructive attitude. A number of new correspondences and etymologies have been dug up in the course of the binary comparisons referred to above, and one should not expect proto-forms to survive in most of the daughter languages. A random scatter is quite good enough and is in any case what is to be expected”. Another possible answer, which I find more compelling, is that the problem lies with the methodology used by the majority of the proponents of Hokan. From this point of view, two types of studies can be delimited. One type uses the method of mass comparison, i.e. comparisons of words based on superficial resemblances rather than on sound correspondences. Studies of this type include Sapir (1917, 1920c, 1925) and Gursky (1974, 1995). At the time of Sapir, there was no alternative to mass comparison in Hokan studies, since there were no phonologically accurate materials on most Hokan languages, and consequently, no possibility to apply the comparative method. As noted above, Gursky used mass comparison because he thought that applying the comparative method on the level of Proto-Hokan would be premature. While mass comparison can be useful as a first step in collecting the materials for further application of the comparative method, its usefulness is limited, and no conclusive demonstration of distant language relationship can be achieved by this method. The second type of studies, mostly authored by Mary Haas and her students, applies a methodology which for the lack of a better name I will call “partial comparative method”. The difference of this methodology from the standard comparative method was not commented upon previously. The traditional comparative method is a cyclic procedure. It involves finding sound correspondences, looking for complementary distributions between them, and then assigning each correspondence or a group of correspondences to a phoneme of the protolanguage. The next step involves getting rid of comparisons that do not fit in the system and adding new comparisons that do show systematic, even if initially unexpected, correspondences. Then one should revise the correspondences in search of new complementary distributions. The initial set of comparisons is thus constantly revised in order to arrive at more precise protoforms. As a result one gets a coherent system of sound correspondences. The “partial 182 Hokan I: A review of comparative studies comparative method” is different. It requires listing sound correspondences found in the data and sometimes grading them by frequency. No attempt is made to revise the correspondences in order to fit them into one system. This allows a reasearcher to retain multiple intersecting correspondences without complementary distributions. Cf. the following subset of SeriYuman correspondences in J. G. Crawford (1976: 321–322): t ~ *t, t ~ *ṭ, t ~ *s, t ~ *š, s ~ *s, s ~ *š, š ~ *s, š ~ *š, š ~*t, š ~ *ṭ. The traditional comparative method would require either finding complementary distributions or rejecting some of the correspondences together with lexical comparanda containing them. Another feature of the “partial comparative method” is the looseness when it comes to the formal accountability of compared words. This is obvious in all those cases where entire wordforms are brought into comparison as evidence for the correspondence of individual segments only, with no account being offered for the residue. The crucial stages missing in the “partial comparative method” are searching for complementary distributions and rejecting those comparisons which do not fit the system. As a result of this, such studies contain many quite improbable or downright wrong comparisons, which is one of the reasons why the Hokan hypothesis lost its appeal to many researchers in the following decades. Only a small minority of Hokan studies attempt to apply the traditional comparative method. Leshchiner & Nikolaev’s (1992) use of the comparative method, while