On the classification of the Ng Yap dialects: somethoughts on the subgrouping of Sinitic languages Текст научной статьи по специальности «Языкознание и литературоведение»

Giorgio Orlandi

independent researcher; [email protected]

On the classification of the Ng Yap dialects:

some thoughts on the subgrouping of Sinitic languages*

The Ng Yap (formerly Sze Yap) dialects are routinely considered a branch of the Yue subfamily. This paper seeks to demonstrate that, contrary to this widespread opinion, these dialects show a wide range of distinctive features which, for formal purposes of language/dialect classification, may warrant their separation from the Yue subfamily. This paper also discusses the criteria which are often at the basis of language subgrouping in the field of Chinese linguistics. Nevertheless, this work should be regarded only as an attempt of stimulating a further discussion into a topic which has been overlooked for far too long.

Keywords: Ng Yap dialects, Yue dialects, language subgrouping, Sinitic languages, Chinese dialects.

«The chief danger to our philosophy, apart from laziness and woolliness, is scholasticism, which is treating what is vague as if it were precise and trying to fit it into an exact logical category» Frank Pulmpton Ramsey, Philosophical Papers, 1929, p. 269.

1. Introduction

The Ng Yap dialects, formerly known as Sze Yap (or Seiyap), are spoken primarily in the Guangdong and Guangxi provinces, as well as in Macao, Hong Kong and in many overseas communities. They are generally considered a branch of the larger Yue subfamily, and may be further divided geographically into two sub-branches, Xin'en ^^ and Kaihe Iff. The most prestigious and perhaps representative Ng Yap dialect1 is Hoishanese ([hDi^san^wa^l] or Toishanese).

* This paper incorporates elements of an unpublished draft presented during the 23rd International Symposium on Yue Dialects, which was held on 15th and 16th December 2018 at Jinan University (Guangzhou, PRC). In some cases, the present writer has followed the advices received; in other cases, this author has, instead, decided to follow his way, perhaps at his own peril. In general, the presentation has met with enthusiasm, though some specialists have stipulated that some revisions, which shall be illustrated and answered in the present paper, must be done.

1 In the present paper terms such as 'language' and 'dialect' are basically interchangeable. The distinction is somewhat arbitrary, and, at times, based on criteria (e.g. prestige, correctness, etc.) which only show a certain degree of linguistic unsophistication (Trask, 2007: 49-50). In China, Mandarin Chinese is the standard language, which is to say that it is the codified variety superimposed over regional dialects, and typically used in formal settings and for education. Nevertheless, it is important to stress that no such thing as modern standard Chinese (pMonghua existed, say, 200 or 300 years ago. Only recently a bunch of social and especially political

measures have combined to give rise to a variety of Mandarin, also referred to as guoyti ИНр- 'national language,' accepted as the standard language in the whole country (Norman 1988: 135-137). In other words, we may agree with Max Weinreich in believing that "a shprakh iz a dialekt mit an armey un flot."

Journal of Language Relationship • Вопросы языкового родства • 17/1 (2019) • Pp. 128-152 • © The authors, 2019

Descriptions and sources of information about the Ng Yap dialects are not abundant, even though they have a long history. One of the first mentions of a Ng Yap dialect (a variety called Llin-nen or Hsin-ning) can be found in Wells Williams's A Tonic Dictionary of the Chinese Language in the Canton Dialect, where it is stated that:

The people from the district of Sinhwui ^^ lying south-westerly from Canton, exhibit the most remarkable

peculiarities in pronunciation, and it is a puzzle to the scholars in the city how they should have originated.

(Williams, op. cit. 1856: ix).

Of much greater interest is the section Tuyan (local speeches) of the Guangdong xinyu ffi^^tn, a work written by the scholar Qu Dajun (1630-1696), where we find scraps of

information about the Xinhui dialect. For example, in this work it is mentioned that the eldest son is called dziaq55 doi55 ^^ or mbwon (?) ® in Xinhui. It is curious to note that this last form is commonly used in Southern Min (ban1) to indicate the youngest son. For instance, it was recorded in both the Lui-im Biau-go' (1800) and the Cheng-po' Lui-im iM^ft^'' (1820)2. In the former, it is assigned to the dan rime ^hM (*-an), with wen initial (*b-) and marked by an upper even tone; in the latter, it is instead described as belonging to the gan rime —Hi (*-an) with the men initial (*b-). It is absent in Douglas' Chinese-English Dictionary of the Vernacular or Spoken Language of Amoy (1873), but it can be found in the Supplement added by Rev. Thomas Barclay (1923). It is also found in Ogawa's dictionary (1908: 551-553).

Returning to Ng Yap dialects, the first real description of the language is attributable to the New Zealand Presbyterian missionary Alexander Don (1857-1934).3 In his two papers dedicated to the "Llin-nen variation of Cantonese," Don describes quite accurately the phonological system and the tonal behaviour4 of the Xinning dialect, with an eye towards comparisons with standard Cantonese, as described by Williams (1812-1884) and Parker (1849-1926). Like Williams, Don is silent on the position of Xinning and on the internal structure of the Yue subfamily in general.

Other studies have been dedicated to the study and the description of Ng Yap. Among these, we may cite Yiu (1946), Chao (1951), Cheng (1973) Him (1980), Light (1986), Lee (1987), Tong (1997), Yue-Hashimoto (2006), Kwok (2006), Takekoshi (2017), etc. Few studies, however, are concerned with the classification of the Ng Yap dialects. Nevertheless, before analysing and discussing subgrouping within the Yue subfamily, the discussion would benefit from a paragraph dedicated to subgrouping within the Sinitic family in general.

2. Dialect classification and subgrouping: what should subgrouping be based on?

First a few questions: out of a group of languages, if two or more languages are more similar to each other, can we safely hypothesise that they are languages of the same group or family? Pace anyone who would answer 'yes,' the correct answer is actually 'no.' As any good textbook in historical linguistics has demonstrated, with ample and documented examples, overall similarities by no means imply genetic relationship or affiliation, because there are many reasons why languages exhibit similar characteristics, and relationship is not necessarily

2 The best treatment of this work is Ang Uijin (Hong Wéirén) . Huiyïn miàowù yu gudài Quânzhou yïn

[The Lui-im Biau-gô ' and the sound system of the ancient Quânzhôu language].

Guôli zhôngyang tushùguân Taiwan fënguân 1996.

3 See, for instance, Don 1883 and Don 1884.

4 In the Guangdong xïnyu it is mentioned that the Xinhui dialect often merges ping tones with zè tones (Xinhui yïn duo yï pingzè xiang yi

one of them. What if two (or more) languages, very close each other, exhibit a common feature? Is it safe to suppose that those languages are related or members of the same family? Surprisingly, still not necessarily. For languages sometimes may exhibit similar or even identical characteristics not because they were inherited by from a common ancestor language, but because of parallel developments, just like Hakka resembles common Mm in that the upper and lower entering tones have switched their places (i.e. the so-called lower entering tone in fact is higher in pitch than the upper entering tone), but this is, in all probability, a parallel development. In evolutionary biology, the former (i.e. an homologous feature) is called homology; the latter (parallel developments) homoplasy. Not differently from biology, linguistics—and it is hoped that the reader will forgive the paternalistic tone of this statement— establishes language classification only on the basis of "synapomorphies," namely on "recent" shared homologies (i.e. shared innovations), because only shared innovations may prove that a closely relation is likely to exist between two or more languages, or may successfully give us information about phylogenies5. Thus, it is up to the competent linguist to ferret out common apomorphic (derived) characters, distinguishing them from plesiomorphic (primitive) features and from parallel developments.

2.1. The subgrouping of the Sinitic family

It seems that dialect classifications within the Sinitic family have generally relied on the phonemic features of medieval Chinese (Wang 1936, cf. Wang 1996: 249, Li 1937: 1-13, Ting 1982: 258).6 According to Li Fang-kuei (1937, 1938), the treatment of medieval Chinese voiced and aspirated stops is a crucial feature for the subgrouping of Chinese dialects. Ting Pang-hsin (1982) suggested to separate "early historical features" (zaoqi lishixing de tiaojian from "late historical features" (wanqi lishixing de tiaojian ^^M^tt^fSfr), although it is not very easy to understand what is meant by these two terms, and how to distinguish them. Norman (1988: 182) has proposed a classification according to phonological, grammatical and lexical items. Lau (2002: 82) has proposed a new classification which takes into consideration a feature which was apparently overlooked by most scholars, i.e. the sound change according to which the "voiced rising tone" (zhuo shang becomes "voiced departing" (zhuo qu One of the most valuable approaches, perhaps, is Simmons (1999), but he is apparently concerned only with Hangzhou and Northern Wu in general8.

It seems to the present writer that it is necessary to remark that the phonological approaches mentioned above represent, more often than not, more an attempt to demonstrate how a given Sinitic language or group of languages have evolved, in a somewhat mechanistic

5 In biology, the scientific methodology which groups organisms on the basis of derived shared characteristics is called 'phylogenetic systematics' (also known as cladistics). The trend of grouping organisms which share derived features was apparently started by the German entomologist Willi Hennig, see Hennig 1950.

6 In the opinion of the present writer, terms such 'early Chinese' or 'medieval Chinese' are more advantageous, so long as they are supposed to indicate a Spracheinheit, intended as an abstraction of linguistic (in this case mainly phonemic) features that may have been common to a given group of speakers, at some time or other, and not a living koine.

7 For further knowledge about the sound change involving the voiced rising tone, see also Ho 1988.

8 However, it seems that a series of related problems have pushed many scholars to adopt different approaches. Among these, we may mention the computational and statistical approaches that have been used by Zhou Zhenhe SfMtl and You Rujie (1985), or by Cheng Chin-chuen (1987, 1994, 1996). Nevertheless, they suffer from serious linguistic flaws (see, e.g., Yue-Hashimoto 1991: 165), and will not be discussed in detail here. A similar, though much better, approach is suggested in Baxter 2006.

fashion, from the sound classes of medieval Chinese (if any), or an attempt to quantify how many medieval Chinese features have been preserved by modern dialects, than an analysis which focuses exclusively on those dialects and on their sound systems. This must be certainly misleading, for if the sound classes of medieval Chinese are used as a phlogiston which act as a constraint on linguistic reconstruction or classification, rather than acting as a device to broaden the horizon of possibilities, then they render the identification or reconstruction of deviant features or of different sound changes impossible a priori, as they introduce a bias toward what is frequent and regular in the sound system of rime tables. This should not be taken to imply that all the analyses which are based on medieval Chinese sound classes must be necessarily wrong, but the great confidence these scholars, or at least part of them, have towards medieval Chinese, as if it were a real language and not a diasystem which is the product of our theoretical abstractions, strikes the present writer as incredible.

In addition, it seems that the field of Chinese linguistics is also plagued with the widespread and undemonstrated belief, according to which unwritten features or deviant features must not be ancient, or must be derived from a sole common source (e.g. a given sound class of medieval Chinese). Many scholars also claim that the comparative method cannot be applied within the Sinitic family (Hirata 1988, Wang 1999, 2004, 2005), because the monosyllabic structure of the Chinese morpheme, and the extensive borrowing throughout the Chinese speaking area render infeasible its application, ignoring the fact that the comparative method, which is independent of "lexical typology," is exactly a tool for eliminating chance resemblance, universals, and borrowings as plausible causes for cross-linguistic similarity.

