Khitan Understanding the Language

Khitan: Understanding the Language Behind the Scripts Juha A. Janhunen

University of Helsinki, Finland

Khitan was the dynastic language of the Liao empire in Manchuria and Northern China (907-1115). Although today extinct, samples of Khitan are preserved in two native scripts, known as the Khitan Large Script and the Khitan Small Script. Both scripts may be classified as “Sinitic” or “Sinoform” in the typological sense, though only the Large Script has a direct connection with the Chinese script. Recent progress in the decipherment of, in particular, the Khitan Small Script allows the lexicon and grammar of the Khitan language to be assessed in much more detail than before. Khitan may be defined as a Para-Mongolic language, meaning that it represents a branch related to, but collateral with, the extant and historically known Mongolic languages. The present paper examines the genetic position of the Khitan language with regard to Mongolic with the help of the methods of comparative linguistics, as applied to the deciphered Khitan language material. Keywords: Sinitic scripts, Khitan Small Script, Khitan language, ParaMongolic, reconstruction, decipherment

In the global context of ancient scripts and their decipherment it is not widely known that some of the last great challenges posed by unknown systems of writing are those connected with the scripts of peripheral mediaeval China. Three non-Chinese ethnic groups—the Khitan, the Jurchen, and the Tangut, corresponding to the “alien” dynasties of Liao 遼 (907-1125), Jin 金 (1115-1234), and Xixia 西夏 (1038-1227), respectively— created scripts and literary languages of their own. After centuries of oblivion, samples of these scripts emerged in the late 19th and early 20th centuries and became objects of intensive study. Even so, we cannot say SCRIPTA, Volume 4 (October 2012): 107-132 © 2012 The Hunmin jeongeum Society

108



SCRIPTA, VOLUME 4 (2012)

that any of these scripts has been fully deciphered. The problems vary from script to script, however, and they are also connected with the level of knowledge we have of the underlying languages—Khitan, Jurchen, and Tangut. It happens that of these three languages, only Jurchen survived till later times, becoming the direct ancestor of Manchu, while both Khitan and Tangut became extinct soon after the fall of the political states that had used them as their dynastic languages.1

1. The Khitan Scripts As far as Khitan is concerned, the situation is complicated by the circumstance that it was written in two different scripts, today known as the Khitan Large Script (大字 Dazi) and the Khitan Small Script (小字 Xiaozi). Like the Tangut Script and the Jurchen Script, the two Khitan scripts have a Chinese “appearance,” meaning that they are composed of the same type of primary and secondary elements (strokes, characters), presupposing the same type of writing instruments (brush, ink), as the Chinese script. We may therefore identify all these scripts as “Sinitic” or “Sinoform” in the typological and aesthetic sense. None of them is, however, identical with the Chinese script, and only two scripts, the Khitan Large Script and the Jurchen Script, are materially “related” to the Chinese script, meaning that they may be viewed as ancient regional derivatives of the latter, a situation which also allows them to be identified as “Sinographic” writing systems, and their characters as “Sinograms.” The Khitan Small Script and the Tangut Script, on the other hand, have no material relationship with the Chinese script, which suggests that their origination may have involved a factor of conscious “invention.”2 We know today that the Khitan Large Script and the Jurchen Script are very closely related with each other, both probably representing a For a more comprehensive survey of the chronological context of the scripts of mediaeval peripheral China, cf. Janhunen (1994).

1

The present paper will not deal in any more detail with the Tangut script. However, from the point of view of graphemic analysis, this script offers a particularly difficult challenge, since nothing is known of the principles of how it was designed. For this reason, the assumptions conventionally made about the identity of the Tangut language are also detached from the graphic reality and cannot be considered as confirmed, as was already pointed out by Kwanten (1984, 1988). 2



Khitan: Understanding the Language Behind the Scripts

109

continuation of a local Manchurian variety of the Chinese script. It is possible, though not verified, that this same variety, or an earlier form of it, was used already before the Liao-Jin period in other political contexts, including, perhaps, Beiwei 北魏 (386-534) and/or Balhae 渤海 (698-926). It is notable that the Khitan Large Script incorporates many Chinese characters, especially simple ones, in unchanged form, while in the Jurchen Script such characters are normally distinguished from their Chinese counterparts by using diacritics. We also know that both the Khitan and the Jurchen used this script as a mixed logo-syllabic system, with some characters functioning as logograms (with Khitan and Jurchen pronunciations) and others as syllabograms (for fixed sequences of sounds, not necessarily corresponding to phonetic syllables). In general, the number of separate characters in the Khitan Large and Jurchen Scripts is considerably smaller than in the Chinese script, and the forms of the characters are less complicated, with no functionally relevant “radical” components. Unfortunately, the Khitan Large Script is still very imperfectly deciphered, with only a small proportion of the characters identified with meanings (logograms) and/or sounds (syllabograms). One important direction of research that has not yet been properly initiated is the systematic comparison of the Khitan Large Script characters with those of the Jurchen Script. Due to the fact that the Jurchen Script survived relatively long (till the 16th century) and was used to write a wellknown language (Jurchen-Manchu), our knowledge of the Jurchen Script is rather close to “decipherment” in the sense that most of its elements have a known linguistic correlate, even if there still remain many open questions.3 On the other hand, the corpus of the Khitan Large Script is much more extensive, and in a much better state of preservation, than that of the Jurchen Script. Therefore, the Khitan Large Script is potentially a crucial source of information also for Jurchen studies. At the present stage, however, its potential cannot yet be exploited. This leaves us only the Khitan Small Script corpus as the most important currently available direct source on the Khitan language. Since several decades, it has been known that this is a basically syllabic script, The principal works on the Jurchen Script are Grube (1896), Kiyose (1978), Kane (1989), Jin Qicong (1984), and Pevnov (2004), as well as, most recently, Golovachev & Ivliev & Pevnov & Rykin (2011). Of these, only Kane comments also on the Khitan Large Script.

3

110



SCRIPTA, VOLUME 4 (2012)

using a limited number of graphically simple syllabic signs (syllabograms), known as the Khitan Small Script “characters.” What makes the graphic image visually complicated is the convention of accumulating the syllabic signs into “blocks” corresponding to linguistic words, a convention which, incidentally, has a close analogy in the Korean Hangeul.4 Since a Khitan word (with derivational and inflectional suffixes) can consist of up to eight “syllables,” a block can have a graphic appearance (stroke composition) even more complex than a Chinese character—a circumstance that caused confusion during the early stages of Khitan studies, when it was not yet clear how the “Small” and “Large” scripts should be properly distinguished.5 The decipherment of the Khitan Small Script has involved both slow progress and rapid steps forward starting with the early 1920s.6 A definitive breakthrough, which for the first time yielded concrete “readings” of Khitan words, was achieved by a team active in China in the ’70s and ’80s.7 Since then, the most important development has been the rapid increase of the corpus, conditioned by the discovery of new texts. The currently known corpus comprises 34 published and a few un­published epigraphic texts plus a diffuse selection of minor materials. The standard of “publication” varies, however, and not all texts have been made available in a form that can be used for textological and linguistic analysis. Even so, and in spite of the fact that the epigraphic texts tend to be stereotypic in form and content, the corpus is large enough to allow relatively detailed conclusions to be made about the Khitan language and the properties of the Small Script. The phonetic value of the Khitan Small Script syllabic signs varies from The differ­ence between the two scripts is, of course, that the blocks in Hangeul are composed of phonograms. It is not known whether the Khitan Small Script can have inspired the creators of the Hangeul, for ultimately the blocks in both the Khitan Small Script and the Hangeul must have been modelled along the principle of Chinese characters. 4

The distinction between the two Khitan scripts was first correctly made and communicated inter­nationally by Toyoda (1964). 5

A large part of the research on the Khitan Small Script has been communicated in scattered papers and a few monographs in Chinese. An important critical summary of all this work was made available for the international readership by Kane (2009). 6

Chingeltei & Liu Fengzhu & Chen Naixiong & Yu Baolin & and Xing Fuli (1985). For a more detailed history of Khitan Small Script studies, see Wu Yingzhe & Janhunen (2010: 20-25). 7



Khitan: Understanding the Language Behind the Scripts

111

one segment (consonant or vowel) to sequences of up to three segments. Some characters seem to be used also for longer words as what may be actual logograms. There are, of course, also undeciphered characters. Of the currently known 459 different characters, “only” 314 have been deciphered in one way or another.8 These include, however, a large proportion of the most frequent characters, while many of the undeciphered items are rare and may even involve mistakes. This means that even if the decipherment is not fully completed, it is already possible to render long sequences of continuous Khitan text a more or less consistent phonetic and/or semantic inter­pretation, with only occasional lacunae. In this situation, attention may gradually be turned away from the script to the language behind the script.

2. The Khitan language The Khitan, or Qidan 契丹, were the dominant ethnic group in the Liaoxi

遼西 region in the period between the Beiwei and the historical Mongols.

The Khitan survived as an ethnic group after the rise of the Jurchen, and written documents in the Khitan language continued to be produced almost till the end of the Jin period. It is well known from Chinese historical sources that Khitan was a distinct language, spoken by virtually all ethnic Khitan, a population whose size may have reached a million people during the height of the Liao dynasty. Khitan was not the only language spoken in the Liao state, however, for other languages, including Chinese and Jurchen, were also widely used especially in the areas conquered by the Khitan during the process of state formation. It is less clear how uniform Khitan was internally. The Khitan were composed of tribes, some of which played a more important political role in the Liao state than others, and it is quite possible that there were tribal forms of speech that deviated from the mainstream language.9 Apart from texts in the two Khitan scripts, there are two other types of sources on the Khitan language. These are, first, the occasional samples of Khitan, mainly lexical items, but also phrases and poems, preserved An up-to-date list of the Khitan Small Script characters is given in Wu Yingzhe & Janhunen (2010: 259-272, cf. also the discussion ibid. 35-48).

8

On the tribes, as well as on the general social history and historical demography of the Khitan, see Wittfogel & Fêng (1949). 9

112



SCRIPTA, VOLUME 4 (2012)

in Chinese transcription and glossing, and, second, the Khitan loanwords transmitted into neighbouring languages, especially Jurchen. As far as can be seen, these sources represent a language identical with that of the Khitan text corpus. The Chinese data are, however, often frustratingly difficult to interpret with any certainty due to the chronological, phonological, and semantic inaccuracies as well as outright errors contained in them.10 Loanword research could potentially yield much more informative results, for it may be taken for certain that Jurchen, as a subordinate language of the Liao state, received a considerable number of Khitan loanwords.11 It is also likely that the two languages were typologically similar, which means that we may occasionally use information from Jurchen to approach the structure of Khitan. The fact that Khitan is, or was, typologically speaking, an “Altaic” language, was known already before the original sources in the language could be “read.”12 This is now confirmed by the actual “readings” of the Khitan Small Script texts. We know, for instance, that Khitan had an elaborate system of nominal and verbal inflectional suffixes, most of which can be phonologically approximated.13 The syntax and morphosyntax, including the basic word order (SOV) and the various types of subordinated (converbial) and embedded (participial) sentence constructions, is also in full accordance with that attested in the other historical and modern “Altaic” languages in the region, including both Jurchen-Manchu and Mongolian. There are, however, some unexpected features, notably, traces of grammatical gender (marked masculine vs. unmarked or generic feminine), a category not typical of “Altaic” typology though, incidentally, present in Middle Mongol.14 10 Recent studies of Khitan words in Chinese sources include Sun Bojun & Nie Hongyin (2008) and Talpe (2010). The only linguistically competent work on the subject is Shimunek (2007). Selected details are also discussed by Vovin (2003) and Róna-Tas (2004).

Some pioneering work on the lexical parallels between Khitan and Jurchen has been carried out by Kane (2006).

11

A “grammar” and a collection of Khitan Small Script “texts” in digital form was compiled by a Russian team in the 1960s and ’70s, cf. Starikov et al. (1970), Arapov (1982). 12

13 So far the only systematic grammatical sketch of Khitan has been compiled by Kane (2009: 131-166), who also gives a vocabulary of 354 items (ibid. 83-130). 14

The role of grammatical gender in Khitan is still far from completely understood.



Khitan: Understanding the Language Behind the Scripts

113

It used to be more difficult to determine what genetic group (language family) Khitan represents. “Altaic” typology as such does not indicate whether a language is Turkic, Mongolic, Tungusic, or something else. In the past, when work on Khitan was solely based on the Chinese transcriptions and glosses of Khitan words, it was common to try to identify these words item by item with the different historical and extant “Altaic” languages. The result was ambiguous, since Khitan turned out to contain lexical items that could be identified variously with either Mongolic or Turkic, or also Tungusic.15 In many cases, parallel identifications were possible, leading to speculations that Khitan might be an “intermediate” idiom between, say, Turkic and Mongolic, or Mongolic and Tungusic. The problem here was that inherited elements were not properly distinguished from borrowed ones. A more reliable method is to operate with genetically diagnostic words, such as items of basic vocabulary, which are likely to have been inherited rather than borrowed. With this method, it was initially possible to identify the language of the Tabghach, or Tuoba 拓跋, the leading ethnic group of the Beiwei state, as basically “Mongolic.” Since the Tabghach were the political and, quite possibly, the linguistic ancestors of the Khitan, the Mongolic identification of Khitan gained ground.16 This identification is now definitively confirmed by the information from the Khitan Small Script texts, which show beyond doubt that Khitan was a language whose basic vocabulary and grammatical resources were related to those attested in the Mongolic languages. It is important to note that this conclusion is exclusive, in the sense that Khitan is not an “intermediate” language and cannot share elements, except borrowings, with other “Altaic” languages than Mongolic.17 The term “Mongolic” needs, however, modification when used about Work on this issue has been carried out by Wu Yingzhe (2005, 2007). In spite of its importance, the issue of grammatical gender is largely unstudied in Middle Mongol, cf. Rybatzki (2003: 75). Menges (1968) made an effort to identify Tungusic elements in Khitan, though we know today that most of his identifications are wrong. He had to admit himself that the material was not conclusive for the determination of the genetic position of Khitan.

15

The Mongolic identification of the Tabghach language was made by Ligeti (1971). A similar identifi­cation for Khitan is implied by Doerfer (1992, 1993).

16

Suggestions that violate this principle are occasionally encountered in the works of Khitan specialists not sufficiently familiar with comparative linguistics.

17

114



SCRIPTA, VOLUME 4 (2012)

Khitan. All other extant and historical Mongolic languages represent a single genetic node, corresponding to the language of the historical Mongols, and linguistically identifiable as Proto-Mongolic. Khitan did not descend from Proto-Mongolic but was, rather, a sister language of the latter. In other words, Proto-Mongolic and Khitan represented two branches of an even older protolanguage, a certain stage of Pre-ProtoMongolic that could also be termed “Khitano-Mongolic.” Khitan itself may also have had sister languages belonging to the same branch, which, in that case, should be termed “Khitanic.” Technically, the most suitable term to describe the position of Khitan (and Khitanic) with regard to ProtoMongolic is to identify the former as “Para-Mongolic,” implying that it is a question of a genetically related, but collateral, branch of the much better known Mongolic language family.18

3. The position of Khitan The availability of direct information on Para-Mongolic in the form of Khitan Small Script texts opens up extremely important new perspectives for studies on the history of the Mongolic language family. Working with the actual Mongolic languages alone, we used to have three kinds of diachronic information: first, the comparative evidence provided by the extant Mongolic languages; second, the information contained in the historical forms of Mongolic, especially Written Mongol and Middle Mongol; and third, the potential conclusions that can be made from the comparative and historical data by the method of internal reconstruction. Only the last type of information allows us to approach the period preceding Proto-Mongolic, which itself is of a rather shallow depth corresponding to no more that 800-900 years. The question is how far backwards the Khitan data allow us to go in the history of Mongolic. This, on the other hand, depends, on how different Khitan was as compared with Proto-Mongolic. Before the decipherment of the Khitan Small Script there was an opinion that Khitan may have been closely related to Proto-Mongolic, so closely that it was thought that The term Para-Mongolic was introduced by Janhunen (1995, 2003) and has since found some support in specialist literature, cf. e.g. Kane (2009: x).

18



Khitan: Understanding the Language Behind the Scripts

115

Khitan could perhaps be “read” in terms of the lexical and grammatical information we have from Middle Mongol. Today we know that this is not so: Khitan is, in fact, a language rather distantly related to the ProtoMongolic branch. To get an idea of how distant the relationship could be we may think of the neighbouring Tungusic language family, in which the southern branch, corresponding to Jurchen-Manchu (Jurchenic), is in many ways strikingly different from the northern branch, corresponding to Ewenki-Ewen (Ewenic).19 Assuming that the difference between Para-Mongolic and ProtoMongolic was of the same chronological scope as that between the southern and northern branches of Tungusic, the breakup of the original protolanguage (Proto-Khitano-Mongolic) would have taken place at least several centuries before the emergence of Khitan as a written language. The last possible historical context for the still uniform proto­language would seem to have been the empire of the Xianbei 鮮卑 (93-234), though the breakup may, of course, have taken place even earlier. A practical consequence of the chronological difference is that Khitan texts, which with one notable exception are not bilingual, are difficult to understand even if the script is no longer a major problem. It is not without reason that the situation has been compared with that of Etruscan: a known script but an unknown language.20 Fortunately, however, Khitan—unlike Etruscan—is not completely unknown, for it is still related to the Proto-Mongolic branch, although the relationship is so distant that it does not substantially facilitate the understanding of Khitan texts. The genetic connection is nevertheless evident from the existence of cognate words and shared morphological elements present in both Khitan and Proto-Mongolic. It is very likely that the number of cognates will continue to grow as more linguistic work is done on the Khitan Small Script texts. We should, however, not be too optimistic about the size of the comparative corpus. The total number of Khitan lexical items that can be assessed both phonetically and semantically today is still less than 500, and many of these items do not The Tungusic family also has two transitional (or mixed) branches, collectively termed “Amur Tungusic.” For a more detailed taxonomy of Tungusic, cf. Janhunen (2012).

19

20 The Etruscan parallel is mentioned by Kane (2009: x), who also discusses the general challenge posed by languages that can be read without being understood. On the unique bilingual Khitan text, cf., most recently, Vovin (2011).

116



SCRIPTA, VOLUME 4 (2012)

belong to the basic vocabulary; moreover, they also comprise a number of Chinese borrowings. As it is, the number of currently known certain or plausible cognates between Khitan and Proto-Mongolic is at the range of a few dozen, that is, about ten per cent of the total known Khitan lexicon.21 The only way to assess the taxonomic status of Khitan with regard to Proto-Mongolic is to analyze this corpus with the methods of comparative linguistics. It is particularly illustrative to see in what respects Khitan, as compared with Proto-Mongolic, is archaic and in what respects it is innovative. Since Khitan became a written language some 200 years before Middle Mongol, one would expect that it might be in some respects more archaic. The picture is not equivocal, however, for in other respects Khitan is surprisingly “modern,” often anti­cipating innovations that took place in the Proto-Mongolic branch only after the breakup of the latter. This is compatible with the picture provided by the Tungusic family, in which Jurchen-Manchu may also be seen as a particularly innovative branch as compared with Ewenki-Ewen.22 Moreover, some of the innovations present in Khitan are likely to have been shared with Jurchen on an areal basis. At this point a word of caution concerning the notation is in place. The language behind the Khitan Small Script can be approached in terms of a Romanized approximation of its phonetic structure. The Romanization is, however, not an exact reconstruction of the segmental properties of the language. This situation is due to two circum­stances: on the one hand, we simply do not know the exact segmental counter­parts of many Khitan Small Script characters, although we can “approximate” them; on the other, we have to consider the fact that the orthography of Khitan is not always consistent with the actual sequences of sounds. In order to show the difference between orthography and sounds, it is useful to let the Romanizations be accompanied by a more phonemically accurate “reading” (marked †), which, in turn, should be distinguished from reconstructions (marked *).23 21 Many of the cognates have been identified over the years by Chinese scholars and are listed by Kane (2009), though his database involves some mistakes and omissions.

The conventional view that Jurchen-Manchu represents an innovative branch of Tungusic against the more conservative Ewenki-Ewen branch has recently been challenged by Alonso de la Fuente (2011). The issue is best seen as unconcluded. 22

23

The current system of Romanization for Khitan was introduced by Kane (2008, 2009)

rd of At caution concerning theofnotation in place. The this point a word caution isconcerning the notation is in place. The language behin Small Script can be approached in terms of a Romanized anguage behind the Khitan Small Script can be approached in terms of a Romanized approximation ic structure. The Romanization is, however, not an exact is, however, not an exact reconstruction o pproximation of its phonetic structure. The Romanization ntal propertiesof thesegmental language.Khitan: This Understanding situation is due to twoThis ofthe the Language Behind the Scripts 117 econstruction properties of the language. situation is due to two circumstances: and, we simplyondothe notone know thewe exact segmental ircumstances: hand, simply do not counterknow the exact segmental counter- parts of many Script characters, we can “approximate” them; arts of many Khitanalthough Small Script characters, although we can “approximate” them; on the other, w the factwethat thetoorthography Khitan always of Khitan is not always consistent with nsider the other, have consider theoffact that is thenot orthography sequenceswith of 4.sounds. In properties order to show the difference onsistent the actual sequences of sounds. In order to show the difference between orthog Lexical unds, it is useful to let the Romanizations be etween orthography and sounds, it is useful toaccompanied let the Romanizations be accompanied by a more pho ccurate “reading” (marked †), which, in turn, shouldlength be which, y a more phonemically accurate “reading” (marked †), in turn, should be distinguished fr As all are memorial in character, 23 Khitan texts of any significant 23 uctions (marked *).reconstructions (marked *). istinguished from containing epitaphs, eulogies, lamentations, and/or historical records, 4. Lexical prop the lexicon used in them is thematically and functionally restricted. This . Lexical properties gives the decipherer the advantage of having many parallel repetitions As all Khitan with are minor variations, as in genealogical lists, which, in turn, allows the significant length memorial in character, containing As all Khitan texts of any significant length are memorial in character, containing epitaphs, eulog identification of the certain crucial phrases multiply ions, and/or historical records, lexicon usedlexical inrecords, themitems isthe and pitaphs, eulogies, lamentations, and/or historical lexicon usedto in be them is thematically an verified. The obvious disadvantage is that large sections of the ly restricted.and This gives the decipherer the advantage hematically functionally restricted. This gives the of decipherer the advantagelexicon of having many pa never occur inasthe texts, or occur so rarely that identification is impossible. ions minor variations, inwith genealogical lists, which, avingwith many parallel repetitions minor variations, as in genealogical lists, which, in turn, allows cation certain crucial lexical items and phrases to beis supplied Athe potentially important source of lexicon by phrases the poems which multiply verifie n turn, ofallows identification of certain crucial lexical items and to be us disadvantage is that large sections of theislexicon never occur in the te often complement epitaphic texts, but the problem is that poetic multiply verified. The obvious disadvantage that large sections of thehere lexicon never so rarely that identification is impossible. A potentially important sour ccur in the texts, or occurare sovirtually rarely that identification impossible. A potentially expressions hopeless to assessiswithout a parallel version in a n is supplied byofthe poemsis which often mportant source lexicon supplied by complement the poems which often complement epitaphic texts, known language. blem heretexts, is that expressions virtually hopeless pitaphic butpoetic problem here are is are virtually hopeless to assess witho Inthe the following, a that few poetic lexicalexpressions spheres relevant to comparative version a known language. In the f o assessinwithout a parallel version in a known language. linguistics are examined in some more detail: ew lexical relevant to lexical comparative linguistics In thespheres following, a few spheres relevant are to comparative linguistics are examined in so ail: xamined in some more detail: Pronouns. For genetic comparisons, personal pronouns would be an Pronouns. For ideal source of diagnostic information. Unfortunately, first and second parisons, personal pronouns would be an ideal source of Pronouns. For genetic comparisons, personal pronouns would be an ideal source of diagnostic info person pronouns do not seem to be in pronouns the extantdoKhitan corpus. ortunately, first and second person pronouns dosecond notattested seem to be attested i iagnostic information. Unfortunately, first and person not seem hó, but this item has no known A possible third person pronoun is Khitan corpus. A possible third person pronoun is ‫ڹ‬ hó, but this item ha o be attested in the extant Khitan corpus. A possible third person pronoun is ‫ ڹ‬hó, cognate in Proto-Mongolic. The proximal demonstrative is, in Proto-Mongolic. The proximal demonstrative is, however,is,fully however, fully ut this item hascognate no known cognate in Proto-Mongolic. The proximal demonstrative with Proto-Mongolic data. with The Khitan shape is ۠ e ‘this’ Proto-Mongolic data. ‘this’ :: ‫ ٻ‬tt ‘these’, s owever, fully comparable Proto-Mongolic The :Khitan shape is ۠ e ‘this’ e readings : pluralsuggesting †e-d, which correspond ProtoMongolic *e-n= ‫ٻ‬ t ‘these’,†esuggesting the readings †e : plural †e-d, which correspond to Proto‘these,’ the readings †e :toplural †e-d, which correspond to Protoe-d=e. As may be seen, the Khitan shapes look like more Mongolic *e-n=e : plural*e-n=e *e-d=e.: plural As may be seen, shapes look like morelook Mongolic *e-d=e. As the mayKhitan be seen, the Khitan shapes

like more simple and, hence, archaic, as they lack the stem extension -n 22 24 conventiona and the postclitic element =e, as attested in the Proto-Mongolic cognates.The chen-Manchu represents an innovative branch represents of Tungusic against the branch of Tungusic against the more conservative The conventional view that Jurchen-Manchu an innovative On the Mongolic side, the proximal pronoun has the oblique stem branch has recently been challenged byhas Alonso de labeen Fuente (2011). by Alonso de la Fuente (2011). The issue is best s more conservative Ewenki-Ewen branch recently challenged 23 *e-xü/n> *üün-, which has been compared with the possibly pronominal The current syst ded. he issue is best seen as unconcluded.

ation for Khitan was of introduced by Kane (2008, 2009) and developed The current system Romanization for Khitan was introduced by Kane (2008, 2009) and developed further by Wu Yin nen (2010). phonetic reliability (2010). of the Romanized shapes varies,of the Romanized shapes varies, and not all Roman urther by WuThe Yingzhe & Janhunen The phonetic reliability endregarded as fully verified. In be thefurther presentby paper, the Khitan&In items not all Romanizations may regarded as Wu fully verified. the are present(2010). paper, the items are quoted both in th and developed Yingzhe Janhunen TheKhitan phonetic reliability given when releva t (linearized) in with avaries, phonetic approximation uoted both in and theof original script (linearized) and in with a phonetic theRomanization, Romanized shapes andRomanization, not all Romanizations may beapproximation regarded as fully iven when relevant or possible. verified. In the present paper, the Khitan items are quoted both in the original script (linearized) and in Romanization, with a phonetic approxi­mation given when relevant or possible.

All Khitan data are quoted from Kane (2009) and Wu Yingzhe & Janhunen (2010). On the demon­strative pronouns, cf. also Wu Yingzhe (2009, 2011).