seemingly more sophisticated than the “partial comparative method” of Haas and her students, is marred by inconsistencies in sound correspondences and semantic looseness, which is why their results cannot serve as a proof of Hokan hypothesis. The only study so far to consistently employ the traditional comparative method is Kaufman (1989), but the absence of actual data makes it quite difficult to verify its results. Overall, while the genetic nature of relationship between Hokan languages seems plausible in the light of some works reviewed above, especially that by Sapir and Kaufman, none of the studies published so far can be said to constitute a definite proof of the Hokan hypothesis. Still, most of these studies contribute at least something to the growing list of probable comparisons between Hokan branches. The fact that in many of Hokan studies such comparisons are buried beneath a heap of “etymological rubbish” should not discourage us from trying to verify or falsify these comparisons in the light of currently available data. Re f e r e nc e s Berman, Howard. 2001. [Sapir’s] Chimariko linguistic material. In: Victor Golla, Sean O’Neill (eds.). The Collected Works of Edward Sapir XIV: Northwest California Linguistics: 1039–1080. Berlin / Boston: De Gruyter Mouton. doi:10.1515/9783110879803.1039. Bright, William. 1954. Some Northern Hokan relationships: a preliminary report. In: Papers from the Symposium on American Indian linguistics held at Berkeley July 7, 1951 (University of California Publications in Linguistics 10): 63–67. Berkeley / Los Angeles: University of California Press. Bright, William. 1956. Glottochronologic counts of Hokaltecan material. Language 32(1): 42–48. doi:10.2307/410651. Bright, William. 1957. The Karok language (University of California Publications in Linguistics 13). Berkeley / Los Angeles: University of California Press. Bright, William. 1970. On linguistic unrelatedness. International Journal of American Linguistics 36(4): 288–290. doi:10.1086/465123. Bright, William & David L. Olmsted. 1959. A Shasta vocabulary. Kroeber Anthropological Society Papers 20: 1–55. Brinton, Daniel G. 1891. The American race: a linguistic classification and ethnographic description of the native tribes of North and South America. New York: N. D. C. Hodges. Campbell, Eric W. 2017a. Otomanguean historical linguistics: Past, present, and prospects for the future. Language and Linguistics Compass 11(4): e12240. doi:10.1111/lnc3.12240. 183 Mikhail Zhivlov Campbell, Eric W. 2017b. Otomanguean historical linguistics: Exploring the subgroups. Language and Linguistics Compass 11(7): e12244. doi:10.1111/lnc3.12244. Campbell, Lyle. 1976. The linguistic prehistory of the Southern Mesoamerican periphery. In: Fronteras de Mesoamérica: 14a mesa redonda: vol. 1, 157–183. Mexico: Sociedad Mexicana de Antropología. Campbell, Lyle. 1997. American Indian languages: the historical linguistics of Native America (Oxford Studies in Anthropological Linguistics 4). New York: Oxford University Press. doi:10.1093/oso/9780195094275.001.0001. Campbell, Lyle. 2024. The indigenous languages of the Americas: history and classification. New York: Oxford University Press. doi:10.1093/oso/9780197673461.001.0001. Campbell, Lyle & David Oltrogge. 1980. Proto-Tol (Jicaque). International Journal of American Linguistics 46(3): 205–223. doi:10.1086/465655. Conathan, Lisa. 2002. Split intransitivity and possession in Chimariko. In: Lisa Conathan, Teresa McFarland (eds.). Proceedings of the 50th Anniversary Conference, June 8-9, 2002, University of California at Berkeley (Survey of California and Other Indian Languages, Report 12): 18–31. Berkeley: Department of Linguistics, University of California. Available online at: https://escholarship.org/uc/item/01c141b1. Crawford, James M. 1976. A comparison of Chimariko and Yuman. In: Margaret Langdon, Shirley Silver (eds.). Hokan Studies: Papers from the First Conference on Hokan Languages held in San Diego, California April 23-25, 1970 (Janua Linguarum: Series Practica 181): 177–191. The Hague / Paris: Mouton. doi:10.1515/9783110819113-012. Crawford, Judith G. 1976. Seri and Yuman. In: Margaret Langdon, Shirley Silver (eds.). Hokan Studies: Papers from the First Conference on Hokan Languages held in San Diego, California April 23-25, 1970 (Janua Linguarum: Series Practica 181): 305–324. The Hague / Paris: Mouton. doi:10.1515/9783110819113-020. Dennis, Ronald K. & Margaret Royce de Dennis. 