max1461

@siberian-khatru-72 to revive an old discussion on neogrammarianism: evidently the trap-bath split was caused by partial progression of a sound change through the lexicon? Is there an alternate explanation for it?

siberian-khatru-72

I agree with what @esoanem​ says in comments. 

What is actually interesting is which words do or do not undergo the split. Wikipedia has a list of such words for RP. Even a cursory look at the list makes clear that the words with /ɑː/ are everyday colloquial words, whereas words with /æ/ belong to more educated and less basic lexicon.

E.g., before /-f/ we have

with /ɑː/:  calf, chaff, giraffe, half, laugh, staff

with /æ/:  Aphrodite, chiffchaff, Daphne, gaff(e), graphic, mafia, scaffold(ing), staph

The neogrammarian explanation would be that words with /æ/ in this group are borrowed from the more conservative variety which did not have the split. (Note that this need not be another "dialect", here this may rather be a different sociolect.) This actually explains the semantic difference between the words: more basic lexicon is less likely to be borrowed. 

(This does not work only for positions before /-l/, /-ʃ/ and /-z/, where it seems that the regular development was the retention of /æ/, and cases with /ɑː/ may be due to hypercorrection or borrowing from other varieties with different distribution.)

The proponents of lexical diffusion would perhaps say that more frequently occurring words are more likely to undergo the change, but this runs counter to the real case where change happens in one word after another - that of analogical change. Analogy (even in neogrammarian view) actually works on principles of lexical diffusion: it spreads from one word to another in the lexicon. But we know that analogical changes are less likely to happen in more frequent words! (See, for example, the paradigm of the verb ‘to be’ in various IE languages.) I think that this proves the correctness of the neogrammarian explanation.  

max1461

I haven't looked at the list myself, but assuming this is a pattern, then it relates to the point I was making here. The question is: is it generalizable? If "exceptions to sound changes are more likely in less basic (=common?) vocabulary" is generally true, then it doesn't really matter to me whether it's "because of interdialectal borrowing" or "because of lexical diffusion"—those things look the same! What matters is that we have systematicity in the apparent exceptions to a change, which lets us do better reconstruction.

On the other hand, if this simple rule isn't generalizable, and indeed the borrowing hypothesis and the lexical diffusion hypothesis make different predictions about the the patterns we should see in exceptions (e.g., that it should depend on sociolinguistic factors under the borrowing hypothesis), then it matters more which is actually true.

siberian-khatru-72

I am not sure that generally exceptions to sound changes are more likely in less basic vocabulary, provided we exclude cases like this, where actual borrowing seems likely. There are other sources of “irregularity”: analogy and sound symbolism (of a language-specific kind), and for each cause the distribution would be different. E.g., for sound symbolism we would expect more “irregular” cases in more expressive lexicon. If, on the other hand, we accept pseudo-explanations like “lexical diffusion”, we deprive ourselves of the possibility to account for different patterns of distribution for different causes of irregularity.

max1461

Yeah, well my claim is that lexical diffusion is a pseudo-explanation iff it makes no predictions about the distribution of exceptions, and it is a bad explanation iff it makes wrong predictions. Explanations like analogy and sound-symbolism make predictions-ish, I think, so they're better than nothing.

siberian-khatru-72

Also, the neogrammarian “dialect borrowing” scenario predicts that rampant irregularity will be found first of all in languages with a developed literary standard, like English. On the other hand, languages spoken by small largely isolated communities, like, say, Eastern Khanty, will be immune to this kind of irregularities. In cases where we know that several small communities recently merged into one, we can expect some amount of irregularities due to dialect borrowing, and this is what we found, e.g., in Tundra Yukaghir.