33.134, Review: Anthropological Linguistics; General Linguistics; Historical Linguistics: Robbeets, Savelyev (2020)

[The LINGUIST List]

LINGUIST List: Vol-33-134. Mon Jan 17 2022. ISSN: 1069 - 4875.

Subject: 33.134, Review: Anthropological Linguistics; General Linguistics; Historical Linguistics: Robbeets, Savelyev (2020)

Moderator: Malgorzata E. Cavar (linguist at linguistlist.org)
Student Moderator: Billy Dickson
Managing Editor: Lauren Perkins
Team: Helen Aristar-Dry, Everett Green, Sarah Goldfinch, Nils Hjortnaes,
      Joshua Sims, Billy Dickson, Amalia Robinson, Matthew Fort
Jobs: jobs at linguistlist.org | Conferences: callconf at linguistlist.org | Pubs: pubs at linguistlist.org

Homepage: http://linguistlist.org

Please support the LL editors and operation with a donation at:
           https://funddrive.linguistlist.org/donate/

Editor for this issue: Billy Dickson <billyd at linguistlist.org>
================================================================


Date: Mon, 17 Jan 2022 11:17:30
From: Geoffrey Sampson [sampson at cantab.net]
Subject: The Oxford Guide to the Transeurasian Languages

 
Discuss this message:
http://linguistlist.org/pubs/reviews/get-review.cfm?subid=36641957


Book announced at http://linguistlist.org/issues/31/31-2664.html

EDITOR: Martine  Robbeets
EDITOR: Alexander  Savelyev
TITLE: The Oxford Guide to the Transeurasian Languages
SERIES TITLE: Oxford Guides to the World's Languages
PUBLISHER: Oxford University Press
YEAR: 2020

REVIEWER: Geoffrey Sampson, University of Sussex

SUMMARY

The term “Transeurasian languages” is novel, and this review had best begin
with an explanation of what the editors mean by it.  They use it to refer to a
large set of languages spoken over an immense range stretching from places in
eastern Europe almost all the way to the easternmost tip of Siberia; five
well-known members of the set are Turkish, Mongolian, Manchu, Korean, and
Japanese.  Broadly speaking, each of these five belongs to a different
subfamily of languages whose members are generally agreed to descend from a
common ancestor (the subfamily containing Manchu is called Tungusic); none of
the five subfamilies is known with certainty to be genetically related to any
of the others, but there are many theories asserting such relationships, some
more controversial than others.  (Some of the five subfamilies contain
numerous languages, often obscure, unwritten ones.  Japanese has as sister
languages only those of the Ryukyu islands; the sole present-day sister of
Korean is the moribund language of Cheju – alias Jeju – island, though there
is shadowy evidence for the past existence of lects on the Korean peninsula
that may have ranked as separate languages rather than dialects.)  

It is generally accepted that common features at all levels of structure run
through the languages of the five subfamilies – a few examples are:  vowel
harmony; lack of words beginning with /r/, except  for loans from other
languages; negation expressed via a negative auxiliary verb; SOV word order;
no finite subordinate clauses; kinship vocabularies which use the same word
for “younger brother” and “younger sister”.  (Not every single language in
these five subfamilies has every one of the shared features, but most of the
languages have traces, at least, of most of the features, and they are less
common among other languages.)  These editors use “Transeurasian” to codify
their belief that all five subfamilies belong to a single superfamily related
by descent – they do not deny that some similarities may be areal phenomena,
but they assert that the evidence is enough to show that others reflect
inheritance from a common proto-language.  Sceptics, on the other hand, see
all the common features as adequately explained as resulting from contact
between languages of separate ancestry, or even from chance coincidence.  At
one sceptical extreme, Sir Gerard Clauson (1956) believed that not even Turkic
and Mongolic share a common ancestor, though this has been seen by some as a
relatively safe linkage – Martine Robbeets quotes Gustaf Ramstedt (1924) as
writing that “Old Turkish and Old Mongolian were no more remote from each
other than are English and German today”.  Alexander Vovin (e.g. 2009) is a
present-day scholar who does not believe that Japanese or Korean share an
ancestor with each other or with other “Transeurasian” languages; András
Róna-Tas remarked in 1988 that if Japanese was related to those other
languages, that relationship must be more remote than the relationship between
any two Indo-European languages.