1983. Diccionario Tol (Jicaque)–Español y Español–Tol (Jicaque). Tegucigalpa: El Instituto Lingüístico de Verano en colaboración con El Instituto Hondureño de Antropología e Historia. Dixon, Roland B. 1905. The Shasta‐Achomawi: a new linguistic stock with four new dialects. American Anthropologist 7(2): 213–217. doi:10.1525/aa.1905.7.2.02a00030. Dixon, Roland B. 1910. The Chimariko Indians and language. University of California Publications in American Archaeology and Ethnology 5(5): 293–380. Dixon, Roland B. & Alfred L. Kroeber. 1903. The native languages of California. American Anthropologist 5(1): 1–26. doi:10.1525/aa.1903.5.1.02a00020. Dixon, Roland B. & Alfred L. Kroeber. 1912. Relationship of the Indian languages of California. American Anthropologist 14(4): 691–692. doi:10.1525/aa.1912.14.4.02a00120. Dixon, Roland B. & Alfred L. Kroeber. 1913a. New linguistic families in California. American Anthropologist 15(4): 647–655. doi:10.1525/aa.1913.15.4.02a00050. Dixon, Roland B. & Alfred L. Kroeber. 1913b. Relationship of the Indian Languages of California. Science 37(945): 225–225. doi:10.1126/science.37.945.225. Dixon, Roland B. & Alfred L. Kroeber. 1919. Linguistic families of California. University of California publications in American archaeology and ethnology 16(3): 47–118. Golla, Victor. 2011. California Indian languages. Berkeley: University of California Press. Good, Jeff. 2002. The vowel systems of California Hokan. In: Lisa Conathan, Teresa McFarland (eds.). Proceedings of the 50th Anniversary Conference, June 8-9, 2002, University of California at Berkeley (Survey of California and Other Indian Languages, Report 12): 32–46. Berkeley: Department of Linguistics, University of California. Available online at: https://escholarship.org/uc/item/1jd6h9mn. Greenberg, Joseph H. 1987. Language in the Americas. Stanford, California: Stanford University Press. Greenberg, Joseph H. & Morris Swadesh. 1953. Jicaque as a Hokan language. International Journal of American Linguistics 19(3): 216–222. doi:10.1086/464222. Gursky, Karl-Heinz. 1964. The linguistic position of the Quinigua Indians. International Journal of American Linguistics 30(4): 325–327. doi:10.1086/464792. Gursky, Karl-Heinz. 1965. Ein lexikalischer Vergleich der Algonkin-Golf- und Hoka-Subtiaba-Sprachen (1). Orbis 14: 160–215. Gursky, Karl-Heinz. 1966a. Ein Vergleich der grammatikalischen Morpheme der Golf-Sprachen und der HokaSubtiaba-Sprachen. Orbis 15: 511–537. Gursky, Karl-Heinz. 1966b. On the historical position of Waikuri. International Journal of American Linguistics 32(1): 41–45. doi:10.1086/464877. 184 Hokan I: A review of comparative studies Gursky, Karl-Heinz. 1968. Gulf and Hokan-Subtiaban: new lexical parallels. International Journal of American Linguistics 34(1): 21–41. doi:10.1086/464993. Gursky, Karl-Heinz. 1974. Der Hoka-Sprachstamm. Eine Bestandsaufnahme des lexikalischen Beweismaterials. Orbis 23: 170–215. Gursky, Karl-Heinz. 1987. Achumawi und Pomo: Eine besondere Beziehung? (Abhandlungen der Völkerkundlichen Arbeitsgemeinschaft 57). Nortorf. Gursky, Karl-Heinz. 1988. Der Hoka-Sprachstamm: Nachtrag I (Abhandlungen der Völkerkundlichen Arbeitsgemeinschaft 58). Nortorf. Gursky, Karl-Heinz. 1989. Der Hoka-Sprachstamm: Nachtrag II (Abhandlungen der Völkerkundlichen Arbeitsgemeinschaft 63). Nortorf. Gursky, Karl-Heinz. 1990. Der Hoka-Sprachstamm: Nachtrag III (Abhandlungen der Völkerkundlichen Arbeitsgemeinschaft 65). Nortorf. Gursky, Karl-Heinz. 