All these prejudices may be understandable in view of the inevitable training to which the general Chinese historical linguist, including the present writer, is routinely submitted, but we must realise, once and for all, that we are in a more advantageous position, and thus we do not need to justify or accept our imprecise—and at times even grotesque—terminology/approach by tracing them back to the Ming (1368-1644) and Qing (1644-1912) philological traditions9.

Unfortunately, many scholars who have recognised the limits of the philological-phonological approach, instead of working within the framework of widely accepted, recommended practices of historical linguistics, have preferred to resort to "mutual intelligibility" as the main criterion for subgrouping, since structural methodologies based on phonological characteristics have been considered "too complex" (Tang 2017: 553). However, this writer is inclined to question mutual intelligibility as a criterion for language subgrouping, otherwise Spanish might well be considered a dialect of Italian, while Bergamasque should be regarded, instead, as a separate, distinct Romance language. In other words, shared innovations, instead of shared retentions (i.e. the retention of features from a common source, in this case the sound system of medieval Chinese) and mutual intelligibility, should be the principal criterion for subgrouping.

2.2. The subgrouping of Yue

The first attempt of subgrouping within the Yue subfamily has been made by Zhan Bohui (1981). Nevertheless, it seems that this attempt was prevalently aimed at validating an earlier

9 It seems, however, that younger scholars are actively applying the 'comparative method' to gain further information about the morphophonological system of the various Chinese dialects. See, for instance, Chen Ruiqing (2018), Wu Rui-wen (2014), Zhang Jingfen (2013), etc. However, they are not concerned with subgrouping, and, in fact, are not entirely liberated from the post-Karlgrenian tradition which they apparently wish to reject. This, of course, holds true in part also for the present paper, but if the aim of this paper is to convince its critics, then it should use arguments that most of them would accept.

proposal by Yuan Jiahua (1960). More worth discussing is the classification of the various Yue languages spoken mainly in the Guangxi province by Yang Huandian (1985). Yang has recognised four branches within Guangxi Yueyu, viz. Guangfu (Wuzhou, Cangwu, Hexian, Dan-zhu, Da'an), Yongxun (Nanning, Yangzhou, Yongning, Chongzuo, Ningming, Hengxian, Guiping, Pingnan, etc.), Goulou (Yulin, Beiliu, Rongxian, Lingxi, Tengxian, Mengshan), and Qinlian (Qinzhou, Hepu, Lianzhou, Lingshan).

Xiong Zhenghui (1987) represents the first attempt of subgrouping based on linguistic features. In his view, the most important feature was the treatment of medieval Chinese 'entirely muddy' initials (quan zhuo yin ^^W). He realised that medieval Chinese voiced initials had, in part, become voiceless aspirated; in part, they had become tenuis. To the former belongs the Wuhua branch, which comprises the dialects of Wuzhou, Huazhou and Zhanjiang; to the latter belongs the Goulou branch, which includes the dialects of Sihui, Guangning, Deqing, Luoding, Yunan, Fengkai, Huaiji, Yangshan and Lianshan. Other three branches were recognised by Xiong on the basis of certain phonological features. For example, they all agree in having aspiration only with level and rising tones, not with entering and departing. Siyi (Sze Yap) dialects (Taishan, Enping, Heshan, Xinhui, Jiangmen, Doumen, Kaiping) agree in having [h] for traditional tou initials MM /*th/. The Gaoyang branch, which includes the dialects of Yangjiang, Yangchun and Gaozhou, differs from Guangfu in showing a voiceless alveolar lateral fricative [4] for traditional xin initials /*s/.

A more recent, and more valuable subgrouping is provided in Yue-Hashimoto (2006). However, she relied mainly on mutual intelligibility, a criterion that the present writer is inclined to question.

If we exclude Yue-Hashimoto (2006), all other classifications have been based on both shared innovations and shared retentions, as dialects have been drawn close to each other both when they retained a feature of medieval Chinese (e.g. aspiration, voicing, etc.) or when they showed a shared innovation (e.g. when showing different phonemes instead of the expected ones, according to the sound classes of medieval Chinese). This must be only partially correct, since classification based on shared retentions is not a recommended practice in historical linguistics. Nevertheless, it is in the opinion of the present writer that these attempts are certainly valid, although better and more reliable results might be obtained by looking at those innovative features in morphology, phonemics and lexicon. As the subgrouping of the entire Yue subfamily is abundantly beyond the scope of the present paper, the analysis shall focus only on the Ng Yap branch.

2.3. In defence of the Stammbaum model

Since subgrouping often involves the internal classification of languages within a family, subfamily or group, normally represented in a family tree, it seems necessary to spend a few words defending the Stammbaum model, whose dismiss, it seems, has become fashionable in these years, especially in the field of Chinese linguistics. This writer is aware of the fact that network models are also recommended, especially when dialects are very close in space and in time, when there has been extensive borrowing between two or more languages, when the speciation of languages is never proceeded by an abrupt separation, or when each internal node of the alleged tree is constrained to represent virtually undifferentiated dialects (Ringe et al. 2002: 106). However, contrary to what many authors believe, the two methods are not mutually exclusive. Recently, network models for the subgrouping of Sinitic have been used by Zhang et al. (2018). Their approach, i.e. an admixture inference to decompose the underlining structure of the diversity of Sinitic languages based on phonemic inventories, is laudable but

does contain some peculiarities. They reject the Stammbaum theory, but feel safe to rely on migrations and on historical socio-genetic speculations. Alas, they do not provide illustrative examples and counter-examples, or make suggestions of what should be done with such language families as Sinitic where Tree models — they say — don't work. The relationship between languages/dialects and socio-genetic history is misleading: language is independent of genes, because social groups change their languages for different reasons and at different epochs. Nevertheless, Zhang et al. (2018: 4) state that Tree models are useless in Sinitic and only network models can work. This author insists that it is not fair, nor historically accurate, to call the entire Tree model theory into question on the grounds that it is based on an unrealistic concept or on an unsuitable scenario, especially when this claim is not discussed in detail, but is plagued with a partial misunderstanding of the Tree model itself, of its application, and of its finalities. Furthermore, since most of the old Sinitic linguistic territory is unknown, the Wellentheorie model cannot be applied in any meaningful way to determine dialectal relationship, at least not as Zhang et al. believe. There have been various attempts to determine how many "dialects" were spoken in Ancient China and to which modern, received languages they correspond. Some scholars even claimed to have discovered the relationship between some of the dialects of the Warring States period (475-221) and their received daughter languages (cf. Xu Wenxian 2001; Zhao Tong, 2006; Hu Haiqiong, 2012). This writer may be mistaken, but what he sees here is an amazing lack of understanding of the linguistic history of what is present-day China (but back in those times was not) and the survivorship bias. We know that, prior to Cristoforo Colombo's first voyage in 1492, there were presumably twice as many languages as there are today. It is not difficult to imagine that, projecting backwards through time, there have reasonably existed thousands and thousands of languages, many of them also on present-day Chinese territory.10 Therefore, in the absence of a clear and detailed description of the languages of the Warring States period (which is lacking), we cannot affirm which was the language of the reign/chiefdom X, Y or Z, and of which received language they were the alleged ancestors. This makes unknown most of the old linguistic territory of present-day China, cum bona pace of anyone who thinks otherwise. Hence, we are left with the Stammbaum model and the hypothesis according to which the various Sinitic groups have gradually dispersed, in a more or less unknown order (Min dialects are typically considered to have split before other dialects, rightly in this writer's opinion, but the split order of other groups is not easy to detect), from a common ancestor (medieval Chinese? Early Chinese?) which can and has been reconstructed by means of a methodology, which, with all its limits and difficulties, has been continuously refined over the years.

Both Tree and Network models are discussed in Mahe and Wang (2006). Mahe and Wang also claim that the tree model is not suitable for Sinitic, because cases of diglossia are attested since ancient times, which is true in principle but only partly so. Diglossia was attested in other parts of the world just as well as in China: Greek, for instance, was spoken in Asia Minor and Southern Italy along with many other languages. Furthermore, it is not clear why they are so sceptical towards the Stammbaum, a method which they seem to misunderstand, and yet feel safe to rely so heavily on lexical items, in spite of the extensive borrowing which has occurred throughout the Sinitic-speaking area. In addition, in Mahe and Wang 2006, trees are rooted using Old Chinese, a language never well defined but always imagined in linguistic terms drawn from later periods, and reconstructed on the basis of sources of evidence which are far from being contemporary with each other. This writer does not seek to take issue with

10 Authors such as Bickel (2014: 120, n.5) have calculated that there may have been half a million languages around 100k years ago, based on current rates of stability of languages and of language death.

the irenic spirit of their work, but it is necessary to voice one's disagreements over misconceptions in a linguistic debate, no matter how laudable one's work might be (and their work certainly is).

The Stammbaum, as a linguistic concept, is not an absolute, and it has been revised and improved many times in its long history. Yet, many critics treat it as if it were a universally-applicable method, static and immutable in time. In fact, as rightly pointed out by Rasmussen (1991: 467), the Tree model theory does not have the scope of telling us how linguistic unity came about or fell apart, but simply to inform us about the existence of unity and disunity of a group of languages. It does not postulate a past without variations, nor does it presuppose a lack of variation11. Thus, contrary to the opinion of many dialectologists, the Stammbaum theory may and does work within the framework of Sinitic (the testimony of other models, nevertheless, may yield even better results), and the huge autonomy of its individual branches plainly reflects the clear-cut splits of the Tree model, further validating both the Stammbaum model per se and its applicability within the Sinitic family. It follows that a subgrouping of the Yue subfamily or of the Ng Yap dialects based on the Tree model is perfectly reliable.

3. Peculiarities and innovations of the Ng Yap dialects

As discussed in section 2.2, it seems that there has always been, in the field of Cantonese linguistics, the general and implicit assumption that Ng Yap must not be a separate branch, despite all the striking divergences it shows.

To the best of this writer's knowledge, the general methodological procedure of stating that a language X is not part of the family Y would be: (a) list certain forms or features that are commonly diagnostic of all or most dialects of a given family; (b) show that a language does not belong to that family because it lacks most of these diagnostic forms and features, preventing it from being classified as such. Nevertheless, it seems that Yue dialects cover a wide range of diverging features, so it is difficult to know what Yue really is, and, in fact, the present writer suspects that Yue is instead a false taxon. For instance, a particularity of Yue dialects (excluding Ng Yap, Yangjiang and Yangchun) is their treatment of medieval Chinese 'entirely muddy' affricates as voiceless unaspirated consonants, and yet many exceptions can be found in Nanhai Jiujiang, Foshan, Gaoming, Sanshui, and other Yue dialects. The lower even tone merges with the lower departing tone in the dialects of Cangwu and Guiping, but merges with the upper departing tone in the dialects of Shunde and Nanhai. Given that no systematic reconstruction of proto-Yue has been done, and that we do not know whether those unusual phonemic features are retentions or innovations, this methodological procedure is probably excluded a priori, and we can only assume that Ng Yap dialects should be regarded as a distinct branch, because they do not behave like all other varieties of Yue are supposed to behave, especially in the treatment of initials and tones. Hence, we can use a sort of "apophatic taxonomy" to reach the conclusion according to which Ng Yap dialects are not Yue, by accumulating a list of features which are absent in any other variety of Yue besides Ng Yap.