“Transeurasian” in this sense overlaps heavily with the far better-established
term “Altaic”.  These editors prefer not to use that term, mainly because it
has often referred to a narrower group of subfamilies, including Turkic and
Mongolic but only one or another subset of the other three (though the wider
sense would not be unprecedented – famously, in 1971 Roy Miller published a
book entitled ‘Japanese and the Other Altaic Languages’, linking in the
subfamily for which evidence of affiliation is thinnest).  Not all
contributors to the book reviewed here are happy with the neologism.  The new
name does not describe superfamily membership uniquely – the Indo-European
family also includes both languages of Asia and languages of Europe. 
“Indo-European” is a satisfying term because it asserts a link between a
purely Indian subfamily and purely European subfamilies, whereas the languages
spoken near both western and eastern extremes of the “Transeurasian” range all
belong to the same uncontroversial subfamily, Turkic; and Indo-European
languages extend over a large part of India as well as almost all of Europe,
whereas (leaving aside recent patterns of emigration into many continents) the
overlap of “Transeurasian” languages into Europe is really marginal – mainly
Turkish in the small area of “Turkey-in-Europe” since the conquest of
Byzantium in 1453.  “Altaic”, from the Altai mountains, does at least roughly
indicate the geographical centre of gravity of the set of languages in
question.  Debates about names are unprofitable, but readers familiar with the
name “Altaic” might be confused without this explanation of how it relates to
Robbeets & Savelyev’s term.

After the editorial introduction, the book comprises 47 chapters by a total of
55 authors or co-authors; six chapters are authored or co-authored by Martine
Robbeets.  One of its purposes is to assemble evidence for the reality of the
“Transeurasian” superfamily, but this is far from the only aim of the book,
and indeed some contributors do not believe in the superfamily, or are
explicitly agnostic.  Much of the book comprises historically-oriented
descriptions of the individual languages and of relationships within the
subfamilies.  (By “historically oriented” I mean that the book is full of
detail on structures and comparison of them across the different languages; it
pays little attention to modern sociolinguistic aspects.)  The book will
undoubtedly be consulted by many scholars concerned with one or another
language subfamily who may have no particular interest in the validity or
otherwise of the Transeurasian hypothesis.

The first two chapters of the book survey the history of the subfamilies and
the sources of evidence for pre-modern periods.  Chapters 3 to 9 set out to
establish family-tree relationships between individual languages and major
dialects within the separate subfamilies, and between the subfamilies within
the hypothetical superfamily.  These chapters (and some later chapters) make
heavy use of computational cladistics, a technique first encountered by many
linguists through writings by Don Ringe (e.g. Ringe et al. 2002) on
relationships between the branches of Indo-European.  Chapters 10 and 11
survey in detail the typological properties distinguishing the “Transeurasian”
languages from others, particularly those such as Uralic languages which are
geographically adjacent.

Chapters 12 to 27 give structural descriptions, organized on a fairly uniform
pattern, of representative languages from the five subfamilies and their main
subdivisions.  Chapters 28 to 38 compare languages across the hypothetical
superfamily with respect to successive aspects of structure, for instance
Chapter 29 deals with vowel systems and vowel harmony, Chapter 34 with
categories of verb inflexion, Chapter 38 with kinship terminologies.  Chapters
39 to 42 examine arguments for and against explanations of shared features in
terms of areal diffusion rather than inheritance; in Chapter 42 Cecil Brown
applies a technique intended to make this aspect of historical linguistics
“compatible with 21st-century norms of scientific inquiry” by replacing
impressionistic judgements with quantitative significance tests.

Finally, Chapters 43 to 47 bring disciplines other than linguistics,
particularly the archaeology of East Asian agriculture, and genomics, to bear
on “Transeurasian” prehistory and identification of the proto-Transeurasian
homeland.  Choongwon Jeong and co-authors find that DNA evidence shows
present-day Tungusic speakers “to be genetically closer to modern Koreans and
Japanese than to Han Chinese or other southern Chinese populations”, tending
to support the Transeurasian hypothesis.

