Yakhontov

Sat Feb 6 16:39:20 UTC 1999

----------------------------Original message----------------------------

Larry Trask recently initiated some discussion of the Yakhontov claims
about how a modified Swadesh list can be used to test for language
relatedness (based on Starostin 1991, where Starostin refers to a written
communication from Yakhontov).   However, some of what Larry says is not
correct, as I will show.

Let's begin at the beginning.  Acc. to Starostin, Yakhontov modified the
Swadesh list and then divided it into two sublists, a 35-word one and a
65-word one.  Larry posted the 35-word one, but not the 65- word one.
There are several problems in figuring what Yakhontov actually claimed,
since Larry (and I) are getting this not from Yakhontov's work but from
Starostin's (1990) book on Altaic and Japanese. Further, there are
problems of translation which I note below, by giving the Russian original
when the English translation is problematic.  Also, Starostin nowhere give
the actual 65-word sublist  or the whole 100-word list. Rather he lists
the 35-word sublist. He then defines the 100-word Yakhontov list by saying
that Yakhontov deletes 10 of Swadesh's items and added ten new ones (all
of these changes are listed), and then defines the 65-word Yakhontov
sublist as what is left from the revised 100-word list when the 35-word
sublist is removed. However, there is a problem since one of the words
said to have been deleted is given as Russian tech', lit. 'to flow', which
is NOT in the Swadesh list.  Starostin (p.c.) tells me this was typo
for zhech' 'to burn'.  So with this in mind it is possible to
reconstruct the Yakhontov 100-word list and the 65-word sublist
(please see below).

Anyway, Larry also relates two claims supposed to have been by Yakhontov
about these lists in relation to language relatedness, one of which are:

Claim 1.
If two languages are genetically related, then the proportion of cognates
in the 35-word list will always be greater than the proportion in the
65-word list.

The other is, acc. to Larry Trask, but not in reality:

Claim 2 acc. to Larry:
"If the proportion of phonetic resemblances in the 35-word list is higher
than the proportion of phonetic resemblances in the 65-word list, then
this is evidence that the languages are related".

Actually, what Starostin (1991) has, though, is something quite different:

"But if the percentage of matches (Russian: sovpadenij) is identical [sc.
in both sublists--AMR] or the percentage of matches (Russian: sovpadenij)
the 35-word list is less than (or equal to) the proportion of phonetic
resemblances in the 65-word list, then the parallelism (Russian:
skhodstvo) between the languages is accidental (i.e., there exists an
accidental coincidence [Russian: sovpadenie] or the result of active
contacts and borrowings."

If I am not mistaken, this means that the second claim is not really
separate claim at all.  It says the same thing as claim 1, viz., that
related
languages are supposed to look a certain way, but it is not stated or
logically implied that languages that look that way must be related.

Moreover, Larry makes an even bigger mistake (or else someone again
misinformed him about what Starostin says).  Specifically, he claims that
the Yakhontov method is to look at superficial phonetic similarities.
However, Starostin (p. 25-26) clearly says that he (Starostin) is looking
at matches based on sound correspondences and all other claims of the
Altaic theory (such as morphological analysis, etymology, etc.), and
indeed relates words which LOOK nothing alike, but are cognate under the
Altaic theory, e.g., Turkic *yapur-gak : Mongolic *lab-c^in : Middle
Korean *nip(h).  Since Yakhontov had not published a detailed description
of his method, I cannot claim that this is how HE would have used it, but
this is certainly how Starostin uses it.

This goes to the heart of the whole fight about what role phonetic
similarity plays in comparative linguistics (Larry claims none at all. I
claim a subordinate but important and indeed probably crucial one).
But note that Starostin is not doing anything that Larry could object
to on this score.  He is NOT using phonetic similarity at all.