24

simple and, hence, as they as lackthey the lack stemthe extension -n simple and, archaic, hence, archaic, stem exten 24 24 4 (2012) SCRIPTA, element =e, asVOLUME attested in the Proto-Mongolic cognates.cognates. element =e, as attested in the Proto-Mongolic simple simple and, and, hence, hence, archaic, archaic, as as Mongolic they theythe lack lack the the stem stem extension extension -n -n and and the thethe postclitic postcliti On the side, the proximal pronoun has obliqu On Mongolic side, pronoun has th 24 24the proximal element element=e, =e,as asattested attested in in the the Proto-Mongolic Proto-Mongolic cognates. cognates. *üün-, which has been compared with the possibly pronominal Kh *üün-, which has been compared with the possibly pronom simple and, archaic, as they lack thepronoun stem the extension -n and stem the simple simple and, hence, hence, archaic, archaic, as as they they lack lack the stem stem extension extension -n -n postclitic and and the the po pp> On Onhence, the theand, Mongolic Mongolic side, side, the the proximal proximal pronoun has has the the oblique oblique stem *e-xü/n*e-xü/nKhitan monosyllable This comparison remains unverified, but it ؕ
ún ún †un. †un. This comparison remains unverified, but it would seem ؕ
ún †un. This comparison 24 remains 24 24 24 unverified, but it wo element =e, as attested in the Proto-Mongolic cognates. element element =e, =e, as as attested attested in in the the Proto-Mongolic Proto-Mongolic cognates. cognates. -n and the postclitic *üün-, *üün-, which which has hasKhitan been been compared compared with with the the‘now’, possibly possibly pronominal pronominal Khitan Khitan monosyllable monosyllabl c, ic, as they lack the the stem stem extension -n -n and and the postclitic postclitic ic,as asthey theylack lack the stem extension -n andthe theas postclitic ؕ۠
ún.e †une which formally corresponds to th would seem to extension correlate with Khitan ún.e †une ‘now,’ which formally Khitan ؕ۠
ún.e †une ‘now’, which formally correspon simple hence, archaic, they lack the stem extension -n and the 24 24 On theand, Mongolic side, the proximal pronoun has the oblique stem *e-xü/n> On On the the Mongolic Mongolic side, side, the the proximal proximal pronoun pronoun has has the the oblique oblique stem stempostc *e*e*e 24 ؕ
ؕ
ún ún †un. †un. This This comparison comparison remains remains unverified, unverified, but but it it would would seem seem to to correlate correlate with wit 24 he he Proto-Mongolic Proto-Mongolic cognates. cognates. the Proto-Mongolic cognates. locative case form *e-xü/n-e ‘upon this’, a case form also attested in locative case form *e-xü/n-e ‘upon this’, a case form also a corresponds to the Proto-Mongolic locative case form *e-xü/n-e ‘upon element =e, as attested in the Proto-Mongolic cognates. *üün-, which has been compared with the possibly pronominal Khitan monosyllable *üün-, *üün-, which which has has been been compared compared with with the the possibly possibly pronominal pronominal Khitan Khitan monos mono mono lique stem *e-xü/n> ؕ۠
Khitan Khitan ؕ۠
ún.e ún.e †une †une ‘now’, ‘now’, which which formally formally corresponds corresponds to todemonstratives the the Proto-Mongolic Proto-Mongoli de, ide, the the proximal proximal pronoun pronoun has has the the oblique oblique stem stem *e-xü/n*e-xü/n> > side, the proximal pronoun has the oblique stem *e-xü/n> (> Modern Mongolian odoo). The demonstratives are, however (> Modern Mongolian odoo). The are, this,’ a case form also attested inside, *e-d/ü-x-e ‘now’ (> Mongolian Onform the This Mongolic the proximal pronoun has oblique stem *e-xü/ ؕ
ún †un. This comparison remains unverified, but itModern would to correlate with ؕ
ؕ
ún ún †un. †un. This comparison comparison remains remains unverified, unverified, but but itseem itthe would would seem seem to to correla correla correl Khitanwith monosyllable locative locative case case form *e-xü/n-e *e-xü/n-e ‘upon ‘upon this’, this’, aaelements case case form form also attested attested in inwhich *e-d/ü-x-e *e-d/ü-x-e ‘now ‘now mpared mpared with the the possibly possibly pronominal pronominal Khitan Khitan monosyllable monosyllable mpared with the possibly pronominal Khitan monosyllable among the most stable ofalso a pronominal language, ismonosyll why th among the most stable elements ofthe a language, which is *üün-, which has been compared with the possibly Khitan odoo). The demonstratives are, however, not necessarily among most Khitan ؕ۠
ún.e †une ‘now’, which formally corresponds to the Proto-Mongolic Khitan Khitan ؕ۠
ؕ۠
ún.e ún.e †une †une ‘now’, ‘now’, which which formally formally corresponds corresponds to to the the Proto-M Proto-M eem to correlate with (> (> Modern Modern Mongolian Mongolian odoo). odoo). The The demonstratives demonstratives are, are, however, however, not not necessarily necessaril on n remains remains unverified, but but it it would would seem seem to to correlate correlate with with son remainsunverified, unverified, but it would seem to correlate with expect a full correspondence between Khitan and Proto-Mongolic. expect a full correspondence between Khitan and Proto-M ؕ
case ún †un. This comparison remains unverified, but itisto would seem correlate stable elements of case acase language, whichthis’, is why there no need expect a isisto locative form *e-xü/n-e ‘upon case form also attested inattested *e-d/ü-x-e ‘now’ locative locative form form *e-xü/n-e *e-xü/n-e ‘upon ‘upon this’, this’, aaiscase case form form also also attested in inno *e-d/ü-x-e *e-d/ü-x*e-d/ü-xonow’, the which Proto-Mongolic among among the the most most stable stable elements elements of aaalanguage, language, which which is why there no need need to t ow’, which formally corresponds corresponds to to the the Proto-Mongolic Proto-Mongolic now’, whichformally formally corresponds to†une the Proto-Mongolic surprising that theof Khitan distal pronoun is why ‫ۇ‬ qi there ‘that’ : plural ‫ٻۇ‬ surprising that the Khitan distal pronoun is ‫ۇ‬ qi ‘that’ : plu Khitan ؕ۠
ún.e ‘now’, which formally corresponds to the Proto-Mong full correspondence between Khitan and Proto-Mongolic. Ithowever, is are, therefore (> Modern Mongolian odoo). The demonstratives are, not necessarily (> (> Modern Modern Mongolian Mongolian odoo). odoo). The The demonstratives demonstratives are, however, however, not not nec nec ed in *e-d/ü-x-e ‘now’ 25 expect expect a a full full correspondence correspondence between between Khitan Khitan and and Proto-Mongolic. Proto-Mongolic. It It is is therefore therefore no no 25 -e e ‘upon ‘upon this’, case form form also attested inin *e-d/ü-x-e *e-d/ü-x-e ‘now’ ‘now’ n-e ‘uponthis’, this’,aaacase case formalso alsoattested attested in *e-d/ü-x-e ‘now’ item totally different from Mongolic *te-. item totally different fromform Mongolic *te-. in *e-d/ü-x-e ‘n locative case form *e-xü/n-e ‘upon this’, aqi case also among the most stable elements of a language, which whyattested there is suggesting no need to among among the the most most stable stable elements elements of of a a language, language, which which is is why why there there is is no noan not surprising that the Khitan distal pronoun is qi ‘that’ plural qi.t, ever, not necessarily surprising surprising that that the Khitan Khitan distal distal pronoun pronoun is is ‫ۇ‬ ‫ۇ‬ qi ‘that’ ‘that’ :::isplural plural ‫ٻۇ‬ ‫ٻۇ‬ qi.t, qi.t, suggesting an doo). doo). The The demonstratives demonstratives are, are, however, however, not not necessarily necessarily odoo). The demonstratives are,Mongolian however, odoo). not necessarily (> Modern The demonstratives are, however, not necess 25 25 25 expect a full correspondence between Khitan and Proto-Mongolic. It is therefore not expect expect a a full full correspondence correspondence between between Khitan Khitan and and Proto-Mongolic. Proto-Mongolic. It It is is theref there y there is no need to an item totally different Mongolic *te-. demonstrations of the relationshi item itemtotally totally different different from from Mongolic Mongolic *te-. *te-. ements ements of of language, which which is isisNumerals. why why there there isisisfrom no no need to lements ofaasuggesting alanguage, language, which why there noneed need tolanguage, One of the most simple Numerals. One of the most simple demonstrations thenee re among the most stable elements of ato which why there isofno surprising that the Khitan distal pronoun is ‫ۇ‬ qi ‘that’ plural qi.t,‫ٻۇ‬ suggesting an surprising surprising that that the the Khitan Khitan distal distal pronoun pronoun is is ‫ۇ‬ ‫ۇ‬:qi qi ‘that’ ‘that’‫ٻۇ‬ :is : plural plural ‫ٻۇ‬ qi.t, qi.t, sugges sugge sugge olic. It is therefore not nce ce between Khitan and and Proto-Mongolic. It It is is therefore therefore not not ncebetween betweenKhitan Khitanexpect andProto-Mongolic. Proto-Mongolic. It is therefore not and Mongolic is offered by the basic numerals. Most numeral roots and Mongolic is offered by the basic numerals. Most nume 25 25 25 25 aOne fullof between Khitan Proto-Mongolic. It is therefore item totally different from Mongolic *te-. item item totally totally different different from from Mongolic Mongolic *te-. *te-. and ‫ٻۇ‬ qi.t, suggesting One ofcorrespondence the the most most simple simple demonstrations demonstrations of of the the relationship relationship between between Khitan Khita Numerals. One the most simple demonstrations of the relationship distal istal pronoun pronoun isisNumerals. ‫ۇ‬ ‫ۇ‬ qian qi ‘that’ :of :plural plural ‫ٻۇ‬ ‫ٻۇ‬ qi.t, qi.t, suggesting suggesting an an distal pronoun isNumerals. ‫ۇ‬surprising qi‘that’ ‘that’ :that plural ‫ٻۇ‬ qi.t, suggesting an written in two ways: by‫ۇ‬ a qi single often forsuggestin cardinal written in twoeither ways: either by acharacter, single character, often for the Khitan distal pronoun is ‘that’ : plural ‫ٻۇ‬ qi.t, 25 25 25 and and Mongolic Mongolic is is offered offered by by the the basic basic numerals. numerals. Most Most numeral numeral roots roots in in Khitan Khitan can can be b 25 for Mongolic Mongolic *te-. between Khitan and Mongolic is offered by the basic numerals. Most Mongolic*te-. *te-. sequence of characters, often ordinals. In both cases, it is a qu sequence of characters, often for ordinals. In both cases, item totally different Mongolic *te-. Numerals. One ofways: the most simple demonstrations ofoften the relationship between Khitan Numerals. Numerals. One One of offrom the theby most most simple simple demonstrations demonstrations of ofcardinals, the the relationship relationship between between written written in in two two ways: either either by a a single single character, character, often for for cardinals, or or in in terms terms of of signscan with phonetic value, though some of asome the single numeral roots in Khitan bea written two ways: either by single signs with a in phonetic value, though of thecharacters single cha and Mongolic is offered by the basic numerals. Most numeral roots in Khitan can be and and Mongolic Mongolic is is offered offered by by the the basic basic numerals. numerals. Most Most numeral numeral roots roots in in Khitan Khitan nship between Khitan sequence sequence of of characters, characters, often often for for ordinals. ordinals. In In both both cases, cases, it it is is a a question question of of syllabic syllabi tsimple simple demonstrations of of the relationship relationship between Khitan Khitan st simpledemonstrations demonstrations ofthe the relationship between Khitan lack athephonetic approximation. From the combined information of lack aterms phonetic approximation. From the combined inform character, often for cardinals, orbetween in of a sequence of characters, often Numerals. One of most simple demonstrations of the relationship between Kh written in two ways: either by a single character, often for cardinals, or in terms of asti written written in in two two ways: ways: either either by by a a single single character, character, often often for for cardinals, cardinals, or or in in ter ter oots in Khitan can be signs signs with with aanumeral phonetic phonetic value, value, though though some some of of the the single single characters characters for fornumeral numerals numerals stil yby ythe the basic numerals. Most Most numeral roots roots in in Khitan Khitan can can be be thebasic basicnumerals. numerals. Most numeral roots in Khitan can be spelling we can establish the following basic numeral stems: 2 ‫ٰا‬ spelling we can establish the following basic stem for ordinals. In both cases, it is a question of syllabic signs with a phonetic and Mongolic ischaracters, offered byoften the basic numerals. roots in Khitan ca sequence of characters, often or for ordinals. In both cases, it numeral is a question of syllabic sequence of of characters, often for for ordinals. ordinals. In InMost both both cases, cases, itof itof is isthe aa question question of of or in terms of lack lack aaasequence phonetic phonetic approximation. approximation. From From the the two two types odo rinals, by by single character, character, often often for for cardinals, cardinals, or terms terms of of acombined acharacter, er byaaasingle single character, often for cardinals, orinin in of a‫ٰٻ‬ .lo†d ٰ
hu.ur†hur-, 4terms ‫ٰٻ‬ t.ur†dur-, 5information ֢ tau 7the ٛ‫
ڽ‬ da ٰ
hu.ur†hur-, 4combined t.ur-information †dur-, 5†tau, ֢ tau †tau, 7intypes 2ٛ‫
ڽ‬ written in two ways: either by a single often for cardinals, or terms value, though some of the single characters for numerals still lack a signs with a phonetic value, though some of the single characters for numerals still signs signs with with a a phonetic phonetic value, value, though though some some of the the single single characters characters for for numer numer aften question of syllabic 26 of spelling spelling we we can can establish establish the the following following basic basic numeral numeral stems: stems: 2 2 ‫
ٰا‬ ‫
ٰا‬ ci.urci.ur†jur-, †jur-, 3 3 ‫؜‬ 26 en ten for for ordinals. In In both both cases, it ititisis aaquestion question of of syllabic syllabic forordinals. ordinals. Insequence bothcases, cases, isaFrom question of syllabic to Mongolic 2 combined *jir(or *jïr-), *gur-, *dör-, *tab-, 75 *dalis7‫؜‬ tothe Mongolic 2 *jir(orIn3*jïr-), 34 *gur-, 45a*dör-, *tab-, of characters, often for ordinals. both itofis question of syll phonetic approximation. information ofcases, the two types lack a phonetic approximation. From the combined information the two types of lack lack a a phonetic phonetic approximation. approximation. From From the the combined combined information information of of the the two two tt ters for numerals still .lo.lo†dal-. †dal-. The The similarity similarit ٰ
ٰ
hu.urhu.ur†hur-, †hur-, 4 4 ‫ٰٻ‬ ‫ٰٻ‬ t.urt.ur†dur-, †dur-, 5 5 ֢ ֢ tau tau †tau, †tau, 7 7 ٛ‫
ڽ‬ ٛ‫
ڽ‬ da da 2 2 ue, e, though some of the the single characters for numerals numerals still still ue,though thoughsome someof ofsigns thesingle singlecharacters characters for numerals still it woulditfor be impossible to explain the situation by assuming borro would be basic impossible explain the situation by assumi awe phonetic though some of55to the single characters for numerals 26 26 ci.urofspelling spelling wewith can establish thevalue, following numeral stems: 2*dalwe can establish the following basic numeral stems: 2 ‫
ٰا‬ ci.ur†jur-, 3 ‫؜‬ spelling spelling we can can establish establish the the following following basic basic numeral numeral stems: stems: 2 2 ‫
ٰا‬ ‫
ٰا‬ ci.urci.ur†ju †ju n of the two types of 3 3 *gur-, *gur-, 4 4 *dör-, *dör-, *tab-, *tab-, 7 7 *dalis is unmistakable, unmistakable, and an to to Mongolic Mongolic 2 2 *jir*jir(or (or *jïr-), *jïr-), ation. tion. From the combined information information of of the the two types types of ation.From Fromthe thecombined combined information of thetwo twopowers types of numerals for the ten 100 jau, ٔ
ming, 10ming 000 numerals for theof powers of ֔
teninformation 100 1000 ֔
jau, ٔ
lack abe phonetic approximation. From the of1000 the two type †dal-. The similarity ٰ
†hur-, 4 ‫ٰٻ‬ t.ur†dur-, 5†jur-, ֢ tau †tau, 7by ٛ‫
ڽ‬ da772.lo33would †hur-, t.ur†dur-, tau ٰ
ٰ
hu.urhu.ur†hur-, †hur-, 44to ‫ٰٻ‬ ‫ٰٻ‬ t.urt.ur†dur-, †dur-, 55 ֢ ֢combined tau †tau, †tau, ٛ‫
ڽ‬ ٛ‫
ڽ‬ da da2222.lo.lo†dal-. †dal-. The The sim si si ‫
ٰا‬ ci.ur- †jur-, †jur-, ‫؜‬ itithu.urwould be impossible impossible to explain explain the the situation situation by assuming assuming borrowing. borrowing. The The highe highe he hefollowing following basic basic numeral numeral stems: stems: 2 2 ‫
ٰا‬ ‫
ٰا‬ ci.urci.ur†jur-, 3 3 ‫؜‬ ‫؜‬ the following basic numeral stems: 2 ‫
ٰا‬ ci.ur†jur-, 3 ‫؜‬ Mongolic cognates: *jaxu-, *mingga-, and *tüme-, though they ar Mongolic cognates: *mingga-, and *tüme-, though 26 the following 26 26 26 26 *jaxu-, spelling we can establish basic numeral stems: 2 ‫
ٰا‬ ci.ur†jur-, to Mongolic 2 *jir(or *jïr-), 3 *gur-, 4 *dör-, 5 *tab-, 7 *dalis unmistakable, and to to Mongolic Mongolic 2 2 *jir*jir(or (or *jïr-), *jïr-), 3 3 *gur-, *gur-, 4 4 *dör-, *dör-, 5 5 *tab-, *tab-, 7 7 *dal*dalis is unmistakab unmistaka †dal-. The similarity †dal-. The similarity to Mongolic 2 *jir(or *jïr-), 3 *gur-, 4 *dör-, 5 *tab-, numerals numerals for for the the powers powers of of ten ten 100 100 ֔
֔
jau, jau, 1000 1000 ٔ
ٔ
ming, ming, 10 10 000 000 ‫ڼ‬ ‫ڼ‬ tum tum also also have hav .lo.lo†dal-. †dal-. The The similarity similarity rr-†dur-, †dur-, ֢ tau †tau, †tau, 7 7 ٛ‫
ڽ‬ ٛ‫
ڽ‬ da da ur†dur-,555֢ ֢tau tau †tau, 7 ٛ‫
ڽ‬ da .lo†dal-. The similarity since they would been more liable to liable beda borrowed. 222 4 ‫ٰٻ‬ since theyhave would have been more to be borrowed. The simil ٰ
ithu.ur†hur-, t.ur5and ֢ tau †tau, 7 by ٛ‫
ڽ‬ 2.lo- †dal-. would be impossible toisisit explain the situation byto assuming borrowing. The higher it would would be be impossible impossible to to†dur-, explain explain the the*tüme-, situation situation bythe assuming assuming borrowing. The The -r-), is262626 unmistakable, and Mongolic cognates: cognates: *jaxu-, *jaxu-, *mingga-, *mingga-, and *tüme-, though though they they are areborrowing. less less diagnostic diagnostic 7it4*dalis unmistakable, and would be impossible explain situation 26 r-), -), 333*gur-, *gur-, 44Mongolic *dör-, *dör-, 5 5 *tab-, *tab-, 7 7 *dal*dalunmistakable, unmistakable, and and *gur-, *dör-, 5 *tab-, 7 *dalis unmistakable, and We also know the approximate phonetic shape of the Khita We also know the approximate phonetic shape of t to Mongolic 2 *jir(or *jïr-), 3 *gur-, 4 *dör-, 5 *tab-, 7 *dalis unmistakable, numerals for the powers of ten 100 ֔
jau, 1000 ٔ
ming, 10 000 ‫ڼ‬ tum also have numerals numerals for for the the powers powers of of ten ten 100 100 ֔
֔
jau, jau, 1000 1000 ٔ
ٔ
ming, ming, 10 10 000 000 ‫ڼ‬ ‫ڼ‬ tum tum als al al orrowing. The higher since since they they would would have have been been more more liable liable to toitem be befor borrowed. borrowed. by assuming The higher numerals powers 100 explain the situation situation by by assuming assuming borrowing. The The higher higher oexplain explainthe the situation byborrowing. assuming borrowing. The higher ֦
isborrowing. †is-. This item is often compared withoftheten Modern ֦
is explain †is-. This is the often compared with the Mongolian Modern Mo it would be impossible to the situation by assuming borrowing. The hi Mongolic cognates: *jaxu-, *mingga-, and *tüme-, though they are less diagnostic, Mongolic Mongolic cognates: cognates: *jaxu-, *jaxu-, *mingga-, *mingga-, and and *tüme-, *tüme-, though though they they are are less less diag dia dia9 000 tum also We We also also know know the theThis approximate approximate phonetic phonetic shape shape of of the the Khitan numeral numeral stem stem jau, ming, Mongolic cognates: *jaxu-, of fof ten ten 100 100 ֔
֔
jau, jau, 1000 1000 ٔ
ٔ
ming, ming, 10 10 000 000 ‫ڼ‬ ‫ڼ‬ tum tum also also have have ten‫ڼ‬ 100 ֔
jau,have 1000 ٔ
ming, 10powers 000 ‫ڼ‬ tum also have This comparison is wrong, however, for the Khitan Mongolian deriv comparison isjau, wrong, however, for the Mongolian ite numerals for the of ten 100 ֔
borrowed. 1000 ٔ
ming, 10 numeral 000 ‫ڼ‬item tum also since they would have been more liable to be since since they they would would have have been been more more liable liable to to be be borrowed. borrowed. y are less diagnostic, ֦
֦
is is †is-. †is-. This This item item is is often often compared compared with with the the Modern Modern Mongolian Mongolian numeral 9 9 is is ~ ~ yis yis -, , *mingga-, *mingga-, and and *tüme-, though though they they are are less less diagnostic, diagnostic, u-, *mingga-, and*tüme-, *tüme-, though they are less diagnostic, *mingga-, and *tüme-, though they are less diagnostic, since they would (as also in 90 *yer-e/n), an innovation of the Proto-Mongolic li (as also in 90 *yer-e/n), an innovation of the Proto-Mo Mongolic cognates: *jaxu-, and *tüme-, though they are less diagno We also We know the approximate phonetic shape of shape the Khitan stem 9 Weisis also also know know the the*mingga-, approximate approximate phonetic phonetic shape of of the thenumeral Khitan Khitan numeral numeral 27 This This comparison comparison wrong, wrong, however, however, for for the the Mongolian Mongolian item item derives derives from from *yer-sü/n *yer-sü/ nnmore more liable to to be be borrowed. borrowed. moreliable liable to be borrowed. confirmed by data from the “teens” in Jurchen-Manchu. Anot have been more liable to have be borrowed. confirmed by data from the “teens” in Jurchen-Manchu since they would been more liable toModern be borrowed. ֦
is †is-. This item is Khitan often with the numeral 9numeral is ~isisyis. ֦
is is †is-. †is-. This This item itemcompared is is often often compared compared with with the theMongolian Modern Modern Mongolian Mongolian numeral 99 Khitan numeral stem 9֦
(as (as also also in in 90 90 *yer-e/n), *yer-e/n), an an innovation innovation of of the the Proto-Mongolic Proto-Mongolic lineage, lineage, as as also als approximate approximate phonetic phonetic shape shape of of the the Khitan numeral numeral stem stem 9numeral approximate phonetic shape of the Khitan numeral stem Mongolic is the numeral 699phonetic *jir-gu-xa/n (2x3), which must have rep Mongolic is theshape 6 shape *jir-gu-xa/n (2x3), which must We also know the approximate phonetic of the Khitan numeral We also know the approximate of Khitan numeral ste 27 27 the This comparison is wrong, however, for the Mongolian item derives from *yer-sü/n This This comparison comparison is is wrong, wrong, however, however, for for the the Mongolian Mongolian item item derives derives from from *y *y *i ian numeral 9 is ~ yis. confirmed confirmed by byMongolian data data from fromnumeral the the “teens” “teens” Jurchen-Manchu. Jurchen-Manchu. Another Anothercharacter innovation innovation nen ncompared compared with with the Modern Modern Mongolian numeral isis~~in yis. yis. comparedstem with the Modern Mongolian numeral 9ishide ~in yis. SIX. 9 Judg numeral likely to99likely behind the Khitan character ‫ך‬ numeral to hide behind the Khitan ‫ך‬is in S 9the is †is-. This item is often compared the Modern Mongolian ֦
is †is-. This item is often with the Modern Mongolian numeral ~ (as also in 90 *yer-e/n), an innovation of the Proto-Mongolic lineage, as is also (as (as also also in in 90 90 *yer-e/n), *yer-e/n), an an innovation innovation of of the the Proto-Mongolic Proto-Mongolic lineage, lineage, as as erives from *yer-sü/n Mongolic Mongolic isis the theitem numeral numeral 66 *jir-gu-xa/n *jir-gu-xa/n (2x3), (2x3), which which must must have have replaced replaced the the origina origina however, for the Mongolian item derives derives from from *yer-sü/n *yer-sü/n g,,however, however,for forthe theMongolian Mongolian item derives from *yer-sü/n 27 27 27 27 This comparison is wrong, however, for character the Mongolian item derives *yerconfirmed by datato from the “teens” in Jurchen-Manchu. Another innovation in confirmed confirmed by data data from from the the “teens” “teens” in in Jurchen-Manchu. Jurchen-Manchu. Another Another innova innov ican as is numeral numeral likely likely toby hide hide behind behind the the Khitan Khitan ‫ך‬ ‫ ך‬SIX SIX .. Judging Judging by byfrom the theinnova corre corre an innovation innovation of of the the Proto-Mongolic Proto-Mongolic lineage, lineage, as as is is also anlineage, innovation ofalso the Proto-Mongolic lineage, asexplained isalso alsocharacter 25 (as also in 90 *yer-e/n), an innovation of the Proto-Mongolic lineage, as is In this case, the discrepancy has also been by drawing the Romanization 27 27 Mongolic is the numeral 6numeral *jir-gu-xa/n (2x3), which must havemust replaced the original Mongolic is is the the numeral 66 *jir-gu-xa/n *jir-gu-xa/n (2x3), (2x3), which which must have have replaced replaced the the 27 Another innovation inMongolic 27 he the “teens” in Jurchen-Manchu. Jurchen-Manchu. Another Another innovation innovation inin Another innovation in the“teens” “teens” in Jurchen-Manchu. ofin the Khitan character into question. Most recently, WuJurchen-Manchu. Yingzhe (2011: 73-76) has Another innovatio confirmed by data from the “teens” in 24 24 numeral likely to likely hide the Khitan character ‫ך‬though SIXand . (2009) Judging the correnumeral numeral likely to to hide hide behind behind the the Khitan Khitan character character ‫ך‬ ‫ך‬ SIX .. by Judging Judging by by&the the replaced the original All behind Khitan data are quoted from Kane (2009) Wu SIX Yingzhe Janhunen (2 All Khitan data are quoted from Kane and Wu& Yingzhe Ja 6e6*jir-gu-xa/n *jir-gu-xa/n (2x3), (2x3), which which must must have have replaced replaced the the original original suggested the Romanization te, congruent Mongolic *te, mun, which *jir-gu-xa/n (2x3), which must have replaced thewith original Mongolic is the numeral 6 *jir-gu-xa/n (2x3), which must have replaced the orig strative pronouns, cf. also Wu Yingzhe (2009, 2011). udging by the correstrative pronouns, cf. also Wu Yingzhe (2009, 2011). would correspond to Mongolic *mön ‘this,’ has also been proposed. Unfortunately, such ehind hind the Khitan character character ‫ך‬ ‫ך‬ SIX SIX . . Judging Judging by by the the correcorreSIX . Judging by the correehindthe theKhitan Khitan character ‫ך‬ 25 to hide 24 24 numeral likely 25 behind the Khitan character ‫ ך‬SIX. Judging by the co In this case, the discrepancy has also been explained by drawing the Roman 118