Some readers may find it surprising that, if the Transeurasian hypothesis is
true, it would not be fairly obviously true.  Once a link between Sanskrit and
European languages was first postulated in the late 18th century, it was
quickly recognized as a real rather than coincidental relationship.  In
Chapter 43, Robbeets and co-authors see the five top-level Transeurasian
protolanguages as having been located in adjacent areas in and around the
Liaoning province of southern Manchuria as late as about 1000 B.C., which
seems rather recent (and hence relatively easy to study) by comparison with
the dispersal of the top-level Indo-European subfamilies (though in the
following chapter Robbeets dates the split of proto-Transeurasian into those
five languages far earlier).

But the research situation is much less favourable in the Transeurasian than
in the Indo-European case.  In the first place, classical languages of Europe
and India became written languages centuries before the Christian Era, and
they were written alphabetically.  No Transeurasian language was written so
early; Marc Miyake dates the earliest examples of continuous writing in any
language of the respective subfamilies as (all centuries A.D.):  Korean 6th
c., Mongolic 6th–7th c., Japanese 7th c., Turkic 8th c., Tungusic 12th c. 
Many Transeurasian languages were not written before the 20th century – some
have no written form now.  Furthermore, the most controversial aspects of the
Transeurasian hypothesis are the relationship between Japonic and Koreanic
subfamilies, and their relationship to the rest of the hypothetical
superfamily:  when Korean and Japanese were first written, the only writing
system available for adaptation to those languages was Chinese script, which
is not alphabetic and was developed for a language which is certainly not
“Transeurasian”, so it is often hard to know what an early Korean or Japanese
inscription is saying about the sequence of phonemes it was intended to
record.  Then, many Transeurasian speakers were nomads (some still are),
making areal-diffusion explanations of long-distance relationships between
languages more plausible than they might be as between languages of settled
European societies.  And in recent centuries pressure from Russian and Chinese
has been pushing many of the Transeurasian languages close to or all the way
to extinction.  (Not long ago, Manchu was a very significant language, being
that of the ruling caste of China after the Manchu conquest of 1644, and
considerable resources were put into trying to maintain it; but it became a
political irrelevance with the Chinese revolution of 1911 – Volker Rybatzki
tells us that by 1992 only about fifty speakers of Manchu remained, and
“nowadays probably no one able to speak the language is left”.)  Even when
plenty of written material in a Transeurasian language existed in the past, it
has not necessarily survived.  After the Mongol Yuan dynasty was conquered by
the Ming in 1368, Rybatzki says that a “rage against anything Mongolic …
resulted in the destruction of every remnant of the Mongolic culture by
Chinese people”, which included most material written in Mongolian. 
Indo-Europeanists do not always appreciate how favourable their research
situation is.

EVALUATION

This is a huge book (almost a thousand large-format pages, and accompanied by
a website offering further data), and it is chock-full of detail.  The senior
editor, Martine Robbeets, is based at the highly-regarded Max Planck
Institute, Jena, and other contributors have similarly respectable
credentials.  The book is surely destined to become a standard reference for
any scholar working on some of the areas it covers, whether or not they are
interested in the overarching Transeurasian hypothesis.  (The publisher’s
blurb claims that the book is the first major reference work on its field
since Nicholas Poppe 1965.)

No one person would be qualified to work through the book language by language
checking all the facts stated – I am certainly not.  I do have some misgivings
about the arguments deployed to support the Transeurasian hypothesis, and for
Linguist List readers who do not happen to be specialists in any of these
languages that might be the most interesting aspect of the book to examine.

Shared language features can only be evidence for a relationship between
languages (whether a relationship of shared ancestry or contact) if the
features are not equally common in other languages.  The features Martine
Robbeets mainly relies on to support the Transeurasian hypothesis are
formalized in her Chapter 10.  Some of the twenty features she lists, e.g.
vowel harmony, clearly are unusual on a world scale; but others seem much less
so.  Feature F9, for instance, is “The imperative is expressed by a bare verb
stem”.  Yet, as Robbeets herself says, imperatives “are cross-linguistically
commonly expressed by the bare root, or stem of the verb” – in which case how
can F9 be relevant to common descent?  Again, F11 is about the use of forms
similar to /mi, Ti/ to express first and second person singular pronouns.  But
forms like this are so common among Indo-European and adjacent
(non-Transeurasian) languages that Merritt Ruhlen suggested “Mitian” as an
alternative to “Nostratic” for the hypothetical superfamily which some
linguists believe to contain Indo-European as one branch!