1995. Some grammatical evidence for the Hokan stock. In: Michael Dürr, Egon Renner, Wolfgang Oleschinski (eds.). Language and culture in Native North America: studies in honor of Heinz-Jürgen Pinnow (LINCOM Studies in Native American Linguistics 2): 138–153. München. Haas, Mary R. 1954. The Proto-Hokan-Coahuiltecan word for “water.” In: Papers from the Symposium on American Indian linguistics held at Berkeley July 7, 1951 (University of California Publications in Linguistics 10): 57–62. Berkeley / Los Angeles: University of California Press. Haas, Mary R. 1963. Shasta and Proto-Hokan. Language 39(1): 40–59. doi:10.2307/410761. Haas, Mary R. 1964. California Hokan. In: William Bright (ed.). Studies in Californian linguistics (University of California Publications in Linguistics 34): 73–87. Berkeley / Los Angeles: University of California Press. Haas, Mary R. 1969. The prehistory of languages (Janua Linguarum Series Minor 57). The Hague: Mouton. doi:10.1515/9783110881646. Harrington, John P. 1943. Hokan discovered in South America. Journal of the Washington Academy of Sciences 33(11): 334–344. Haynie, Hannah J. 2012. Studies in the history and geography of California Languages. University of California, Berkeley Ph.D. Available online at: https://escholarship.org/uc/item/03c2z9tr. Haynie, Hannah J. 2014. Deep relationships among California languages. Diachronica 31(3): 407–447. doi:10.1075/dia.31.3.04hay. Jacobsen, Jr., William H. 1958. Washo and Karok: an approach to comparative Hokan. International Journal of American Linguistics 24(3): 195–212. doi:10.1086/464456. Jacobsen, Jr., William H. 1979. Hokan inter-branch comparisons. In: Lyle Campbell, Marianne Mithun (eds.). The languages of Native America: historical and comparative assessment: 545–591. Austin / London: University of Texas Press. doi:10.7560/746244-011. Jany, Carmen. 2013. Hokan languages. In: Oxford Bibliographies Online: Linguistics. Oxford University Press. doi:10.1093/obo/9780199772810-0113. Jany, Carmen. 2017. The Northern Hokan area. In: Michael Fortescue, Marianne Mithun, Nicholas Evans (eds.). The Oxford Handbook of Polysynthesis: 260–283. 1st edn. Oxford University Press. doi:10.1093/oxfordhb/9780199683208.013.34. Kaufman, Terrence. 1989. A research program for reconstructing Proto-Hokan: first gropings. In: Papers from the 1988 Hokan-Penutian Languages Workshop, Held at the University of Oregon, Eugene, Oregon, June 16–18, 1988: 50–168. Eugene, OR: Department of Linguistics, University of Oregon. Kaufman, Terrence. 2015. Some hypotheses regarding Proto-Hokan grammar. Unpublished manuscript. Available online at: https://www.researchgate.net/publication/340925991_Some_hypotheses_regarding_proto-Hokan_ grammar_Kaufman_2015. Kaufman, Terrence. 2016. Tlapaneko-Sutiaba, OtoMangean, and Hokan: where Greenberg went wrong. Unpublished manuscript. Available online at: https://www.researchgate.net/publication/340925787_subtitle_TlapanekoSutiaba_OtoMangean_and_Hokan_where_Greenberg_went_wrong. Kroeber, Alfred L. 1904. Languages of the coast of California south of San Francisco. University of California publications: American archaeology and ethnology 2(2): 29–80. Kroeber, Alfred L. 1915. Serian, Tequistlatecan, and Hokan. University of California Publications in American Archaeology and Ethnology 11(4): 279–290. Langdon, Margaret. 1974. Comparative Hokan-Coahuiltecan studies: A survey and appraisal. The Hague: Mouton. doi:10.1515/9783110887839. 185 Mikhail Zhivlov Langdon, Margaret. 1979. Some thoughts on Hokan with particular reference to Pomoan and Yuman. In: Lyle Campbell, Marianne Mithun (eds.). The languages of Native America: historical and comparative assessment: 592– 649. Austin / London: University of Texas Press. doi:10.7560/746244-012. Langdon, Margaret. 