All AllKhitan Khitanare data datanot are arecompatible quoted quotedfrom from Kane Kane (2009) (2009) and Wu WuYingzhe Yingzhe & Janhunen Janhunen (2010). (2010). On the demon demon In thisthecase, theand discrepancy has also been explained bythe drawing t reinterpretations with evidence supporting the& Romanization qi, On into question. recently, (2011: has73-76) suggested strative strativepronouns, pronouns, cf. cf.character also also Wu Yingzhe Yingzhe (2009, (2009, 2011). 2011). character into Most question. MostWu recently, Yingzhe (2011: has as established on the basis of Wu another (ethnonymic) attestation by Yingzhe Kane Wu (2009: 73),73-76) cf. 25 24 25 24 24 24case, congruent with Mongolic *te, though mun, which would correspond to the Mongolic In In this this case, the the discrepancy discrepancy has has also also been been explained explained by by drawing drawing the the Romanization Romanization of of theOn Khitan Khita congruent with Mongolic *te, though mun, which would correspond to also Wu Yingzhe &Khitan Janhunen 65-66). All Khitan data are quoted from Kane (2009) and Wu Yingzhe & Janhunen On(2010). the demonAll All Khitan data data(2010: are are quoted quoted from from Kane Kane (2009) (2009) and and Wu Wu Yingzhe Yingzhe & &(2010). Janhunen Janhunen (2010). On the th th

been proposed. Unfortunately, such reinterpretations are not the compatible with the character character into intoquestion. question. Most Most recently, recently, Wu Wu Yingzhe Yingzhe (2011: (2011: 73-76) 73-76) has hassuggested suggested theRomanization Romanization tet 26strative been proposed. Unfortunately, such reinterpretations pronouns, cf. also Wu Yingzhe (2009, 2011). strative strative pronouns, pronouns, cf. cf. also also Wu Wu Yingzhe Yingzhe (2009, (2009, 2011). 2011). Since the Proto-Mongolic reconstructions are, in principle, well known, the presentare not compatib en (2010). On the demon25and 25 25 25 om m Kane Kane (2009) and Wu Wu Yingzhe Yingzhe & & Janhunen Janhunen (2010). (2010). On On the the demondemonthe Romanization qi, as established on the basis of another (ethnonymic) attestatio 24 congruent congruent with with Mongolic Mongolic *te, *te, though though mun, mun, which which would would correspond correspond to to Mongolic Mongolic *mön *mön ‘this’, ‘this’, has has also als rom Kane(2009) (2009) and Wu Yingzhe & Janhunen (2010). On the demonthe Romanization qi, as established on the basis of another (ethnonymic In will this not case, the has also by drawing the Romanization of theOn Khitan In In this thisdiscrepancy case, case, the the discrepancy discrepancy has has explained also also been been explained explained by by drawing drawing the the Romanization Romanization of of de th th All Khitan data from Kane (2009) and Wu Yingzhe & Janhunen (2010). the paper discuss themare in quoted the detail. Itbeen may nevertheless be noted that the numeral ngzhe ingzhe(2009, (2009, 2011). cf. also Wu Yingzhe &vocalism Janhunen (2010: 65-66). been been proposed. proposed. Unfortunately, Unfortunately, such such reinterpretations reinterpretations are not not(2011: compatible compatible with with the the evidence evidence supporting supportin Yingzhe (2009,2011). 2011). cf. also Wu Yingzhe & Janhunen (2010: 65-66). character question. Most recently, Wu Yingzhe (2011: 73-76) has suggested the Romanization te, character character into into question. Most Most recently, recently, Wu Wu Yingzhe Yingzhe (2011: 73-76) has has suggested suggested the the Romani Romani pronouns, cf. also Wu Yingzhe (2009, 2011). root for 2strative isinto harmonically ambiguous: palatal isare suggested by73-76) *ji-tüxer ‘second omanization of the Khitan 26 question. 26 has also been explained explained by drawing drawing the the Romanization Romanization of of the the Khitan Khitan Since the Since Proto-Mongolic are, inattestation principle, well known, thekn p 25 by the theand Romanization Romanization qi, qi, as as established established on on the the basis basis of ofreconstructions another another (ethnonymic) (ethnonymic) attestation by‘this’, Kane Kane (2009: (2009: 73 73 yhas hasalso alsobeen been explained by drawing the Romanization of the Khitan the Proto-Mongolic reconstructions are, in by principle, well congruent with Mongolic *te, though mun, which would correspond to Mongolic *mön has also congruent congruent with with Mongolic Mongolic *te, *te, though though mun, mun, which which would would correspond correspond to to Mongolic Mongolic *mön *mön ‘this’, ‘this’ ‘this’ (wife)’ *jirin ‘two (females),’ while velar vocalism isbysuggested by *jïr-a/n Inpossibly this case, the discrepancy has also been explained drawing the Romanization of the K ested the Romanization te, cently, cently, Wu Wu Yingzhe (2011: (2011: 73-76) 73-76) has has suggested suggested the the Romanization Romanization te, te, discuss them in the detail. It may nevertheless be noted that the numeral root cf. cf. also also Wu Wu Yingzhe Yingzhe & & Janhunen Janhunen (2010: (2010: 65-66). 65-66). ecently, WuYingzhe Yingzhe (2011: 73-76) has suggested the Romanization te, discuss them in the detail. It may nevertheless be noted that the num been proposed. Unfortunately, such reinterpretations are not compatible with the evidence supporting been been proposed. proposed. Unfortunately, Unfortunately, such suchWu reinterpretations reinterpretations are are73-76) not not compatible compatible with with the the theRomanizati evidence evidence su s ‘sixty’ character into question. Most recently, Yingzhe (2011: has suggested golic mun, *mönwhich ‘this’, has also 26 26(2x3x10). hough ough would would correspond correspond Mongolic Mongolic *mön *mön ‘this’, ‘this’, has has also also ambiguous: palatal vocalism isprinciple, suggested by *ji-tüxer (wife)’ and Since Since the theRomanization Proto-Mongolic Proto-Mongolic reconstructions are, are, in in principle, well known, known, the the‘second present present paper paper will will no no houghmun, mun,which which would correspond to Mongolic *mön ‘this’, has also ambiguous: palatal vocalism iswell suggested by *ji-tüxer ‘second (w the Romanization qi, to asto established on the basis of another (ethnonymic) attestation by Kane (2009: 73), the the Romanization qi, qi,reconstructions as as established established on on the the basis basis of of another another (ethnonymic) (ethnonymic) attestation attestation by by Kane Kane (2 ( congruent with Mongolic *te, though mun, which would correspond to Mongolic *mön ‘this’, ha h the evidence supporting uch such reinterpretations are are not not compatible compatible with with the evidence evidence supporting supporting (females)’, while velar vocalism isvocalism suggested by *jïr-a/nroot ‘sixty’ (2x3x10). discuss discuss them them ininWu the the detail. detail. It& may may nevertheless nevertheless be be noted noted that thatnot the thesuggested numeral numeral root for for 22 is is harmonically harmonicall suchreinterpretations reinterpretations are not compatible withItthe the evidence supporting (females)’, while velar is by *jïr-a/n ‘sixty’ (2x3x1 cf. also Wu Yingzhe &Yingzhe Janhunen (2010: 65-66). cf. cf. also also Wu Yingzhe & Janhunen Janhunen (2010: (2010: 65-66). 65-66). been proposed. Unfortunately, such reinterpretations are compatible with the evidence suppo station by Kane (2009: 73), 27 27 by 26 26 hed ed on on the basis ofof another another (ethnonymic) (ethnonymic) attestation attestation by Kane Kane (2009: (2009: 73), 73), On theestablished teens inthe Jurchen-Manchu, cf.in Janhunen (1993). Although these items we ambiguous: ambiguous: palatal palatal vocalism vocalism isisOn suggested suggested by by *ji-tüxer *ji-tüxer ‘second (wife)’ and and possibly possibly *jirin *jirin ‘two ‘tw shed onthe thebasis basis26 of another (ethnonymic) attestation by Kane (2009: 73), teens in cf.(wife)’ Janhunen (1993). Although thes Since the26 Proto-Mongolic reconstructions are, inJurchen-Manchu, principle, well known, theknown, present paper will not Since Since the the Proto-Mongolic Proto-Mongolic reconstructions reconstructions are, in‘second principle, principle, well well known, the the present present paper paper pape the Romanization qi, as on the basis ofare, another (ethnonymic) attestation by Kane (2009 n(2010: (2010: 65-66). from a Mongolic language, we do not know whether the borrowing happened befo (females)’, (females)’, while while velar velar vocalism vocalism is is suggested suggested by by *jïr-a/n *jïr-a/n ‘sixty’ ‘sixty’ (2x3x10). (2x3x10). en (2010:65-66). 65-66). from a Mongolic language, we do not know whether the borrowing hap discuss them inWu thethem detail. Itthe nevertheless be noted that the numeral root for 2 root is harmonically discuss discuss them in in the detail. detail. ItIt(2010: may may nevertheless nevertheless be be noted noted that that the the numeral numeral root for for 22 isis harm harm also Yingzhe & may Janhunen 65-66). the present paper will not 27 27 cf. constructions onstructions are, in in principle, well well known, known, the the present present paper will will not not the Proto-Mongolic Para-Mongolic lineages. The structure of the teens in Ju 26the On On the teens teens ininProto-Mongolic Jurchen-Manchu, Jurchen-Manchu, cf. cf.paper Janhunen Janhunen (1993). (1993). Although Although these these items items were were certainly borrowed borrowe constructionsare, are, inprinciple, principle, well known, the present paper will not the Proto-Mongolic and lineages. The structure of the ambiguous: palatal vocalism isvocalism suggested by *ji-tüxer ‘second (wife)’ and possibly *jirin ‘two ambiguous: ambiguous: palatal palatal vocalism is isand suggested suggested by by *ji-tüxer *ji-tüxer ‘second ‘second (wife)’ (wife)’ and and possibly possibly *ji *jt Since the reconstructions are, inPara-Mongolic principle, well known, thecertainly present paper wi root for 2 is harmonically may ay nevertheless nevertheless be noted noted that that the numeral root for 2not 22is is harmonically harmonically fromroot that attested in the language of the Khitan texts. On the reconstruction from from aaMongolic Mongolic language, language, we wefor do do know whether whether theborrowing borrowing happened happened before before or or after afterthe the therecons split splitan oo may neverthelessbe be noted thatthe thenumeral numeral root fornot isknow harmonically from that attested in the language of the Khitan texts. On

ongolic 2 *jir- (or *jïr-),26 3 *gur-, 4 *dör-, 5 *tab-, 7 *dal- is unmistakable, and uld be impossible to explain the situation by assuming borrowing. The higher rals for the powers of ten 100 ֔
jau, 1000 ٔ
ming, 10 000 ‫ ڼ‬tum also have Khitan: Understanding the Language Behind the Scripts 119 golic cognates: *jaxu-, *mingga-, and *tüme-, though they are less diagnostic, they would have been more liable to be borrowed. We also know the approximate phonetic shape of the Khitan numeral stem 9 numeral 9 is ~ yis. This comparison is wrong, however, for the Mongolian †is-. This item is often compared with the Modern Mongolian numeral 9 is ~ yis. derives from for *yer-sü/n (as also initem 90 derives *yer-e/n), an *yer-sü/n innovation of the comparison isitem wrong, however, the Mongolian from Proto-Mongolic lineage, as is also confirmed by data from “teens” in lso in 90 *yer-e/n), an innovation of the Proto-Mongolic lineage, as isthe also 27 27 innovation in Mongolic is the numeral rmed by dataJurchen-Manchu. from the “teens” Another in Jurchen-Manchu. Another innovation in 6 *jirgu-xa/n (2x3), which(2x3), must which have replaced originalthe numeral golic is the numeral 6 *jir-gu-xa/n must havethe replaced originallikely to hide behind behind the the Khitan character SIX. Judging Judging by the the corre­ corre-sponding ral likely to hide character ‫ ך‬SIX.

“teen” in Jurchen-Manchu, the Khitan item may have been †nil-, but this shape cannot yet be confirmed by the Khitan Small Script. It may be noted further that the Khitaninbasic numerals seem toKhitan lack sponding “teen” in & Jurchen-Manchu, thetheKhitan item may item havemay bee sponding “teen” the Khitan data are quoted from Kane (2009) and Wu Yingzhe Janhunen Jurchen-Manchu, (2010). On demonthe stem extension *-pA/n > *-bA/n ~ *-xA/n, which is present in the shape cannot yetcannot be confirmed by the Khitan Small Script. ehu, pronouns, cf. also Wubeen Yingzhe (2009, 2011). an item may have but this shape yet be confirmed by the Khitan Small Script. the Khitan item may†nil-, have been †nil-, but this his the Small discrepancy hasitems also been explained by be drawing thefurther the Khitan Mongolic for the first decade 3 *gur-ba/n, 4 further *dör-be/n, 7the *dal.u-xa/ It may that theof Khitan basic numerals mall Script. Itnoted may beRomanization noted that Khitan basic seem nume he case, Khitan Script. ter into question.n,Most recently, Wu Yingzhe (2011: 73-76) has suggested the an Romanization te, specific extension *-pA/n > *-bA/n ~ *-xA/n, which is present in the Mongo and others. This element might, then, represent innovation basic numerals seem to lack the stem extension *-pA/n > *-bA/n ~ *-xA/n, which is present in t at the Khitan basic numerals seem to lack the stem uent with Mongolic *te, though mun, which would correspond to Mongolic *mön ‘this’, has also first decade 3 *gur-ba/n, 4 *dör-be/n, 7 *dal.u-xa/n, and others. Thio to the Proto-Mongolic lineage, though it is also possible that the Khitan present in the Mongolic items for the first decade 3 *gur-ba/n, 4 *dör-be/n, 7 *dal.u-xa/n, and A/n, which is present such in the Mongolic items the proposed. Unfortunately, reinterpretations are notfor compatible with the evidence supporting then, represent an innovation specific to the Proto-Mongolic lineage, -xa/n, others. This element might, then, represent anininnovation specific to thewhat Proto-Mongoli e/n, 7 and *dal.u-xa/n, andsystem others. This element might, was secondarily simplified; fact,bywe do not know manization qi, asnumeral established on the basis of another (ethnonymic) attestation Kane (2009: 73), that Khitan numeral system was secondarily i oific Wuto Yingzhe &lineage, Janhunen (2010: 65-66). oto-Mongolic itpossible also possible that the Khitan numeral system was secondarily the Proto-Mongolic lineage, though itthe is also the exactthough shapes ofis the Khitan numerals were in absolute cardinal use. Onsimplified;sim e the Proto-Mongolic reconstructions are, in principle, well known, the present paper will not know whatknow thedo exact of3 shapes the Khitan numerals were condarily insimplified; fact, wethe doin not what exact the Khitan numerals were in ystem wassimplified; secondarily fact, we not shapes the other Khitan numerals 2the†jur-, †hur-, 4of †durcontain thein absolute s them in the detail. It may hand, nevertheless be noted thatthe the Khitan numeral root for 2 is harmonically the other hand, numerals 2numerals †jur-, 3 2†hur-, 43†durcontain erals were in absolute cardinal use. On the other hand, the Khitan †jur-, †hur-, 4 †du Khitan numerals were in absolute cardinal use. On segment which by most probably is a(wife)’ derivative element. This‘two element is uous: palatal vocalism is *r, suggested *ji-tüxer ‘second and possibly *jirin which most probably is a derivative element. This element is also presi †hur-, 4 †durcontain the segment *r, which most probably is a derivative element. This element ls 2 †jur-, 3 †hur-, 4 †durcontain the segment *r, es)’, while velar also vocalism is suggested by *jïr-a/nmeaning ‘sixty’ (2x3x10). present in Mongolic, that it must represent an innovation meaning that it must represent an innovation common to both the Pro his element is also present in Mongolic, meaning that it must represent an innovation common to bo he teens in Jurchen-Manchu, cf. Janhunen (1993). Although these items were certainly borrowed e element. This element is also present in Mongolic, 28 common the Proto-Mongolic and the28Para-Mongolic 28 thelineage. Mongolic language, we do to notboth knowthe whether the borrowinglineage. happened before or after split of Para-Mongolic

mmon to both the Proto-Mongolic and the Para-Mongolic lineage. nnovation common to both the Proto-Mongolic and

oto-Mongolic and Para-Mongolic lineages. The structure of the teens in Jurchen-Manchu differs hat attested in the language of the Khitan texts.Among On the the reconstruction and connections the Terms for the four seasons. surprises offered bysurprises theofcomparative Terms forTerms the four the byoffered the com forseasons. the fourAmong seasons. Among the offered surprises by n numerals, cf. also Chingeltei (1997).

material are thematerial terms the fourterms seasons, which virtually identical are thefor terms for the four which arewhich virtually es offered by the comparative are the for seasons, the fourare seasons, are identical virtually in id g the surprises offered by the comparative material heu.úr †haur ==*kabur in Khitan (†) andProto-Mongolic Proto-Mongolic (*): ‫)*(
צپ‬: heu.úr †haur *kabur ju3 identical in Khitan (†) Proto-Mongolic ‫
צپ‬ heu.úr †haur‘spring,’ =‘spring’, *kabur ‫
ڕؘ‬ ‘spring’ s,virtually which are virtually identical inand Khitan (†) (*): and ‫צؽٿ‬ n.am.úr *namur== ‘autumn’, ؈۲
u.ul ؈ †jun ‘summer,’ †namur *namur‘autumn’, bur ‘spring’, ‫
ڕؘ‬ ju33.un †jun == ‘summer’, ‫†צؽٿ‬namur n.am.úr=†namur *namur †haur = *kabur ‘spring’, ‫‘
ڕؘ‬summer’, ju*jun .un †jun = *jun 3*jun *ebül) ‘winter’. Since the possibility of a wholesale borrowing is ext ‘autumn’, ؈۲
u.ul †uul = *öbül (< *ebül) ‘winter’. Since the possibility of a wholesale borrow ‘autumn,’ ؈۲
u.ul ur = *namur ‘autumn’, u.ul †uul †uul = = *öbül *öbül (< (< *ebül) ‘winter.’ Since the possibility it must be a question of cognates. Moreover, the Khitan items, esale extremely is unlikely, must be aunlikely, questionitofmust cognates. Moreover, ity of borrowing a wholesale extremelyitis unlikely, of is a borrowing wholesale borrowing extremely be a question of the Khita identical with Proto-Mongolic counterparts, show a somewhat g r, Khitancognates. items, when not fully identical with theirnot Proto-Mongolic show a so es.the Moreover, the Khitan items, when nottheir fully Moreover, the Khitan items, when fully identicalcounterparts, with their phonological innovation. This would seem to confirm Khitan waK show a somewhat degree of phonological This would seem tothat confirm that cs, counterparts, show greater a somewhat greater degree of innovation. innovative, or also more “worn”, than Proto-Mongolic. 27 onfirm was generally more innovative, or also more “worn”, than Proto-Mongolic. uld seemthat to Khitan confirm that Khitan was generally more On the teens in Jurchen-Manchu, cf. Janhunen (1993). Although these items were The terms for the seasons also provide anprovide important clue ongolic. The terms the fourwhether seasons an impo han Proto-Mongolic. certainly borrowed from a Mongolic language, wefour dofor not know thealso borrowing the innovations shared by Proto-Mongolic and Para-Mongolic. The happened before or after the split of the Proto-Mongolic and Para-Mongolic lineages. ide an important clue to understanding the innovations shared by Proto-Mongolic and Para-Mong ons also provide an important clue to understanding TheThe structure of the teens in Jurchen-Manchu differs from that represent attested in segment the language in *namur ‘autumn’ must represent the same as the medial Para-Mongolic. medial nasal *m in *namur ‘autumn’ must the same segment as t Mongolic and Para-Mongolic. The medial nasal *m of the Khitan texts. On the reconstruction and connections of the Khitan numerals, cf. *kabur ‘spring’. In other words, it is a question of a suffix, possibly segment as the medialasobstruent *b obstruent in *kabur In other words, it is a question of a suffix,re nt the same segment the(1997). medial *b‘spring’. in also Chingeltei Pre-Proto-Mongolic *-pUr, which is represented as *-mUr as after a of aa question suffix, possibly reconstructable as Pre-Proto-Mongolic *-pUr, which is represented *-mU 28 a suffix, is of as Using the possibly method ofreconstructable internal reconstruction, the *r in 3 *gur- may be compared nasal onset, and asan *-bUr elsewhere. Possibly, the element in * *-mUr with after a syllable an nasal onset, and as *-bUr elsewhere. Possibly, the*(-)bül element hedis as represented asthe *-mUr afterwith syllable with *t present in aMongolic 30 *guc-i/n < *gut-ï/n, though the background of this also relevant inrelevant this any may case, change *pchange > *m, which he Possibly, element *(-)bül in *öbül ‘winter’ is The also in :this context. Inbethe any case, the *p > *m correspondence is unknown. case of 4iscontext. *dör40In *döcin more complicated re. the element *(-)bül in *öbül ‘winter’ since this> basic digit is also in some Mongolic languages (Shirongolic) *der-, case ofaattested progressive assimilation, is likely toas place > *m, which may seen as ofas progressive distant assimilation, is have likelytaken to have ta yange case,*pthe change *p *m,be which may becase seen adistant 29 29 while the corresponding decade is present, as a loanword, inand Tungusic in the unexpected protolanguage of Proto-Mongolic Para-Mongolic. ly to have taken place in the common protolanguage of Proto-Mongolic and Para-Mongolic. ation, is likelyshape to have taken place in the common (*)deki(n), cf. Doerfer (1985: 78-79). 29 ngolic. and Para-Mongolic.29 Kinship terms. In view numerals the terms four Kinship terms.ofInthe view of the and numerals and for the the terms forsea th that there is no overall correspondence in the realm of kinship termsand for the the terms four seasons it is curious there is no overall correspondence in the realm ofterm kin erals for the four seasons itthat is curious Khitan terms, including ‫
׿‬ai ‘father’, ia ‘elder realm in of the kinship Severalkinship Khitan kinship terms, including ‫
׿‬ai‫
ۓ‬ ‘father’, ‫
ۓ‬brother’, ia ‘elder dence realmterminology. of kinship terminology. Several have obvious on the Mongolic side. In s  ‘elder brother’, and brother’, ‫
ئ‬ausister’, ‘elder sister’, have no cognates obvious cognates on the Mongolic and ‫
ئ‬ au no ‘elder ‫
ۓ
׿‬ aiia‘father’, ‫
ۓ‬ia ‘elder