The first scholar who systematically used Sze Yap data to gain further knowledge about the Yue subfamily and to reconstruct certain aspects of the sound system of proto-Yue is

11 The very concept of 'variation' is somewhat misleading. Of course, for a discipline such as dialectology, which is all about variation, this concept is inevitably maximised, but general experience tells us that "variation is generally short-lived and territorially restricted" (Rasmussen 1991: 464).

McCoy (1966). Nevertheless, it is quite regrettable that McCoy's work showed no hint of interest in such a recommended practice of historical linguistics as internal reconstruction (rather, his reconstruction is fundamentally an assessment of the sound classes of these dialects on the basis of the ones found in rime tables), which may be extremely useful for exploiting many important data available since the end of the nineteenth century, when the systematic study of Ng Yap took place under the hands of Alexander Don (1883). It is likewise regrettable, in the opinion of the present writer, that McCoy (and others) did not show any knowledge of (or interest for) lexicostatistics, a method which is certainly not a recommended practice (with notable exceptions) in historical linguistics, but that, with all its uncertainties and shortcomings, could still be as good a way for dealing with data which are not easily verifiable.12

Generally speaking, the Ng Yap dialects show a wide range of features and innovations which are not shared by standard Cantonese and other Yue dialects. Apart from the set of correspondences argued in the previous sections, Hoishanese, Enping, etc. have prenasalised stops /mb/ /md/ /mg/. This peculiar feature appears to be a recent innovation, since there is no trace of it in Don's article. Prenasalised stops are also found elsewhere within the Yue subfamily, though they show some differences13. But Ng Yap, especially Hoishanese, also shows another type of nasalisation, which was recorded by Don, and that still occurs today, especially in those words which, according to traditional terminology, are classified as yi initials (yimu

In the version described by Don (1883), 'word' and 'play' are homophonous, while today we can observe that assimilation of the precedent velar nasal has occurred. This may lead us to talk about some of the many lenitions and fortitions that occurred in Hoishanese. Many words that originally had a velar nasal in initial position have changed their initial into a plain voiced velar plosive /g/, followed by the formation of a vocalic diphthong. Other varieties of Ng Yap, such as the Heshan dialect, show a lenition of the bilabial plosive /p/ to voiced labiodental

12 This is not to be taken to imply that the present writer recommends lexicostatistics and glottochronology, two methodologies of dating which are in fact partly independent of each other, as a way to establish genetic relationship. It is true that most linguists reject lexicostatistics and glottochronology (Campbell & Poser, 2008: 303 footnote), but to interpret them as two tools for establishing genetic relationship is neither fair to those scholars, such as Rafinesque and Broca, who contributed to invent them, nor historically accurate. Both lexicostatistics and glottochronology do not involve questions of proof, but can be pursued entirely within the frameworks of accepted linguistic families. In this specific case, they might be a useful tool to help establish a certain course for the study of the Yue family as a whole, and not to establish whether Ng Yap dialects do or do not belong to the Yue subfamily. Lexicostatistics has been clearly remodelled on the basis of radiometric dating, a scientific method which has proved successful in other fields of science. However, just like radiometric dating works only on rocks which cool from a liquid melt, such as basalts or granites, both of which solidify from lava, and not on fossils which were formed from dumped sediments, in the same way both lexicostatistics and glottochronology are not universally applicable practices. But to deny that they may have their advantages is a violence to the history of these methods and to all the serious practitioners who have helped to create, develop and adjust these techniques. For a lexicostatistical attempt to estimate the time depths of five major Sinitic languages, see Wang Yude 1960; however, see Matisoff 2000 for a rejection of glottochronology in Sino-Tibetan linguistics in general.

13 For this reason, scholars such as Ting Pang-hsin & Zhang Shuangqing (2002: 207), or Liu Xinzhong (2010, personal communication) distinguish bi guan seyin (nasalised stops) from hou se biyin ^kMM^ (prenasa-lised stops). Although the present writer does not agree wholeheartedly with this terminology, he thinks that they are right in pointing out that a difference likely exists: in the case of Ng Yap we observe the articulation of a plosive segment which is realised with a brief period of air flow through the nasal cavity; in other cases, we observe a phonemic process where a segment, which does not involve oral closure (with consequent lowering of the velum), acquires nasalisation.

Table 1. Comparative table of Hoishanese words in traditional yi initials. Hoishanese pronunciations have been taken, with minor revisions, from Deng Jun 2006.

character English gloss Don's notation pronunciation [IPA]

H14 moon ngut ['gut32]

H15 day ngit ['git21]

word ngun ['gun21]

u play ngun [gu3n32]

m refined nga ['ga21]

* cow ngeu ['geu11]

m16 I ngoe [guD21]

m eye ngan ['gan55]

m 17 fish absent ['gui11]

^ 18 outer absent ['gai32]

fricative /v/, e.g., *pek19 > pak > viak 'one hundred.'20 Traditional pang initials jjM /*ph-/ have become /h/ in both Hèshan and Kaiping.

In most of Ng Yap dialects, the palatal approximant /j/ is an allophone of /3/, but can also be an allophone of the close front vowel /i/ when used as a glide. Similarly, /w/ can be an allophone of the vowel /u/. The palatal sibilants are allophones of their equivalent alveolar sibilants in cases such as when the first vowel of the final consonant is a vowel which, according to the Jakobson-Halle distinctive feature system (1956), either is acute and non-flat /i/ or is characterised by a low second formant /u/. Like standard Cantonese, Hoishanese and other Ng Yap dialects aspirate in the lower rising tone and in the even tone, and routinely confine occlusive initials in the lower rising tones, but unlike Cantonese they do not develop aspirate stops into fricatives (Cantonese shows instead a marked predilection for fricatives over aspirates, even though the voiceless aspirated occlusive generally remains so in lower level and rising tones). The treatment of coronal sibilant /*s/ and postalveolar sibilant /*g/ is very peculiar. These sibilant phonemes have presumably become a voiceless alveolar lateral fricative /4/, as in Welsh.21 However, this phoneme appears to be an areal feature: some linguists, such as

14 In Hoishanese the word for 'moon' can be pronounced in the shàng yang rù (high lower entering) tone ['gut32] as well as in the xià yang rù (low lower entering) tone ['gut21].

15 The word ngit 'day' in modern Cantonese can be found only in the xià rù tone, while in the Hoishanese version described by Don it could be found in shàng rù, zhong rù and xià xià rù tones as well (Don 1884: 479).

16 The 1st person pronoun can be pronounced in the yïn ping tone as [guD33] (literal reading) and [guDi33] (colloquial reading) and in the yang shàng tone as [guDi21].

17 The literal reading of 'fish' is ['gui11], while its colloquial reading is ['gui21]. 'Fish' can also be pronounced ['gui11-55] or ['gui21-55] as a result of changed tone.

18 'Outer' deserves a special mention. It is a yang qù word which is pronounced as ['gai32] (and which sometimes exhibits a changed tone phenomenon, ['gai32-55]) or as [guDi32], with assimilation of the velar nasal.

19 This "proto-Ng Yap" form has been reconstructed by the present author.

20 In Hoishanese the phoneme /v/ may have evolved from an ancient *w, see Lau 2007: 169-74. For further information about lenitions in Ng Yap dialects, see Zëng 2014: 96-104.

21 In his monumental Études (Chap. VI), the great Swedish sinologist Bernhard Karlgren doubted about the existence of this phoneme: "[l]atérale dentale, orale, sourde, le 'll' du dial. celtique du pays de Galles, par ex. dans Llewellyn, existerait selon M. A. Don (China Review, Vol. XI) dans le parler de Sin-ning du groupe Yue, comme représentant d'un ancien s, renseignement qu'il faudra vérifier." For further reading, see Karlgren 1915-1926: 270.

Li Jinfang (2002) and Mai Yun (2010), attribute the distribution of /4/ to a Kradai substratum, even though others (cf. de Sousa 2015) find this scenario problematic. This phoneme, however, is found also in Southern Gan and in other dialects of North Fujian. The change from *s to 4 is typologically unusual, but does have some parallels. Most Central Tai and Northern Tai languages show this sound change (probably via an intermediate stage of n, viz. *s > n > 4). In addition, Arapaho exhibits the extremely unexpected sound change *s > n, via an intermediate stage of 4 in the following way: Proto-Algonquian *s > 4 > l > n (Jacques 2013)22. Furthermore, a sound change such as */s/ > /4/ implies the simultaneous change of only few phonetic properties, viz. [lateral] [continuant]. Phonemes such as /*ts-/ and /*tg-/ have merged into a plain voiceless dentalveolar stop /t/, while /*t-/ and /*d-/ have completely disappeared. The aspirated dentalveolar stop /*th-/ has debuccalised into the abutting voiced segment /h/, but this sound change is also observed in Xinhui Hecun, Jiangmen, Doumen (Ng Yap) and Nanhai (non-Ng Yap, see Peng 1990). According to Zhang Weigang (1943), in the Taihe dialect of Ji-angxi both duan (*t-) and tdu initials have apparently become /h/.

Since Yue dialects exhibit only a limited number of morphological processes, in this paper much attention has been given to phonemics, because "phonemic mergers are clearly innovations" (Ringe et al. 2002: 70). If in biology two species are shown to be greatly diverse based on all those features which are controlled by anatomy, such as growth, metabolism, behaviour, etc., then mutatis mutandis two languages may be proven to be highly diverse (i.e. separated) on the grounds of those features which are controlled by phonemics, such as aspiration, tonal behaviour, etc.

3.1. Ng Yap innovations (tones)

A description of the tone classes of the Xinning dialect was already provided by Don (1883). It seems that very little, if anything, has changed since his times. The tonal behaviour of the various Ng Yap dialects resembles much more that of the Northern varieties of Chinese, than that of other Yue dialects. Like Guangfu and Gaoyang, Ng Yap dialects aspirate in the lower rising and in the even tones, and generally confine occlusive initials to the lower rising tones, but unlike other varieties of Yue, the upper even tone has not assumed a falling cadence, and the upper rising tone has not become a very high level tone. Guangfu has developed a middle tone for words in the entering tone, while in Ng Yap they are in the upper series, where they presumably belonged at an older stage.

Traditional yang tones are higher in pitch than yin tones,23 a feature rarely observable in Yue, but quite common in Hakka, Gan, Wu and Xiang dialects (Yue-Hashimoto 1988, 1991). A major innovation, observable exclusively in Ng Yap, is that the upper even tones have merged with the lower departing tones24. This feature is not observed elsewhere (see Table 2 below).

This phoneme should not be confused with the coronal lateral phoneme /1/, which instead is realised with the back of the tongue raised towards the velum.

22 For different solutions, see Picard 1994.

23 The two terms are generally considered to be two labels for 'high' and 'low' (Bauer & Benedict 1997: 121). In fact, the two terms are misnomers, because there are dialects, such as Hakka and Min, in which 'low tones' are actually higher in pitch. Thus, the two terms are in fact two impressionistic labels formerly applied to a given quality of the toneme, probably 'height,' but they should not be absolutised.

24 Zhan & Cheung (1987) observed that the mid-rising tone of the Yayao dialect (Heshan) is also high-pitched. However, it seems to the present writer that this is, in all probability, a case of changed tone (bian yin S^), as it only concerns few words (see also Yue 1991).