There are also questions about how closely a feature needs to match across
languages for them to count as sharing it.  Robbeets’s chapter 36 addresses a
standard objection by sceptics, that these languages contain too few shared
items of basic vocabulary to make descent from a common ancestor plausible. 
Using the Leipzig–Jakarta hundred-item basic vocabulary list (Tadmor et al.
2010), she finds a satisfactory number of shared words obeying regular
sound-correspondences.  But item (12) is the concept “breast”, which she
claims to be represented by an etymon shared by all the five subfamilies
except Koreanic, but where /kɨkɨrǝ/, offered as the proto-Japonic version of
the etymon, actually meant “heart”; she supports this with a reference to an
(unrelated) Ryukyuan word which shifted between the meanings “breast” and
“heart”.  It is easy to agree that a word might slide between the senses
“heart” and “breast”, singular, that is the part of a man’s or woman’s body
containing the heart.  However, the forms in the other subfamilies correspond
only to /kɨkɨ/; Robbeets explains the /rǝ/ as a plural suffix.  But plural
“breasts” must surely refer to a woman’s mammary glands, and a semantic slide
between that idea and “heart” feels less convincing.  Doubtless such a slide
is possible, but can this possibility be counted as positive evidence for a
Japanese–Altaic relationship?

Then I am unclear about how much these authors are claiming to infer from
their use of computational cladistics.  As Robbeets points out, “the
‘mathemagic’ it involves is not made sufficiently transparent for classically
trained historical linguists”, and while I might not count as “classically
trained” I have only a limited understanding of the technique; it is taken for
granted rather than explained in this book, and I have not found suitable
tutorials on the Web.  (At the first point in the book where a specific
application of the technique is cited, Robbeets simply says “I applied
Bayesian phylogenetic methods using BEAST …  Through a statistical procedure,
we determined that the pseudo Dollo covarion model with relaxed clock was the
most suitable evolutionary model for our data”; she gives no explanations or
citations to clarify the technical terms.)  My – possibly mistaken –
impression is that, given the assumption that a set of languages descend from
a common ancestor, cladistics will identify a “best” tree-structure of
successive splits among proto-languages leading to the observed languages at
the leaf nodes (which is how several contributors use it), but that it does
not say anything about whether that tree-structure is “good” enough to make
genetic relationship likely; and furthermore, since the technique treats a
language simply as a set of binary properties (“a sequence of 1s and 0s”, as
Robbeets puts it), it does not seem that it could say anything about whether
shared properties derive from common ancestry rather than areal diffusion.  If
this is wrong and the technique can yield inferences on one or both of these
latter issues (and hence support the Transeurasian hypothesis), that is not
spelled out clearly in this book.  And indeed it is not quite clear whether
the contributors are claiming to use the technique for those purposes – but
repeatedly one finds remarks that suggest they are.  For instance, Nataliia
Hübler sums up her cladistic analysis applied to 226 binary properties of 38
“Transeurasian” languages by saying “The internal structure of the
Transeurasian unity achieved in this study goes in line with the proposal of
Robbeets & Bouckaert (2018) on the Transeurasian tree” – the words “unity
achieved” sound like a claim that the cladistic analysis supports the
Transeurasian hypothesis, rather than merely showing what the likely internal
structure of the superfamily would be if one assumes that it is one.  And
Robbeets says that Hübler “shows that the concrete shape of the tree [produced
by Hübler’s application of cladistics] is more suggestive of inheritance than
of areal influence”, but I cannot see which passage in Hübler’s chapter
relates to this (possibly because it is veiled by unfamiliar technical
terminology).  I wonder whether readers who share my ignorance of
computational cladistics may take the book to have demonstrated more than it
can, and I suspect that the soundest arguments it contains for the
Transeurasian hypothesis could be those couched in terms familiar to
“classically trained” linguists.