which most numerals †jur-, 3 †hur-, 4 †durcontain thecontain segmentthe *r,segment he Khitan2numerals 2 †jur-, 3 †hur-, 4 †dur*r, probably is a derivative elem meaning must represent an innovation common toan both the P ent. element isThis also present Mongolic, that it must represent innovation commo that it mustan represent innova erivative element isinalso present initmeaning Mongolic, ablyThis is a element. derivative element. This element isthat also present in meaning Mongolic, 28 28 28 28 the Para-Mongolic lineage. on common to both the Proto-Mongolic and Para-Mongolic lineage. lineage. ent anrepresent innovation common tocommon both the to Proto-Mongolic and the Para-Mongolic must an innovation both thethe Proto-Mongolic and 8 ic lineage.28120 SCRIPTA, VOLUME 4 (2012) Terms for the Terms four seasons. Among the four surprises offered by the co cs for the four seasons. Among the Among surprises Terms for the seasons. theoff are the terms for the four seasons, which are virtually identical urprises offered by the comparative material are the terms for the four seasons, which are virtu are the terms for the four seasons, whi Among theAmong surprises by offered the comparative material material ur seasons. theoffered surprises by the comparative Proto-Mongolic (*): ‫
צپ‬Proto-Mongolic heu.úr =(*): *kabur ‘spring’, ‫‘
ڕؘ‬ ch are four virtually identical incounterparts, Khitan (†) in and ‫
צپ‬heu.úr †haur = *kabur ‫
צپ‬ heu.úr †haur which are which virtually identical Khitan (†) and rseasons, the seasons, are virtually identical in Khitan (†)(*): and†haur Proto-Mongolic show aProto-Mongolic somewhat greater degree of phono­ ‘summer’, ‫צؽٿ‬ n.am.úr †namur = *namur ‘autumn’, ؈۲
= *kabur ‘spring’, ‫
ڕؘ‬ ju .un †jun = *jun ‘summer’, ‫צؽٿ‬ n.am.úr = *namur 3 * ‘summer’, ‫צؽٿ‬ n.am.úr †namur ‘autu = u. *n heu.úr †haur = *kabur ‘spring’, ‫
ڕؘ‬ ju3.un †jun *jun†jun (*): ‫
צپ‬ heu.úr †haur = *kabur ‘spring’, ‫
ڕؘ‬ = *jun 3.un logical innovation. This would seem toju= confirm that Khitan was†namur generally *ebül) ‘winter’. Since the possibility of a wholesale borrowing is a amur ‘autumn’, ؈۲
u.ul †uul = *öbül (< *ebül) ‘winter’. Since the possibility of a wholesale *ebül) ‘winter’. Since the possibility of †namur = more *namur ‘autumn’, ؈۲
u.ul †uul = *öbül (< ‫צ‬ n.am.úr †namur = *namur ‘autumn’, ؈۲
u.ul †uul = *öbül (< innovative, or also more “worn,” than Proto-Mongolic. it unlikely, must be ais cognates. Moreover, theMoreover, Khitan item wholesale isaextremely itquestion must beofitaunlikely, question cognates.of th must be aof question cognates. M ossibility ofborrowing a wholesale borrowing is extremely unlikely, Since the possibility of wholesale borrowing extremely The terms for the four seasons also provide an important clue to identical with their Proto-Mongolic counterparts, show a somewh oreover, theMoreover, Khitan items, when the not fully identical Proto-Mongolic counterparts, sh identical with their Proto-Mongolic coun cognates. the Khitan items, when not when fullywith uestion of cognates. Moreover, Khitan items, nottheir fully understanding the innovations by Proto-Mongolic and Para-seem phonological innovation. This would towould confirm erparts, show a somewhat degree of shared phonological innovation. This toKhitan confirm innovation. Thisthat would see Mongolic counterparts, showgreater a somewhat degree of phonological eir Proto-Mongolic counterparts, show a greater somewhat greater degree of seem Mongolic. The medial nasal *m in *namur ‘autumn’ must represent the innovative, or also more “worn”, than Proto-Mongolic. m to confirm that Khitan was generally more innovative, or also more “worn”, than Proto-Mongol Pr innovative, or also more “worn”, than Pro his wouldThis seemwould to confirm thatconfirm Khitanthat wasKhitan generally ovation. seem to wasmore generally more The terms for the four seasons also provide an important cla same segment as the medial obstruent *b in *kabur ‘spring.’ In other words, to-Mongolic. The terms for the four seasons also provide The terms for the four seasons als orn”, than Proto-Mongolic. so more “worn”, than Proto-Mongolic. the innovations shared by Proto-Mongolic and Para-Mongolic. T our provide an important clue to understanding it is a question of a suffix, possibly reconstructable as Pre-Proto-Mongolic the shared by Proto-Mongolic and Para Proto-Mongo the innovations shared by Proto-Mongol seasons alsoseasons providealso an provide important to understanding s for the four anclue important clueinnovations to understanding in *namur ‘autumn’ must represent the same segment as the med me c and Para-Mongolic. The medial nasal *m *‑pUr, which is and represented *-mUrin after a syllable with an nasal onset, *namur ‘autumn’ represent same segm innasal *namur ‘autumn’ mustthe represent the Proto-Mongolic and Para-Mongolic. Theasmedial nasal *m shared by Proto-Mongolic Para-Mongolic. The medial *mmust other question ame segment asasthe*-bUr medial obstruent *b in *kabur ‘spring’. Init‘spring’. other words, itofis awords, a suffix, question a *kabur In ‘winter’ other itpossibl isof a qu and elsewhere. Possibly, theInelement *(-)bül epresent the same segment as the*kabur medial obstruent *b in words, mn’ must represent the same segment as‘spring’. the medial obstruent *b inisina *öbül Pre-Proto-Mongolic *-pUr, which is represented as *-mUr after estion possibly reconstructable as Pre-Proto-Mongolic *-pUr, which is represented as re Pre-Proto-Mongolic *-pUr, which is rep ords, itofisawords, asuffix, question of a suffix, possibly reconstructable as In other it is a question of a suffix, possibly reconstructable as is also relevant in this context. In any case, the change *p > *m, which nasal onset, and as *-bUr elsewhere. Possibly, the element *(-)bül resented as *-mUr after a syllable with an nasal onset, and as *-bUr elsewhere. Possibly, the el nasal onset, and as *-bUr elsewhere. Pos which is represented as *-mUr after a syllable with an olic *-pUr, may which represented *-mUr afterdistant a syllable with an is likely to have be is seen as a case as of progressive assimilation, also relevant in this context. any case, *pIn >any *m, whi ibly, the element *(-)bül in *öbül ‘winter’ is also inInthis context. In change any case, the change also relevant in the this context. case, sewhere. Possibly, the element *(-)bül inproto­ *öbül ‘winter’ is as *-bUr elsewhere. Possibly, element *(-)bül in relevant *öbül is taken place in thethe common language of ‘winter’ Proto-Mongolic and Paracase of progressive distant assimilation, is likely to have taken pl he change *p > *m, which may be seen as a ofasprogressive distant assimilation, is likely to 29 distant assimilation, .his In context. any case, *p >change *m, which becase seen abecase Inthe anychange case, the *p > may *m, which may seenofasprogressive a 29 Mongolic. protolanguage of Proto-Mongolic and Para-Mongolic. sive likely to have taken place in the common of Proto-Mongolic and Para-Mongolic of Proto-Mongolic and Pa assimilation, is likely to ishave taken placetaken in theprotolanguage common distant assimilation, likely to have place in theprotolanguage common 29 29 a-Mongolic. and Para-Mongolic. fngolic Proto-Mongolic and Para-Mongolic.29 Kinship terms. In view Kinship of the numerals terms for the and seasons terms.Kinship Inand viewthe ofKinship theIn numerals the terms for the terms. view of four the numerals the four termas terms. In view of theand numerals it is curious that there is no overall correspondence in the realm of kinship that there overall the realm of tei nd the terms for the seasons itfour is curious that there itcorrespondence isthat overallis correspondence in kinship the realm there noinoverall correspondence he numerals thefour terms for seasons itno isseasons curious n view of theand numerals and thethe terms for the is four isnocurious ‘father,’ ia terminology. Several Khitan kinship terms, including  ‘elder brothe Khitan kinship terms, ‫
׿‬ ai ‘father’, ‫
ۓ‬ ia nrespondence the realm of kinship terminology. Several Khitan kinship terms, including ‫
׿‬ ai ‘father’, ‫
ۓ‬ ia Khitan kinship terms, including ‫
׿‬ ai ‘fa in the realm terminology. Several Several overall correspondence in of thekinship realm of kinship terminology. sister’, have no obvious cognates onnocognates the I ‘elder and sister,’ nosister’, obvious cognates onMongolic the on ‘elder brother’, and‫
ۓ‬ ‫
ئ‬ ‘elder her’, ‫
׿‬ ‫
ۓ‬ ia‘father’, have no obvious theside. Mon have obvious cognates on brother’, and ‫
ئ‬sister’, au have ‘elder ding ai ‫‘
ۓ‬father’, ia ‘elder  ‘elder brother’, and ‫
ئ‬ au ‘elder erms, including ‫
׿‬brother,’ ai iaau correspondences are uncertain: ֚
mo ‘woman, mother, wife’ (pl the Mongolic side. In other cases the Mongolic side. In other cases the correspondences are uncertain: mo ‘woman, are ֚
uncertain: ֚
momo‘w the Mongolic side. In other cases the correspondences o cognates obvious on cognates on the Mongolic side. correspondences In other cases are theuncertain: ‫
ى‬ bu and ‘grandfather’ (in ‫
ى‬ ‫
׿‬  (in ai id.) have compared w oman, mother, wife’ (plural ֚‫ٻ‬ mo.t) ‫
ى‬ bu ֚‫ٻ‬ ‘grandfather’ ‫
׿
ى‬ bu been ai‫
ى‬id.) have wife’ (plural and bu ‘grandfather’ (in bubu ‫
ى‬ ‘grandfather’ (in ‫
׿‬ bu aibeen id. ain: mo ‘woman, ‘woman, mother, wife’ (plural ֚‫(ٻ‬plural mo.t) and are ֚
uncertain: ֚
momother, ‘woman, mother, wife’ mo.t) and ‘woman’ resp. *ebü-gen ‘old man’, but the issue cannot be have compared with Mongolic *eme ‘woman’ resp. *ebü-gen but the issue bc ‘woman’ resp. ‘old *ebü-gen man’, bu ‫
׿‬ bu ai have been compared with Mongolic *eme bu have been with Mongolic *eme ‘woman’ resp.man’, *ebü-‘old her’ (inbeen ‫
ى‬id.) ‫
׿‬ buaiaiid.) id.) have beencompared compared with Mongolic *eme 30 told theman’, issue‘old cannot bebut considered fully but the man,’ issue cannot beascannot considered as fully as *ebü-gen man’, thethe issue fully concluded. The gen ‘old but issue cannotbebeconsidered considered as fully 30 30 3030 comparison between b.qo ‘child’ (plural b.hu.án) and Mongolic 30 ded. d. The The comparison comparison between between ‫
حڻ‬ ‫
حڻ‬ b.qo b.qo ‘child’ ‘child’ (plural (plural b.hu.án) b.hu.án) and and 30 ‫
ڦ؜ڻ‬ concluded. concluded. The The comparison comparison between between ‫
حڻ‬ ‫
حڻ‬ b.qo b.qo ‘child’ ‘child’ (plural (plural ‫
ڦ؜ڻ‬ ‫
ڦ؜ڻ‬ b.hu.án b.hu.á concluded. The comparison between ‫
حڻ‬ b.qo‫
حڻ‬ ‘child’ (plural ‫ڻ‬ concluded. The‫
ڦ؜ڻ‬ comparison between b.qo ‘child’ 28 28 28 28 Using the method of internal reconstruction, the *r in 3 *gurmay*rbeincompare compar olic c *baga *baga ‘small’ ‘small’ > ‘child’ > ‘child’ is is also also problematic. problematic. *baga ‘small’ > ‘child’ is also problematic. Using the method of internal reconstruction, the 3 *gurUsing the method of internal reconstruction, th Mongolic Mongolic *baga *baga ‘small’ ‘small’ > > ‘child’ ‘child’ is is also also problematic. problematic. ld’ (plural ‫
ڦ؜ڻ‬ b.hu.án) and Mongolic *baga ‘small’ > ‘small’ ‘child’ is also problematic. ‫
ح‬b.qo ‘child’ (plural ‫
ڦ؜ڻ‬b.hu.án) and30 Mongolic >the ‘child’ is alsoofproblematic. Mongolic *guc-i/n go.er >‘household’ ‘household’ =‘tent’ *ger. *ger. Mongolic, as in the =case of‘tent’ ‫ۙۃ‬ ‘tent’ > =‘household’ = monosyllabic *ger. however, also m gnates in‘household’ Mongolic, as in the case of ‫ۙۃ‬ go.er > many ‘household’ =Khitan *ger.33has, Khitan has,monosy howev ems stems((C)V), ((C)V),a type a type absent absent in inProto-Mongolic Proto-Mongolic (except (except pronouns).In Inthethe vowel vowel stems stems ((C)V), ((C)V),aatype type absent absentin inininpronouns). Proto-Mongolic Proto-Mongolic (except (exceptininpronouns). pronouns).I

concluded.30 The comparison be Mongolic *baga ‘small’ ed. ed. concluded. The The comparison comparison between between ‫
حڻ‬ ‫
حڻ‬ b.qo b.qo ‘child’ ‘child’ (plural (plural ‫
ڦ؜ڻ‬ ‫
ڦ؜ڻ‬ b.hu.án) b.hu.án) and and The comparison between ‫
حڻ‬b.qo ‘child’ (plural ‫
ڦ؜ڻ‬ b.hu.án) and > ‘child There are, however, also ic c *baga *baga ‘small’ ‘small’ > > ‘child’ ‘child’ is is also also problematic. problematic. Mongolic *baga ‘small’ > ‘child’ is also problematic. Khitan: Understanding the Language Behind the Scripts 121 ָ‫
ڦ‬ na.ha †naha (plural ָ‫ۙڦ‬ There There are, are, however, however, also good good and and non-trivial correspondences. These These include include There are,also however, alsonon-trivial good and correspondences. non-trivial correspondences. These include 30 k.iú ~ ۘ‫
ڥ‬ x.iú †keu ed. Thena.ha comparison between ‫
حڻ‬b.qo ‘child’ (plural b.hu.án) and id., ‫‘
ڥڟ‬younger s .ha ha ָ‫
ڦ‬ †naha †naha (plural (plural ָ‫
ؑۙڦ‬ ָ‫
ؑۙڦ‬ na.ha.án.er) na.ha.án.er) ‘maternal ‘maternal uncle’ uncle’ =‫
ڦ؜ڻ‬ = *naga-cu *naga-cu id., ‫
ڥڟ‬ ‫
ڥڟ‬ †naha (plural ָ‫
ؑۙڦ‬ na.ha.án.er) ‘maternal uncle’ =id., *naga-cu (genitive ‫
ڕת‬ ic *baga ‘small’ > ‘child’ is also problematic. ۘ‫
ڥ‬ ۘ‫
ڥ‬ x.iú x.iú~†keu †keu ‘younger sister’ == *keü *keü ‘child’, ‘child’, and and‘child’, very very probably probably ‫
ת‬ deu deu †deu †deu k.iú ۘ‫‘
ڥ‬younger x.iú †keusister’ ‘younger sister’ = *keü and very‫
ת‬ probably ‫
ת‬ deudeu.un) †deu ‘younger 31 31 31 to haveseem *de(x)ü id.31good In two items, Khitan would seem a include monosyllabic root toKhitan have a would monosyllabic roo are, however, also and= These In In two two items, items, Khitan Khitan would would eThere e ‫
ڕת‬ ‫
ڕת‬ deu.un) deu.un) ‘younger ‘younger brother’ brother’ =non-trivial *de(x)ü *de(x)ü id. id. In two items, (genitive ‫
ڕת‬ deu.un) ‘younger brother’ =correspondences. *de(x)ü id. (CV) where Mongolic a bisyllable, aapparently derivative: een ‫
حڻ‬ b.qo ‘child’ (plural ‫
ڦ؜ڻ‬ b.hu.án) and .ha †naha (plural na.ha.án.er) ‘maternal uncle’ = *naga-cu id., ‫
ڥڟ‬ have have aa monosyllabic monosyllabic root root (CV) (CV) where where Mongolic Mongolic has has aaapparently bisyllable, bisyllable, aapparently a ‫
ط‬ku †kua ‘person’ (g seem to have aָ‫
ؑۙڦ‬ monosyllabic root has (CV) where Mongolic has a apparently bisyllable, 32 32 32 32 = ‘spouse’ *nö-kür id., ‘friend’. ku.û.un) = *kü-xün id.,*kü-xün = ‘person’ also problematic. ۘ‫
ڥ‬ x.iú ‘younger sister’ =‫ط‬۸‫
ڕ‬ *keü ‘child’, and very probably ‫‘
ת‬spouse’ deu ve: ve: ‫
ط‬ ‫
ط‬ ku ku†keu †ku †ku ‘person’ ‘person’ (genitive (genitive ‫ط‬۸‫
ڕ‬ ku.û.un) ku.û.un) == *kü-xün *kü-xün id., id., ‫
ؿٿ‬ ‫
ؿٿ‬ n.o n.o †no †no derivative: ‫
ط‬ku †ku ‘person’ (genitive ‫ط‬۸‫
ڕ‬ ku.û.un) = id.,†deu ‫
ؿٿ‬ n.o †no 31 both branches have od and non-trivial correspondences. These include In†naiji two items, Khitan would deu.un) ‘younger *de(x)ü id.‫
ا׿ٿ‬ ’e==‫
ڕת‬ *nö-kür *nö-kür id., id., ‘friend’. ‘friend’. In Inbrother’ the the item item=In ‫
ا׿ٿ‬ ‫
ا׿ٿ‬ n.ai.ci n.ai.ci †naiji ‘friend’ ‘friend’ == *nai-ji *nai-ji id. id. id., ‘friend.’ the n.ai.ci †naiji ‘friend’ = ‘spouse’ = *nö‑kür *nö-kür id., ‘friend’. In the item n.ai.ci †naiji ‘friend’ = *nai-ji id.a derivative; and it is also attested na.ha.án.er) = where *naga-cu id.,*nai ‫
ڥڟ‬ have ahave monosyllabic root (CV) Mongolic has apreserved bisyllable, apparently a Mongolic, anches nches have aa‘maternal derivative; derivative; the the root root *nai *nai ‘friendship’ ‘friendship’ is is preserved in Mongolic, Mongolic, both branches have auncle’ derivative; root ‘friendship’ isin preserved both branches have a the derivative; the root *nai ‘friendship’ is in preserved in in other der 32 er’ = *keü ‘child’, and very probably ‫
ת‬ deu‘concord, †deu ve: ‫
ط‬ ku †ku ‘person’ (genitive ‫ط‬۸‫
ڕ‬ ku.û.un) = other *kü-xün id., ‫
ؿٿ‬ n.o*nai-r †no ‘concord, also also attested attested in in other other derivatives derivatives like like *nai-r ‘concord, festivity’. festivity’. and it is also attested in other derivatives like *nai-r ‘concord, festivity’. Mongolic, and it is also*nai-r attested in derivatives like structure. Khitan seems to two Khitan n.ai.ci would †naiji ‘friend’ = Root = *de(x)ü id.31 In In ’other’ = *nö-kür id.,festivity.’ ‘friend’. theitems, item ‫
ا׿ٿ‬ *nai-ji id. CV) where Mongolic has athe bisyllable, apparently a by anches have a derivative; root *nai ‘friendship’ is the preserved in Mongolic, ucture. ucture. Khitan Khitan seems seems to to have have been been characterized characterized by the same same type of of basic basic Root structure. Khitan seems to have been characterized bytype thephonotactic same type restrictions of basic as the o 32 Proto-Mongolic. itive ‫ط‬۸‫
ڕ‬ *kü-xün id., ‫
ؿٿ‬ n.o †noincluding also attestedku.û.un) in restrictions other like *nai-r ‘concord, festivity’. ctic ctic restrictions restrictions as as the the=derivatives other other “Altaic” “Altaic” languages, languages, including both both Jurchen Jurchen and andJurchen This phonotactic the other “Altaic” including both and means, am Root structure.asKhitan seems to havelanguages, been characterized by the same type syllable-final consonant the item ‫
ا׿ٿ‬ n.ai.ci †naiji ‘friend’ = *nai-ji id. Mongolic. ongolic. This This means, means, among among other other things, things, that that there there were were no no syllable-initial syllable-initial or or Proto-Mongolic. This means, among other things, that there were no syllable-initial or clusters of basic phonotactic restrictions as the other “Altaic” but languages, including there eucture. root *nai ‘friendship’ preserved initems Mongolic, Khitan seems toisMost have been characterized by theare same type of((C)V(C)CV(C)), basicare also conspicuously -final final consonant consonant clusters. clusters. Most lexical lexical items are are bisyllabic bisyllabic ((C)V(C)CV(C)), ((C)V(C)CV(C)), syllable-final consonant clusters. Most lexical items bisyllabic both Jurchen and Proto-Mongolic. This the means, among other things,((C)VC) that can hav in latter, aending consonant atives like *nai-r ‘concord, festivity’. restrictions as the other “Altaic” languages, including both Jurchen and eectic are are also also conspicuously conspicuously many many monosyllables. monosyllables. Among Among the latter, latter, those those ending but there are also conspicuously many monosyllables. Among the those ending there were no syllable-initial or syllable-final consonant clusters. Most go.er ‘tent’ = *g Mongolic. This means, among other things, that there were or sonant sonant ((C)VC) can can have have direct direct cognates cognates in Mongolic, Mongolic, asno in in syllable-initial the the case case of of the ‫ۙۃ‬ ‫ۙۃ‬ in a ((C)VC) consonant ((C)VC) can have directin cognates in as Mongolic, as in case>of‘household’ ‫ۙۃ‬ 33 33 bisyllabic 33 ((C)V(C)CV(C)), but there are lexical are also stems conspicuously vowel ((C)V), a type abs ave been characterized byMost the type of basic -final lexical items arehas, bisyllabic ((C)V(C)CV(C)), Khitan Khitan has, has, however, however, also also many many monosyllabic monosyllabic ent’ ent’ >>consonant ‘household’ ‘household’ ==items *ger. *ger. Khitan however, also many monosyllabic go.er ‘tent’ >clusters. ‘household’ =same *ger. monosyllables. Among theAmong latter, those ending a consonant comparative r are “Altaic” languages, including both Jurchen and etems also conspicuously many those ending tems ((C)V), ((C)V), amany a type type absent absent in in monosyllables. Proto-Mongolic Proto-Mongolic (except (exceptthe in inlatter, pronouns). pronouns). In In the the context vowel stems ((C)V), a type absent in Proto-Mongolic (except inin pronouns). In ((C) the it seems th archaic trait eliminated VC) can have direct cognates in Mongolic, as in the case of go.er ‘tent’ gtive other things, that there were no syllable-initial or sonant ((C)VC) can have direct cognates in Mongolic, as in the case of ‫ۙۃ‬ ative context context it it seems seems that that the the monosyllabic monosyllabic vowel vowel stems stems in in Khitan Khitan are are an an comparative context it seems that the monosyllabic vowel stems in Khitan are an on the M 33 33 the above-mentioned Most lexical items are bisyllabic ((C)V(C)CV(C)), ent’ > ‘household’ = *ger. Khitan has, however, also monosyllabic trait traitarchaic eliminated eliminated on the the Mongolic Mongolic side side by byKhitan way way of ofhas, secondary secondary suffixation, as asmonosyllabic in in >on ‘household’ however, also many trait eliminated on= the*ger. Mongolic side by way many of suffixation, secondary suffixation, as in items †ku ‘ Importantly, any monosyllables. Among the latter, those ending tems ((C)V), a type absent in†ku Proto-Mongolic (except in pronouns). In the ve-mentioned e-mentioned items items †ku †ku ‘person’, ‘person’, †no †no ‘spouse’. ‘spouse’. the above-mentioned items †no ‘spouse’. vowel stems ((C)V), a‘person’, type absent in Proto-Mongolic (except in pronouns).the method stems to be established irect cognates Mongolic, as inmonosyllabic the case of ‫ۙۃ‬ ative context itIninmethod seems that the vowel stems inallows Khitan are an mportantly, mportantly, the the method of of internal internal reconstruction reconstruction allows allows monosyllabic monosyllabic vowel vowel Importantly, the method of internal reconstruction monosyllabic vowel the comparative context it seems that the monosyllabic vowel stems also for 33 collective by *-d ( however, eliminated on the by way ofstage secondary suffixation, innouns otrait be beKhitan established established also also for forMongolic aaalso very very early early stage stage of ofeliminated the the Mongolic Mongolic lineage, lineage, as as lineage, in inas the the stems tohas, beinestablished also forside a monosyllabic very early ofon thethe Mongolic informed the Khitan are anmany archaic trait Mongolic side byasway of these isof so farone identified in the te in Proto-Mongolic (except in*-d pronouns). In the ve-mentioned items †ku ‘person’, †no ‘spouse’. ve nouns nouns formed formed by by *-d *-d (for (for countables) countables) and and *-s *-s (for (for liquifiables). liquifiables). Only Only one one of collective nouns formed by (for countables) and *-s (for liquifiables). Only of secondary suffixation, as in the above-mentioned items †ku ‘person,’ †no ‘blood’ > ‘blood relative’ = *ci-s the monosyllabic vowel stems in Khitan are an mportantly, thefar method internal reconstruction allows monosyllabic vowel so sothese far far identified identified in in the theofKhitan Khitan Small Small Script Script database: database: ‫
֦ۗا‬ ‫
֦ۗا‬ ci.i.is ci.i.is †cis †cis is so identified in the Khitan Small Script database: ‫
֦ۗا‬ ci.i.is †cis ‘spouse.’ Asroot may be seen, this item show side way secondary suffixation, ongolic established also for a very early stage of as thein Mongolic lineage, as in*ci. the >>be ‘blood ‘blood relative’ relative’ == of *ci-s-ü/n *ci-s-ü/n ‘blood’, ‘blood’, presupposing presupposing the the monosyllabic monosyllabic root *ci. ‘blood’ >by‘blood relative’ = *ci-s-ü/n ‘blood’, presupposing the monosyllabic root *ci. Importantly, the method of internal reconstructionthat allows monosyllabic this group of nouns was fo son’, †no ‘spouse’. ve formed byshows *-d countables) *-s*-s (foralso liquifiables). one of suggesting be benouns seen, seen, this this item shows the the collective collective suffix suffix *-s also in in *-s Khitan, Khitan, suggesting suggesting As may beitem seen, this(for item shows theand collective suffix alsoOnly in Khitan, vowel stems to be established also for a very early stage of the Mongolic Proto-Mongolic and reconstruction allows monosyllabic vowel so far this identified in Khitan Small database: ‫
֦ۗا‬ci.i.is sinternal group group of of nouns nouns was formed formed already already in inScript the the common common protolanguage of of †cis the the that groupwas of the nouns was formed already in theprotolanguage common protolanguage of thePara-Mongo 34 34 34formed by *-d (for countables) and *-s lineage, as in the collective nouns very early stage of=the Mongolic lineage, as in the the monosyllabic root *ci. > ‘blood relative’ *ci-s-ü/n ‘blood’, presupposing Mongolic ongolic and and Para-Mongolic Para-Mongolic lineages. lineages. Proto-Mongolic and Para-Mongolic lineages. (for liquifiables). Only one of one these so far the Khitanphonology Small 5.inDiachronic countables) *-s (for liquifiables). Only ofisalso be seen, thisand item shows the collective suffix *-s in identified Khitan, suggesting Script database: ‘blood’protolanguage > ‘blood relative’ Khitan Small Script database: ci.i.is †cis sronic group of nouns was formed ‫
֦ۗا‬ already ci.i.is in the†cis common of the= *ci-s-ü/n hronic phonology phonology 5. Diachronic phonology 34 No be systematic n ‘blood’, presupposing the lineages. monosyllabic root *ci. root *ci. As may Mongolic and Para-Mongolic ‘blood,’ presupposing the monosyllabic seen, thispicture item of Khitan instance, reconstruct he No collective suffix *-s also in Khitan, suggesting ematic ematic picture picture of of Khitan Khitan phonology phonology is is yet yet possible possible to to draw. draw. We We cannot, cannot, for for systematic picture of Khitan phonology is yet possible to draw. We cannot, forthe exact co This is to the nature of the med already in the common protolanguage of the hronic phonology ,, reconstruct reconstruct the the exact exact configuration configuration of of the the systems systems of of vowels vowels and and consonants. consonants. instance, reconstruct the exact configuration of the systems of vowels anddue consonants. 31 34 The only problem in the item for ‘younger brother’ is that the Romanization of the lineages. due due to to the the nature of the script, script, which involves involves multiple cases cases of of both both underunderThis is nature due to of thethe nature of which the script, whichmultiple involves multiple cases of both undercharacter is not verified by independent information. The genitive form confirms, 30 ematic picturehowever, of Khitan is syllable yet possible to†u, draw. We cannot, for Wuthe Yingzhe (2009) argues, certainl that phonology the underlying ended in suggesting that Romanization is “hidden” initial vowel. If the Khitan 30 , reconstruct the exact configuration of the systems of vowels and consonants. correct. ngzhe ngzhe Wu (2009) (2009) argues, argues, certainly certainly correctly, correctly, that thatcorrectly, in in some some cases casesinaasome Khitan Khitan character character can cancharacter imply imply aa can imply a Yingzhe (2009) argues, certainly that cases a Khitan cognateship with Mongolic would be in 32the due to the nature of the script, which involves multiple cases of both underinitial initial vowel. vowel. If If the Khitan Khitan phonemic phonemic shapes shapes were were something something like like †eme †eme resp. resp. †ebu †ebu the the “hidden” initialMongolic vowel. If the Khitan shapes(as were something like †eme resp.suggesting †ebu the *küxün has thephonemic variant *kümün in Written Mongol Oirat), 31 and The only problem in the item for ‘y ip ip with with Mongolic Mongolic would would beoriginal in in no no doubt. doubt. cognateship with Mongolic would be inWe no doubt. honology is yet possible to draw. cannot, (with for early dialectal nasalization of the medial that thebe shape was *kü-pün verified by independent information. T 31 ly y problem problem in in the the item item for for ‘younger ‘younger brother’ brother’ is is that that the the Romanization Romanization of of the the character character is is not not The only problem in the for ‘younger brother’ is that the Romanization of the character is not consonant). iguration of argues, the systems ofitem vowels and consonants. (2009) certainly correctly, that confirms, in some cases a Khitan character can syllable imply ended in †u,a suggesting that the Roman yngzhe y independent independent information. The The genitive genitive form form confirms, however, however, that that the the underlying underlying syllable 33 verified byinformation. independent information. The genitive form confirms, however, that the underlying syllable 32 TheKhitan exact “reading” of theboth Khitan word is difficult to reconstruct. ript, which multiple cases of underinitial vowel.involves If the phonemic shapes were something like †eme resp. †ebuThe the comparative Mongolic *küxün has the variant * †u, u, suggesting suggesting that that the the Romanization Romanization isis correct. correct. ended in †u, suggesting that the Romanization is correct. 30 30

30

evidence suggest the shape †ger, but the orthographical image is exceptional ip with Mongolic would bewould in no doubt. original shape *kü-pün (with early 32 lic ic *küxün *küxün has has the the variant variant *kümün (as (as in in Written Written Mongol and andespecially Oirat), Oirat), suggesting suggesting that that the the was Mongolic *küxün has*kümün the variant *kümün (asMongol in Written Mongol and Oirat), suggesting that the 33as the and might imply something more complex, as character far vowel quality is the Khitan w ly problem in the item for ‘younger brother’ is that the Romanization of the is not The exact “reading” of hape hape was was that *kü-pün *kü-pün (with (withcases early earlyadialectal dialectal nasalization nasalization of the the medial medial consonant). consonant). original shape was *kü-pün (with early dialectal nasalization of the medial consonant). orrectly, ininformation. some Khitan character canof imply a concerned. yctindependent The genitive form confirms, however, that the underlying syllable suggest the shape †ger, 33 act “reading” “reading” of of the the Khitan Khitan word word is difficult difficult to to reconstruct. reconstruct. The comparative comparative evidence would would Theshapes exact “reading” of the is Khitan word is difficult toThe reconstruct. The evidence comparative evidence wouldbut the orthog onemic were something like †eme resp. †ebu the †u, suggesting that the Romanization is correct. complex, especially as far as the vowel he edoubt. shape shape †ger, †ger, but the the †ger, orthographical orthographical image image isis exceptional exceptional and might might imply imply something more more suggest thebut shape but the orthographical image isand exceptional and something might imply something more 34 lic *küxün has the variant *kümün (as in Written Mongol and Oirat), suggesting that the On the Pre-Proto-Mongolic collect especially especially as asis far farthat as as the the vowel quality isisof concerned. concerned. complex, especially as farquality as the vowel quality is concerned. ger brother’ thevowel Romanization the character is not hape Mongolic side, these nouns are always 34was *kü-pün (with early dialectal nasalization of the medial consonant). Pre-Proto-Mongolic Pre-Proto-Mongolic collectives collectives in in *-d *-d resp. resp. *-s, *-s, cf. cf. Janhunen Janhunen (1996: (1996: 210-215). 210-215). On On the the On the Pre-Proto-Mongolic collectives in *-d resp. *-s, cf. Janhunen (1996: 210-215). On the genitive form confirms, however, that the underlying syllableThe comparative evidence act “reading” of theare Khitan word is difficult to element reconstruct. for be ‘blood’ alsothe theitem base for the Khita side, side, these nouns nouns always always expanded expanded by by the the element *-U/n. ItIt may may*-U/n. be be noted noted that that the thewould item itemisthat Mongolic side, are these nouns are always expanded by*-U/n. the element It may noted ion is these correct. he shape †ger, but the orthographical image is exceptional and(2004), might imply something more ’’ isis for also also‘blood’ the the base base for for the the Khitan Khitan term term for for ‘filial ‘filial piety’, piety’, see see Kane Kane (2004), Shimunek Shimunek (2007: (2007: 71). 71). is also the base for the Khitan term for ‘filial piety’, see Kane (2004), Shimunek (2007: 71).