Hence, generally speaking, Ng Yap dialects show certain similarities in tones with other Yue dialects (especially with the Yangjiang and Yangchun dialects25), but also show distinct innovations.

3.2. Ng Yap innovations (rimes)

Concerning rimes, Ng Yap innovations are not numerous. The two most important innovations are as follows (Yue-Hashimoto 1991):

(a) traditional Grade II, III, IV rimes of the xie she have merged. They show no vowel length contrast; thus, characters such as ^ 'street' and ^ 'chicken,' which in standard Cantonese are respectively /ka:il/ and /keil/, in Ng Yap dialects such as Hoishanese are gai21-55 and gai33.

(b) xiao she Grade II rimes have merged with Grade I rimes of liu she Hence, Ng Yap dialects show no long vowel vs. short vowel contrast in words which are traditionally assigned to these categories.

Nevertheless, there are two things that we need to make clear at the outset: first, these sound changes occur also in the Yue dialects spoken at Tengxian, Enhe and Shinan; second, although it is stated that sound class X has changed into Y, we cannot be on a firm footing regarding the actual time when this change occurred. Actually, we cannot even be sure that this sound change did really occur. As stated in sections 2.1 and 3, medieval Chinese classes should not be used as a phlogiston which could provide a mechanistic explanation for ways in which sound changes have occurred in all Chinese dialects. In fact, medieval Chinese sound classes represent a North-South mixtum compositum of literary pronunciations from different epochs. Many scholars still continue to work within the framework of medieval Chinese sound classes, but it is a mere scholar convention, and medieval Chinese does not reflect a living language. In fact, it is hard to imagine that all those scholars who are mainly concerned with the assessment of medieval Chinese sound classes have never heard that "chaque mot a son histoire."

Since prevocalic glides have been considered traditionally a part of the rime,26 vocoid ap-proximants should be discussed in this section. If it is true that the four Grades of rime tables indicated the presence of certain glides, then it seems that in standard Cantonese they have disappeared; in most Ng Yap dialects, the two vocoid approximants /j/ and /w/ are clearly vowel-depending, which is to say that they occur respectively only before /e/ and /d/. This means that they are predictable phonetic onglides. In other Yue dialects which exhibit medial glides, these two vocoid approximants are not so predictable (cf. Lin Qinjuan 2008).

3.3. Ng Yap innovations (initials)

As discussed in section 3, the Ng Yap dialects show a wide range of fortitions and lenitions. Nevertheless, most of these peculiarities, such as prenasalised occlusives or the presence of a voiceless alveolar lateral fricative which contrast with standard Cantonese /s/, are widely diffused among other Yue dialects. Another peculiarity which characterises most of Ng Yap dialects is the merger between traditional kaikou (with no *-w-) and hekou (with *-w-) jian M /*k-/ and qi M /*kh-/ initials. But, again, this feature is observed also in Cangwu, Guiping, Lingxi, Rongxian, Xinyi, Yangjiang, Yangchun, Zhongshan, Zhuhai, etc.

25 Yangchun and Yangjiang dialects are very similar to each other but they do show differences: for example, the Yangchun dialect has only the yang ru tone 53, while the dialect of Yangjiang has both high upper entering (shang yang ru 54) and low upper entering (xia yang ru 43) tones. See Liu Weimin 2012: 17.

26 In historical Chinese phonology, the rime (yun M) may include a medial glide (yuntou a nucleus

(yunfu MM) and a coda (yunwei M®).

Table 2. Distribution of this feature within the Yue subfamily. Data are taken, with minor revisions, from Yue-Hashimoto (1991) and Zhan & Cneung (1987, 1988, 1990, 1994, 1998). The grey slots indicate the Ng Yap branch. Regarding the presence of this feature in the Yangjiang dialect, Yue-Hashimoto (1991: 169) includes it in her table 1, but does not include it in her further discussion of this feature. To the best of this author's knowledge, this feature is absent in Yangjiang and Yangchun, as well as in other non-Ng Yap varieties, with the sole exception being Xinjie, clearly a case of parallel development.

feature

merger of the upper even tone with the lower

departing (MS)

dialect presence/absence

Mffi Guangzhou -

^^ Macau -

Zengcheng -

Hua xian -

f^ib Conghua -

fgg Xinyi -

Nanhai jiujiang -

Hii^S Shunde daliang -

^^ Gaoyao -

Gaoming cheng -

fM Hua xian -

SfnO^ffi) Cangwu (Cangzhou) -

Yulin -

HS Shinan -

ImM Heng xian -

XM Binyang -

S^^gg Nanning pinghua -

^^ Dongguan -

^^ Bao'an -

f^ Xinjie +

Lianjiang -

^ffl Huizhou -

ItM Beihai -

^ffl Qinzhou -

^^ Zhongshan -

Zhuhai -

Jiangmen -

Xinhui cheng +

Xinhui hecun -

^^ Taishan +

^^ Kaiping +

Enping -

H^ Heshan -

^P^ffi Doumen zhen +

Yangjiang + (?)

Yangchun -

H£(SS) Luoding (Sihe) -

Guiping jiangkou -

AS Bobai -

There is, however, a feature which is exhibited only by Ng Yap dialects such as Hoishan, Kaiping and Heshan. Whereas all other dialects have t- or other consonantal initials (traditional duan initials /*t-/), Ng Yap has zero. Furthermore, if we exclude the dialect of Nanhai, and if we are forced to justify and rescue medieval Chinese consonantism, then we are advised to regard the vowel-like abutting segment h- of most Ng Yap dialects (including Doumen, Enping, Jiangmen, Xinhui) as a result of the following sound change: *th- > h-. This trend of dropping tenuis while preserving only their suprasegmental feature /*-h/ is found only in Ng Yap varieties.

3.4. Ng Yap innovations (morphology)

One of the most interesting features exhibited by Ng Yap is the absence of plural markers for expressing plurality in personal pronouns.

Table 3. Hoishanese and Cantonese personal pronouns

singular plural

person Hoishanese Cantonese Hoishanese Cantonese

romanisation IPA jyut pin romanisation IPA jyut pin

1st ngoi (S) [jpiH] ngo5 ngoi (g/M/S) foaiJJ] ngo5 dei6 (S*)

2nd ni №) [niH] nei5 niek [nrek-IJ] nei5 dei6

3rd kui (®) [khuiH] keoi5 kiek (M/fe) [khiek-IJ] keoi5 dei6

As can be seen from Table 3, plurality is expressed in Hoishanese (as well as in other Ng Yap dialects) by a change in tone. This phenomenon is observable also in the dialect of Yangji-ang, but contrary to Ng Yap, which shows an "anomaly" in the first person plural, the Yangji-ang dialect expresses 'we' with /qDkV. Long ago, Antoine Meillet (1925: 27) had already stressed the importance of "les formes anomales," therefore it is in the opinion of the present writer that this 'exception' deserves much more attention than it has received.

3.5. Ng Yap innovations (lexicon)

Lexical analysis is often overlooked in historical linguistics. Although this author agrees on the fact that lexical analysis alone is not sufficient as a criterion for subgrouping, and that the testimony of morphology and phonemics is also required, the dismissal of lexical analysis a priori is certainly exaggerated.

Ng Yap dialects show a different set of interrogative pronouns for 'who' and 'which':

Table 4. Comparison of Ng Yap interrogative pronouns. Data are taken, with some revisions, from Yue-Hashimoto (1991: 175).

pronoun Ng Yap Yángjiang Zhongshan Gaozhou Huáxián Téngxián

who sui55 /m^tl sey-IJ/ ti^ /pinl sey-IJ / pin É /metl sey-IJ/ ti^ /metl sey-IJ/ ti^ /metl sey-IJ/ ti^

which nai21 /pin-IJ kD:H/ pinffi /pinl ka:H/ pinffi /syre: nitl/ /senl tse:kH/ sen— /binl kD:H/ binffi

Hoishanese (and Ng Yap dialects in general) makes use of the morpheme hau22 ^ for 'head hair,' like Southern Gan, instead of the neutral lexeme /fa:tV The use of the demonstrative pronoun koi21 M for 'this,' instead of standard Cantonese /ni:l/ ^ (probably a Taic loanword, viz. nii < Proto-Tai *najc), may suggest that Hoishanese has either replaced the old borrowing or that it has not borrowed the demonstrative pronoun for some unclear reason.

4. Final considerations

It is interesting to note that whichever are the criteria (including mutual intelligibility) utilised to determine which languages are more closely related to one another within the Yue subfamily, Ng Yap dialects appear to be a distinct, sui generis branch. Incidentally, no scholar, to the best of this author's knowledge, has ever demonstrated that the Ng Yap dialects are effectively a branch of the Yue family: it is, therefore, left to the sceptical audience to prove otherwise, a fallacious type of argumentation (onus probandi incumbit ei qui dicet non ei qui negat). Given that the Yue family has never been classified by means of the standard methodologies and procedures recommended by historical linguistics, one cannot but wonder why the Ng Yap dialects have always been implicitly considered a branch of Yue, in spite of the wide range of distinctions they show27. Since proto-Yue has never been reconstructed, and since no scholar, not even McCoy, has ever demonstrated how Ng Yap forms are effectively later, changed versions of earlier proto-Yue forms, we cannot just claim that Ng Yap is a branch which has simply undergone more radical changes than the other varieties of Yue28.

But languages do not develop in a vacuum: they are socially, culturally and politically connected with their speakers. Thus, the question which now arises is: where do speakers of Hoishanese come from? The answer is unclear and certainly premature. In biology, when naturally selected features become so differentiated that two subsets are unable to reproduce with each other, we are forced to consider the two types to have developed into two separate species. Similarly, in linguistics, when a given language, which is spoken over any significant area, gradually differentiates and ends breaking up into rather distinct varieties, we may encounter regional dialects of that language which, given sufficient time, may become so different from one another that we are forced to regard them as separate languages. Hence, either

27 During the 23rd International Conference on Yue dialects, many specialists agreed with the present writer in thinking that Ng Yap may not be a variety of Yue, with some of them even claiming that /khey.H m-J hei-l jy:t-l jy:JV 'it is not Yue.' Others (few) completely rejected the idea, claiming that it cannot be proved that Ng Yap are not a branch of Yue, a claim which only shows an amazing lack of understanding, for scholars of such calibre, of the scope of the present paper. Those who were broadly sympathetic towards the argumentations of the present manuscript appeared to be, nonetheless, reluctant to the idea that Ng Yap may not be a variety of Yue, although they had to admit that the distinctions and the idiosyncratic features exhibited by Ng Yap dialects are not easily explainable. This writer may be mistaken, but it seems that the refutation of such a conclusion is motivated only by the desire of preserving at all costs the traditional subgrouping, which recognises the existence of Xiang, Gan, Min, Wu, Hakka and Yue as the only varieties of Southern Sinitic.

28 Perhaps, regionalism and provincialism have played a role in this game, reinforced by some lingering adumbrations of the normally involved and generally rather special political unity that has allegedly existed in many parts of the present-day Chinese territory. Moreover, empty concepts, invented in the recent years in the — nonetheless courageous and remarkable — attempt of replacing the English mistranslation of the Chinese word fangyan ^m, such as 'regionalect' (cf. DeFrancis 1984: 57) and 'topolect' (cf. Mair 1991: 7, 2008) have probably added further fuel to the fire, with the result that, paradoxically, this undemonstrated classification has hardened into an orthodoxy that none have dared to challenge.

