Transcript Reconstruction_3_CD_Corr_Sets

CLASSIFYING THE INTERESTING
CORRESPONDENCE SETS AND DOING
SOMETHING ABOUT THEM
PART II
The correspondence sets should be arranged
to show either contrast or complementary
distribution.
Contrasting sets look like this:
k=k=Ɂ=Ɂ /__#
Ɂ=Ɂ=Ɂ=Ɂ / __#
An apparent contrast:
a=a=i=a /CVC__(C)#
a=a=a=a /CVC__(C)#
Adam’s observation
His data set covered all the words ending with a simple vowel.
The a=a=i=a correspondence set is peculiar, because the /i/ always ends
the word, whereas the /a/s never do—they are regularly closed by a
velar nasal, as the next display illustrates.
# GLOSS BELAWI
MATUDARO
DALAT
KANOWIT CORR
181 bone
tulaŋ
tulaŋ
tuli
tulaŋ
a=a=i=a
190 wing
kə-payaŋ
kə-payaŋ
lə-payi l-
n.c.
a=a=i=4
215 spider
bəlagaŋ
332 look for
piñaŋ
150 star
bitaŋ
bəlagi
a=2=i=4
piñaŋ
piñi
a=a=i=4
bintaŋ -nt-
biti
a=a=i=4
What is the implication of the
observed regularity of patterning?
a=a=i=a /CVC__(C)# < aŋ=aŋ=i=aŋ /__#
a=a=a=a /CVC__C#
:where C# is not ŋ#
# GLOSS BELAWI
MATUDARO
DALAT
KANOWIT CORR
181 bone
tulaŋ
tulaŋ
tuli
tulaŋ
aŋ=aŋ=i=aŋ
190 wing
kə-payaŋ
kə-payaŋ
lə-payi l-
n.c.
aŋ=aŋ=i=4
215 spider
bəlagaŋ
332 look for
piñaŋ
150 star
bitaŋ
bəlagi
aŋ=2=i=4
piñaŋ
piñi
aŋ=aŋ=i=4
bintaŋ -nt-
biti
aŋ=aŋ=i=4
Crowley says on p. 103
“What you must do is look for
evidence of complimentary
distribution before you do
your final reconstruction.”
Talk until you understand.
Alexander Graham Bell’s diagram for a
telephone, 1876.
(go to next slide)
Some more correspondence sets in
complementary distribution should have
been generated by the homework
assignments distributed so far. But not
everyone will be so blessed.
THE MIKE IS OPEN !
To be continued ...
LING 485/585
Winter 2009
What to do with complementary
distribution and contrast
Take our earlier sets, and assume contrast.
k=k=Ɂ=Ɂ /__#
Ɂ=Ɂ=Ɂ=Ɂ / __#
An apparent contrast:
a=a=i=a /CVC__(C)#
a=a=a=a /CVC__(C)#
You must reconstruct two proto-phonemes. Call them *k and *Ɂ. You must
write rules for each of the four dialects, e.g. *k > /__#in B & M-D; *k > Ɂ/__#
in D & K; and finally, * Ɂ > Ɂ/__# in B, M-D, D and K.
Assuming contrast, you must reconstruct two proto-phonemes. What
proto-phonemes? a=a=a=a surely must reflect *a. So a=a=i=a must be
*a2. If you are a uncomfortable with that, dig deeper. If later you find
CD then *a2 will dissolve into *a. But if the contrast holds up, then *a2
may be further postulated to be *[æ], *[ᴧ] ,*[ɑ] or whatever.
Proto-phonemes imply rules,
and some rules imply changes.
Continuing to assume contrast re: a=a=i=a and a=a=a=a,
we have reconstructed *a to account for for a=a=a=a
and *a2 to account for a=a=i=a. This implies that rules
(often called diachronic correspondences) are needed
to derive the dialect data from the proto-language.
*a > a=a=a=a
*a2 > a=a=i=a
B
a
M-D D
a
a
K
a
B
a
M-D D
a
i
K
a
Question: Why *a2? Why not reconstruct
*i to account for a=a=i=a?
Why not reconstruct *i to account for
a=a=i=a?
 Because Proto-Melanau *i is already taken.
 Remember we are reconstructing a language,
constrained by Realism, meaning our protolanguage
must have properties of a real language. A real
language does not have two phonemes *i.
Question: Then why not reconstruct *i2 to
account for a=a=i=a?
Why not reconstruct *i2 to account for
a=a=i=a?
 No problem, formally considered. However, let us consider the
consequences.
 What is simpler: to derive a=a=i=a from *i2 or from *a2?
 What is meant by “simpler” in this context?
Which is simpler?
 *a2 requires one change: *a2 > i in Dalat.
 *i2 requires three changes: *i2 > a in B, M-D and K
Is simplicity an absolute value?
Absolutely not! But ...
 A scientist will not choose the more complicated
solution over a simpler one, all things being equal.
 To overturn a simpler solution, strong evidence is
needed.
 Einstein: A scientific theory should be as simple as
possible, but not simpler.
Ergo
*a2 is formally better than *i2 because it is
simpler.
A scientist will stop there unless some new
evidence appears that forces them to reopen the investigation.
But of course this whole line was
an exercise in methods, using a
counter-factual assumption.
Not only did we feel uncomfortable with *a2 (and *i2), we knew from Adam’s
observation that there was complementary distribution, at least in the case of
aŋ=aŋ=i=aŋ/__# vs. aC=aC=i=aC/__#, where C=not ŋ. That observation led
us earlier to posit a more satisfying explanation.
This is what CD looks like
*a > a=a=a=a/__C# where C is not *ŋ
*a > a=a=i=a /__ŋ#
alternatively (and with perhaps more clarity):
*a > aC=aC=aC=aC/__# where C is not *ŋ
*a > aŋ=aŋ=iŋ=aŋ /__#
Question: Where does the
environment belong, with the
dialects or the protolanguage?
Is the environment in the dialects or
is it part of the proto-language?
Proto-language.
*a > aC=aC=aC=aC/__# where C is not *ŋ
*a > aŋ=aŋ=iŋ=aŋ /__#
Ø
*a > aŋ=aŋ=i =aŋ /__#
Another rule is needed to
account for Dalat.
STUDY THIS CHART UNTIL YOU UNDERSTAND THE TWO CHANGES IN DALAT
# GLOSS BELAWI
MATUDARO
DALAT
KANOWIT CORR
181 bone
tulaŋ
tulaŋ
tuli
tulaŋ
aŋ=aŋ=i=aŋ
190 wing
kə-payaŋ
kə-payaŋ
lə-payi l-
n.c.
aŋ=aŋ=i=4
215 spider
bəlagaŋ
332 look for
piñaŋ
150 star
bitaŋ
bəlagi
aŋ=2=i=4
piñaŋ
piñi
aŋ=aŋ=i=4
bintaŋ -nt-
biti
aŋ=aŋ=i=4
However, there is more to a=a=i=a, as
many of you have observed.
A Valentine’s Teddy goes to the student who figures
out the next part of the problem.
I’ll come up with another prize for the student who
finds hidden implications that follow from the next
part .
Sufficient unto the day is the
evil thereof.
--Matthew 6:34
LING 485/585
Winter 2009