122



SCRIPTA, VOLUME 4 (2012)

shows the collective suffix *-s also in Khitan, suggesting that this group of nouns was formed already in the common protolanguage of the ProtoMongolic and Para-Mongolic lineages.34

5. Diachronic phonology No systematic picture of Khitan phonology is yet possible to draw. We cannot, for instance, reconstruct the exact configuration of the systems of vowels and consonants. This is due to the nature of the script, which involves multiple cases of both under- and over-differentiation, as well as a considerable amount of orthographical variation. We also have to reckon with the possibility that Khitan changed in the course of time: the currently available corpus of Khitan Small Script texts covers a period of over a hundred years (c. 1041-1171),35 while some of the orthographical con­ventions are likely to date from the time when the script was created (924-925). In general, work on the Khitan phonological system is closely connected with the reconstruction of Liao Chinese, a little known form of speech, transitional between local Late Middle Chinese and Early Mandarin.36 On the other hand, Khitan phonology is likely to share many features with Jurchen on an areal basis. In the following some general points of relevance in the diachronic context are discussed in comparison with the Proto-Mongolic lineage: Vowel Rotation. As a working hypothesis it is reasonable to assume that Khitan originally had a vowel system similar to Pre-Proto-Mongolic. This system would have comprised 8 vowels, divided into 4 back (*a *o *u *ï) and 4 front (*e *ö *ü *i), as well as 4 lower (*a *e *o *ö) and 4 higher (*ï *i *u *ü) vowels. A similar system, with minor modifications, may be On the Pre-Proto-Mongolic collectives in *-d resp. *-s, cf. Janhunen (1996: 210-215). On the Mongolic side, these nouns are always expanded by the element *-U/n. It may be noted that the item for ‘blood’ is also the base for the Khitan term for ‘filial piety,’ see Kane (2004), Shimunek (2007: 71). 34

These datings are based on the most recent summary of Khitan Small Script epitaphic documents by Wu Yingzhe (2012). 35

On Liao Chinese, cf. Kane (2009: 227-264). Tentative outlines on Khitan phonology are given by Shimunek (2007: 38-52) and Takeuchi (2007). 36

chen on an areal basis. many features with Jurchen onfeatures an arealwith basis. many many features withJurchen Jurchenon onan anareal arealbasis. basis. g some general points of relevance inInthe are points of relevance in the diachron thediachronic followingcontext some general In Inthe thefollowing followingsome somegeneral generalpoints pointsof of n with the Proto-Mongolic lineage: discussed in comparisondiscussed with the Proto-Mongolic lineage: discussedinincomparison comparisonwith withthe theProto-Mongol Proto-Mongo Khitan: Understanding the Language Behind the Scripts 123

working hypothesis it is reasonable to assume Khitanhypothesis it is reasonable to assum Vowel Rotation. As athat working Vowel Vowel Rotation. Rotation. As As aa working working hypothesis hypothesis itit system similar to Pre-Proto-Mongolic. This system would have originally had a vowel system similar toaavowel Pre-Proto-Mongolic. This syste originally originally had had vowel system system similar similar totoPre-Pro Pre-Pr reconstructed for Proto-Tungusic, even Proto-Koreanic. ivided into 4 back (*a *o *ucomprised *ï)Proto-Turkic, and 4 8front (*e *ö *ü *i), as and vowels, divided into 488back (*a divided *o *u *ï) and 4back front (*e comprised comprised vowels, vowels, divided into into 44back (*a (*a * *o *ö) and 4All higher *u *ü) system, over (*ï the *i region, thevowels. has, however, undergone a process of vowelvowels. A s well as 4system lower A (*asimilar *e *o *ö) 4 higher (*ï *i *u well well as asand 44 lower lower (*a (*a*e *e*o *o*ö) *ö)*ü) and and44 higher higher (* (* ons, may be reconstructed for Proto-Tungusic, rotation, in which the original front vowelsmay raised and velarized, withProto-Turkic, minor modifications, bebeen reconstructed for Proto-Turkic, Pr with withhave minor minor modifications, modifications, may may be be reconstruct reconstruc nic. ic. All over the region, the system has, however, undergone a while the original have been lowered and, in some languages, and back even vowels Proto-Koreanic. All over the region, the system has, however and andeven evenProto-Koreanic. Proto-Koreanic.All Allover overthe theregion, region, ion, in which pharyngealized. the original front have been and ofthe A vowels typical example ofraised theineffect vowel rotation the the process of vowel rotation, which original front vowels have be process process of of vowel vowel rotation, rotation, ininis which which the origin origin ginal back vowels have been lowered inthe some languages, development ofvelarized, the pair and, *u - *ü = phonetically [u y] to o u = [o u], as while original back vowels have been lowered and, in so velarized, velarized,while whilethe theoriginal originalback backvowels vowelshave have cal example of the effect of pharyngealized. vowel rotation isAthe development 37 typical example theAAeffect ofexample vowel rotation iseffec the observed in languages such as Manchu, Dagur, andofKorean. pharyngealized. pharyngealized. typical typical example of ofthe theeffec honetically [u - y] to o - u =of[othe - u], as *u observed in languages [u - y] to o - u = [o - u], as observed - *ü = phonetically the thepair pair*u *u--*ü *üphenomenon, ==phonetically phonetically [u [u--y] y]totoooNortheast Asia is of aofrelatively recent not 37 37 Vowel rotation in pair 37 r,, and Korean. such Manchu,inDagur, and such suchKorean. as asManchu, Manchu, Dagur, Dagur,and and Korean. Korean.3737 yet observable withas certainty Proto-Mongolic. It is, therefore, important in Northeast Asia is a relatively recent phenomenon, not yet in Northeast Asia is a relatively recent phenom Vowel rotation rotation ininthe Northeast Northeast Asia Asia isis aa r to note that seems Vowel to haverotation beentopresent in Vowel Khitan. Although vowel nty ty in Proto-Mongolic. It is,ittherefore, note that it observableimportant with certainty in Proto-Mongolic. It is, therefore, important observable observable with with certainty certainty in in Proto-Mongolic. Proto-Mongolic. ItI letters contained in the Romanizations should not be taken at face value, resent esent in Khitan. Although the vowel lettersbeen contained ininthe seems to have present Khitan. Although theinin vowel letters co seems seems to to have have been been present present Khitan. Khitan. Althou Althou there examples suggesting the developments *ü > †u and *u > †o, not be taken at faceare value, there are examples suggesting the at Romanizations should not be taken face value, are at examples Romanizations Romanizations should should not notthere be be taken taken at face faceval vas in ‫ ׀‬uni uni †uni †uni ‘ox’ = *üni-xe/n ‘cow,’ on†on‘to ride’ = *unuid. and *u > †o, as in ‘ox’ = *üni-xe/n ‘cow’, ‫
؟‬ ondevelopments *ü > †u and *u > †o, as*ü in >‫׀‬ uniand †uni = as *üni-xe/n developments developments *ü >†u †u and *u *u‘ox’ >>†o, †o, asinin‫׀‬ ‫׀‬un u d. The ultimateThe impact of vowel is=the reduction thereduction ultimate impact ofride’ vowel rotation is of the of of thevowel paradigm †on-rotation ‘to *unuid. The ultimate impact rotation is the re †on†on‘to ‘to ride’ ride’ = = *unu*unuid. id. The The ultimate ultimate impact impact o o 38 We dosystem nota know whether mple five-vowel systema (a e paradigm i o five-vowel u).38 towards simple (a paradigm eparadigm i five-vowel o u).38 towards We do not know whether towards simple system (a e ifive-vowel o u).38 We system do not ( towards aasimple simple five-vowel system iss stage, but there arehad indications at stage, least, abut neutralization Khitan reached this arehad indications of, stage, at least, Khitanof, had reached thisthere stage, but there are this indications of,athere at least, Khitan Khitan had reached reached this stage, but but there are areaini ‫
׀ٸ‬s.uni ‘night’ = *söni id. On the other hand, iss.uni also ‘night’ neutralization between between *ö *ö and and *ü, *ü, cf. cf.there = *söni id. On the ‫
׀ٸ‬ s.uni ‘night’ = *söni id. On the other between between*ö *öand and*ü, *ü,cf. cf.‫
׀ٸ‬ ‫
׀ٸ‬s.uni s.uni‘night’ ‘night’=hand =*sö *s neral neutralization of rounded vowels, cf. e.g. the numerals 3 evidence of a more general neutralization of rounded vowels, other hand, there 39 is also evidence ofevidence a more of general neutralization of cf. e.g. evidence ofaamore more general generalneutralization neutralization of ofro rt 39 39 ur- vs, 4 ‫ ٰٻ‬t.ur- †dur- = *dör-. ‫
ٰ؜‬hu.ur= *gurvs, 4hu.ur‫†ٰٻ‬hurt.ur= *dör-. rounded vowels, cf. e.g. the†hurnumerals 3‫
ٰ؜‬hu.ur†hur= *gurvs, ‫
ٰ؜‬hu.ur†hur-†dur==*gur*gurvs, vs,44‫ٰٻ‬ ‫ٰٻ‬t.urt.ur-†dur†dur t.ur- †dur- = *dör-.39

Preservation of *p. Another widespread areal feature in Northeast Asia

on n the the most most recent recent summary summary of of Khitan Khitan Small Small Script Script epitaphic epitaphic documents documents by 35 35 most by These datings based on35the recentlabial summary ofon(*p), Khitan Small Script epitaph concerns the original strongare(voiceless/aspirated) stop which These Thesedatings datings are arebased based on the themost most recent recent summary summary o Wu Yingzhe (2012). Wu Wu Yingzhe Yingzhe (2012). (2012). hasTentative been variously spirantized (> f), velarized (> x), pharyngealized ne ne (2009: (2009: 227-264). 227-264). Tentative outlines outlines on on Khitan Khitan phonology phonology are are given given by by 36 On Liao Chinese, cf. Kane3636(2009: 227-264). Tentative on KhitanTentativ phonolo On OnLiao Liao Chinese, Chinese, cf. cf.Kane Kaneoutlines (2009: (2009:227-264). 227-264). Tentativ dd Takeuchi Takeuchi (2007). (2007). (> h), or lostShimunek (> Ø) in almost and language families of (2007: 38-52)all andlanguages Takeuchi (2007). Shimunek Shimunek (2007: (2007: 38-52) 38-52) and and Takeuchi Takeuchi (2007). (2007). of of the the areal areal and and typological typological background background of of vowel vowel rotation rotation in in the the languages languages 37 For a general discussion of3737 the areal and typological vowel rotationb For For aageneral general discussion discussionbackground ofofthe theareal arealof and and typological typological unen nen (1981). (1981). This This model model of of vowel vowel rotation rotation has has recently recently been been challenged challenged by by of Northeast Asia, cf. Janhunen (1981). This model of vowel (1981). rotation has model recently of of Northeast Northeast Asia, Asia, cf. cf. Janhunen Janhunen (1981). This This model ofofbe vv claims claims that that the the diachronic diachronic process process took took place place in in the the opposite opposite direction, direction, that that Ko Seongyeon (2011), who claims that the diachronic process took place in the oppos Ko Ko Seongyeon Seongyeon (2011), (2011), who who claims claims that that the the diachronic diachronic prp pharyngeal pharyngeal (RTR (RTR == “retracted “retracted tongue tongue root”) root”) harmony harmony to to palatal palatal harmony. harmony. is, from an apertural and/or pharyngeal = “retracted tongue root”) harmony to is, is,from from an an(RTR apertural apertural and/or and/orpharyngeal pharyngeal (RTR (RTR ==“retracte “retrac inor nor relevance relevance to to37the the number of of oppositions oppositions in in the the the paradigm, paradigm, and and the the two two Fornumber a general discussionthe of typological background of vowel rotation in the paradi Although issueareal is ofand minor relevance to the number of oppositions Although Althoughin the issue issueisisofofminor minorrelevance relevancetotothe thenumber numbe alatal alatal and and apertural) apertural) coexist coexist in in Northeast Northeast Asia, Asia, the the arguments arguments presented presented intheThis in the languages of Northeast cf. Janhunen (1981). model ofinvowel rotation types of vowelAsia, harmony (palatal and apertural) coexist(palatal Northeast Asia, thecoexist argumi types types ofofvowel vowelharmony harmony (palatal and andapertural) apertural) coexist do do not not seem seem to to be behas tenable. tenable. recently beenfavour challenged by Ko Seongyeon (2011), who claims that the diachronic of the revised model do not seem torevised be tenable. favour favour the the revised model modeldo donot notseem seemtotobe betenable. tenable. darily darily increased increased by by the the introduction introduction ofinaa new new set set of of palatal palatal vowels, vowels, as asofof has has 38 of process took place the opposite direction, that is, from anintroduction apertural of and/or 3838 The system can be secondarily increased by the a newby set of introd palata The The system system can can be besecondarily secondarily increased increased bythe the intro l as as in in several several varieties varieties of of Mongolic. Mongolic. pharyngeal (RTRhappened = “retracted tongue root”) harmony to palatal harmony. Although the in Korean, as well as in several varieties of Mongolic. happened happened in in Korean, Korean, as as well well as as in in several several varieties varieties of of Mo Mo f course, course, to to consider consider also the possibility possibility of of to analogical analogical levelling levelling between in the paradigm, and the two 39 issue also is of the minor relevance the of 39 oppositions 39 between In numerals wenumber have, of course, to consider also the possibility of analogical In In numerals numerals we we have, have, of of course, course, to to consider consider also also thle th types of vowel harmony (palatal and apertural) coexist in Northeast Asia, the arguments consecutive items. consecutive consecutive items. items. presented in favour of the revised model do not seem to be tenable. The system can be secondarily increased by the introduction of a new set of palatal vowels, as has happened in Korean, as well as in several varieties of Mongolic.

38

In numerals we have, of course, to consider also the possibility of analogical levelling between consecutive items.

39

124



SCRIPTA, VOLUME 4 (2012)

feature Northeast Asia feature concerns the fspread *p. Another widespread arealwidespread feature in areal Northeast Asia concerns the ation ofareal *p. Another widespread areal infeature Northeast Asia concerns the Preservation of *p. in Another in Northeast Asia concerns the 40 ed) labial stop (*p), which has been variously the region, including Turkic, Mongolic, Tungusic, and even g original (voiceless/aspirated) labial labial stop (*p), which has been variously strong (voiceless/aspirated) stop (*p), has beenhasvariously strong (voiceless/aspirated) labial stopwhich (*p), which been Japonic. variously pharyngealized (>vowel h), x), orrotation lost Ø) f), velarized (> f), x), pharyngealized (> inh),almost lost (> Ø) all While toor proceeded from east toall west, the zed (> f), velarized (> pharyngealized (>have h),all or (>oralmost Ø) in(>almost spirantized (> velarized (> (> x), seems pharyngealized (>lost h),in lost Ø) in almost all flanguage the region, including Turkic, Mongolic, Tungusic, families of the region, including Turkic, Mongolic, Tungusic, eslanguages and language families of Turkic, Mongolic, Tungusic,direction, spirantization ofthe theregion, strong labial stop moved in theMongolic, opposite and language families of theincluding region, including Turkic, Tungusic, 40 40 40 seems to have proceeded from east to (Turkic) west, nic. While vowel rotation seems to have from east to east west, notation Japonic. While vowel rotation seems toproceeded have proceeded from to west, and even Japonic. While vowelin rotation seems to have proceeded from east to west, being most ancient the west and most recent in the east ial stop moved in the opposite direction, being most on of the strong labial stop moved in the opposite direction, being most antization of the strong labial stop moved in the opposite direction, being the spirantization of the labialThere stop moved thelanguage oppositefamily direction, being most (Tungusic andstrong Japonic). is only in one ofmost the “Altaic” ost recent the east (Tungusic and Japonic). There 41 west (Turkic) and most recent in the east (Tungusic There in the west (Turkic) and mostand recent inthis the east (Tungusic Japonic). There ancient inin the west (Turkic) most recent in the and eastJaponic). (Tungusic and Japonic). There sphere, Koreanic, in which phenomenon is and totally absent. fanguage the “Altaic” sphere, Koreanic, in which this family of the “Altaic” sphere, Koreanic, in which this language family of the “Altaic” sphere, Koreanic, in which this isoneonly one Importantly, language family of the “Altaic” sphere, Koreanic, in which this the41strong labial stop is also preserved in Khitan. The same 41 41 totally enon is absent. totally isabsent. phenomenon totally absent. seems to haveinbeen the The situation in early Jurchen, while spirantization (> ial stop also same seems ntly, theisImportantly, strong labial stop isKhitan. also preserved Khitan. The same seems mportantly, thepreserved strongthe labial stop is also preserved in Khitan. sameThe seems strong labial stop isinalso preserved inThe Khitan. same seems f) place in later Jurchen, attested inf) Manchu. Theinstatus of Khitan ybeen while spirantization (> f)Jurchen, tookasplace in(>spirantization he situation intook early while spirantization took thebeen situation inJurchen, early in Jurchen, while spirantization (> f)place took in place in toJurchen, have the situation early while (> place f) took inManchu. this respect was known already from therespect lexical dataknown preserved in hu. The status of Khitan in this respect was known as attested in The status of Khitan in this respect was known chen, as attested in Manchu. The status of Khitan in this was later Jurchen, as attested in Manchu. The status of Khitan in this respect was known 頗rendering, Chinese rendering, in枿 ‘time’ = *pon served infrom Chinese rendering, asas in †poin= =Chinese ‫ ڛ‬as po in he lexical data preserved in Chinese 枿 = as ‫ڛ‬in po=枿‫†ڛ‬po from the lexical data preserved in Chinese rendering, as †po in Pre-Proto-Mongolic 枿 †po po = ‫ڛ‬ already the lexical data preserved rendering, po 42 (= Jurchen †po) > Proto-Mongolic *xon > Modern (= Jurchen †pon) > Proto-Mongolic *xon > Modern Mongolian on ‘year.’ oto-Mongolic *pon (= Jurchen †po) > Proto-Mongolic *xon > Modern Pre-Proto-Mongolic *pon (= Jurchen > Proto-Mongolic *xon > Modern ‘time’ = Pre-Proto-Mongolic *pon (= †po) Jurchen †po) > Proto-Mongolic *xon > Modern 42 42 42 eian have a Today, few examples, including ‫؀ګ‬ p.ar Today, have afew few examples, including p.ar †par(a) ‘people’ wewe have a have more examples, including ‫ ؀ګ‬including p.ar ‘year’. on ‘year’. Today, we amore few more examples, including ‫؀ګ‬ p.ar Mongolian onmore ‘year’. Today, we have a few more examples, ‫؀ګ‬ p.ar a/n : (plural) *xara-d > ar(a)d id., ‫ګ‬۲‫
ر‬ p.ul.uh e’ = *para > *xara/n :*xara/n (plural) > ar(a)d id., ‫ګ‬۲‫
ر‬ p.ul.uh *para : *xara-d (plural) *xara-d ar(a)d id., †pulu(-h) ‘people’ == *para *xara/n (plural) >>ar(a)d ‫ګ‬۲‫
ر‬ p.ul.uh †par(a) ‘people’ =>> *para > :*xara/n : *xara-d (plural) *xara-d >id., ar(a)d id.,p.ul.uh ‫ګ‬۲‫
ر‬ p.ul.uh more’ =month’ *püle-xü ‘more, well ‫
ٸצګ‬ rcalary < ‘more’ =extra’, *püle-xü ‘more, extra’, extra’, as‘more, wellasas ‫
ٸצګ‬ h) ‘intercalary month’ x) and pharyngealization (> h), (> h), (> h), tended to undergo spirantization (> x) and pharyngealization stopalso *k has also tended to undergo spirantization (> x) and pharyngealization corresponding velar stop *k has also tended to undergo spirantization (> ofin Northeast Asia. tendency isoftypically suages (> loss Ø), the languages ofthe Northeast Asia. This isThis typically rely (> Ø), in(> the languages of Northeast Asia.tendency This tendency is typically more rarely loss Ø), inThis languages Northeast Asia. tendency islanguages typically x) and pharyngealization (> h), more rarely loss (> Ø), in the hose languages that preserve, have recently ore prominent in those thatuntil preserve, orthat have d/or more prominent inlanguages thoseorin languages that preserve, oruntil haverecently until recently older and/or more prominent those languages preserve, or have until recently of Northeast Asia. This tendency typically older and/or more prominent entact. find the development *k > h as a relatively early we find thefind development *kisdevelopment > h *k as >a relatively early d, *p Therefore, intact.*pTherefore, we the development h as a relatively early preserved, intact. Therefore, we find the *k > h as a relatively early 43 43 have43until recently in those languages that preserve, or preserved, *p 43 Korean and Jurchen-Manchu, while in the Protowhile in the Protonenon languages like Korean and Jurchen-Manchu, while in the Protoin languages like Korean and Jurchen-Manchu, phenomenon in languages like Korean and Jurchen-Manchu, while in the Protourkic) is(as much Depending onlater. the intact. Therefore, find theitlanguage, development > hthe as language, a on relatively early ge (as itwell aswell in later. Turkic) it as is we much Depending on *k the language, ic lineage as inwell Turkic) is much later. Depending on Mongolic lineage (as in itTurkic) is much later. Depending the language, 43 tual restrictions; it is particularly common that it is phenomenon in languages like Korean and Jurchen-Manchu, while in nt may have contextual restrictions; it is particularly common that it is elopment may havemay contextual restrictions; it is particularly commoncommon that it is that itthe the development have contextual restrictions; it is particularly is vowels. fore original back vowels. only before vowels. present onlyoriginal before back original back vowels. ct, examples ofhave, preserved original front an, we have, fact, of preserved *k original front n Khitan, inexamples fact,*kexamples of preserved *k before front 40 in In we Khitan, we have, inbefore fact, examples ofbefore preserved *koriginal before original front The phenomenon is also attested in a few Uralic languages. Starting with Pelliot genitive ‫ط‬۸‫
ڕ‬ ku.û.un) = *kü-xün id. The variation ‫
ط‬as kuin†ku ‘person’ (genitive ‫ط‬۸‫
ڕ‬ ku.û.un) = *kü-xün id. The variation ‫
ط‬ku †ku‫
ط‬ ‘person’ (genitive ‫ط‬۸‫
ڕ‬ ku.û.un) = *kü-xün id. The vowels, as (1925), in ku †ku ‘person’ (genitive ‫ط‬۸‫
ڕ‬ ku.û.un) = has *kü-xün id. The variation particular attention in comparative Altaic studies been variation paid to the h-stage, unger sister’ =~*keü might suggest that ‘child’ there ۘ‫
ڥ‬ †keu ‘younger sister’ =sister’ might suggest that there though the‘child’ implications of *keü the issue have not always beensuggest correctly understood. k.iú ~x.iú ۘ‫
ڥ‬ x.iú †keu ‘younger =‘child’ *keü might that therethat there in ‫
ڥڟ‬ k.iú ۘ‫
ڥ‬ x.iú †keu ‘younger sister’ = *keü ‘child’ might suggest 41 so in this position, another possible example being yndency to spirantize *k also in this position, another possible example being Another language family in which the phenomenon is not attested is Amuric, being today to spirantize *k also in this position, another possible example being was a tendency to spirantize *k also in this position, another possible example may or may not correspond to Proto-Mongolic *kürrepresented by the single language (isolate) of Ghilyak (Nivkh), but earlier distributed ‘to arrive’, which may or may not correspond to Proto-Mongolic *kürx.ui.ri ‘to arrive’, which may or may not correspond to Proto-Mongolic *kürۘ‫
ٯڊ‬x.ui.ri ‘to arrive’, which may or may not correspond to Proto-Mongolic *kürmore in Manchuria, including, quite the northern neighbourhood of spirantization is widely confirmed by ‫
צپ‬ heu.úr †haur =possibly, ginal back vowels, spirantization isspirantization confirmed byis‫
צپ‬ †haur =†haur ore back vowels, is confirmed byheu.úr ‫
צپ‬ heu.úr = †haur = id.44original Before original back spirantization vowels, confirmed by ‫
צپ‬ heu.úr Korean. 42