Ng Yap once was really a branch of proto-Yue whose evolution, nonetheless, involved linguistic changes so fundamental that now they should be considered to be different groups, or is a branch of another subfamily which was radically transformed under a Yue substratum when its speakers moved to the south-western coast of the Yue speaking area. Of course, another possible answer is that all the doubts expressed in this paper are circumstantial and not worth considering, and that Ng Yap is effectively and without any hint of doubt a branch of Yue. Solution of this problem goes vastly beyond the scope of the present paper, but if this work is allowed to dwell a little longer on this topic, then one might also hypothesise that the parent language of the various Ng Yap dialects was the language historically spoken by Song soldiers who were stationed in today's Guangxi, Fujian and Guangdong provinces during the Southern Song epoch (1127-1279), possibly as a result of the loss of the capital of Hangzhou at the hands of Mongol invaders. The presence of words such as on33 yin22 ^A for 'husband's mother,' which is the courtesy form used for officials' family during the Song dynasty,29 is very suggestive, though still hardly sufficient to prove this scenario.

Be it as it may, Ng Yap vocalism resembles Hakka, its tonal behaviour is closer to the Northern varieties of Mandarin than to that of other Yue dialects, and it shows predictable phonetic onglides, unlike any other variety of Yue. Its consonantism also shows two unique features, namely the loss of tenuis and voiced dentalveolar plosives and the debuccalisation of an aspirated dentalveolar stop. The use of personal and interrogative pronouns also is much closer to Northern varieties of Mandarin than to any other Yue dialect.

The humble aim of this paper is to claim that Ng Yap shows a wide range of distinct features which may warrant its separation. Subgrouping or the placing of a given language within a family is inevitably a matter of weighing criteria on an arbitrary basis: one has to choose which features of a given language are the most important, and of course the perspective that one adopts inescapably changes the weighing that one gives. The present paper has chosen to give more emphasis to specific features of tonal behaviour, as well as to the morphological process involving the pluralisation of personal pronouns. Other phonemic features, such as the treatment of certain initial consonants and the presence of prevocalic glides, and a few lexical features, such as the use of interrogative pronouns, have also received special attention.

In concluding, although further effort is needed to strengthen the conclusions drawn in this paper,30 it is in the opinion of the present writer that any other conclusion will require a whole lot of special pleading.

References

Ang, Uijin 1996. Huiyin miaowu yu gudai Quanzhou yin (^W^M^S^^-ffiW [The Lui-im Biau-go ' and

the sound system of the ancient Quanzhou language]). Guoli zhongyang tushuguan Taiwan fenguan

Atkinson, Quentin, Russell Gray. 2006. In: Peter Forster, Colin Renfrew (eds.). How old is the Indo-European language family? Illumination or more moths to the flame? Phylogenetic methods and the prehistory of languages: 91-110. Cambridge: MacDonald Institute Press. Ball, James Dyer. 1890. The Tung-Kwun Dialect. The China Review 18(4): 284-299.

29 See the Da Song xuanhe yishi http://open-lit.com/html/lit/562/19250.html.

30 A good way might be to write down on a spreadsheet all the peculiar features exhibited by the various Yue

languages, e.g. aspiration in the upper range of tones 1, merger of upper even with upper departing tones 2, etc.,

and to put it into a computer software, such as PAUP, which uses an algorithm to create a phylogenetic tree. This

method is widely diffused in many areas of science, such as palaeontology (see, for instance, Brusatte et al. 2010).

Ball, James Dyer. 1907. Cantonese Made Easy: A book of simple sentences in the Cantonese dialect, with free and literal translations, and directions for the rendering of English grammatical forms in Chinese. Singapore: Kelly & Walsh.

Bauer, Robert S., Paul K. Benedict. 1997. Modern Cantonese phonology. Vol. 102. Berlin / New York: Walter de Gruyter.

Baxter, William H. 2006. Mandarin dialect phylogeny. Cahiers de Linguistique — Asie Orientale 35 (1): 71-114.

Bickel, Balthasar. 2014. Linguistic diversity and universals. In: N. J. Enfield, Paul Kockelman, Jack Sidnell (eds.). The Cambridge Handbook of Linguistic Anthropology: 101-124. Cambridge University Press.

Brusatte, Stephen L., Michael J. Benton, Julia B. Desojo, Max C. Langer. 2010. The higher-level phylogeny of Ar-chosauria (Tetrapoda: Diapsida). Journal of Systematic Palaeontology 8(1): 3-47.

Campbell, Lyle, William J. Poser. 2008. Language classification: history and method. Cambridge University Press.

Chao, Yuen-Ren. 1928. Studies in the Modern Wu Dialects. Tsinghua College Research Institute Monograph, 4. Beijing.

Chao, Yuen-Ren. 1943. Languages and dialects in China. The Geographical Journal 102(2): 63-66.

Chao, Yuen-Ren. 1951. Taishan Yuliao (ПШвМ4 [Corpus of Toishanese]). Bulletin of the Institute of Historical Philology (Academia Sinica) 23: 25-76.

Chen, Ruiqing . 2018. HuangkSnghua yü zaoqi Min bei yü (м^Абйй^ШКУЫо [The Huangkeng dialect and

early Northern Min]). Talk given at the The 2nd Li Fanggui Young Scholars Seminar (Institute of Languages, Academia Sinica).

Cheng, Chin-Chuan. 1982. A quantification of Chinese dialect affinity. Studies in the Linguistic Sciences 12(1): 29-47.

Cheng, Chin-Chuan. 1986. Quantifying Affinity Among Chinese Dialects. Language Learning Laboratory Technical Report LLL-T -10-86. Urbana, Illinois: University of Illinois.

Cheng, Chin-Chuan. 1987. A quantification of Chinese dialect affinity. In: Chinese Language Society of Hong Kong (ed.), Wang Li Memorial Volumes English Volume: 73-100. Hong Kong: Joint Publishing.

Cheng, Chin-Chuan. 1988. Hanyü fängyan qinshü guänxi de jiliang yanjiü [Quanti-

tative studies of Chinese dialect affinity]). Zhongguo yüwen (ФВШо^ [Studies of the Chinese Language]) 203: 87-102.

Cheng, Chin-Chuan. 1991. Quantifying affinity among Chinese dialects. Journal of Chinese Linguistics Monograph Series 3: 76-110.

Cheng, Chin-Chuan. 1992. Syllable-based dialect classification and mutual intelligibility. In: Chinese Languages and Linguistics (I): Chinese Dialects. Symposium Series of the Institute of History and Philology, Academia Sinica 2: 145177.

Cheng, Chin-Chuan. 1994. Hanyü fängyan göutöng du de jisuan [Calculation of Chinese

dialectal mutual intelligibility]). Zhongguo yüwen (ФВШо^ [Studies of the Chinese Language]) 238: 35-43.

Cheng, Chin-Chuan. 1996. Quantifying dialect mutual intelligibility. In: C.-T. James, Huang Yen-hui, Audrey Li (eds.). New horizons in Chinese linguistics: 269-292. Dordrecht: Springer.

Cheng, Teresa M. 1973. The Phonology of Taishan. Journal of Chinese Linguistics 1(2): 256-322.

D'Andrea, John A., Timothy Light. 1981. Toishan Glossary Project, an Interim Report. Journal of Chinese Linguistics 9(1): 130-137.

Defense Language Institute. 1964. Chinese-Cantonese (Toishan) Basic Course. Washington, DC: Defense Language Institute.

DeFrancis, John. 1984. The Chinese language: Fact and fantasy. Honolulu: University of Hawaii Press.

Deng, Jün (Щ^). 2006. Taishän fängyin zidian (пШ^Ш?^ [Phonetic dictionary of Hoishanese]). Changshä: Hunan jiaoyu chübanshe

Deng, Jün (ed.). 2000. Käiping fängyan (fu. yindai) [The Kaiping dialect (with an appendix)]).

Changshä: Hunan dianzi yinxiang chüban she (-ft^^^SiW^ffi^tt).

Don, Alexander. 1883. The Llin-nen variation of Cantonese. China Review 11: 236-247.

Don, Alexander. 1884. The Llin-nen variation of Cantonese. China Review 12: 474-481.

Douglas, Carstairs. 1873. Chinese-English dictionary of the vernacular or spoken language of Amoy, With the principal variations of the Chang-chew and Chin-chew dialects. Trübner [Suppl.:] Commercial Press in Shanghai.

Dunn, Michael, Robert Foley, Stephen Levinson, Ger Reesink, Angela Terrill. 2007. Statistical reasoning in the evaluation of typological diversity in Island Melanesia. Oceanic Linguistics 46(2): 388-403.

Eiichi, Chishima —). 2007. Honkon Kanton-go no kenkyü: Goi to gengo bunka o chüshin ni

^^¿вюХЛ^Ф^^ [A study of Hong Kong Cantonese from a viewpoint of vocabulary, language and culture]). Ph.D. thesis: Kanazawa Daigaku daigakuin shakai kankyö kagaku kenkyü-ka (^^^^^^l^tts [Graduate School of Social and Environmental Sciences, Kanazawa University]).

Eska, Joseph F., Donald A. Ringe. 2004. Recent work in computational linguistic phylogeny. Language 80(3): 569-582.

Evans, Steven N., Don Ringe, Tandy Warnow. 2006. Inference of divergence times as a statistical inverse problem. In: Peter Forster, Colin Renfrew (eds.). Phylogenetic methods and the prehistory of languages: 119-130. Cambridge: McDonald Institute for Archaeological Research.

Faith, Daniel P., Catherine A. Lozupone, David Nipperess, Rob Knight. 2009. The cladistic basis for the phyloge-netic diversity (PD) measure links evolutionary features to environmental gradients and supports broad applications of microbial ecology's "phylogenetic beta diversity" framework. International Journal of Molecular Sciences 10(11): 4723-4741.

Farris, James S., Arnold G. Kluge, Michael J. Eckardt. 1970. A numerical approach to phylogenetic systematics. Systematic Zoology 19(2): 172-189.

Feng, Guoqiang (//H5S) (ed.). 2012. Shaoguan shiqu Yueyu yuyin bianyi yanjiu (bSM^E^mm^SMW^ [Research on Cantonese Phonetic Variation in Shaoguan City]). Wanjuanlou chubanshe (M^Sffi^tt).

Forrest, Robert Andrew Dermod. 1948 [1965]. The Chinese Language. London: Faber & Faber.

Gan, Yu'en (^SSS), Zeng Jiansheng (MB^). 2010. Guangdong dill yuyanxue yanjiu zhi ruogan sikao (M^tftiMtoWSW^^^^®^ [Some Thoughts on the Study of Geographical Linguistics in Guangdong]). Jinan xuebao (zhexue shehui kexue ban) 3: 20-27.

Gan, Yu'en (^SSS). 2003. Siyi fangyan de xingtai biandiao (Ee^WKKSS^ [The morphological changes of the Sze Yap dialects]). In: Zhan Bohui (ed.). Di ba jie guoji Yue fangyan yantaohui lunwenji (^ASHIS-^AWW B^tX® [Proceedings of the 8th International Cantonese Dialect Symposium]): 227-250. Beijing: Zhong-guo shehui kexue chubanshe №Htt"#?4¥ffiM±).

