CLASSIFYING THE INTERESTING CORRESPONDENCE SETS AND RECONSTRUCTING PROTO-PHONES PART I
This presentation has been revised starting at slide 11.
CONSIDER THE FOLLOWING aw=aw=aw=aw / __#Hmwk#2 aw=o=o=əw / __#Hmwk#3 ay=ay=ay=ay /__#Hmwk#2 ay=e=əy=əy /__#Hmwk#3 (99, 118, 120, 121, 212, 213, 245, 254, 301, 313)
And check these out s=s=s=s / #__ (4) and V__V (2) but not __# (Hmwk#2) s=s=h=h /__# (11) (Hmwk#3)
And these === / __# (12) Hmwk #2 k=k== / __# (13) Hmwk #3
Are these sets strong or weak?
Answer: They are strong—just look at the numbers.
Are all these sets in contrast or in complementary distribution? If they are in contrast, what would be the implication? If they are in complementary distribution, what would be the implication?
Ans: they are in contrast; they all occur in word-final position. What is the implication?
The contrasting sets must derive from separate phonemes in the protolanguage. It’s nice to have principles to guide us in our work. The next step is theoretical: you must posit two proto-phonemes and write rules deriving the data from these.
Back to correspondence sets that are in contrast === / __# (12) Hmwk #2 k=k== / __# (13) Hmwk #3
What are the proto-phones? There are various considerations. What we know so far is that they must be different, so we need two proto-phones (or phonemes if you prefer). The criteria for proto-phones were given in the textbook: they must be realistic, uniformitarian, simple, and phonetically plausible. For example: === / __#< *- k=k== / __# <*-k
=== / __#< *- k=k== / __# <*-k Prima faciae, this solution meets all four criteria. However, it must still be tested empirically against the other correspondence sets found in our data. For example, is there a set that looks like this? k=k=k=k /__# If the answer is “No” then the above solution still stands.
In fact, the answer is “No” There is no correspondence set in the Melanau data that looks like k=k=k=k /__#. Therefore, the solution stands. BUT JUST SUPPOSE: WHAT IF THERE WERE?
THIS WOULD CAUSE A PROBLEM THEN WE WOULD HAVE ANOTHER PAIR OF CORRESPONDENCE SETS IN CONTRAST, AND THEREFORE, WE WOULD HAVE TO RECONSTRUCT ANOTHER PHONE TO ACCOUNT THEM. AND THIS WOULD CAUSE A PROBLEM FOR THE ANALYSIS, NAMELY: === /__#<*- k=k== / __#< *- k=k=k=k / __# <*-k
The only way out We must reconstruct three proto-phones in contrast. We can never be sure of the exact phonetics, but we can make educated guesses based on Realism, Uniformitarianism, Simplicity and Phonetic Plausibility. For example: = = = < *- k=k=k=k < *-k k=k== < *-k 2
What is k 2 ? Does it satisfy the four reconstruction criteria? Realism: Are there languages with two velar stop positions? Uniformitarianism: Are there languages in the family with two velar stop positions? Simplicity: Is k 2 the simplest solution available? What phonetic assumptions do we make for k 2, and are they plausible?
What is k 2 ? Does it satisfy the four reconstruction criteria? Realism: Are there languages with two velar stop positions?Yes: Hindi has front and back velars. Uniformitarianism: Are there languages in the family with two velar stop positions? Yes, several Formosan languages have front and back velars. Simplicity: Is k 2 the simplest solution available? Yes. What phonetic assumptions do we make for k 2, and are they plausible? If k 2 is a back velar, no problem.
Back to Melanau This last part of the discussion has been academic, in that there is no contrast between k=k==/__# and k=k=k=k /__# in the Melanau data, hence no need to reconstruct two *k’s.
Conclusion When reconstructing protolanguages, linguists do posit “superscripted” phones to account for contrasting correspondence sets, especially in the early stages of the work, but sometimes these elements are kept for years and published as such. Just remember that theoretical work is tentative until all possibilities are tested against all the facts. And when dealing with a large language family, “all the facts” may be many years in coming.
This Ppt has dealt with strong correspondence sets in contrast LING 485/585 Winter 2009