On this item, cf. Kane (2009: 68, 122-123). The Jurchen word is, of course, a

borrowingStarting from some early Mongolic form of speech. It may be noted that the Khitan few with Pelliot (1925), particular 40 is Uralic on also attested in aisfew languages. Starting with Pelliot (1925), particular enomenon islanguages. also attested inUralic a few Uralic languages. Starting with Pelliot (1925), particular The phenomenon also attested inMongolic a few Uralic languages. Starting Pelliot (1925), particular “reading” is though †po, while and Jurchen suggest the with presence of a final nasal, i.e. snrative been paid to the h-stage, the implications of the issue Altaic studies has been paid to the h-stage, though the implications of the the implications issue comparative Altaic studies has been paid to the h-stage, though the implications of the issue attention in comparative Altaic studies has been paid to the h-stage, though of thediffer issue *pon. There are also other cases in which Khitan and the Proto-Mongolic lineage d.have een correctly understood. always been correctly understood. not always been correctly understood. in use of theisfinal nasaltoday *n inrepresented nominals. The reasons underlying these differences are phenomenon is the not attested Amuric, by ge family which the phenomenon isthe notphenomenon attested is Amuric, today represented by r41 language in which the phenomenon is not attested is attested Amuric, represented by Anotherinfamily language family in which is not istoday Amuric, today represented by not known. (Nivkh), distributed more widely in Manchuria, gethe (isolate) of43earlier Ghilyak (Nivkh), but earlier distributed more widely Manchuria, language (isolate) of(isolate) Ghilyak but earlier distributed more in widely in widely Manchuria, singlebut language of(Nivkh), Ghilyak (Nivkh), but earlier distributed more in Manchuria, The neighbourhood spirantization of of *k in Jurchen-Manchu has been the object of some dispute. ghbourhood of Korean. ossibly, the northern quite possibly, the northern of Korean.of Korean. including, quite possibly, theneighbourhood northern Korean. neighbourhood However, the development seemsis, toof have beena regular initial position, as maintained 123). The Jurchen word is, of course, a borrowing from some 42 . item, Kane (2009: 68, The 68, Jurchen word course, borrowing from some Kane (2009: 68, (2009: 122-123). The Jurchen word is, ofword course, ainborrowing from somefrom some On cf. this item, cf.122-123). Kane 122-123). The Jurchen is, of course, a borrowing erm noted that the Khitan “reading” is †po, while Mongolic and of speech. It may be noted that the Khitan “reading” is †po, while Mongolic and ngolic form of speech. It may be noted that the Khitan “reading” is †po, while Mongolic early Mongolic form of speech. It may be noted that the Khitan “reading” is †po, while and Mongolic and asal, i.e.the *pon. are also other cases in which Khitan and eJurchen presence of aThere final nasal, i.e.nasal, There areThere also other cases in also which and uggest presence a final i.e. *pon. are also other cases inKhitan which Khitan andKhitan and suggest theofpresence of*pon. a final nasal, i.e. *pon. There are other cases in which use of the differ final nasal *n inof nominals. The reasons underlying ic lineage in differ thelineage use the nasal *n in The Mongolic lineage in the usefinal of the the final *n innasal nominals. The underlying reasons underlying the Proto-Mongolic differ in use ofnasal thenominals. final *nreasons in nominals. The reasons underlying re not differences known. erences are not known. these are not known. Manchu hasinof been object ofhas some However, the 43 of *k on Jurchen-Manchu beendispute. thebeen object some However, the irantization *k the in Jurchen-Manchu has the ofdispute. some the The spirantization of *k in Jurchen-Manchu hasofobject been the object dispute. of someHowever, dispute. However, the

†pur(e)s ‘descendants’ = (plural) *püre-s ‘seeds’. Spirantization SpirantizationSpirantization ofof*k. *k.Very Verymuch much like like the thestrong stronglike labial labial stop stop *p, *p,much the thecorrespondin correspondi of *k. Very much strong labial stopthe *p,str th Spirantization ofthe *k. Very like stop stop *k *k has has also also tended tended to to undergo undergo spirantization spirantization (> (> x) x) and and pharyngealization pharyngealizatio stop *k has also tended to undergo spirantization (> x) and pha *k has also tended ntization of *k. Very much like the strong labial stop *p, stop the corresponding velar to undergo spirant more more rarely rarely loss loss (>(>Ø), Ø),in(> inthe the languages languages ofNortheast Northeast Asia. Asia. This This tendency tendency isisof tyt more rarely loss (> Ø), in of the languages ofh),Northeast Asia. This more rarely loss (> Ø), in the125 languages *k has also tended to undergo spirantization x) and pharyngealization (> Khitan: Understanding the Language Behind the Scripts older older and/or and/or more more prominent prominent in in those those languages languages that that preserve, preserve, or or have have until until reo older and/or more prominent in those languages that preserve, older and/or is more prominent in those langu rarely loss (> Ø), in the languages of Northeast Asia. This tendency typically preserved, preserved, *p *p intact. intact. Therefore, Therefore, we we find find the the development development *k *k > > h h as as a a relatively relative preserved, *ppreserve, intact. preserved, Therefore, weintact. find the development *k >theh *p Therefore, we find and/or more prominent in those languages that or have until recently 4343 44 phenomenon phenomenon in in languages languages like like Korean Korean and and Jurchen-Manchu, Jurchen-Manchu, while while in inthe the phenomenon in languages like Korean and Jurchen-Manchu, phenomenon in languages like Korean an rved, *p intact.Proto-Mongolic Therefore, we lineage find the(as development *k > h as a relatively early well as in Turkic) it is much later. Depending Mongolic Mongolic lineage lineage (as (as well well asasininTurkic) Turkic) itisisinmuch much later. later. on onDepend the thelan 43 itas Mongolic lineage (as well Turkic) it Depending isDepending much Depen Mongolic lineage (as well as inlater. Turkic) itlai omenon in languages like Korean Jurchen-Manchu, while in therestrictions; Protoon the the language, the and development may haverestrictions; contextual it is common the development development may may have have contextual contextual restrictions; it it is is particularly particularly common tht the development may have contextual restrictions; it is particula the development may have contextual restr golic lineage (as well as incommon Turkic) that it is it much later. only Depending on the language, particularly isonly present before back original back vowels. present present only only before before original original back back vowels. vowels. present before original vowels. presentcommon only before back vowels. evelopment may have contextual restrictions; it is particularly that original it is In Khitan, we have, inwe fact, examples preserved *kpreserved before original InInKhitan, Khitan, we have, have, ininfact, fact, examples ofof preserved *k *kin before before origin In Khitan, weofexamples have, in Khitan, fact, examples of preserved *k In we have, fact, origina examp nt only before original back vowels. front vowels, as in ku †ku ‘person’ (genitive = *küvowels, vowels, as as in in ‫
ط‬ ‫
ط‬ ku †ku †ku ‘person’ ‘person’ (genitive (genitive ‫ط‬۸‫
ڕ‬ ‫ط‬۸‫
ڕ‬ ku.û.un) ku.û.un) = = *kü-xün *kü-xün id. id. The The va v vowels,ofaspreserved in ‫
ط‬ku †ku ‘person’ ‫ط‬۸‫
ڕ‬ ku.û.un) = *kü vowels, in(genitive ‫
ط‬kufront †ku ‘person’ (genitive ‫ط‬ In Khitan, we have, in fact, examples *k beforeasoriginal in in ‫
ڥڟ‬ ‫
ڥڟ‬ k.iú k.iú ~ ~ ۘ‫
ڥ‬ ۘ‫
ڥ‬ x.iú x.iú †keu †keu ‘younger ‘younger sister’ sister’ = = *keü *keü ‘child’ ‘child’ might might suggest suggest tha th xün id. The variation in k.iú ~ x.iú †keu ‘younger sister’ = *keü ‫
ڥڟ‬k.iú ~ ۘ‫
ڥ‬ x.iú †keuid. ‘younger sister’ = *keü ‘child’sist mi in ‫
ڥڟ‬ k.iú ~ ۘ‫
ڥ‬ x.iú †keu ‘younger ls, as in ‫
ط‬ku †ku ‘person’ (genitivein‫ط‬۸‫
ڕ‬ ku.û.un) = *kü-xün The variation was was a a tendency tendency to to spirantize spirantize *k *k also also in in this this position, position, another another possible possible example examp was a tendency to spirantize *k also in this position, another po ‘child’ might suggest that there was a tendency to spirantize *k also in this was a tendency to spirantize *k also in this ‫
ڥ‬k.iú ~ ۘ‫
ڥ‬x.iú †keu ‘younger sister’ = *keü ‘child’ might suggest that there ۘ‫
ٯڊ‬ ۘ‫
ٯڊ‬ x.ui.ri x.ui.ri ‘to ‘to arrive’, arrive’, which which may may or or may may not not correspond correspond to to Proto-Mongolic Proto-Mongol ۘ‫
ٯڊ‬ x.ui.ri ‘to arrive’, which may or may not correspond to P position, another possible example being arrive,’ ۘ‫
ٯڊ‬ x.ui.ri ‘to arrive’, which may or ma a tendency to spirantize 44*k position, another possible example being 44 also in this44 44 44 vowels, is 44 Before Before original original back back vowels, vowels, spirantization spirantization is confirmed confirmed by by ‫
צپ‬ ‫
צپ‬ heu.úr heu.úr † id. id. id. Before original back spirantization is confirmed by Before original backoriginal vowels, spirantiza may orwhich may not Proto-Mongolic *kür- id. Before ‫
ٯ‬x.ui.ri ‘to arrive’, maycorrespond or may not to correspond toid.Proto-Mongolic *kür†haur = Before originalback backvowels, vowels,spirantization spirantization isis confirmed confirmed by ‫
צپ‬heu.úr †haur = *kabur

‘spring.’4040Again, caution be exercised with regard to the letters used 40 40 should 40‘spring’. The Thephenomenon phenomenon isisalso also attested attested inin few few Uralic Uralic languages. languages. Starting Starting with with Pelliot Pelliot (1925), (1925), p The The phenomenon phenomenon is isa aalso also attested attested in in aa few few Uralic Uralic languages. Starting with with *kabur Again, caution should befew exercise The phenomenon is also languages. attested in aStarting Uralic in the Romanization (), the exact phonetic values and phonemic attention attentioninincomparative comparative Altaic studies studies has has been been paid paid to to the the h-stage, h-stage, though though the the implications implications of of attention attention in in comparative comparative Altaic Altaic studies studies has has been been paid paid to to the the h-stage, h-stage, though though th th attention in(1925), comparative has been paid the Romanization (),Altaic the studies exact phonetic vat phenomenon is also attested in a always few Uralic languages. Starting with Pelliot have have not not always been been correctly correctly understood. understood. have have not not always always been beenItcorrectly correctly understood. understood. have not always been correctly understood. relationships of which are not known. may be noted that, at least at the which are notexercised known. It may be noted that, at leu 4141 ‘spring’. caution should be with regard toattested the letters on in comparative Altaic*kabur studies has been paid tofamily the h-stage, though thein ofattested the issue 41 41 Again, 41 Another Another language language family in inwhich which the thephenomenon phenomenon is isnot not attested isin isAmuric, Amuric, today today represe Another Another language language family family inimplications which whichlanguage the the phenomenon phenomenon is is not not attested is isrepre Amu Amu Another family which the phenomen level ofthe Romanization, a spirantized velar in Khitan can also correspond to ot always been correctly understood. spirantized velar in Khitan can also correspond (), the exact phonetic values andbut phonemic relationsh the theRomanization single single language language (isolate) (isolate) of of Ghilyak Ghilyak (Nivkh), (Nivkh), but but earlier earlier distributed distributed more moredistributed widely widely ininto Ma Mbt the single single language (isolate) (isolate) of of Ghilyak Ghilyak (Nivkh), (Nivkh), but earlier earlier distributed mo m should be exercised with regard totothe the letters used inin nther should be exercised with regard the letters used the single language (isolate) of Ghilyak (Nivkh), language family in which which the phenomenon is not attested is Amuric, today represented by the Mongolic weak velar stop *g, as 3 hu.ur†hur= *gur-. including, including, quite quite possibly, possibly, the the northern northern neighbourhood neighbourhood of of Korean. Korean. in 3 ‫
ٰ؜‬hu.ur†hur= *gur-. not known. Itquite maypossibly, be that, quite atneighbourhood least at the level ofneighbourhoo Romanizat including, including, quite possibly, the the northern northern neighbourhood of Korean. Korean. including, possibly, the of northern phonemic relationships ofofnoted eexact exactphonetic phoneticvalues valuesand andare phonemic relationships

ngle language (isolate) of4242 Ghilyak (Nivkh), but earlier distributed more widely in word Manchuria, 42 42Kane 42 On Onthis thisitem, item, cf.cf. Kane (2009: (2009: 68, 68, 122-123). The Jurchen Jurchen is,is, of ofcourse, course, a aborrowing borrowing frJ On On this this item, item, cf. cf.122-123). Kane Kane (2009: (2009: 68, 68, 122-123). 122-123). The The Jurchen Jurchen word word is, is, of of course cours OnThe this item, Kane (2009: 68, 122-123). Thefro spirantized velar in Khitan can also correspond to cf. theword Mongolic weak velar stop e noted that, least at the level of Romanization, a be noted that,atatthe least at the level of Romanization, a ing, quite possibly, northern neighbourhood of Korean. early earlyMongolic Mongolicform form ofof speech. speech.Itform Itmay mayof bebe noted noted that that the theKhitan Khitan “reading” “reading” isismay †po, †po,be while while Mong Mon early early Mongolic Mongolic form of speech. speech. It It may may be be noted noted that that the the Khitan Khitan “reading” “reading” is is early Mongolic form of speech. It noted tha Vowel contraction. The Proto-Mongolic lineage h in 3 ‫
ٰ؜‬hu.ur†hur=stop *gur-. Vowel contraction. The Proto-Mongolic lineage a medial velo-laryngeal his item, cf. Kane (2009: 68, 122-123). The Jurchen word of course, ahad borrowing from some lso correspond to Mongolic weak velar as also correspond tothe the Mongolic weak velar stop as Jurchen Jurchen suggest suggest the the presence presence ofof*g, a*g, ais, final final nasal, nasal, i.e. i.e. *pon. There are also alsoother other cases inalso in which which Kh Kcc Jurchen Jurchen suggest suggest the the presence presence of of aa*pon. final finalThere nasal, nasal, i.e. i.e. *pon. *pon. There are arenasal, also other other Jurchen suggest the are presence ofThere a cases final i.e. *p *x [h], which is represented as a segment innominal Writte Mongolic form ofspirant speech.the It may be noted that the Khitan “reading” is †po, while Mongolic and *x [h], which the islineage represented as ause segment in Written Mongol, but theProto-Mongolic Proto-Mongolic lineagediffer differinin the the use of ofthe thefinal final nasal nasal *n *n in innominals. nominals. The The reasons unuf the Proto-Mongolic Proto-Mongolic lineage lineage differ differ in in the the use use of of the the final final nasal nasal *n *n use in inreasons nominal the Proto-Mongolic lineage differ in the of the n suggest the presence ofVowel athese final nasal, i.e.are *pon. There are in also other caseslineage in whichhad Khitan and extant Mongolic languages, and even in Mids contraction. The Proto-Mongolic a medial velo-laryngeal these differences differences are not not known. known. these differences differences are areall not not known. known. which has been lost inthese all extant Mongolic languages, and even in Middle these differences are not known. 4343in the use of the oto-Mongolic lineage differ nasal in nominals. The reasons underlying 43 43 final 43 The Thespirantization spirantization ofofspirantization *k *kinin*n Jurchen-Manchu Jurchen-Manchu has has been been the the object object of of some some dispute. dispute. Howe How (long vowels and diphthongs) as a result. The seg The The spirantization of of *k *k in in Jurchen-Manchu Jurchen-Manchu has has been been the the object object of of som som *x [h], which is represented as a segment in Written Mongol, but which has bee The spirantization of *k Jurchen-Manchu has Mongolic lineage had aawith medial velo-laryngeal spirant Mongolicare lineage haddevelopment medial velo-laryngeal spirant vowel contraction (long vowels and diphthongs) as in a result. differences notMongol, known. development seems seems to to have have been been regular regular in in initial initial position, position, as as maintained maintained by by Näher Näher (1999 (199 development development seems seems to to have have been been regular regular in in initial initial position, position, as as maintained maintained development seems to have been regular in initia origin and may in some cases even have been “o in all extant Mongolic languages, and even in Middle Mongol, with vowel contr segment Mongol, but which has been lost of *k in Jurchen-Manchu has been the object of some dispute. However, the aspirantization segmentininWritten Written Mongol, but which has been lost The segment itself was ofcf.cf. aVovin heterogeneous origin and may in some (1997). cases different different explanation, explanation, Vovin (1997). (1997).cf. different different explanation, explanation, cf. possibility Vovin Vovin (1997). (1997). different explanation, cf. Vovin by the that there may also have existed 4444 regular (long vowels and diphthongs) as a result. The segment itself was of a toheteroge seems toMiddle have been in initial position, as maintained byKhitan Näher (1999); aberead 44 44vowel 45 would 44 s,pment and Mongol, with vowel contraction es, andeven eveninin Middle Mongol, with contraction This This correspondence correspondence would bebe valid valid if ifThis the the Khitan item is isfor to to be read †hur-, †hur-, but the This This correspondence correspondence would would be be valid valid if if the the Khitan Khitan item item is is to be be read read even have been “original.” The issue is complicated byitem the possibility that correspondence would be valid ifbut thethe Kh 45 nt explanation, cf. Vovin (1997). though the comparative material does not allow the origin and may in some cases even have been “original”. The issue isto compl orthographical orthographical image image may may actually actually point point to to a a different different shape. shape. In In any any case, case, there there must must have hav orthographical orthographical image image may may actually actually point point to to a a different different shape. shape. In In any any case, case s aaresult. segment itself was of a heterogeneous as result.The Thethere segment itself was of a heterogeneous orthographical image may actually point a diffe also have existed “original” diphthongs (like *aipeculiar *au), though s correspondence wouldmay be valid if theregard Khitan item ismay to 3be read †hur-, but the 45possibility 45 with sequences with awith *x *axi *axu). difference difference with regard the the numeral numeral 3†hur-. †hur-. by the that there also have existed “original” diphthongs (like *a difference difference with with regard regard to to the the numeral numeral 3medial 3 †hur-. †hur-. difference regard to (like thea numeral 3 †hur-. ven have been “original”. The isto even have been “original”. The issue istocomplicated complicated raphical image may actually point to aissue different shape. In any case, there must have been

the comparative material does not allow them to be distinguished reliably

*x be is,distinguished in any case, normal though the comparative doesProto-Mongolic not allow them to reliably also have existed diphthongs (like *ai ynce also have existed “original” diphthongs (likematerial *ai*au), *au), with regard to the“original” 3 †hur-. fromnumeral sequences with aamedial *x (like *axi *axu). probably be understood as diphthongs in Khitan, sequences with medial *x (like *axi *axu). ldoes doesnot notallow allowthem themtotobe bedistinguished distinguishedreliably reliablyfrom from Proto-Mongolic *x is, in any*xcase, represented ascounterpart zero in what the segmentalrepresented Proto-Mo Proto-Mongolic is,However, innormally any case, normally asof zero in wha e*axi *axi*axu). *axu). may probably be understood as diphthongs in Khitan, as in 100 jau Khitan when followed by a rounded vowel, be as understood as diphthongs in Khitan, as in 100 ֔
jau †jauas=in*j5 represented may nany anycase, case,normally normallyprobably represented aszero zeroininwhat what may †haur =counterpart *kabur ‘spring’. The theretoisbenol However, the segmental counterpart of Proto-Mongolic *bfact alsothat seems †jau = *jaxu/n. However, segmental of Proto-Mongolic hthongs asas inin100 jau †jau ==*jaxu/n. hthongsininKhitan, Khitan, 100֔
֔
jau †jauthe *jaxu/n. confirmed by the use of the characters ֔
jau andh Khitan when followed by a rounded vowel, as in 5 ֢
tau †tau = *tabu/n, ‫
צپ‬ *b also seems to be lost in Khitan when followed by a rounded vowel, as rpart ofofProto-Mongolic erpart Proto-Mongolic*b *balso alsoseems seemstotobe belost lostinin jau.tau †jautau ‘punitive commissioner’ = ᤋ䀾 †haur = *kabur ‘spring’. The fact that there is no medial consonant in these ca heu.úr ded vowel, inin55֢
tau ==*tabu/n, nded vowel,asasin ֢
tau tau†tau †tau= *tabu/n,‫
צپ‬ ‫
צپ‬heu.úr heu.úr †haur = *kabur ‘spring.’ The fact character ֢
tau in the word ֢ٓ‫
ق‬ tau.li.a ‘har confirmed by the use of the characters ֔
jau and ֢
tau for the Chinese term ct that medial isis cases is confirmed by the use of act thatthere thereisthat isno nothere medial consonant inthese thesecases cases is consonant no medial in consonant in these though background of this jau.tau †jautau ‘punitive commissioner’ = ᤋ䀾 *jawtaw (zhaotao). The“rebus use racters ֔
tau the term ֔
֢
aracters ֔
jau jauand and֢
֢
taufor for theChinese Chinese term ֔
֢
the phonological The fate of *x in monophthongoid contex character ֢
tau inThe theuse word tau.li.a ‘hare’ = *taula-i id. is also signi issioner’ ==ᤋ䀾 *jawtaw (zhaotao). ofof֢ٓ‫
ق‬ the missioner’ ᤋ䀾 *jawtaw (zhaotao). The use the by Näher (1999); for a different explanation, cf. comparisons Vovin (1997). like ؕ
ún †un = ?*e-xü/nwhy are u though the phonological background of this “rebus” remains unclear.46 ٓ‫
ق‬tau.li.a = *taula-i id. is also significant, ֢ٓ‫
ق‬ tau.li.a‘hare’ ‘hare’ = *taula-i id. is also significant, 44 This correspondence would be valid if the Khitan item is to be read †hur-, but the 46 46 be offered by ֺ
qa ‘emperor’, butisthis is amb The fate of *x in monophthongoid contexts in Khitan lessitem clear, wh ound unclear. oundofofthis this“rebus” “rebus” remains unclear. peculiarremains orthographical image may actually point to a different shape. In any case, there either *ka.n ‘prince’ or *kaxa.n ‘emperor’ on the M why comparisons like ؕ
ún †un = ?*e-xü/nare uncertain. A possible example phthongoid ininbeen Khitan isisless clear, which isisnumeral 3 †hur-. phthongoidcontexts contexts Khitan lesswith clear, which must have a difference regard to the qa.ha.án ‘of the emperor’ is also enigmatic since it be offered by ֺ
qa ‘emperor’, but this item is ambiguous, since it could corresp 45 uncertain. AApossible example would n==?*e-xü/n?*e-xü/n-are areOn uncertain. possible example would *x in Mongolic, cf. Janhunen (1999). In earlier research, this segment was often ‫ָڦ‬ na.ha †naha ‘maternal uncle’ *naga-cu either *ka.n orgraphically *kaxa.n ‘emperor’ on inthe Mongolic side. =The genitiveid.ֺ with since *g, with is indinguishable Written Mongol. ut this ambiguous, it‘prince’ could toto but thisitem itemisisconfused ambiguous, since itwhich coulditcorrespond correspond readings should be †kaga : †kaga-n, in which case qa.ha.án ‘of the emperor’ is also enigmatic since it seems to contain a medial *g ‘emperor’ ‘emperor’on onthe theMongolic Mongolicside. side.The Thegenitive genitiveָֺ֐
ָֺ֐
to Turkic *kaga.n, rather than to Mongolic *kaxa ‫ ָڦ‬na.ha †naha ‘maternal totocontain aamedial asasinin = *naga-cu id. It is possible that the actual K oenigmatic enigmaticsince sinceititseems seems contain medial*g, *g,uncle’ seems toininvolve an unknown network of borrowin should be †kaga-n, which case the Khitan word would corre le’ ==*naga-cu isispossible that actual cle’ *naga-cuid. id.ItItreadings possible thatthe the†kaga actual:Khitan Khitan to Turkic *kaga.n, rather than to Mongolic *kaxa.n. Altogether, this is a wor a-n, ininwhich ga-n, whichcase casethe theKhitan Khitanword wordwould wouldcorrespond correspond Vowel elision. The loss of vowels in non-initial sy