Gan, Yu'en (^SSS). 2008. Guangdong liang Yang Yueyu yuyin tedian gaishuo [Gen-

eral description of the main charachteristics of the Yangchun and Yangjiang dialects]). Guilin shifan gaodeng zhuankexuexiao xuebao 22(2): 1-6.

Gan, Yu'en (^SSS). 2010. Guangdong Siyi fangyan yufa yanjiu (M^EeJbwSW^ [Research on the Grammar of Guangdong Siyi Dialect]). Jinan daxue chubanshe

Gray, Russell D., Quentin D. Atkinson. 2003. Language-tree divergence times support the Anatolian theory of Indo-European origin. Nature 426(6965): 435-439.

Hashimoto, Mantaro 1978. Gengo ruikei-chiri ron (W^^Mffiffit [Language type geography theory]).

Tokyo: Kobudo

Hashimoto, Mantaro 2010. The Hakka dialect: A linguistic study of its phonology, syntax and lexicon. Vol. 5.

Cambridge University Press.

He Wenhui (f^B). 2009. Yueyin pingze rumen-Yueyu zhengyin shili (he ding ben) [In-

troduction to Cantonese Tones — Cantonese Correct Sounds]). Ming chuang (^S).

Hennig, Willi. 1950. Grundzuge einer Theorie der phylogenetischen Systematik. Berlin: Deutscher Zentralverlag. Translated into English in 1966 as Phylogenetic Systematics, Annual Review of Entomology 10: 97-116.

Him, Kam Tak. 1980. Semantic-Tonal Processes in Cantonese, Taishanese, Bobai and Siamese. Journal of Chinese Linguistics 8(2): 205-240.

Ho, Dah-An (f^^). 1988. Zhuo shang gui qu yu xiandai fangyan (^iMS^S^^W [Ascription of Voiced Rising to Departing Tone and Modem Dialects]). Zhongyang yanjiuyuan lishiyuyan yanjiusuo jikan 59: 115-140.

Hou, Xingquan (f^M^). 2016. Yueyu gu quanzhuo shengmu jindu zhuoyin de leixing ji chengyin — jianlun gu quanzhuo shengmu jindu zai fangyan fenqu zhong de zuoyong (■^■gn^^^Sffi'^iH^^WtMMX^H —

[The Types and Causes of the Voiced Sounds of the Ancient Cantonese Voiced Initials — On the Role of the Ancient Voiced Initials in the Dialect Division]). Yuyan ji yuyanxue (^bSMbS) 17(5): 661-678.

Hsie, Hsin-I (§ffg—). A new method of dialect subgrouping. Journal of Chinese Linguistics 1(1): 64-89.

Hu, Haiqiong 2012. Cong chutu wenxian kan shanggu Chuyu de xingzhi

[On the nature of ancient Chu language from the unearthed literature]). Xuexi Yuekan 16: 10-16.

Huang, Jianyun (SMS) (ed.), Ye Guoquan (^H^) (rev.). 1990. Taishan fangyan (^l^W [The Taishan Dialect]). Guangzhou: Zhongshan daxue chubanshe

Jacques, Guillaume. 2013. The sound change *s- > n- in Arapaho. Folia Linguistica Historica 34(1): 43-58.

Jacques, Guillaume, Johann-Mattis List. 2018. Save the trees: Why we need tree models in linguistic reconstruction (and when we should apply them). Journal of Historical Linguistics 8(2): 128-166.

Karlgren, Bernhard. 1915-1926. Études sur la phonologie chinoise. Archives d'Etudes Orientales.

Krishnamurti, Bhadriraju, Lincoln Moses, Douglas G. Danforth. 1983. Unchanged cognates as a criterion in linguistic subgrouping. Language 59(3): 541-568.

Kwok, Bit Chee 2006. Yiduàn wéiguan de Yuèyu shï: Zhujiang sanjiâo jiang ba zhong yuèyu fangyan de

fënqun [A microscopic Cantonese history: the grouping

of eight Cantonese dialects in the Pearl River Delta]). In: Ho Dah-an (ed.) Shan gao shut châng: Ding Bangxin xiànshëng qi zhi shouqing lùnwénji (lL^/K—: T^ff^^tffsStfX® [High as mountains and long as rivers: Festschrift in honor of Ting Pang-hsin for his seventeenth birthday]): 701-739. Taibëi: Zhongyang yanjiuyuàn yuyanxué yanjiu suo

Kwok, Bit Chee, Andy C. Chin, Benjamin K. Tsou. 2016. Grammatical diversity across the Yue dialects. Journal of Chinese Linguistics 44(1): 109-152.

Lau, Chun-fat (M№#), Liang Huimin (^SS). 2011. Zhujiang sânjiâozhôu fangyan yin rùshëng shou Guâng-zhouhuà yïngxiâng de yânbiàn qingkuàng [The Cantonese

influence over the evolution of the lower entering tone in the Pearl River Delta dialects]). Yuyân yanjiu [Studies in Language and Linguistics]) 4: 92-97.

Lau, Chun-fat (M№#). 2002. Hànyu fangyan de fënlèi biaozhun yu "kèjia huà" zài hànyu fangyan fënlèi shàng de wènti [Criteria for Chinese dialect classification

and the problem of the position of the Hakka dialect in Chinese dialect grouping]). Journal of Chinese Linguistics 30(1): 82-96.

Lau, Chun-fat (M№#). 2007. Xiàndài fangyan de [j] zhuôcahuà — fùlùn Zhonggu xiému de xingchéng

[The voicing and fricativisation of [j] in the modern dialects— and the formation of the xié initial in Middle Chinese]). Zhongguo yuwén 317(2): 169-174.

Lee, Gina. 1987. A Study of Toishan F0. Ohio State University Working Papers in Linguistics 36: 16-30.

Lï, Dongxiang 2010. Yuèbëi tuhuà xiâochëng biànyin de biànyi yânjiu (^itijg/J^SWISMW^ [Re-

search on Northern Cantonese changed tones]). Jinan xuébào (zhéxué shèhui këxué bàn) (H^ÇfS 32(1): 111-116.

Li, Fang-Kuei. 1937. Languages and Dialects of China. The Chinese Yearbook. Shanghai: Commercial Press (Reprinted in Journal of Chinese Linguistics, 1:1-13).

Lï, Jïnfang 2002. Dong Tâi Yuyân yu Wénhuà [Tai-Kadai Language and Culture]). Beijing:

Minzu chubânshè (KRffiMi).

Lï, Jingzhong 1990. Yuèyu shi Hànyuzù qun zhong de duli yuyan (fillilgÎKSÈis [Can-

tonese as an independent language in the family of Chinese languages].) In: Zhan Bohui (ed.). Di èr jiè guoji Yuè fangyan yântào hui lùnwén ji (^—SHIiS^-^WisÎ^iroX® [Proceedings of the 2nd International Cantonese Dialect Symposium]): 22-29. Guângzhou: Jinan dàxué chubânshè (HSASffi^tt).

Lï, Lilin 2010. Dongguàn Yuèyu yuyin yanjiu (^^-^-gogo^i^ [Dongguan Cantonese Speech Research]).

Ph.D. Thesis. Guângzhou: Jinan University.

Light, Timothy. 1986. Toishan affixal aspects. In: John McCoy, Timothy Light (eds.). Contributions to Sino-Tibetan studies: 415-425. Leiden: Brill.

Lin, Bâisong (#fgfè). 1990. Shùndéhuà zhong de "biàn yin" (M^iSW'SW" [Changed tones in Shunde dialect].). In: Zhan Bohui (ed.). Di èr jiè guoji Yuè fangyan yantào hui lùnwén ji (^—SHIiS^-^WisÎ^iroX® [Proceedings of the 2nd International Cantonese Dialect Symposium]): 86-88. Guângzhou: Jinan dàxué chubânshè (BSASffiMi).

Lin, Qinjuan 2008. Qinzhouhuà tôngyinzihui [The homophony of the Qinzhou dialect]).

Guilin shifàn gaodëng zhuankë xuéxiào xuébào 22(1): 1-7.

Liu, Fu (Liu Bannong). 1925. Étude expérimentale sur les tons du chinois. Paris: Société d'Éditions "Les Belles Lettres."

Liu, Wëimin (Mf$K). 2012. Yangchùnhuà yu Yangjiànghuà de yuyin chayi Xiàndài

yuwén, yuyan yanjiu (Sf^toX, toWW^) 6: 16-18.

Liu, Xinzhong (Mf^), Yang Wèi (^Bf), Cui Shuhui (S^S). 2003. Guanyu hànyu fangyan fënqu de yuyan tiaojiàn (M^^S^W^EISWfS'fr [Linguistic criteria for the classification of Chinese dialects]). Yuyan yanjiu (SWi^) 3: 113-118.

Luô, Yanfa (HW#). 2013. Àomén Yuèyu yinxi de lishï biànqian ji qi chéngyin [The

Historical Changes of the Macao Cantonese Phonetic System and Its Causes]). Ph.D. Thesis: Bëijing University.

Maddieson, Ian, Sandra Ferrari Disner. 1984. Patterns of sounds. Cambridge University Press.

Mai, Yun 2010. Yueyu de Xingcheng, fazhan yu Yueyu he Pinghua de Guanxi (l^fif®'

[The formation and development of Yue, and the relationship between Yue and Pinghua]). In: Pan Wuyun Shen Zhongwei (>^§Sff) (eds.). Yanjiu zhi le — qingzhu Wang Shiyuan

xiansheng qishiwu shouchen xueshu lunwenji (W^^^K^ii^^^t+EsMSffjiiX® [The Joy of Research II—A Festschrift in Honor of Professor William S-Y. Wang on His Seventy-fifth Birthday]): 227-243. Shanghai: Shanghai jiaoyu chubanshe

Mair, Victor H. 1991. What is a Chinese "dialect/topolect"?: Reflections on Some Key Sino-English Linguistic Terms. Philadelphia: Department of Oriental Studies, University of Pennsylvania.

Mao, Yuhang (^^ff), Zhang Guozhen 2001. Guangdonghua wenben yuyin zhuanhuan xitong de sheji

yu shixian (M^BSX^toWWrR^t^Wf^ffMiS [Design and Implementation of Cantonese Text-to-speech

System]). In: Huang Changning (ed.). Ziran yuyan lijie yu jiqi fanyi--quanguo di liu jie jisuan yuyanxue lianhe

xueshu huiyi lunwenji (g^tWifijMf^Si? — [Natural Lan-

guage Understanding and Machine Translation — Proceedings of the 6th National Conference on Computational Linguistics]): 443-447. Qinghua daxue chubanshe (>f ^^^ffi^tt).

Matisoff, James A. 2000. On the uselessness of glottochronology for the subgrouping of Tibeto-Burman. In: Colin Renfrew, April McMahon, Larry Trask (eds.). Time depth in historical linguistics: 333-372. Cambridge: The McDonald Institute for Archaeological Research.

McCoy, John. 1966. Szeyap Data for a First Approximation of Proto-Cantonese. Ph.D. Thesis: Cornell University.

Marcet-Houben, Marina, Toni Gabaldon. 2011. TreeKO: a duplication-aware algorithm for the comparison of phylogenetic trees. Nucleic acids research 39(10): e66.

Minett, James W., William S.-Y. Wang. 2003. On detecting borrowing: distance-based and character-based approaches. Diachronica 20(2): 289-330.

Mueller, Laurence D., Francisco J. Ayala. 1982. Estimation and interpretation of genetic distance in empirical studies. Genetics Research 40(2): 127-137.