1990. Morphosyntax and problems of reconstruction in Yuman and Hokan. In: Philip Baldi (ed.). Linguistic change and reconstruction methodology (Trends in Linguistics Studies and Monographs 45): 57–72. Berlin / New York: Mouton de Gruyter. doi:10.1515/9783110886092.57. Langdon, Margaret & Pamela Munro. 1980. Yuman numerals. In: Kathryn Klar, Margaret Langdon, Shirley Silver (eds.). American Indian and Indoeuropean studies: papers in honor of Madison S. Beeler (Trends in Linguistics Studies and Monographs 16): 121–135. The Hague / Paris / New York: Mouton Publishers. doi:10.1515/9783110808681.121. Leščiner, D. R. 1989. Rekonstrukcija nazvanij častej tela v jazykax semʹi xoka. In: Lingvističeskaja rekonstrukcija i drevnejšaja istorija Vostoka. Materialy k diskussijam na Meždunarodnoj konferencii (Moskva, 29 maja — 2 ijunja 1989 g.). Častʹ 1: 159–166. Moskva: Izdatelʹstvo “Nauka.” Leščiner, D. R. 1990. Rekonstrukcija tonov i dolgot v prajazyke semʹi xoka. In: Sravnitelʹno-istoričeskoe jazykoznanie na sovremennom ètape. Pamjati V. M. Illič-Svityča. 6-9 fevralja 1990 g. Tezisy dokladov: 34–36. Moskva: Institut slavjanovedenija i balkanistiki AN SSSR. Leshchiner, Dmitry & Sergei Nikolaev. 1992. Proto-Hokan roots. In: Vitaly Shevoroshkin (ed.). Nostratic, DeneCaucasian, Austric and Amerind: materials from the First International Interdisciplinary Symposium on Language and Prehistory, Ann Arbor, 8.-12. November 1988 (Bochum Publications in Evolutionary Cultural Semiotics 33): 366–402. Bochum: Universitätsverlag Dr. Norbert Brockmeyer. Marlett, Stephen A. 1981. The structure of Seri. La Jolla CA: University of California at San Diego Ph.D. Available online at: http://lengamer.org/admin/language_folders/seri/user_uploaded_files/links/File/Marlett_1981_Seri_ Dissertation.zip. Marlett, Stephen A. 2007. Las relaciones entre lenguas “hokanas” en México: ¿Cuál es la evidencia? In: Cristina Buenrostro, Samuel Herrera Castro, Yolanda Lastra, Fernando Nava, Juan José Rendón, Otto Schumann, Leopoldo Valiñas, María Aydeé Vargas Monroy (eds.). Clasificación de las lenguas indígenas de México: Memorias del III Coloquio Internacional de Lingüística Mauricio Swadesh: 165–192. Mexico City: Universidad Nacional Autónoma de México & Instituto de Investigaciones Antropológicas & Instituto Nacional de Lenguas Indígenas. Marlett, Stephen A. 2008. The Seri–Salinan connection revisited. International Journal of American Linguistics 74(3): 393–399. doi:10.1086/590087. McLendon, Sally. 1964. Northern Hokan (B) and (C): a comparison of Eastern Pomo and Yana. In: William Bright (ed.). Studies in Californian linguistics (University of California Publications in Linguistics 34): 126–144. Berkeley / Los Angeles: University of California Press. McLendon, Sally. 1973. Proto Pomo (University of California Publications in Linguistics 71). Berkeley: University of California Press. McLendon, Sally. 1975. A grammar of Eastern Pomo (University of California Publications in Linguistics 74). Berkeley / Los Angeles / London: University of California Press. Miller, Amy. 2001. A grammar of Jamul Tiipay (Mouton Grammar Library 23). Berlin / New York: Mouton de Gruyter. doi:10.1515/9783110864823. Miller, Amy. 2018. Phonological developments In Delta-California Yuman. International Journal of American Linguistics 84(3): 383–433. doi:10.1086/697588. Mithun, Marianne. 2007. Grammar, contact and time. Journal of Language Contact 1(1): 144–167. doi:10.1163/000000007792548378. Mixco, Mauricio J. 1997. Haas’s Hokan: dead end, or gateway to the future? Anthropological Linguistics 39(4): 680–694. Moser, Mary Beck & Stephen A. Marlett. 2010. Comcaac quih yaza quih hant ihiip hac: cmiique iitom — cocsar iitom — maricaana iitom = Diccionario seri — español — inglés: con índices español — seri, inglés — seri. 2nd edn. Hermosillo, Son.: Editorial UniSon : Plaza y Valdés Editores. Nevin, Bruce. 1998. Aspects of Pit River phonology. University of Pennsylvania PhD Thesis. Nevin, Bruce. 2019. Why Proto-Palaihnihan is neither. Unpublished manuscript. doi:10.13140/RG.2.2.33917.20966. Nevin, Bruce. 