ay also have existed “original” diphthongs (like *aialso *au), the comparative ility that there may also the have possibility existed that “original” there may diphthongs have(like existed “original” *au), diphthongs (like *ai *m result. (long The vowels segment and itself diphthongs) was a*axi as heterogeneous a*axu). result. The segment itself*aiwas of a though heterogeneous lt.sequences The segment itself was ofsequences aaofheterogeneous with aby medial *x (like sequences medial *x (like *axi *axu). with a medial *x (like *axi 45towith 45 *axu). ial does not allow them be distinguished reliably from 45 sequences with a medialfr* mparative material though does not the allow comparative them material benormally not allow reliably them from to be have origin been and “original”. may in some The cases issue is even complicated been “original”. The issue complicated ve been “original”. The issue is complicated Proto-Mongolic *x is, in any case, represented asrepresented zero inisdistinguished what may in reliably Proto-Mongolic *xtohave is, indistinguished anydoes case, asrepresented zero what may Proto-Mongolic *x is, normally in any case, normally as zero ke *axi *axu). Proto-Mongolic aexisted medial *x understood (like sequences *axi *axu). with a medial *x (like *axi *axu). oth have by existed the “original” possibility “original” thatbe diphthongs there may (like also have *ai *au), existed “original” diphthongs (like *ai†jau *au),= *jaxu/n.* veprobably diphthongs (like *ai *au), be asunderstood diphthongs in Khitan, as in 100 ֔
jau †jau = *jaxu/n. probably as diphthongs in Khitan, as in 100 ֔
jau probablyasbe understood as diphthongs in Khitan, as in be 100understood ֔
jau †ja innot any case, normally represented zero inis, what may probably VOLUME 4 zero (2012) SCRIPTA, *x is, any case, Proto-Mongolic normally represented *x in any as case, in normally what may represented as in zero in what 126 in es though allow the them comparative todistinguished be distinguished material does reliably not from allow them to*b be distinguished reliably from otMongolic allow them to be reliably from However, segmental counterpart of Proto-Mongolic also be lost However, the segmental counterpart ofcounterpart Proto-Mongolic *btoHowever, also seems to be lost inm However, the segmental ofseems Proto-Mongolic *b also seems iphthongs in Khitan, as in 100 ֔
jau †jau = *jaxu/n. the segmental understood as diphthongs probably in be Khitan, understood as in as 100 diphthongs ֔
jau †jau in Khitan, = *jaxu/n. as in 100 ֔
jau †jau = *jaxu xi *axu). sequences with a medial *x (like *axi *axu). axu). Khitan when followed by afollowed rounded vowel, as in 5by ֢
a tau †tau *tabu/n, ‫
צپ‬ heu.úr Khitan when by afollowed rounded vowel, as in 5=vowel, ֢
tauas †tau *tabu/n, ‫
צپ‬ heu.úr Khitan when rounded in 5=֢
tau †tau = *tabu/n of Proto-Mongolic *b also seems to*bmay be lost inof Khitan when followed by segmental counterpart However, of Proto-Mongolic the segmental counterpart also seems Proto-Mongolic to be lost in *b also seems to be ynterpart case, normally Proto-Mongolic represented *x as is, zero in any in what case, normally represented as zero in what may se, normally represented as zero in what may †haur = *kabur ‘spring’. The fact that there is no medial consonant in these cases is †haur =֢
*kabur ‘spring’. The‫
צپ‬ fact that therefact is no in these‘spring’. caseslost †haur =*tabu/n, *kabur ‘spring’. The thatmedial there consonant is †haur no medial consonant inis unded vowel, as in 5 tau †tau = heu.úr = *kabur the characters jau and tau for the Chinese term   jau.tau †jautau followed by a rounded Khitan vowel, when as followed in 5 ֢
by tau a †tau rounded = *tabu/n, vowel, ‫
צپ‬ as in heu.úr 5 ֢
tau †tau = *tabu/n, ‫
צپ‬ heu ngs probably in Khitan, asunderstood in ֔
the as jau diphthongs †jau =֔
*jaxu/n. in Khitan, as inforand 100 ֔
jaufor †jau =Chinese *jaxu/n. in Khitan, as in 100 ֔
jau †jau =of*jaxu/n. confirmed bybe the use100 of by the characters jau and tau the֢
Chinese term ֔
confirmed use the characters ֔
jau tau the֢
֔
of ֢ confirmed by the use of ֢
the characters ֔
jau and tau֢
forterm theuse Chines fact that there isfact no medial consonant in these cases isin confirmed by The †haur that = there *kabur is ‘spring’. no medial The consonant fact that there these is cases no medial is consonant ininthe these cases 招討 tur of‘spring’. However, Proto-Mongolic the segmental *b also counterpart seems to be of lost Proto-Mongolic in *b also seems to be lost ‘punitive commissioner’ = *jawtaw (zhaotao). The use of the character Proto-Mongolic *b also seems to be lost in jau.tau †jautau ‘punitive commissioner’ = ᤋ䀾 *jawtaw (zhaotao). The use of the jau.tau †jautau ‘punitive commissioner’ = ᤋ䀾 *jawtaw (zhaotao). The use of the jau.tau †jautau ‘punitive commissioner’ = ᤋ䀾 *jawtaw (zhaotao). T haracters ֔
and ֢
tau for the Chinese term ֔
֢
‘punitive the useinasof the characters confirmed ֔
by jau the and use ֢
of tau the for characters Chinese ֔
term and ֔
֢
֢
taujau.tau for‫
צپ‬ theis†jautau Chinese term ֔ vowel, Khitan when 5jau ֢
followed tau †tau by =tau *tabu/n, a֢ٓ‫
ق‬ rounded ‫
צپ‬ vowel, heu.úr asthe in tau.li.a 5= taujau †tau ==isis *tabu/n, heu.úr tau in the tau.li.a ‘hare’ *taula-i id. also significant, wel, as 5in ֢
†tau =֢
*tabu/n, ‫
צپ‬ heu.úr character ֢
word tau.li.a ‘hare’ =֢
*taula-i id. also significant, character in the word ֢ٓ‫
ق‬ ‘hare’ *taula-i id. also significant, character ֢
tau in the word ֢ٓ‫
ق‬ tau.li.a ‘hare’ = *taula-i id. is als mmissioner’ =*kabur ᤋ䀾 *jawtaw (zhaotao). The use themedial 46 incharacter tau in 46 cases au ‘punitive commissioner’ jau.tau †jautau =in ᤋ䀾 ‘punitive *jawtaw commissioner’ (zhaotao). The = unclear. ᤋ䀾 use of*jawtaw the (zhaotao). The usetheofw 46 ֢
is hat there †haur is=medial no medial ‘spring’. consonant The in fact these that cases there isof isofthis noremains consonant these there is no consonant these cases is“rebus” though the phonological background of this though the phonological background “rebus” remains unclear. though the phonological background of this “rebus” unclear. though the phonological background ofremains this “rebus” remains unclear.46 b ֢ٓ‫
ق‬ tau.li.a ‘hare’ = *taula-i id. is also significant, though the phonological taujau in the word ֢ٓ‫
ق‬ character ֢
tau ‘hare’ in the =֔
word *taula-i ֢ٓ‫
ق‬ id. tau.li.a also significant, ‘hare’ =clear, *taula-i id. ֔
is also significa ers ֔
confirmed jau and by ֢
the tau use for of the the characters term ֔
֔
jau ֢
andis ֢
tauKhitan for Chinese term ֢
which ֔
and ֢
fate tau for the Chinese term ֢
The of *x intau.li.a monophthongoid contexts in46 Khitan isthe less which is 46 The fate of *x inChinese monophthongoid in is less clear, which The fate of *x in monophthongoid contexts in Khitan is less clear, is The fate of *xcontexts in monophthongoid contexts inThe Khitan less cl 46fate is ground ofᤋ䀾 this “rebus” remains unclear. of *x in onological background though of the this phonological “rebus” remains background unclear. of this “rebus” remains unclear. oner’ jau.tau = †jautau *jawtaw ‘punitive (zhaotao). commissioner’ The use of = the ᤋ䀾 *jawtaw (zhaotao). The use of the r’why = ᤋ䀾 *jawtaw (zhaotao). The use of the comparisons like ؕ
ún †un = ?*e-xü/nare uncertain. A possible example would is why comparisons like ún †un = ?*e-xü/nare uncertain. A possible why comparisons ؕ
ún = ?*e-xü/nare uncertain. A possible example would why comparisons like ؕ
ún †un = ?*e-xü/n- arewhy uncertain. A possible ex nophthongoid contexts in Khitan is less clear, which is comparisons ؕ
ate of *x in monophthongoid The fate contexts of *xsignificant, inintau.li.a Khitan monophthongoid is less contexts which isissince Khitan is less clear,like which ‫
ق‬ character ‘hare’ =offered tau in id. the id. word is also ֢ٓ‫
ق‬ ‘hare’ *taula-i id.in also it significant, au.li.a ‘hare’ =֢
*taula-i is also significant, betau.li.a offered by ֺ
qa*taula-i ‘emperor’, but this isbut ambiguous, since could tosince be bybe ֺ
qa this item=clear, is ambiguous, could correspond to example would offered ‘emperor,’ but thisit item iscorrespond ambiguous, be byitem ֺ
qa qa ‘emperor’, but is ambiguous, it could 46 ‘emperor’, 46 un = ?*e-xü/nare uncertain. A possible example would 46 be offered by ֺ
qa ‘emp sons like ؕ
ún †un why = ?*e-xü/ncomparisons are uncertain. like ؕ
ún A †un possible = ?*e-xü/nexample are would uncertain. A possible example wo dthis of this though “rebus” the phonological remains background of this “rebus” remains unclear. “rebus” remains unclear. either *ka.n ‘prince’ or unclear. *kaxa.n ‘emperor’ on ‘emperor’ the Mongolic side. The genitive ָֺ֐
either *ka.n ‘prince’ or *kaxa.n on the Mongolic side. The genitive ָֺ֐ either *ka.n ‘prince’ or *kaxa.n ‘emperor’ on the Mongolic side. The ge since it could correspond to either *ka.n ‘prince’ or *kaxa.n ‘emperor’ ,ֺ
butqathis item isfate ambiguous, since itclear, could correspond to item either *ka.n or * ‘emperor’, be butKhitan offered this is ֺ
ambiguous, qa ‘emperor’, since this could correspond is itambiguous, to since itas ongoid contexts The in of *xitem in isby monophthongoid less which contexts isbut insince Khitan is aless clear, which iscorrespond oid contexts in Khitan is less clear, which is qa.ha.án ‘of the emperor’ is also enigmatic since ititseems to contain medial into‘prince’ qa.ha.án ‘of the emperor’ is enigmatic seems to emperor’ contain acould medial *g, as in qa.ha.án ‘of thealso emperor’ is also enigmatic since it*g, seems contain a me on the Mongolic side. The genitive qa.ha.án ‘of the is .n ‘emperor’ on the Mongolic side. The genitive ָֺ֐
qa.ha.án ‘of the emperor’ prince’ orcomparisons *kaxa.n either ‘emperor’ *ka.n ‘prince’ the Mongolic *kaxa.n side. ‘emperor’ The genitive onid. the ָֺ֐
Mongolic side. The genitive ָֺ *e-xü/nwhy are uncertain. like A possible ؕ
on ún †un example ?*e-xü/nwould are uncertain. A possible example would xü/nare uncertain. A possible example would ‫ָڦ‬ na.ha †naha ‘maternal uncle’ ==or *naga-cu id. It is uncle’ possible that the actual Khitan ‫ָڦ‬ na.ha †naha ‘maternal uncle’ = *naga-cu It is possible that the actual Khitan ‫ָڦ‬ na.ha †naha ‘maternal = *naga-cu id. It is possible that the lso enigmatic it seems to contain athis medial *g, as inaa medial also enigmatic since it to contain medial *g, ascould na.haa†naha na.ha ‘mater he emperor’ issince also qa.ha.án enigmatic ‘of since the emperor’ itseems seems isto toalso contain enigmatic since *g, itas inin ‫ָڦ‬ to contain medial *g, as his item be offered is should ambiguous, by ֺ
qa since ‘emperor’, it could but correspond item to is ambiguous, since itseems correspond tocorrespond tem is ambiguous, since it could correspond readings be †kaga : readings †kaga-n, in which case the Khitan word would correspond readings should be †kaga : †kaga-n, in which case the Khitan word would should be †kaga : †kaga-n, in which case the Khitan word wou ncle’ = *naga-cu id. It is possible that the actual Khitan readings should be †kaga †naha ‘maternal uncle’ = *naga-cu id. It is possible that the actual Khitan naha ‘maternal uncle’ ‫ָڦ‬ = na.ha *naga-cu †naha id. ‘maternal It is possible uncle’ that = *naga-cu the actual id. Khitan It is possible that the actual Khi peror’ either on the *ka.n Mongolic ‘prince’ side. or *kaxa.n The genitive ‘emperor’ ָֺ֐
on the Mongolic side. The genitive ָֺ֐
or’ the Mongolic side. The genitive ָֺ֐
to on Turkic *kaga.n, rather than torather Mongolic *kaxa.n. Altogether, this is*kaxa.n. a wordthis thatis a wordthis to case Turkic *kaga.n, thancorrespond to Mongolic *kaxa.n. Altogether, thati to Turkic *kaga.n, rather than to Mongolic Altogether, aga-n, in which the Khitan word would to Turkic *kaga.n, rathe readings should be †kaga : †kaga-n, in which case the Khitan word would ld be †kaga : †kaga-n, readings in which should case be †kaga the Khitan : †kaga-n, word in would which correspond case the Khitan word would correspo gmatic qa.ha.án since it ‘of seems the emperor’ to contain is a also medial enigmatic *g, as since in it seems to contain a medial *g, as in atic sincetoitinvolve seems to contain a medial *g, as in seems an unknown network of borrowings. seems to involve an to unknown of borrowings. seems involve anword unknown network of borrowings. to‫ָڦ‬ Mongolic *kaxa.n. Altogether, this isKhitan anetwork that involve unknt aga.n, rather totopossible Turkic Mongolic *kaga.n, rather than to id. Mongolic this a word *kaxa.n. thatAltogether, this is aanword correspond tothat Turkic *kaga.n, rather than toItMongolic *kaxa.n. Altogether, =naga-cu *naga-cu id.than is ‘maternal that uncle’ the actual = Altogether, *naga-cu isispossible that theseems actualtoKhitan id.na.ha It isIt†naha possible the*kaxa.n. actual Khitan network of borrowings. lve an unknown network seems toof of involve borrowings. an network of Khitan borrowings. in which readings case should the be †kaga word :seems †kaga-n, would correspond in which case the word would correspond this is aKhitan word that involve an unknown network of borrowings. which case the Khitan word would correspond Vowel elision. The loss vowels intounknown non-initial syllables is a in trivial process observed Vowel elision. The loss of vowels in non-initial syllables is aVowel trivial process observed Vowel elision. The loss of vowels non-initial syllables is a trivial proc elision. o Mongolic to*kaxa.n. Turkic *kaxa.n. *kaga.n, Altogether, rather than this is to a Mongolic word that *kaxa.n. Altogether, this is a word that Theorloss golic Altogether, this is a word that in many languages. The phenomenon may concern all short (single) vowels, or only in many languages. The phenomenon may concern all short (single) vowels, only in many languages. The phenomenon may concern all short (single) vo wels in non-initial syllables is a trivial process observed in many languages. The .fthose The loss of vowels Vowel incertain non-initial elision. The syllables loss of vowels is anon-initial trivial in non-initial process observed syllables is aan trivial process rk ofseems borrowings. to involve unknown network ofthose borrowings. borrowings. in certain positions, or those with certain qualities. In Northeast Asia, earlyAsia, Vowel elision. The loss of vowels in syllables a trivial process those inan positions, or certain qualities. In those Northeast anobserv earlyA those in certain positions, or those withiscertain qualities. In Northeast omenonThe may concern allmay short (single) vowels, orwith only in certain positions uages. phenomenon in many languages. concern The all short phenomenon (single) may vowels, concern or only all short (single) vowels, or o vowel elision may be established for several language families, including Turkic and observed in many languages. The phenomenon may concern all short vowel elision may be established for several language families, including Turkic and vowel elision may be for several language families, includin with elision. certain qualities. In vowels Northeast Asia, anestablished early vowel elision may be est in positions, or those those with in certain certain positions, qualities. or In those Northeast with Asia, certain an qualities. early In Northeast Asia, an ea nthose non-initial Vowel syllables The is loss a trivial of process in observed non-initial syllables is a trivial process observed n-initial syllables is vowels, a trivial process observed Korean(ic), while other families, including Mongolic andMongolic Tungusic, show a less Korean(ic), while other families, including and Tungusic, aother less (single) or only those in Turkic certain positions, or those with certain Korean(ic), while other families, including Mongolic andshow Tungusic, hed for several language families, including and 47 Korean(ic), while may be established vowel for elision several may language be established families, including for several Turkic language and families, including Turkic a 47 on may in many concern languages. all short The (single) phenomenon vowels, or may only concern all short (single) vowels, or only may concern all short (single) vowels, or only A special systematic, or at least chronologically later, tendency to lose vowels. systematic, orand at Tungusic, least later, to lose vowels. A special qualities. In Northeast Asia,chronologically an at early vowel elisiontendency may be established forto orlose systematic, or least chronologically later, tendency vowels ilies, including Mongolic show a less systematic, at least c while other families, Korean(ic), including while Mongolic other families, and Tungusic, including show Mongolic a less and Tungusic, show a l those certain in qualities. certainInpositions, In Northeast orAsia, those Asia, an47early certain qualities. In Northeast Asia, an early hwith certain qualities. Northeast anwith early 47 Korean(ic), while other 47 several languagelater, families, including Turkic and Alose special ologically later, tendency to lose vowels. 45 45 systematic, AIn special spec rveral atOn least chronologically or at tendency least chronologically to later, tendency to lose vowels. *x inlanguage Mongolic, cf. in Janhunen (1999). In earlier research, this segment was often confused with *g, several vowel elision families, be established for Turkic several and language families, including Turkic and wasA 45including language families, including Turkic and Onmay *x Mongolic, cf. Janhunen (1999). Invowels. earlier research, this segment often confused with *g 45 was On *x in Mongolic, cf. Janhunen (1999). earlier research, this segment often On *xorin at Mongolic, cf. Janhc including Mongolic and Tungusic, show a less systematic, withKorean(ic), which itfamilies, is graphically indinguishable in Written Mongol. ncluding Mongolic while and other Tungusic, families, show including a less Mongolic and Tungusic, show a less with which it is graphically indinguishable in Written Mongol. uding Mongolic and Tungusic, show a less with which it is graphically indinguishable in Written Mongol. 46 47 with which ittois graphically ind 45 on what (1999). earlier research, this segment was often confused with *g,earlier 46 47Janhunen 47 Thecf.In issue depends part the Proto-Mongolic numeral Khitan †tau to: only 46 of 47what Acorresponds special feature of least chronologically later, tendency to lose vowels. golic, Janhunen (1999). Inlose in earlier Mongolic, research, cf. this (1999). was often In research, with this *g, segment was often confused The depends on part oflater, the Proto-Mongolic numeral Khitan †tau corresponds to: corresp onlywith to 46numeral The issue on what part ofconfused theto Proto-Mongolic Khitan †tau A segment special special ally systematic, later, tendency or atissue to*x least chronologically vowels. tendency lose vowels. later, toOn lose vowels. A depends special TheAissue depends on what shable intendency Written Mongol.

the root *tab, orthe to the bisyllabic sequence *tabu. The word *taulai, the *tabu. other hand, istheone of hand, graphically indinguishable with which in Written is graphically Mongol. in Written Mongol. rootentire *tab, oritthe to the entire bisyllabic sequence *tabu. The word *taulai, onword other is one of root *tab, orindinguishable to the entire bisyllabic sequence The *taulai, othe Jurchen-Manchu (and certain forms ofonly Mongolic) isonthe loss of vowels in the root *tab, or toon thethe entire b 46an of the Proto-Mongolic numeral Khitan †tau corresponds to: to the items containing “original” diphthong with no medial consonant in Written Mongol, a situation ends 45on what partthe of items the The Proto-Mongolic issue the depends numeral on what Khitan part of“original” †tau the Proto-Mongolic corresponds to:consonant numeral onlyno to medial Khitan †tau Mongol, corresponds to: Mo onl containing an “original” diphthong with no medial in Written a situation items containing an diphthong with consonant in Written Infurther earlier On research, *x in Mongolic, this segment cf. Janhunen was often (1999). confused In earlier with *g, research, this segment was often confused with *g, the items containing an “origi open second syllables (the so-called “Mittelsilbenschwund”). abic sequence *tabu. The word *taulai, on the other hand, is one of arlier research, this segment was often confused with *g, complicated bysequence the fact that Turkic counterpart has aThe However this*b. to the entire bisyllabic the root *tab, *tabu. or the to the The entire word bisyllabic *taulai, on*tabïsh-gan sequence thecounterpart other *tabu. hand, ismedial one word of*b. *taulai, on the other hand, isthis on* complicated by the fact that the Turkic *tabïsh-gan has a medial However further complicated by athe fact that the Turkic counterpart *tabïsh-gan has aby medial nmay Written with which Mongol. itfurther is graphically indinguishable in Written Mongol. further complicated the fac diphthong with no medial consonant Written Mongol, situation ritten Mongol. be,“original” the connection between ‘five’ and was among the ‘five’ strongest pieces of evidence inWritten favour ning an diphthong the items with containing noinmedial an‘hare’ “original” consonant diphthong inbetween Written with Mongol, no medial a situation consonant inthe Mongol, afavour situa may be, the connection between ‘five’ and ‘hare’ was among the strongest pieces of evidence in 46 may be, the connection and ‘hare’ was among strongest pieces of ev roto-Mongolic The issue numeral depends Khitan on what †tau part corresponds the to: only to phases numeral Khitan †tau corresponds to may be,to: theonly connection betwee the Turkic counterpart *tabïsh-gan hasofaby medial *b. However thisthe Mongolic numeral Khitan †tau corresponds to:Proto-Mongolic only to 46 of the Mongolic identification of Khitan during the early of the decipherment of the ated by the fact that the further Turkic complicated counterpart *tabïsh-gan the fact that has the aTurkic medial counterpart *b. However *tabïsh-gan this has aKhitan medial *b. However of Mongolic identification of Khitan during early phases of the decipherment of the Khitan The issue depends on what part of the Proto-Mongolic numeral Khitan †tau of the Mongolic identification of Khitan during the early phases of the deciphermen uence the *tabu. root The word or toKara *taulai, the entire onbisyllabic the other sequence hand, is of one *tabu. of Theofword *taulai, on the other is one ofidentification of hand, the Mongolic ve’ and ‘hare’ was among the strongest pieces of in favour eSmall *tabu. The *tab, word *taulai, on the other hand, is evidence one Script, cf.‘five’ e.g. (1975: On the diachrony the Turko-Mongolic item forpieces ‘hare’, nection between may and be, ‘hare’ the connection was among between the ‘five’ pieces and ‘hare’ of evidence was among in favour the strongest evidence in fav Small Script, cf. e.g. Kara 165-166). On the diachrony ofthe thediachrony Turko-Mongolic item for ‘hare’, corresponds to: only to165-166). the root(1975: *tab, or to the entire bisyllabic sequence *tabu. The Small Script, cf.strongest e.g. Kara (1975: 165-166). On of theofTurko-Mongolic ng with the no items medial containing consonant an “original” in Written diphthong Mongol, awith situation no medial consonant in Written Mongol, aword situation Small Script, cf. e.g. Kara (19 Khitan during the early phases of the decipherment of the Khitan th no medial consonant in Written Mongol, a situation cf., most recently, Rybatzki (2010: 188-189). c 47identification*taulai, ofcf., Khitan of the during Mongolic the early identification phases of ofthe Khitan decipherment during theof early the Khitan phasesdiphthong of the decipherment of the Kh most recently, Rybatzki (2010: 188-189). on the other hand, is one of the items containing an “original” with cf., most recently, Rybatzki (2010: 188-189). rkic counterpart further complicated byvowel the has fact a47elision medial that the *b.aTurkic However counterpart this *tabïsh-gan has a and medial *b. most However this Rybatzki (2 cf., recently, 65-166). On the *tabïsh-gan diachrony of the Turko-Mongolic item for counterpart *tabïsh-gan has aOn medial *b. However this 47 fact of The diachronic in language can belanguage easily established demonstrated byestablished e.g. Kara (1975: 165-166). Small Script, the cf. diachrony e.g. Kara of(1975: the Turko-Mongolic 165-166). On the item diachrony for ‘hare’, of established the Turko-Mongolic item for ‘ha The diachronic fact vowel elision inamong a‘hare’, can easily and demonstrated by no medial consonant in Written Mongol, aof situation further that 47the The diachronic vowel elision incomplicated abelanguage can befact easily ‘hare’ may wasbe, among the connection the strongest between pieces ‘five’ ofof evidence and ‘hare’ infact favour was the strongest pieces ofby evidence in favour The diachronic fact ofand vowd 188-189). e’ was among the strongest pieces of evidence in favour external comparisons with other languages with preserved vowels. In Turkic and Korean, for instance, y, Rybatzki (2010: 188-189). cf., most recently, Rybatzki (2010: 188-189). external comparisons with other languages with preserved vowels. In Turkic and Korean, for instance, the Turkic counterpart *tabïsh-gan has a medial *b. However this may be, the connection external comparisons with other languages with preserved vowels. In Turkic and Kore during the the early Mongolic phases of decipherment the decipherment of Khitan of during the Khitan the early phases of the decipherment of the Khitan external comparisons with oth sion inearly can be established and demonstrated gc fact theof phases of47identification the of of the Khitan ofa language vowel elision The in aeasily diachronic language fact easily vowel established elision in aby and language demonstrated can be easily byin favour established and demonstrated between ‘five’ and ‘hare’can wasbe among strongest pieces evidence ).guages OnSmall thewith diachrony Script, cf.ofe.g. the Kara Turko-Mongolic (1975: 165-166). item for Onthe ‘hare’, the diachrony of theofTurko-Mongolic item of forthe ‘hare’, preserved vowels. In Turkic and Korean, for instance, the diachrony of the Turko-Mongolic item for ‘hare’, isons with other languages external with comparisons preserved with vowels. other In languages Turkic andwith Korean, preserved for instance, vowels. In Turkicofand identification Khitan during the early phases of the decipherment theKorean, for instan ). cf., most Mongolic recently, Rybatzki (2010:of188-189). Khitan Small Script, cf. e.g. Kara (1975: 165-166). On the diachrony of the Turko47 a language The caneasily be easily factestablished of vowel and demonstrated in a language can be easily established and demonstrated by nguage can diachronic be and elision demonstrated by by Mongolicestablished item for ‘hare,’ cf., most recently, Rybatzki (2010: 188-189). with external preserved comparisons vowels. In with Turkic other andlanguages Korean, for with instance, preserved vowels. In Turkic and Korean, for instance, preserved vowels. In Turkic and Korean, for instance, 47 The diachronic fact of vowel elision in a language can be easily established and demonstrated by external comparisons with other languages with preserved vowels. In Turkic and Korean, for instance, the phenomenon is confirmed by comparisons such as Turkic *er ‘man’ = Mongolic *ere, Korean kom ‘bear’ = Japanese kuma. Another language that has lost vowels is Ghilyak, as may be seen from examples such as Ghilyak camng ‘shaman’ = Manchu saman. Note that all these examples involve areal contacts (loanwords), rather than cases of genetic relationship.