Nakhleh, Luay, Don Ringe, Tandy Warnow. 2005. Perfect phylogenetic networks: A new methodology for reconstructing the evolutionary history of natural languages. Language 81(2): 382-420.

Nicholls, Geoff K., Russell D. Gray. 2008. Dated ancestral trees from binary trait data and their application to the diversification of languages. Journal of the Royal Statistical Society: Series B (Statistical Methodology) 70(3): 545-566.

Nichols, Johanna, Tandy Warnow. 2008. Tutorial on computational linguistic phylogeny. Language and Linguistics Compass 2(5): 760-820.

Norman, Jerry. 1988. Chinese. Cambridge University Press.

Ogawa, Naoyoshi (/JIH®). 1931-1932. Jittai dai kotobaten (Hl^®^ [Great Japanese-Taiwanese Dictionary]). Minnanyu jingdian cishu huibian [Compilation of Southern Min Classical dictionaries]),

Vol. 5. Taibei: Taiwan zongdufu minzhengbu zongwuju xuewuke faxing (□^li^^KKoBII®

Peng, Hsin-yi. 2014. The Yue Sub-Dialect Yayao: The Process of and Phonetic Motivation for the Loss of Its Bilabial Obstruent Initials. Hanxue yanjiu 32(3): 227-255.

Peng, Xiaochuan (^/J\H[). 1990. Guangdong Nanhai (sha tou) fangyan yin xi ^WW^

[The sound system of the Nanhai dialect (Shatou)]). Fangyan (^W) 1(1990): 23-33.

Picard, Marc. 1994. Principles and methods in historical phonology: From Proto-Algonkian to Arapaho. Montreal & Kingston: McGill-Queen's University Press.

Pike, Kenneth Lee. 1943. Phonetics: a critical analysis of phonetic theory and a technique for the practical description of sounds. Ann Arbor: University of Michigan Press.

Qu, Dajun 2006. Guangdong xinyu (M^ffto). Beijing: Zhonghua shuju

Rasmussen, Jens Elmegard. 1991. Trubetzkoy's Thoughts about Indo-European: A myth come into fashion. In: H. A. Sigurdsson (ed.). Papers From the Twelfth Scandinavian Conference of Linguistics: Reykjavik, June 14-16, 1990: 345-356. Linguistic Institute, University of Iceland.

Ringe, Don, Tandy Warnow, Ann Taylor. 2002. Indo - European and computational cladistics. Transactions of the philological society 100(1): 59-129.

Robinson, David F., Leslie R. Foulds. 1981. Comparison of phylogenetic trees. Mathematical biosciences 53(1-2): 131147.

Rong, Huihua (§§¥)• 2009. Hetanghua yu Siyihua de bijiao yanjiu (^ft^^Ei^^fc^W^ [A comparative study of the Hetang dialect and the Sze Yap dialects]). Xiandai yuwen (Sf^toX) 33: 103-105.

Rong, Huihua 2010. Hetanghua xi shu wenti chütan (ff^tSffMPP®®-® [A preliminary study on the

problem regarding the affiliation of the Hetang dialect].) Jiämüsi jiaoyu xueyuan xuebao [Journal of Jiamusi Education College]) 4: 265-266.

Sagart, Laurent. 1993. Les dialectes gan. Revue bibliographique de sinologie 11: 261-262.

Sagart, Laurent. 1998. On Distinguishing Hakka and non-Hakka Dialects Journal of

Chinese linguistics 26(2): 281-302.

Sagart, Laurent. 2002. Gan, Hakka and the formation of Chinese dialects. In: Dah-an Ho (ed.). Dialect Variations in Chinese: 129-154. Academia Sinica.

Shao, Huijün (SPSS), Gan, Yu'en (^SSS). 1999. Guangdong siyi fangyan de yuyin tedian JsfilaSS [The phonetic features of the Sze Yap dialects]). Fangyan = ) 3(1999): 128-135.

Shao, Huijün (SSS). 2010. Lun Yue fangyan i, u jie yinyunmu (II-^AW i, u ^WbS® [On the medial i, u in modern Cantonese]). Jinan xuebao (zhexue shehui ke xue ban) 6: 116-122.

Simmons, Richard VanNess. 1999. Chinese dialect classification: a comparative approach to Harngjou, Old Jintarn, and Common Northern Wu. Amsterdam/Philadelphia: John Benjamins.

de Sousa, Hilario. 2015. Language contact in Nanning: Nanning Pinghua and Nanning Cantonese. In: Hilary Chappell (ed.). Diversity in Sinitic languages: 157-189. Oxford University Press.

Stöver, Ben C., Kai F. Müller. 2010. TreeGraph 2: combining and visualizing evidence from different phylogenetic analyses. BMC Bioinformatics 11(1): 7.

Takekoshi, Minako (ttS^^i). 2017. Kindai etsugo gü Tsutomu Osamu itto ji no hensen

[Changes of Grade I xiao rimes words in Modern Cantonese]). Toho-gaku-shi 46(2): 1-25.

Tan, Yutian. 2011. The origin and nature of high rising diminutive changed tone in Siyi dialect. In: Zhuo Jing-Schmidt (ed.). Proceedings of the 23rd North American Conference on Chinese Linguistics (NACCL-23): 190-207. Eugene: University of Oregon.

Tang, Chaoju, V. J. J. P. van Heuven. 2007. Mutual intelligibility and similarity of Chinese dialects: Predicting judgments from objective measures. Linguistics in the Netherlands 24(1): 223-234.

Tang, Chaoju, V. J. J. P. van Heuven. 2008. Mutual intelligibility of Chinese dialects tested functionally. Linguistics in the Netherlands 25: 145-156.

Tang, Chaoju, V. J. J. P. van Heuven. 2009. Mutual intelligibility of Chinese dialects experimentally tested. Lingua 119: 709-732.

Tang, Chaoju, V. J. J. P. van Heuven. 2015. Predicting mutual intelligibility of Chinese dialects from multiple objective linguistic distance measures. Linguistics in the Netherlands 53(2): 285-312.

Tang, Chaoju. 2017. Dialects of Chinese. In: Charles Boberg, John Nerbonne, Dominic Watt (eds.). The Handbook of Dialectology: 547-558. Wiley-Blackwell Publishers.

Tang, Yingying 2011. Shünde daliang fängyan de yüyin yanjiü (IMito^S^WWIoWW^ [Speech Research

of Shunde Daliang Dialect]). Master's Thesis, South China Normal University.

Ting, Pang-hsin (TiPf^). 1982. Hanyu fangyan qüfen de tiaojian (^goAWE^Wfft'fr [Phonological features for classification of the Chinese dialects]). Tsing Hua Journal of Chinese Studies 14((1-2): 257-273.

Ting, Pang-hsin (TiPf^), Chang Song-hing (^^U) (eds.). 2002. Minyü yanjiü ji qi yü zhoubiän fängyan de guänxi

[The study of Min and their relationship with surrounding dialects]). Hong Kong: Chinese University of Hong Kong Press.

Tong, Choi Lan 1997. Siyi fängyan shengyün chütan (Hb^WSbM^S [A preliminary study of the pho-

nology of Sze Yap dialects]). PhD diss., National Taiwan University Department of Chinese Literature.

Tong, Choi Lan 2004. Guangdong Xinhui Siqianhua de yinyün yanjiü (M^I^^^OTbS^WbSW^ [The pho-

nology of the Siqian dialect of Xinhui (Guangdong)]). PhD diss., Jinan University.

Trask, Robert Lawrence. 2007 [1999]. Language and linguistics: The key concepts. 2nd edition. London/New York: Routledge.

Wang, Li (i^), Qian Songsheng (fl^^). 1930. Taishan fangyin (^^^W [Sounds of Taishanese]). Lingnan xuebao 10(2): 67-104.

Wang, Li (i^), Qian Songsheng (Hfi^). 1950. Zhüjiang sanjiaozhou fang yin zong lun (^IH^i'H^WIIII [General description of the Pearl River Delta dialects]). Lingnan xuebao (^^^f!) 10(2): 57-66.

Wang, Li (i^). 1936. Zhongguo yinyünxue (^hwbss [Phonology of Chinese]). Beijing: Shangwu yinshüguan

Wang, Mäolin Liu Xinzhong (M^^)- 2010. Liang Yang Yueyu Yangjianghuä ji ge fuyin de yuyinxue

yanjiü [The Phonetic Study of Several Consonants in Yangjiang Dia-

lect of Liangyang Cantonese]). In: Di jiü jie Zhöngguö yuyinxue xueshu huiyi lunwenji SSffi^iSiXÄ [Proceedings of the 9th Chinese Conference on Phonetics]). Tianjin.

Wang, William S.-Y., Shen Zhongwei (>^®f$). 1992. Fangyan guanxi de jiliäng biaoshu (^WHflf^^ffs^M [Numerical representations of dialectal relationships]). Zhöngguö yüwen (iBro^) 2: 81-92.

Wang, William S. - Y., James W. Minett. 2005. Vertical and horizontal transmission in language evolution 1. Transactions of the Philological Society 103(2): 121-146.

Wang, William S.-Y. 1987. Representing languages relationships. In: Henry M. Hoenigswald, Linda F. Wiener (eds.). Biological Metaphor and Cladistic Classification. An Interdisciplinary Perspective: 243-256. Philadelphia: University of Pennsylvania Press.

Wang, William S.-Y. (ed.). 1991. Languages and dialects of China (Journal of Chinese Linguistics Monograph Series No. 3). Berkeley: Project on Linguistic Analysis.

Wang, William S.-Y. 1994. Glottochronology, lexicostatistics, and other numerical methods. In: R. E. Asher, J. M. Y. Simpson (eds.). The Encyclopedia of Language and Linguistics Vol. 3: 1445-1450. Oxford: Pergamon Press.

Wang, William S.-Y. 1996. Linguistic diversity and language relationships. In: C.-T. James, Huang Yen-hui, Audrey Li (eds.). New horizons in Chinese linguistics: 235-267. Dordrecht: Springer.

Wang, Yude. 1960. Chügoku go dai hogen no bunretsu nendai no gakuteki shitan ®

^WlSff [The lexicostatistic estimation of time depths of the five main Chinese dialects]). Gengo kenkyü (Wt§W^) 38: 33-105.

Williams, Samuel Wells. 1856. A Tonic Dictionary of the Chinese Language in the Canton Dialect. Canton: Printed at the Office of the Chinese Repository.

Wu, Rui-wen (^i®^). 2014. Cong bijiäo Minyu de guandian lun Nanchao Jiangdong fangyan qi yun de nice

[The reconstruction of the rimes of the Nanchao Jiangdong dialect]). Döngfäng yüyanxue (^^goWS) 14: 60-80.

Wu, Wei (^tt). 2016. Qinzhou Yueyu yu Guangzhou Yueyu yuyin xitong duibi yanjiü

■^■gogD^^tSithW^ [Comparative Study of Qinzhou Cantonese and Guangzhou Cantonese Phonetic System]). Guängxi jiaoyu xueyuan xuebao (MB^WSI^SfS) 1(2016): 45-47.

Xiong, Zhenghui 1987. Guangdong fangyan de qüfen [The classification of Guangdong

dialects]). Fängyan (^W) 5: 161-165.