2020. Achumawi Dictionary. Available online at: https://www.webonary.org/odissi/. [Accessed 13 February, 2024]. O’Connor, Loretta. 2013. Latyaygi — English — Español: A trilingual dictionary of Lowland Chontal of Oaxaca (Languages of the World/Dictionaries 48). Muenchen: LINCOM GmbH. 186 Hokan I: A review of comparative studies Olmsted, D. L. 1956. Palaihnihan and Shasta I: labial stops. Language 32(1): 73–77. doi:10.2307/410654. Olmsted, D. L. 1957. Palaihnihan and Shasta II: apical stops. Language 33(2): 136–138. doi:10.2307/410725. Olmsted, D. L. 1959. Palaihnihan and Shasta III: dorsal stops. Language 35(4): 637–644. doi:10.2307/410603. Olmsted, D. L. 1964. A history of Palaihnihan phonology (University of California Publications in Linguistics 35). Berkeley / Los Angeles: University of California Press. Olmsted, D. L. 1966. Achumawi dictionary (University of California Publications in Linguistics 45). Berkeley / Los Angeles: University of California Press. Oswalt, Robert L. 1977. The Hokan word for “tongue”: the Pomo evidence. In: James E. Redden (ed.). Proceedings of the 1976 Hokan-Yuman languages workshop, held at University of California, San Diego, June 21-23, 1976 (University Museum Studies 11): 69–78. Carbondale, Illinois: University Museum and Art Galleries, Southern Illinois University. Oswalt, Robert L. 1981. Southern Pomo word list and map of native place names in the Warm Springs area. San Francisco: U.S. Army Corps of Engineers. Peiros, Ilia. 1997. Macro families: can a mistake be detected? In: Irén Hegedűs, Peter A. Michalove, Alexis Manaster Ramer (eds.). Indo-European, Nostratic, and Beyond: Festschrift for Vitalij V. Shevoroshkin (JIES Monograph 22): 265–292. Washington DC. Poser, William J. 1992. The Salinan and Yurumanguí data in Language in the Americas. International Journal of American Linguistics 58(2): 202–229. doi:10.1086/ijal.58.2.3519756. Poser, William J. 1995. Binary comparison and the history of Hokan comparative studies. International Journal of American Linguistics 61(1): 135–144. doi:10.1086/466247. Rivet, Paul. 1942. Un dialecte hoka colombien: le Yurumangí. Journal de la Société des Américanistes 34: 1–59. doi:10.3406/jsa.1942.2334. Sapir, Edward. 1913. Southern Paiute and Nahuatl, a study in Uto-Aztekan. Journal de la Société des Américanistes 10(2): 379–425. doi:10.3406/jsa.1913.2866. Sapir, Edward. 1917. The position of Yana in the Hokan stock. University of California publications in American archaeology and ethnology 13(1): 1–34. Sapir, Edward. 1919. Southern Paiute and Nahuatl, a study in Uto-Aztekan, Part II. Journal de la Société des Américanistes 11(1): 443–488. doi:10.3406/jsa.1919.3856. Sapir, Edward. 1920a. A note on the first person plural in Chimariko. International Journal of American Linguistics 1(4): 291–294. doi:10.1086/463727. Sapir, Edward. 1920b. The Hokan and Coahuiltecan languages. International Journal of American Linguistics 1(4): 280–290. doi:10.1086/463726. Sapir, Edward. 1920c. The language of the Salinan Indians. International Journal of American Linguistics 1(4): 305–309. doi:10.1086/463730. Sapir, Edward. 1921. A supplementary note on Salinan and Washo. International Journal of American Linguistics 2(1/2): 68–72. doi:10.1086/463735. Sapir, Edward. 1925. The Hokan affinity of Subtiaba in Nicaragua. American Anthropologist 27(3–4): 402–435, 491– 527. doi:10.1525/aa.1925.27.3.02a00040. Sapir, Edward & Morris Swadesh. 1960. Yana dictionary (University of California Publications in Linguistics 22). Berkeley / Los Angeles: University of California Press. Shaul, David Leedom. 2019. Esselen studies: language, culture, and prehistory (LINCOM Studies in Native American Linguistics 83). Muenchen: Lincom GmbH. Shaul, David Leedom. 2020a. Salinan and English lexicon (Languages of the World Dictionaries 74). Muenchen: LINCOM GmbH. Shaul, David Leedom. 2020b. Salinan language studies (LINCOM Studies in Native American Linguistics 85). Muenchen: LINCOM GmbH. Silver, Shirley. 