Khitan: Understanding the Language Behind the Scripts

127

feature of Jurchen-Manchu (a second syllables (the of so-c ure of Jurchen-Manchu (and certain forms of Mongolic) the loss ofopen vowels in certain featuremakes ofisfeature Jurchen-Manchu (and forms M of Jurchen-Manchu (and certain f In Khitan, the nature of the script it often difficult to verify In Khitan, the nature of n second syllables (the so-called “Mittelsilbenschwund”). open second syllables (the so-called “Mittelsilbensch open second syllables (the so-called “Mitt the nature presence or script absence of vowels all positions, including evenofwordor absence vowels in all p In Khitan, the of the makes it oftenindifficult to verify the presence In Khitan, the nature of the script it ofm In Khitan, the nature of makes the script initially. Many Khitan characters, irrespective of what Romanizational characters, irrespective of wha bsence of vowels in all positions, including even word-initially. Many Khitan urchen-Manchu of Jurchen-Manchu (and certain (and (and certain forms forms of Mongolic) of Mongolic) theisloss the of loss of vowels ature of Jurchen-Manchu certain forms of Mongolic) isof the loss in all positions, oris absence vowels allinpositions, including in e or absence ofin vowels shapeofiswhat usedRomanizational for them, refer shape basically to afor consonantal which may tocore, abasically consonantal core, which m acters, irrespective is used them, refer econd syllables syllables (the so-called (the so-called “Mittelsilbenschwund”). en second syllables (the “Mittelsilbenschwund”). so-called “Mittelsilbenschwund”). characters, irrespective of what Romanizational shap characters, irrespective of what Romaniza be (VC) and/or followed (CV) by vowel. such as ‫
سڻ‬b.as ‘again Items such as consonantal core, which may be script preceded and/or followed (CV) by vowel. hitan, In Khitan, theKhitan, nature the nature ofpreceded the ofscript the makes makes it often it(VC) often difficult difficult verify totoaverify presence theItems presence In the nature ofscript the makes it often difficult toconsonantal verify thea presence to ato consonantal core, which maywhich be preceded athe core, may be (VC) prece may, therefore, well have con b.as ‘again, also’ = *basa id. and p.ar ‘people’ = *paraid. may, sabsence such ‫
سڻ‬ b.as ‘again, also’ = *basa id. and ‫؀ګ‬ p.ar ‘people’ = *paraid. ence of vowels ofasvowels in all in positions, all positions, including including even even word-initially. word-initially. Many Many Khitan Khitan of vowels in all positions, including Items even word-initially. Many Khitan suchItems as ‫
سڻ‬ ‘again, also’ = *basa suchb.as as ‫
سڻ‬ b.as ‘again, also’id. = an *b presence of ahand, vowel in ,ers, therefore, well have final in Khitan. On the other hand, the therefore, well havea contained ashape final in On the other rrespective irrespective of what of Romanizational what Romanizational shapevowel shape is used ismay, for used them, for may, them, refer basically refer basically aracters, irrespective ofcontained what Romanizational isvowel used forKhitan. them, refer basically therefore, well have contained a finalletter vowel therefore, well have contained a fin was aain segmental vowel in the ence of a core, vowel letter inpreceded the does notfollowed necessarily that there the presence ofRomanization abevowel letter in the Romanization does not necessarily nsonantal ntal core, which which may be may be preceded (VC) (VC) and/or and/or followed (CV) by amean vowel. by a by vowel. a consonantal core, which may preceded (VC) and/or followed (CV) vowel. presence of a(CV) vowel letter the Romanization doeK presence of vowel letter in the Romani such as, for instance, ۗ‫
ٯ‬ i.ri ams segmental vowel in the Khitan word. We cannot, therefore, be certain that items mean that there segmental vowel inp.ar the=‘people’ word. We cannot, such ‫
سڻ‬ as ‫
سڻ‬ b.as ‘again, b.as ‘again, also’ also’ = *basa =was *basa id. id.‫؀ګ‬ and ‫؀ګ‬ ‘people’ ‘people’ *para= *paraid. id. such as ‫
سڻ‬ b.as ‘again, also’ =a and *basa id. p.ar and ‫؀ګ‬ =in*paraid. was ap.ar segmental vowel the Khitan word. We word cann was aKhitan segmental vowel in the Khitan ؊‫
ٯ‬m.ri ‘horse’ = *mori/n id. as, for instance, ۗ‫
ٯ‬ i.ri ‘name’ = *nere id. (with a secondary initial nasal) and herefore, re, therefore, well have well well have contained contained a final a vowel final vowel initems Khitan. in such Khitan. the On other the other hand, hand, the the ‘name’ ay, have contained athat final vowel inOn Khitan. On thefor other hand, the i.ri such as, for instance, ۗ‫
ٯ‬ i.ri ‘name’ = *nere i.ri = therefore, be certain as, for instance, such as, instance, ۗ‫
ٯ‬ ‘name’id.= (w * Even so, the circumsta ‫
ٯ‬m.ri ‘horse’ = *mori/n id. ended in a vowel. ce a vowel of a of vowel letter letter in the inRomanization the does initial not doesnecessarily not necessarily mean mean that there that that there esence a vowel letter in Romanization the Romanization does notand necessarily there ؊‫
ٯ‬m.ri ‘horse’ = mean *mori/n ended in vowel.in a *nere id. (with a secondary nasal) m.ri ‘horse’ = id.aended ؊‫
ٯ‬m.ri ‘horse’ =id.*mori/n suggests that that Khitan pr so, circumstance that final vowels are retained inbe Jurchen-Manchu ntal egmental vowel vowel inthe the in Khitan the word. word. We word. cannot, We cannot, therefore, therefore, be Even certain be certain that items that as aEven segmental vowel inKhitan Khitan We cannot, therefore, certain that so, the circumstance finalmost vowels Even so,items theitems circumstance that fin ended in athe vowel. suggested by the fact that cert gests Khitan most probably also retained them. is,initial incidentally, also s, instance, for instance, ۗ‫
ٯ‬ i.riEven ‘name’ i.riso, ‘name’ =the *nere = *nere id.= (with id. (with aid. secondary afinal secondary initial nasal) nasal) and and ch as,that for ۗ‫
ٯ‬ instance, ۗ‫
ٯ‬ i.ri ‘name’ *nere (with aThis secondary nasal) and suggests that Khitan most probably retained suggests that Khitan most also probably also circumstance that vowels areinitial retained in Jurchenalternating variants, although gested the fact that suffixes in suggested a vowel (-CV) positionally orse’ .ri ‘horse’ =by *mori/n = *mori/n ended id.certain ended in vowel. inthat a vowel. ‫
ٯ‬m.ri ‘horse’ =id.*mori/n id.aended inKhitan aending vowel. byalso thehave factby that certain suffixes in suggested thethem. fact48that certain suffixes Manchu suggests most probably retained This is, ending harmony. It is also importan nating variants, although it is not necessarily a question of regular vowel Even so, the so, circumstance the circumstance that final that vowels final vowels are retained are retained in Jurchen-Manchu in Jurchen-Manchu Even so, the circumstance that final vowels are retained in Jurchen-Manchu alternating variants, although it is not necessari alternating variants, although it is not 48 incidentally, also suggested by the fact that certain suffixes ending in 48 Mongolic elements (diphthongs and co mony. It is also important to note the counterparts of long vowel 48also ts t Khitan that that Khitan most most probably probably also retained also that retained them. them. This This is, incidentally, is, ggests Khitan most probably also retained them. This is, also incidentally, also Itincidentally, is important to note that harmony. It is also important to the notecou th harmony. a vowel (-CV) have positionally alternating variants, although itvowel is notletters, as additional ments (diphthongs and contracted vowels) are in Khitan rendered by using yggested ted thebyfact the fact that that certain suffixes suffixes ending ending in ending a vowel in a in vowel (-CV) (-CV) have positionally have positionally bythat thecertain fact certain suffixes a vowel (-CV) have positionally elements (diphthongs and contracted vowels) elements (diphthongs and contracted ar v 48 necessarily ainquestion of regular vowel Itvowel is also to *moga-i id mu.ho.o ’snake’ tional vowel letters, as itnot ‫ح‬
t.qo.a ‘chicken’ = regular *takï-xa id., ٤asimportant ‫ב‬ ‫=ح ٻ ؿ‬ variants, ting variants, although although it is is‫ٻ‬ not necessarily a question a question of of regular vowel ernating variants, although itnecessarily is ‫ق‬ not necessarily a harmony. question of vowel additional vowel letters, in additional regular vowel letters, as ‫ق‬in
t.qo.a ‫‘ق ح ٻ‬c time being we cannot reconstr ho.o ’snake’ *moga-i id., and also ֢ٓ‫
ق‬ tau.li.a ‘hare’ *taula-i id. For theand that the counterparts of long (diphthongs ny. t is48also It 48 is important important to note to that note thatcounterparts the counterparts of Mongolic of Mongolic long vowel long vowel rmony. Italso is=note also important to the note that theMongolic counterparts of=vowel Mongolic long vowel mu.ho.o ’snake’ = elements *moga-i also mu.ho.o ’snake’ =id., *moga-i id.,֢ٓ‫
ق‬ and alsotau but the suggests we cannot reconstruct the Khitan of Khitan such items with any certainty, contracted vowels) areshapes in rendered additional vowel iphthongs tsbeing (diphthongs andand contracted and contracted vowels) vowels) are in areKhitan Khitan intime rendered rendered byusing using by using ements (diphthongs and contracted vowels) are in Khitan rendered by orthography using being weby cannot reconstruct the Khitan shapes time being we cannot reconstruct the Khi (diphthongs or long vowels) the orthography suggests that they may have involved complex vowel elements t.qo.a ‘chicken’ = *takï-xa id., mu.ho.o ‘snake’ letters, as in nal owelvowel letters, letters, asletters, in as‫ٻ‬in ‫ح‬ ‫ٻ‬in
‫ح‬t.qo.a ‫حق‬
t.qo.a ‘chicken’ = *takï-xa *takï-xa id., ‫ב‬ ‫ؿ‬ ٤ ‫ב‬٤ ‫בؿ‬ ditional vowel as ‫ق‬ ‫ٻ‬ ‫‘ق‬chicken’
t.qo.a ‘chicken’ = id., *takï-xa id., ‫ؿ‬they may but the= orthography suggests that have al but the ٤ orthography suggests that they mi hthongs or=long vowels) also Khitan. ake’ o ’snake’ = ’snake’ *moga-i *moga-i and id.,also and ֢ٓ‫
ق‬ also ֢ٓ‫
ق‬ tau.li.a tau.li.a ‘hare’ ‘hare’ = *taula-i =(diphthongs *taula-i For id. the For the = *moga-i id., and tau.li.a ‘hare’ = *taula-i id. For the time u.ho.o =id., *moga-i id.,in also ֢ٓ‫
ق‬ tau.li.a ‘hare’ =id. *taula-i id. For (diphthongs or long vowels) also in Khitan. or long vowels) also in Khitan we eingcannot we cannot reconstruct reconstruct thecannot Khitan the Khitan shapes shapes of shapes such of items such items with any withcertainty, any certainty, me being we cannot reconstruct the Khitan of such items with anyitems certainty, being we reconstruct the Khitan shapes of such with any *have ography suggests suggests that they that may they have may involved involved complex complex vowel vowel elements elements t orthography the orthography suggests that they may have involved complex vowel elements certainty, but the orthography suggests that they may have involved * It has to be stressed that only or ongs long or vowels) long vowels) also inalso Khitan. in Khitan. phthongs or long vowels) also in Khitan. complex vowel elements (diphthongs or long vowels) also in Khitan. properties of Khitan has been s to be stressed that only a small selection of diagnostic and phonological It has to lexical beIt stressed only athat small selection di has to bethat stressed only a small of sele even more remains to be disco erties of Khitan has been discussed Much more remains to be said, and * * above. * properties of Khitan has been discussed above. Mu properties of Khitan has been discussed * consistent the status of n more remains to be discovered. What is, however, is that the material iswith What evencertain moreeven remains be discovered. is, howe moretoremains to be discovered. Wha between Khitan and Proto-Mo istent with the status of Khitan as a Para-Mongolic language. The distance stressed o betostressed only that aonly small a small selection selection of diagnostic of diagnostic lexicallexical and phonological andthe phonological has bethat stressed that only a small selection of diagnostic lexical andwith phonological consistent with status of status KhitanofasKhitan a Para-M consistent the as Ithas has to be stressed that only a small selection of diagnostic lexical and communication between the t ween and Proto-Mongolic isabove. clearly large enough to have made immediate fies Khitan ofKhitan Khitan hasKhitan been been discussed discussed above. Much Much more more remains remains to be to said, be said, and and operties of has been discussed above. Much more remains to be said, and between Khitan and Proto-Mongolic is clearly large between Khitan and Proto-Mongolic is cl similarities are conspicuous e phonological properties of Khitan has been discussed above. Much more munication between the two speech communities impossible. At the same, the more emains remains toremains be to discovered. betodiscovered. What What is, however, is, however, certain is that iscommunication the that the material istwoisspeech en more be discovered. What is, certain however, certain ismaterial that theismaterial communication between the communi between the two speech speaker. The real situation ma larities conspicuous enough to and have evident evenlanguage. forbe the naïve native remains be said, even more remains to discovered. What is, with ent with the are status the status of status Khitan oftoKhitan asKhitan a as Para-Mongolic aasPara-Mongolic language. language. The distance The distance nsistent with the of abeen Para-Mongolic The distance similarities are conspicuous enough to have similarities are conspicuous enoughbeen to h idioms of which we have no di The real situation may have been modified the of intermediate however, certain that the material is consistent with thereal status ofhave Khitan nker. tan Khitan and Proto-Mongolic and Proto-Mongolic is clearly isisclearly large large enough enough toby have to presence made have made immediate immediate tween Khitan and Proto-Mongolic is clearly large enough to have made immediate speaker. The real situation may beenhave modifie speaker. The situation may bee As farProtoas the position o ms ofbetween which we no direct information. ashave atwo Para-Mongolic language. The distance between Khitan and unication ion between the the speech two speech communities communities impossible. impossible. Atwhich the Atsame, the same, the same, the mmunication between the two speech communities impossible. Atwhich the the idioms of we have no direct information. idioms of we have no direct inform the situation is ambiguous. Ev As far as the position of Khitan on the archaic—innovative scale is concerned, are itiesconspicuous are conspicuous enough enough have to have been been evident evident even even forAs the for the naïve native native milarities are conspicuous enough to have been even forthe the Mongolic istoclearly large enough toevident have made immediate communication farnaïve as position ofposition Khitan on Khitan the archa As farnaïve as thenative of on terized as an innovative langu situation is ambiguous. Even so, at least phonologically, Khitan may be characeeaker. r. real Thesituation real real situation may have maythe have been been modified modified by thebythe presence the presence of intermediate of intermediate The situation may have been modified by theimpossible. presence of At intermediate situation issituation ambiguous. Even so,the at leastso,phono between two speech communities same, the is the ambiguous. Even at le thanlanguage, the lineage leading to Prr ed aswe innovative which in most respects evolved more rapidly hich of which have we no have nolanguage, direct information. information. oms ofanwhich wedirect have no direct information. terized had as terized an innovative which in most as an innovative language, which earlier than the Proto-Mongo the lineage toofProto-Mongolic. Many areal innovations rAs asfar the position Khitan Khitan on Khitan theonarchaic—innovative the scale scale is concerned, isreached concerned, Asasposition farthe asleading theofposition of onarchaic—innovative the archaic—innovative scale is concerned, than the lineage leading toKhitan Proto-Mongolic. Many than the lineage leading to Proto-Mongo 48 evolution must be searched in than theis ambiguous. Proto-Mongolic lineage. The reasons for the different speed of An example is offered by least the dative case ending, Romanized as -de ~ -do ~ -du, cf. uation ambiguous. is ambiguous. Even Even so, at so, least atso,phonologically, least phonologically, Khitan Khitan may be may characbe characeerissituation Even at phonologically, Khitan may be characearlier than the Proto-Mongolic lineage. The rea earlier than the Proto-Mongolic lineage the centuries precedingdem th must searched in136-138). thewhich geographical, demographical and political situation. Kane (2009: The ofrespects “vowel harmony” inevolved Khitan isIn still anin open nution as innovative anas innovative language, language, which in most innature most respects had evolved had evolved more rapidly rapidly ized an be innovative language, which in most respects had more rapidly evolution must bemore searched the issue. geographical, evolution must be searched in the geograp rapidly growing and highly mo he centuries preceding the Liao period, Khitan had become the language of a eage e lineage leading toleading Proto-Mongolic. to Proto-Mongolic. ManyMany areal areal innovations innovations reached Khitan Khitan an theleading lineage to Proto-Mongolic. Many reached Khitan Inareal the innovations centuries preceding the Liao period, In thereached centuries preceding the LiaoKhitan peri dly andProto-Mongolic highlylineage. mobile population, which, moreover, contacted on speed a mobile wide than thegrowing Proto-Mongolic the Proto-Mongolic lineage. The reasons The The reasons for the for different the different speed speed of of highly rlier than the lineage. reasons for the different of population, rapidly growing and highly whic rapidly growing and mobile popula ust on be must searched be searched theingeographical, theingeographical, demographical demographical and political and political situation. olution must bein searched the geographical, demographical andsituation. political situation. the phenomenon is confirmed by com ries centuries preceding preceding the Liao the Liao period, period, Khitan Khitan had become had had become the language the language of a a of a kuma. Another l the centuries preceding the Liao period, Khitan become the language ‘bear’of = Japanese

henomenon is confirmed by comparisons such as Turkic *er ‘man’ = Mongolic *ere, Korean kom

128



SCRIPTA, VOLUME 4 (2012)

similarities are conspicuous enough to have been evident even for the naïve native speaker. The real situation may have been modified by the presence of intermediate idioms of which we have no direct information. As far as the position of Khitan on the archaic—innovative scale is concerned, the situation is ambiguous. Even so, at least phonologically, Khitan may be charac­terized as an innovative language, which in most respects had evolved more rapidly than the lineage leading to ProtoMongolic. Many areal innovations reached Khitan earlier than the ProtoMongolic lineage. The reasons for the different speed of evolution must be searched in the geographical, demographical and political situation. In the centuries preceding the Liao period, Khitan had become the language of a rapidly growing and highly mobile population, which, moreover, contacted on a wide scale with a variety of other speech communities. For the ProtoMongolic lineage, such a period of intensive growth and contacts was yet to come.

References Alonso de la Fuente, José (2011). Tense, Voice and Aktionsart in Tungusic: Another Case of “Analysis to Synthesis”? Tunguso-Sibirica 32. Wiesbaden: Harrassowitz Verlag. Arapov, M. V. (1982). Leksika i morfologiia tekstov malogo kidan’­skogo pis’ma. In: Zabytye sistemy pis’ma: Materialy po deshif­rovke. Moskva: Izdatel’stvo “Nauka.” 211-239. Chingeltei 清格爾泰 [Qinggeertai] (1997). Qidanyu shuci ji Qidan Xiaozi pindufa 契丹語數詞及契丹小字拼讀法. Altai Hakpo 7. 143-152. Chingeltei 清格爾泰, Liu Fengzhu 劉鳳翥, Chen Naixiong 陳乃雄, Yu Baolin 于寶林, and Xing Fuli 邢復禮 (1985). Qidan Xiaozi Yanjiu 契丹小字研 究. Beijing 北京: Zhongguo Shehui Kexue Chubanshe 中國社會科學出版社. Doerfer, Gerhard (1985). Mongolo-Tungusica. Tungusica 3. Wiesbaden: Otto Harras­sowitz. Doerfer, Gerhard (1992). Mongolica im Alttürkischen. In Bruno Lewin zu Ehren 3. Bochum. 30-56. Doerfer, Gerhard (1993). The Older Mongolian Layer in Ancient Turkic. Türk Dilleri Araştırmaları 1993:3. 79-86. Golovachev, V. C., A. L. Ivliev, A. M. Pevnov, and P. O. Rykin (2011).



Khitan: Understanding the Language Behind the Scripts

129

Tyrskie stely XV veka: Perevod, kommentariï, issledovanie kitaiskix, mongol’skogo i chzhurchzhen’skogo tekstov. Sankt-Peterburg: “Nauka.” Grube, Wilhelm (1896). Die Sprache und Schrift der Jučen. Leipzig: Kommissions-Verlag von O. Harrassowitz [Reprints: Peking 1936, Tientsin 1941]. Janhunen, Juha (1981). Korean Vowel System in North Asian Perspective. Han-geul 172. 129-146. Janhunen, Juha (1993). The Teens in Jurchen and Manchu Revisited. In Festschrift für Raija Bartens. Ulla-Maija Kulonen (ed.) Mémoires de la Société Finno-Ougrienne 215. Helsinki. 169-184. Janhunen, Juha (1994). On the Formation of Sinitic Scripts in Medi­aeval Northern China. Journal de la Société Finno-Ougrienne 85. 107-124. Janhunen, Juha (1995). Kittanjin ha nanigo wo hanashite ita ka 契丹人はなに 語をはなしていたか. Minpaku Tsuushin 民博通信 68. 82-85. Janhunen, Juha (1996). Prolegomena to a Comparative Analysis of Mongolic and Tungusic. In Proceedings of the 38th Permanent Inter­ nation­al Altaistic Conference (PIAC) (Kawasaki, Japan: August 7-12, 1995). Giovanni Stary (ed.). Wies­baden: Harrassowitz Verlag. 209218. Janhunen, Juha (1999). Laryngeals and Pseudolaryngeals in Mongolic: Problems of Phonological Interpretation. Central Asiatic Journal 43. 115-131. Janhunen, Juha (2003). Para-Mongolic. In The Mongolic Languages, Juha Janhunen (ed.), Rout­ledge Language Family Series 5. Routledge: London & New York. 391-402. Janhunen, Juha (2012). The Expansion of Tungusic as an Ethnic and Linguistic Process. In Recent Advances in Tungusic Studies, Andrej L. Malchukov & Lindsay J. Whaley (eds.) Turcology 89. Wies­baden: Harrasso­witz Verlag. 5-16. Jin Qicong 金啓孮 (1984). Nüzhen wen cidian 女真文辞典. Beijing 北京: Wenwu Chubanshe 文物出版社. Kane, Daniel (1989). The Sino-Jurchen Vocabulary of the Bureau of Interpreters. Indiana University Uralic and Altaic Series 153. Bloomington. Kane, Daniel (2004). A Note on *Čisdeben. Central Asiatic Journal 48. 223225. Kane, Daniel (2006). Khitan and Jurchen. In Alessan­dra Pozzi, Juha

130



SCRIPTA, VOLUME 4 (2012)

Janhunen, and Michael Weiers (eds.) Tumen Jalafun Jecen Akū: Man­chu Studies in Honour of Giovanni Stary. Tun­guso-Sibiri­ca 20. Wiesbaden: Harrassowitz Verlag. 121-132. Kane, Daniel (2008). A Transcription System for Texts in the Qidan Small Script. Handout presented at International Conference on Qi­dan, Tan­ gut, Jurchen and Mongolian Historical Documents 遼夏金元歷史文獻國際 研討會 Liao Xia Jin Yuan Lishi Wen­x ian Guoji Yantaohui (Beijing, 2008.11.3-5). Kane, Daniel (2009). The Kitan Language and Script. Handbook of Orien­tal Studies VIII, 19. Brill: Leiden & Boston. Kara, György (1975). Apropos de l’inscription khitane de 1150. Annales Universitatis Scientiarum Budapestinensis. Sectio Linguistica 6. 163167. Kiyose, Gisaburo N. (1977). A Study of the Jurchen Language and Script: Recon­struction and Decipherment. Kyoto: Hôritsubunka-sha. Ko Seongyen (2011). Vowel Contrast and Vowel Harmony Shift in the Mongolic languages. Language Research 47:1. 23-43. Kwanten, Luc (1984). The Phonological Hypothesis of the Hsi Hsia (Tangut) Language. T’oung Pao 70. 159-184. Kwanten, Luc (1988). The Structure of the Tangut (Hsi Hsia) Characters. Journal of Asian and African Studies 36. 69-105. Ligeti, Louis (1970). Le tabghatch, un dialecte de la language sien-pi. In Louis Ligeti (ed.) Mongolian Studies. Bibliotheca Orientalis Hungarica 14. Buda­­­pest. 265-308. Menges, Karl H. (1968). Tungusen und Ljao. Abhandlungen für die Kunde des Morgenlandes 38:1. Wiesbaden: Franz Steiner GmbH. Näher, Carsten (1999). Der urtungusische stimmlose velare Plosiv im Mandschu. Journal de la Société Finno-Ougrienne 88. 113-130. Pelliot, Paul (1925). Les mots à h initiale, aujourd’hui amuie, dans le mongol des XIIIe et XIVe siècles. Journal Asiatique 206. 193-263. Pevnov, A. M. (2004). Chtenie chzhurchzhen’skix pis’men. Sankt-Peterburg: Rossii­skaia Akademiia Nauk, Institut lingvisticheskix issledovanii & “Nauka.” Róna-Tas, András (2004). A Khitan Word for Marmot. Acta Orientalia Academiae Scien­tiarum Hungaricae 57. 27-29. Rybatzki, Volker (2003). Middle Mongol. In Juha Janhunen (ed.) The Mongolic Languages, Rout­ledge Language Family Series 5. Routledge:



Khitan: Understanding the Language Behind the Scripts

131

London & New York. 57-82. Rybatzki, Volker (2011). Classification of Old Turkic loanwords in Mongolic. In Mehmet Ölmez, Erhan Aydın, Peter Zieme, and Mustafa S. Kaçalin (eds.) From Ötüken to Istanbul: 1290 Years of Turkish. Istanbul Büyükşehir Belediyesi. 185-202. Shimunek, Andrew A. (2007). Towards a Reconstruction of the Kitan Lan­ guage, with Notes on Northern Late Middle Chinese Phono­­logy. MA Thesis, Indiana Uni­versity, Depart­ment of Lin­guistics & Department of Central Eurasian Studies. Starikov, V. S., M. Arapov, A. Karapet’ianc, Z. Malinovskaia, and M. Probst (1970). Materialy po deshifrovke kidan’skogo pis’ma 1-2. Moskva: Aka­ de­miia Nauk SSSR. Sun Bojun 孫伯君 & Nie Hongyin 聶鴻音 (2008). Qidanyu Yanjiu 契丹語研究, Zhongguo Shehui Kexueyuan Wenku Wenxue Yuyan Yanjiu Xilie 中國 科學院文庫文學語言研究系列. Beijing 北京. Takeuchi Yasunori 武內康則 (2007). Kittan Shouji de Hyouki Sa­re­ta Kanjion kara Mita Kittango On’in Taikei no Kenkyuu 契丹小字で表記された漢字音 から見た契丹語音韻体系の研究. MA Thesis. Kyoto University: Kyo­t o Daigaku Daigakuin Bungaku Kenkyuuka 京都大學大學院文學研究科. Talpe, Lode (2010). Some Qidan Words in Chinese Poems. Central Asiatic Journal 54. 79-91. Toyoda Gorou [Gorō] (1964). An Analysis of the Major Ch’i-tan Char­acters. Memoirs of the Research Department of the Toyo Bunko 23. 119-135. Toyoda Gorou [Gorō] 豐田五朗 [Fengtian Wulang] (1998). Qidan Xiaozi dui siji de chenghu 契丹小字对四季的称呼. Minzu Yuwen 民族語文 1998:1. 7881. Vovin, Alexander (1997). Voiceless Velars in Manchu. Journal de la Société Finno-Ougrienne 87. 263-280. Vovin, Alexander (2003). Once Again on Khitan Words in Chinese-Khitan Mixed Verses. Acta Orientalia Academiae Scien­tiarum Hungaricae 56. 237-244. Vovin, Alexander (2011). A Modest Proposal on the Decipherment of the Khitan-Chinese Bilingual Text of 1134 (the Langjun In­scription). In Michael Knüppel & Aloïs van Tongerloo (eds.). Life and Afterlife & Apocalyptic Concepts in the Altaic World, Proceedings of the 43rd Annual Meeting of the Permanent International Altaistic Conference (PIAC), Tunguso-Sibirica 31. Wiesbaden: Harras­so­witz Verlag. 123-

132



SCRIPTA, VOLUME 4 (2012)

130. Wittfogel, Karl A. & Fêng Chia-shêng (1949). History of Chinese Society: Liao (907-1125). Trans­actions of the Amer­ic­ a ­ n Philosophical Society. New Series 36. Philadelphia. Wu Yingzhe 吳英喆 (2005). Qidan Xiaozi «xing» de yufa fanchou chutan 契丹小字 “性” 的語法範疇初探. Nei Menggu Daxue Xuebao (Zhexue Shehui Kexueban) 內蒙古大學學報 (哲學社會科學版) 2005:3. 25-28. Wu Yingzhe (2007). Qidanyu Jingci Yufa Fanchou Yanjiu 契丹語静詞語法範 疇研究 [Re­search on the grammatical cat­egories of the noun in the Khitan language]. Huhe-haote 呼和浩特: Nei Menggu Daxue Chu­ban­she 內蒙古大学出版社. Wu Yingzhe (2009). A Brief Discussion on the Vowel Attachment in the Khitan Small Script. Voprosy filologiï, Seriia “Uralo-altaiskie issle­ dovaniia” 1. 26-30. Wu Yingzhe (2011). Deciphering Some Demonstrative Pronouns in Khitan Small Script. Altai Hakpo 21. 69-77. Wu Yingzhe (2012). Organization of the Khitan materials kept in Inner Mongolia, Unpub­lished Powerpoint presentation (Tokyo University of Foreign Studies). 1-67. Wu Yingzhe & Juha Janhunen (2010). New Materials on the Khitan Small Script: A Critical Edition of Xiao Dilu and Yelü Xiangwen. Corpus Scriptorum Chitanorum 1, The Languages of Asia Series. Folkestone, Kent: Global Oriental. Juha A. Janhunen Institute for Asian and African Studies University of Helsinki, Finland

[Received 10 May 2012; accepted 10 August 2012]