A different issue with this book relates to audience.  The book ought to
appeal to historical linguists working on any of the language subfamilies it
covers, few of whom will be expert on all the other Transeurasian subfamilies.
 Thus contributions should not require readers to be familiar with technical
concepts and terminology relevant only to one language or subfamily.  This
principle is not always observed.  For instance, Chapter 1 is Miyake’s chapter
on “Historical sources of Japonic and Koreanic” – since the chapter is placed
first, it will be influential in potential readers’ decisions on whether the
book is for them.  Comparing early written Korean and Japanese, Miyake says (I
correct a misprinted example number) that  “ ソ, a ‘kugyŏl’ abbreviation of the
‘idu’ semantogram 爲 <ho> ‘to be, do’ seen in (7), was a ‘katakana’
abbreviation of the ‘man’yōgana’ 曾 <so>.”  Miyake’s point here is that both
Korean and Japanese scribes formed phonographic signs representing syllables
by taking small distinctive-looking parts of complete Chinese graphs, but this
sometimes led to the same sign having different syllabic values in the two
languages.  ‘Idu’ was an earlier method of writing Korean with complete
Chinese graphs, representing roots by Chinese translation-equivalents, and
Korean grammatical inflexions (which have no equivalent in Chinese) by Chinese
words with broadly relevant meanings; in ‘idu’ a verb-forming Korean suffix
/ho/ could be written with the Chinese graph 爲 meaning “do”.  In practice much
documentation in Korea was in Chinese, which has a different word-order from
Korean; ‘kugyŏl’ was a method of clarifying Chinese sentence-structure for
Korean readers by adding small Chinese graphs to mark the grammatical function
of the Chinese words written full-size, and when 爲 was used in this way it was
reduced to ソ.  ‘Man’yōgana’ was an early method of writing Japanese comparable
to ‘idu’ for Korean, in which the syllable /so/ could be written as 曾 (Chinese
“past”); ‘katakana’ was (and is) a Japanese syllabic script formed from
elements of Chinese graphs, and when 曾 was abbreviated for this purpose the
reduced form was again ソ, so the same sign came to represent /ho/ in Korean
but /so/ in Japanese.  (It may not be easy to see ソ as an abbreviation of the
respective Chinese graphs as they appear in modern printing fonts, but the
sign is a natural reduction of either graph as they are handwritten.)

Miyake’s glosses of ‘idu’ and the other technical concepts here are so cursory
that I believe someone unfamiliar with these languages – a specialist in
Turkish, say – might find the quoted passage more or less impenetrable and be
put off the book altogether.  This would be a pity, particularly since the
passage is actually fairly peripheral to the topic of Miyake’s chapter (but if
you don’t understand it you won’t know that).

The book contains a number of inconsistencies.  I do not underestimate the
large editorial problems that must be involved in co-ordinating contributions
by so many separate authors, but the most noticeable anomalies occur in
material for which Martine Robbeets herself was responsible.  On p. 37 she
refers to blue and green elements in Fig. 3.7, but although the book contains
many colour plates Fig. 3.7 is printed in black and white.  The opening
paragraph of the introduction to the book tells us that the Transeurasian
languages stretch “from the Pacific in the East to the Baltic, the Black Sea,
and the Mediterranean in the West”, which comes as a surprise since the
previous page contains a large and detailed map (Fig. 1) of the “Distribution
of the Transeurasian languages” that shows none of them anywhere near the
Baltic.  (Turkish accounts for the Black Sea and Mediterranean.)  This puzzle
is partly solved by a later map, Fig. 3.1, also labelled “The distribution of
the Transeurasian languages”, which differs considerably from Fig. 1 and does
show a Transeurasian language near the south-east shore of the Baltic; however
this language is identified only as “.UP”, a member of the family “7XUNLF”. 
The passage on p. 31 which discusses Fig. 3.1 says that its abbreviations are
explained in the list of abbreviations in the prelims, but they are not. 
(Comparing the two maps, it is clear that “7XUNLF” is Turkic.  I believe “.UP”
must be Lithuanian Karaim, the Turkic language of a small Jewish community
which originated in the Crimea and now live in Trakai, west of Vilnius;
according to Éva Csató & Lars Johanson in chapter 23, fewer than twenty of
them still speak it.)

