Zompist Bboard Again

The oft-repeated Neogrammarian saw is that sound laws suffer no exceptions. But sound change often starts in a restricted pool of words, if I remember correctly, and only then diffuse out to the general lexicon. How do these get reconciled?

(This is something I should know already but my college days are almost a decade behind me already.)

Man in Space wrote: ↑Mon Jul 03, 2023 8:35 pm The oft-repeated Neogrammarian saw is that sound laws suffer no exceptions. But sound change often starts in a restricted pool of words, if I remember correctly, and only then diffuse out to the general lexicon. How do these get reconciled?

(This is something I should know already but my college days are almost a decade behind me already.)

The Neogrammarian answers to exceptions in sound change are analogy and dialect borrowing.

Man in Space wrote: ↑Mon Jul 03, 2023 8:35 pm The oft-repeated Neogrammarian saw is that sound laws suffer no exceptions. But sound change often starts in a restricted pool of words, if I remember correctly, and only then diffuse out to the general lexicon. How do these get reconciled?

(This is something I should know already but my college days are almost a decade behind me already.)

I believe the neogrammarian principles of sound change are generally regarded as a simplification nowadays--whether a useful one or a misleading one depends on what context you're interested in. When reconstructing distant unattested stages of languages/proto-languages, having a strong presumption for the universality of sound changes helps restrict the space of possible etymologies. Lexical diffusion is kind of difficult to demonstrate in that context.

When trying to describe relatively well-attested stages of languages, it may be necessary to admit exceptions to the immediacy of sound changes to give an accurate account of the observed facts. There also may be enough data, whether in written texts or audio recordings, to attempt to do things like statistically analyzing whether there is a relationship between a lexeme's frequency and its susceptibility to an in-progress sound change.

I agree with Estav. Real sound changes are not perfectly Neogrammarian, precisely because they spread via lexical diffusion.

Thank you all!