Xiong, Zhenghui Zhang Zhenxing (®fM®). 2008. Hänyu fangyan de fenqü (Äto^W^^E [The classi-

fication of Chinese dialects]). Fängyan (^W) 30(2): 97-108.

Xu, Jiehong (tt>X^I). 2013. Guängdöng Yangshän Yueyü yüyin yanjiü Om^MI^IdId^W^ [A phonetic study of Yangshan Cantonese]). Ph.D. diss., Jinan University (Guangzhou).

Xu, Wenxiän (tt^lR). 2001. Zhanguö Chüxi duö shengfüzi yanjiü (^H^^^S^^W^ [Study of multi-reading characters of Warring States Chu]). Master's thesis, Department of Chinese Literature, Changhua Normal University.

Yang, Huändian 1990. Nanning Yueyu yuyin yanbiän qüxiäng fenxi [Analy-

sis of the Evolution of Cantonese Pronunciation in Nanning]). Jinan xuebao (zhexue shehui kexue bän)

1(1990): 79-84.

Yang, Huändian (iü), Liang Zhenshi (Sf!f±), Li Puying Liu Cünhän (MfäÄ). 1985. Guängxi de

hänyu fangyan (MB^Äto^W [The Chinese dialects of Guangxi]). Fängyan (^W) 3(1985): 181-190.

Yao, Yumin 2009. Xianggang Yueyu shängsheng biänhuä chütän: Yu yin shiyän yanjiü

roWÄMW^ [A Preliminary Study on the Change of Cantonese in Hong Kong: A Study of Speech Experiments]). Yüyan ji yüyanxue (goWHgoWS) 10(2): 269-291.

Yiu, Tung. 1996. The Tai-Shan Dialect. Ph.D. diss., Princeton University.

Yuan, Jiahua (S^J). 2001 [1960]. Hanyü fängyan gaiyao (Äto^W-fö^ [Overview of Chinese dialects]). Second Edition. Beijing: Yuwen chübanshe

Yue-Hashimoto, Anne O. (^B^). 1973. A Guide to the Tai-shan Dialect. Washington, D.C.: Educational Resources Information Center.

Yue-Hashimoto, Anne O. (^B^). 1982. The differences between the literary and the vernacular readings in the Sui-hsi dialect. Bulletin of the Institute of History and Philology, Academia Sinica 53(2): 353-366.

Yue-Hashimoto, Anne O. 1985. The Suixi dialect of Leizhou: A study of its phonological, lexical and syntactic

structure. Ng Tor-tai Chinese Language Research Centre, Institute of Chinese Studies, Chinese University of Hong Kong.

Yue-Hashimoto, Anne O. 1988. A preliminary investigation into the subclassification problem of the Yue

dialects. Computational Analyses of Asian and African Languages 30: 7-41.

Yue-Hashimoto, Anne O. 1991. Yuèyu fangyan fënqu wènti chutàn (^^^"g^EPolffi^S [A Prelimi-

nary Study on the Subgrouping of Cantonese Dialects]). Fângyân (^"g) 3: 164-181.

Yue-Hashimoto, Anne O. 1992. Guangdong Kaiping fangyan de zhongxing wénjù

AttA^ [Neutral sentence of the Kaiping dialect of Canton]). Zhôngguô yùwén (^B^J) 4: 279-286.

Yue-Hashimoto, Anne O. 1993. Comparative Chinese dialectal grammar: handbook for investigators. Paris:

Centre de Recherches linguistiques sur l'Asie orientale (École des hautes études en sciences sociales).

Yue-Hashimoto, Anne O. 2002. Development of the stop endings in the Yue dialects. In: Dah-An Ho

(ed.). Dialect Variations in Chinese, Papers from the Third International Conference on Sinology, Linguistics Section: 217-245. Taipei: Academia Sinica.

Yue-Hashimoto, Anne O. 2006. Tâishàn Dàncùn fângyân yânjiù (n^^MAsW^ [The Dancun Dialect of

Taishan]). Hong Kong: Language Information Sciences Research Centre, City University of Hong Kong.

Yue-Hashimoto, Anne O. 2015. The Yue Language. In: William S.-Y. Wang, Chaofen Sun (eds.). The Ox-

ford Handbook of Chinese Linguistics: 173-188. Oxford University Press.

Zëng, Jiànshëng (MB^). 2014. Guangdong Siyi fangyan gu qi, tou, pangmu cayinhuà M,

ASfft^b [Fricativization of kh, th, and ph initials in Sze Yap dialects].) Yuyân këxué (goA?4S) 13(1): 96-104.

Zhan, Bohui (ftffS). 1981. Xiàndài hànyu fângyân (SA^inAW [Modern Chinese Dialects]). Wuhàn: Hubëi rénmin chubânshè (>№AKffiM±).

Zhan, Bohui (ftffS), Zhang Rishëng (^0A). 1987. Zhùjiâng sânjiaozhôu fângyân ziyïn duizhào

As^WSM [Comparative pronunciation of Pearl River Delta dialects]). Guangzhou: Guangdong rénmin chubânshè (JgmAKffiMi).

Zhan, Bohui (ftffS), Cheung Yat-Shing (^0A). 1988. Zhùjiâng sânjiaozhôu fângyân cihui duizhào

Asl^ftS'M [Comparison of vocabularies of Pearl River Delta dialects]). Guangzhou: Guângdong rénmin chubân shè (JgmAKffiMi).

Zhan, Bohui (ftffS), Zhang Rishëng (^0A). 1990. Zhùjiâng sânjiaozhôu fângyân zongshù (^IH^AAWfêM [Summary of the Pearl River Delta dialects]). Guângzhou: Guângdong rénmin chubânshè (M^AKffiMA).

Zhan, Bohui (ftffS), Zhang Rishëng (^0A). 1994. Yuè bëi shixiànshi Yuè fângyân diàochâ bàogào (-^-jb+SfA ■^A^IMS^S^ [Report of fieldwork in ten counties of northern Guangdong]). Jinan dàxué chubânshè

(BS*SffiM±).

Zhan, Bohui (ftffS), Zhang Rishëng (^0A). 1998. Yuè xï shixiànshi Yuè fângyân diàochâ bàogào (^HASfA ■^A^IMS^S^ [Report of fieldwork in ten counties of western Guangdong]). Jinan dàxué chubânshè

(BS*SffiM±).

Zhang, Jingfën 2013. Mïnnân fângyân de lishï bïjiào ji qi yuyïn gounï

[The historical comparison of the Minnan dialect and its phonetic reconstruction]). Master's thesis, Bëijïng University.

Zhang, Meng-han, Pan Wuyun, Shi Yan, Li Jin. 2018. Phonemic evidence reveals interwoven evolution of Chinese dialects. Ms., available at: https://www.semanticscholar.org, accessed on May 2018.

Zhang, Wèigang 1943. Tàihé tou ding èr mu de liâng zhong tèbié biànhuà zhi yanjiu (^ftMÊ

[Special changes in tou and ding initials in the Tàihé dialect]). Zhôngshân xuébào

1(8): 72-82.

Zhào, Tong (MA). 2006. Zhànguo Chu fângyân yïnxi [The sound system of the Warring States Chu

dialect]). Bëijïng: Zhongguo xijù chubânshè (^HSMffiMA).

Zhou, Zhènhè (MfMH), You Rujié (^AiS). 1985. Hunan shëng fangyan quhuà ji qi lishï bèijïng

AsEM^S^M^^S [The classification of Hunan dialects and their historical background]). Fângyân (As) 4(1985): 257-272.

Appendix I. Developments of initial consonants in Yue dialects.

dialects 1 2 3a 3b 4 5a 5b 6a 6b 7a 7b 8 9 10 11a 11b 11c

Guangzhou (Mffl) +

Huaxian (?hE) +

Conghua №) + +

Xinyi M) + +

Gaozhou + +

Nanning baihua wet© + +

Foshan +

Nanling jiujiang + + + +

Nanhai shatan + + +

Shunde daliang + + + + +

Sanshui (HK) + +

Gaoyao + + +

Wuzhou (^ffl) + + + + +

Huaxian (» + + + + + +

Cangwu («!§) + + + + + +

Tengxian + + + + + +

Rongxian + + + + + +

Yulin + + + + + +

Shinan (ES) + + + +

Binyang (TO + + + +

Nanning pinghua + + +

Dongwän + + +

Bäo'an +

QïnzhOu (sWM) + +

ZhOngshan №m) + +

Zhühäi (« + + +

Jiangmén (OT) + +

E E Xïnhuichéng + + + +

Xïnhuihécùn + + +

Táishan (êm) + + + + +

Kaipíng (M¥) + + + + + +

Enpíng (S¥) + + +

Hèshan dim) + + + + +

Douménzhèn + + +

Yángjiang + + +

Yángchün «) + + +

Sihè (SS) + +

Guipíng + + + +

Píngnán M) + + +

Data taken, with minor revision, from Yue-Hashimoto 1991.

List of features

Number description

1 medieval Chinese voiced consonants > tenuis

2 medieval Chinese voiced consonants > voiceless aspirated

3a medieval Chinese *k- becomes palatalised, only in Grade III rimes of traditional liú shè shën shè zhën shè ^S

3b medieval Chinese *k- becomes palatalised, only in Grade III rimes of traditional liú she shen she

4 medieval Chinese jing zü ffffi (*ts-, *tsh-, *dz-, *s-, *z-) > ts-, ts-, s-; zhí zü *|;h-, *4~, and zhao zü Mffi (*tg-, *tgh-, *d^-, *g-, *z¡-) > t?-, V-,

5a medieval Chinese jing zü ffffi (*ts-, *tsh-, *dz-, *s-, *z-) > t-, t11-

5b medieval Chinese jing zü ffffi (*ts-, *tsh-, *dz-, *s-, *z-) >

6a medieval Chinese *t- > 0

6b medieval Chinese *th- > h

7a medieval Chinese *p- > b-, *t- > d-

7b medieval Chinese *p- > v-

8 medieval Chinese *ñ- > h-

9 medieval Chinese *k(w)- and *kh(w)- have merged

10 medieval Chinese *s- (in some dialects also *dz-, *z-, *?-) > i- or 9

11a medieval Chinese > q-

11b medieval Chinese > r[-

11c medieval Chinese > j-

It is interesting to note that there is no other branch, except for Ng Yap, which would exhibit features that are not shared by any other Yue dialect. Features 6a and 7a are unique distinctions of the Ng Yap branch. Features 4, 5ab, in the opinion of the present writer, are highly speculative; there is no real evidence that these phonemic mergers have really occurred.

Джорджио Орланди. Классификация диалектов нг-яп: к вопросу о принципах выделения подгрупп внутри синитической семьи

Диалекты группы нг-яп (ранее — сэй-яп) обычно считаются подветвью диалектной группы юэ. В настоящей статье предпринята попытка показать, что, несмотря на это широко распространенное мнение, диалекты нг-яп обнаруживают целый ряд дистинк-тивных особенностей, которые с точки зрения формальной языковой классификации вынуждают нас обособить их от группы юэ. В статье также обсуждаются общие критерии, используемые для диалектной классификации в современной синологии. Работа ориентирована в первую очередь на то, чтобы стимулировать дальнейшую дискуссию по данной теме, долгое время игнорировавшейся в китайской диалектологии.

Ключевые слова: диалекты нг-яп, диалекты юэ, языковая классификация, синитические языки, китайские диалекты.