Some minor errors I spotted are:  p. 90, note to Fig. 6.2, “centuries before
the present” should be “years before the present”; p. 156, Fig. 11.3, the
legend defines the meaning of two shades of grey in the Figure, but it uses
three shades; p. 443 says that “In contrast to what is found in Turkic” Sakha
(Yakut) and Dolgan express the concept “have” in a certain way, yet the
chapter identifies Sakha and Dolgan as Turkic languages; p. 694, first
paragraph of sec. 38.3.2, “do not make an age distinction” should be “do not
make a sex distinction”.

This ‘Oxford Guide’ is a valuable, comprehensive historically-oriented survey
of the languages and language subfamilies it covers. About the extended
argument for the Transeurasian hypothesis which is grafted onto the survey I
feel less sure. I wonder how many readers whose prior position was agnostic or
sceptical will find themselves shifted to a significantly different posterior
assessment of the hypothesis.

The wisest view may be that expressed in chapter 41 by Edward Vajda, who sees
it as unhealthy for linguists to think that “proving a genetic relationship
between languages represents some sort of pinnacle of achievement …, with
evidence of borrowing or other forms of language contact often treated as
chaff to be winnowed away”. Modern historical linguists, Vajda urges, should
be interested in all aspects of language history.

REFERENCES

Clauson, G.  1956.  “The case against the Altaic theory”.  Central Asiatic
Journal 2.181–7.

Miller, R.  1971.  Japanese and the Other Altaic Languages.  University of
Chicago Press.

Poppe, N.  1965.  Introduction to Altaic Linguistics.  Harrassowitz
(Wiesbaden).

Ramstedt, G.J.  1924.  “A comparison of the Altaic languages with Japanese”. 
Transactions of the Asiatic Society of Japan, 2nd ser. 7.41–54.

Ringe, D., et al.  2002.  “Indo-European and computational cladistics”. 
Transactions of the Philological Society 100.59–129.

Robbeets, M.& R. Bouckaert.  2018.  “Bayesian phylolinguistics reveals the
internal structure of the Transeurasian family”.  Journal of Language
Evolution 3.145–62.

Róna-Tas, A.  1988.  “Turkic influence on the Uralic languages”.  In D. Sinor,
ed., The Uralic Languages.  Brill (Leiden).  [NB I have not seen this item; I
infer it to be the probable location of the view attributed to Róna-Tas but
not fully referenced in the Wikipedia article “Altaic languages”, accessed 17
Oct. 2020.]

Tadmor, U., et al.  2010.  “Borrowability and the notion of basic vocabulary”.
 Diachronica 27.226–46.

Vovin, A.  2009.  “Japanese, Korean, and other ‘non-Altaic’ languages”. 
Central Asiatic Journal 53.71–132.


ABOUT THE REVIEWER

Geoffrey Sampson graduated in Chinese from Cambridge University in 1965. After
several years studying linguistics and computing as a graduate student at
Yale, he began his career as a fellow of Queen's College, Oxford University.
He went on to teach linguistics at the London School of Economics, at
Lancaster University (where he was a member of Geoffrey Leech's corpus
linguistics group), and Leeds University, with sabbatical periods at Swiss and
South African universities and British Telecom; he later moved to Sussex
University to teach informatics. After reaching retirement age in 2009 he
spent several years as a research fellow in South Africa. Sampson has
published books and articles on most areas of linguistics, on computer
science, and on several other subjects; his most recent book is ''Voices from
Early China: the 'Odes' Demystified'' (2020).





------------------------------------------------------------------------------

***************************    LINGUIST List Support    ***************************
 The 2020 Fund Drive is under way! Please visit https://funddrive.linguistlist.org
  to find out how to donate and check how your university, country or discipline
     ranks in the fund drive challenges. Or go directly to the donation site:
                   https://crowdfunding.iu.edu/the-linguist-list

                        Let's make this a short fund drive!
                Please feel free to share the link to our campaign:
                    https://funddrive.linguistlist.org/donate/
 


----------------------------------------------------------
LINGUIST List: Vol-33-134	
----------------------------------------------------------