"Signals and Symbols in Linguistic Variation and Change"
« previous post | next post »
This afternoon I'm scheduled to give a talk at the CUNY Graduate Center (365 Fifth Avenue, Room 9205), with the title "Signals and Symbols in Linguistic Variation and Change". The abstract:
Words are digital symbols transmitted as acoustic signals. The word sequence in an utterance is encoded by a phonological system whose symbol-facing side connects to morpho-syntax, while its signal-facing side controls articulation and perception. This "duality of patterning" (Hockett) or "double articulation" (Martinet) has crucial and little-recognized benefits for accurate transmission, lexical learning, and community convergence. It also raises serious and rarely recognized questions for phonological theory, including the nature of phonetic interpretation and the role of extra-phonological communication. This talk will explore these aspects of phonology, while also discussing the end-to-end nature of many contemporary AI systems.
There's a site to register for a Zoom link.
My slides are here.
Stephen Goranson said,
March 30, 2025 @ 1:49 pm
I hope for comments here; some of this is over my head, people.
Chris Button said,
March 31, 2025 @ 5:58 pm
I recall that Edwin Pulleyblank wrote a series of three articles on the matter back in the 1980s.
Unfortunately I don't have time right now to go back and review any of them. But I can see them sitting here on my computer and tempting me. I will post something if I get the chance before the comments thread times out.
Rod Johnson said,
March 31, 2025 @ 6:16 pm
Great ending. Says it all.
Nice to see someone taking up Hockett's work, which has been in eclipse since the rise of generative theory in the early sixties. It feels like the road not taken, in a way, kind of a synthesis of the previous three decades (since Bloomfield's Language in 1933) of American structuralism, which was in the process of being abandoned by a new generation of Chomsky's students.
Benjamin E. Orsatti said,
April 1, 2025 @ 10:43 am
(from slides): "[M]ost published phonology is probably wrong".
Wowie! I always like to see what shakes out of these sorts of paradigm-shifting observations. Where's my popcorn?…
J.W. Brewer said,
April 2, 2025 @ 9:59 am
This had caught my eye because the talk was happening only a few blocks from my office (7-minute walk if you believe google maps), but it turned out to be an inopportune day to try to leave work early. I was initially puzzled by the claim that words were "digital symbols" but then eventually realized that of course the relevant sense of "digital" is, to quote wiktionary, "Property of representing values as discrete, often binary, numbers rather than a continuous spectrum" and that is a serious mistake to slide unconsciously from "often binary" to "necessarily binary."
So an English word can be conceptualized as a string of "digits" in base-44, at least if one agrees with the conclusion of google's "AI Overview" that English has exactly 44 phonemes, no more and no less and assumes a mental lexicon with no prosodic features encoded into it.
That different varieties of English appear in practice to have different numbers of phonemes presents difficulties of course. That we can generally understand people whose dialects have mergers that ours doesn't (and that they can likewise generally understand us) means something, but I'm not sure what it means. (Like a lot of Americans, I have the NORTH-FORCE merger but don't have the LOT-THOUGHT merger, so I'm somewhere in the middle of the range of variation.)
I have I believe on prior occasions expressed some regret-in-hindsight that I paid no serious attention to phonology when I was an undergraduate linguistics major, because for whatever reasons I can't really reconstruct it didn't sufficiently grab my attention at the time. Although perhaps I would have just been taught a lot of stuff that was "probably wrong"! My alma mater's department did at the time benefit from connections to the Haskins Labs folks (including myl's late father), which may have meant you could have learned something substantive other than whatever the Chomskyans were saying about phonology during that particular phase of their history.
Mike Maxwell said,
April 12, 2025 @ 10:38 pm
I came too late to this post. Was a recording made? I read the slides, but it feels like reading an extended abstract; I suspect that there must have been a lot more words to the spoken version.