1964. Shasta and Karok: a binary comparison. In: William Bright (ed.). Studies in Californian linguistics (University of California Publications in Linguistics 34): 170–181. Berkeley / Los Angeles: University of California Press. Silver, Shirley. 1974. Some Northern Hokan plant-tree-bush forms. The Journal of California Anthropology 1(1): 102–109. Silver, Shirley. 1976. Comparative Hokan and the Northern Hokan languages. In: Margaret Langdon, Shirley Silver (eds.). Hokan Studies: Papers from the First Conference on Hokan Languages held in San Diego, California April 23-25, 1970 (Janua Linguarum: Series Practica 181): 193–202. The Hague / Paris: Mouton. doi:10.1515/9783110819113-013. 187 Mikhail Zhivlov Swanton, John R. 1915. Linguistic position of the tribes of Southern Texas and Northeastern Mexico. American Anthropologist 17(1): 17–40. doi:10.1525/aa.1915.17.1.02a00030. The Santa Ynez Band of Chumash Indians. 2007. Samala-English dictionary: A guide to the Samala language of the Ineseño Chumash people. Turner, Paul R. 1967. Seri and Chontal (Tequistlateco). International Journal of American Linguistics 33(3): 235–239. doi:10.1086/464966. Turner, Paul R. 1976. Pluralization of nouns in Seri and Chontal. In: Margaret Langdon, Shirley Silver (eds.). Hokan Studies: Papers from the First Conference on Hokan Languages held in San Diego, California April 23-25, 1970 (Janua Linguarum: Series Practica 181): 297–303. The Hague / Paris: Mouton. doi:10.1515/9783110819113-019. Turner, Paul R. & Shirley Turner. 1971. Dictionary: Chontal to Spanish-English, Spanish to Chontal. Tucson, Arizona: The University of Arizona Press. Waterhouse, Viola Grace. 1976. Another look at Chontal and Hokan. In: Margaret Langdon, Shirley Silver (eds.). Hokan Studies: Papers from the First Conference on Hokan Languages held in San Diego, California April 23-25, 1970 (Janua Linguarum: Series Practica 181): 325–343. The Hague / Paris: Mouton. doi:10.1515/9783110819113-021. Webb, Nancy M. 1979. The semantic domain “tree”: Hokan lexical evidence. In: James E. Redden (ed.). Proceedings of the 1978 Hokan languages workshop, held at the University of California, San Diego, June 27-29, 1978 (Occasional Papers on Linguistics 5): 1–7. Carbondale, Illinois: Department of Linguistics, Southern Illinois University. Webb, Nancy M. 1980. Esselen-Hokan relationships. In: James E. Redden (ed.). Proceedings of the 1979 Hokan languages workshop, held at the University of California, Los Angeles, June 26-28, 1979 (Occasional Papers on Linguistics 7): 72–77. Carbondale, Illinois: Department of Linguistics, Southern Illinois University. Webb, Nancy M. 1981. Some notes on Hokan person terms. In: James E. Redden (ed.). Proceedings of the 1980 Hokan languages workshop, held at the University of California, Berkeley, June 30-July 2, 1980 (Occasional Papers on Linguistics 9): 78–85. Carbondale, Illinois: Department of Linguistics, Southern Illinois University. Zhivlov, Mikhail. 2018. Some morphological parallels between Hokan languages. Journal of Language Relationship / Вопросы языкового родства 16(2): 138–161. М. А. Живлов. Хока I: обзор исследований по сравнению хоканских языков В статье рассматривается история хоканской гипотезы, от её формулировки Диксоном и Крёбером в 1912-1913 годах до наших дней. Несмотря на более чем сто лет исследований, до сих пор нет единого мнения относительно обоснованности хоканской гипотезы. В статье утверждается, что одной из причин этого является тот факт, что многие попытки сравнения хоканских языков использовали нестандартную методологию, в которой исследование останавливается на перечислении наблюдаемых фонетических соответствий, вместо классического сравнительного метода, требующего поиска дополнительных распределений и достаточно полной реконструкции фонологии праязыка. Ключевые слова: сравнительный метод; языки хока; индейские языки; генеалогическая классификация языков.