Swadesh List

1. all   2. ash(es) 3. bark   4. belly  5. big 6. bird  7. bite   8. black
9. blood 10.bone  11.breast (female) 12.burn  13.claw  14.cloud 15.cold
16.come 17.die 18.dog  19.drink  20.dry  21.ear  22.earth 23.eat  24.egg
25.eye  26.feather   27.fire 28.fish 29.fly (vb) 30.foot  31.full  32.give
33.go 34.good 35.grease  36.green 37.hair  38.hand  39.head  40.hear
41.heart  42.horn  43.hot    44.human being 45.I, me  46.kill 47.knee
48.know 49.leaf  50.lie down, recline   51.liver  52.long  53.louse 54.man
55.many 56.meat  57.moon 58.mountain 59.mouth  60.name  61.neck  62.new
63.night 64.nose  65.not  66.one 67.path 68.rain (noun) 69.red  70.root
71.round  72.sand 73.say 74.see 75.seed   76.sit 77.skin 78.sleep(vb)
79.small  80.smoke  81.stand 82.star  83.stone  84.sun  85.swim 86.tail
87.that  88.this  89.tongue  90.tooth 91.tree  92.two  93.water  94.we, us
95.what  96.white  97.who  98.woman 99.yellow 100.you (sg)

Yakhontov deletes:  all, bark, bite, burn, claw, feather,
hot, lie down, seed, we

Yakhontov adds: 101. blizkij 'close, near (adj.)' 102. daljokij 'far,
distant (adj.)', 103. tjazholyj 'heavy' 104. sol' 'salt'  105.
korotkij 'short' 106. zmeja 'snake' 107. tonkij 'thin' 108. veter 'wind'
109. cherv' 'worm' 110. god 'year'

Hence, Yakhontov 100-word list must be:

1.--  2. ash(es) 3. --  4. belly  5. big 6. bird  7. --  8. black  9.
blood 10.bone  11.breast (female) 12. -- 13.claw 101. close, near
(adj.), 14.cloud 15.cold  16.come 17.die 18.dog  19.drink  20.dry  21.ear
22.earth 23.eat  24.egg  26.--   102. far, distant (adj.) 27.fire 28.fish
29. fly  30.foot  31.full  32.give 33.go 34.good 35.grease
36.green 37. hair  38.hand  39.head  40. hear  41.heart  103. heavy
42.horn  43.--  44.human being  46.kill 47.knee 48.know 49.leaf  50.--
51.liver  52.long  53.louse 54.man  55.many 56.meat  57.moon 58.mountain
59.mouth  60.name  61.neck  62.new  63.night 64.nose  65.not  66.one
67.path 68.rain (noun) 69.red  70.root  71.round 104. salt 72.sand 73.say
74.see 75.-- 105. short 76.sit 77.skin 78.sleep(vb) 79.small  80.smoke
106. snake 81.stand 82.star  83.stone  84.sun 85. swim
86.tail  87.that 107.thin 88. this   90.tooth 91.tree  92.two
93.water  94. --  95.what
96.white  97.who 108. wind 98.woman 109 worm 99.yellow 110 year 100. you
(sg)

Yakhontov's 35-word sublist (order reflects the alphabetical order of the
Russian translations):

wind, water, louse, eye, year, give, two, know, tooth, name, stone, bone,
blood, who, moon,
new, nose, fire, one, full, horn, hand, fish, dog, sun, salt, you (sg.),
die, ear, tail, what, this, I, tongue, egg

And hence Yakhontov's 65-word sublist is:

1.--  2. ash(es) 3. --  4. belly  5. big 6. bird  7. --  8. black
11.breast (female)  13.claw 101. close, near (adj.), 14.cloud
15.cold  16.come  19.drink  20.dry   22.earth 23.eat  24.egg  26.--   102.
far, distant (adj.) 29. fly 30.foot   33.go 34.good 35.grease  36.green
37.hair   39.head  40. hear  41.heart  103. heavy  43.--  44.human being
45.I, me  46.kill 47.knee 49.leaf  50.--  51.liver  52.long  54.man
55.many 56.meat  58.mountain 59.mouth   61.neck   63.night  65.not
67.path 68.rain (noun) 69.red  70.root  71.round 72.sand 73.say 74.see
75.-- 105. short 76.sit 77.skin 78.sleep (vb) 79.small  80.smoke 106.
snake 81.stand 82.star    85.swim  87.that  107. thin  89.tongue
91.tree    94. --   96.white  98.woman 109 worm 99.yellow