Grammatical developments propelling sound change

[Ulrike Meinhof]

It's common for sound changes to have sometimes radical effects on grammar, such as the loss of case distinctions in Romance for example, but are there any cases of grammatical changes leading to changes in the phonology? Perhaps there are textbook examples of this that I just can't recall for the moment, but surely it's not as common?

[Ser]

It's common for sound changes to have sometimes radical effects on grammar, such as the loss of case distinctions in Romance for example, but are there any cases of grammatical changes leading to changes in the phonology? Perhaps there are textbook examples of this that I just can't recall for the moment, but surely it's not as common?

The usual way to think about case loss in Romance nowadays goes the other way: a continuous increase of the use of prepositions made case endings more and more useless. The argument goes that various Latin endings should've remained distinct after sound changes (sound changes alone do not make the cases fully merge, and Old French and Old Occitan show Latin could cope with having very similar cases and different syncretisms across the declensions). Plus, Latin during the Republic was already a language that used prepositions for most things, and nonetheless there are also attested uses of prepositions to mark locatives, datives, instrumentals, and nouns modifying nouns (not just substantivized quantifiers)—even if the last two were still rare at that point.

Even then, I don't personally know how accurate it is to say that [phonology/grammar] affected [phonology/grammar]. Surely they're always very much intertwined. Why phonology vs. grammar in particular? Could also be phonology vs. morphology, morphology vs. syntax, marking vs. coordination... The usual focus in this aspect of the history of Latin is whether prepositions were used in this or that, but there were also other related things that got probably affected, like the order of arguments, the presence of noun phrase discontinuity (with adverbs and verbs put in between a noun and its adjective/relative clause), of which little is usually talked about.

[Ulrike Meinhof]

The usual way to think about case loss in Romance nowadays goes the other way: a continuous increase of the use of prepositions made case endings more and more useless. [...] Even then, I don't personally know how accurate it is to say that [phonology/grammar] affected [phonology/grammar]. Surely they're always very much intertwined.

Ah, alright. But I only took Romance case loss as an example; let's say the loss of distinct personal verb endings in French and the subsequent obligatory subject marking. That seems to be a pretty clear case of sound change forcing a change in grammar (morphology). If consensus has agreed on another view in that case also, it's easy nonetheless to imagine the hypothetical scenario:

pak 'tree.NOM'
pak-a 'tree-ACC'
pak-o 'tree-DAT'
pak-i 'tree-GEN'

V > @ / _#[-stressed]

pak 'tree.NOM'
pak-@ 'tree-ACC'
pak-@ 'tree-DAT'
pak-@ 'tree-GEN'

And voilà, sound change had an inevitable impact on morphology. My question is whether changes in morphosyntax (if 'grammar' is too vague a word) can just as easily effect sound change, and if there are any good examples of this attested.

[hwhatting]

but are there any cases of grammatical changes leading to changes in the phonology?

What kind? The way how phonological changes can affect morphology is clear, by eradicating contrasts or creating a new contrast that can be used in morphology. And all examples for interaction I've seen is morphology reacting to phology changes, never triggering them. Perhaps in a few cases you could argue that morphological change helps to widen the distribution of a phoneme, e.g. Slavic /x/ (from PIE */s/), which developed after /i/, /u/, /r/, /k/ - in the loc. pl. it appeared regularily in the o-stems (PIE *-oisu), i-stems, and u-stems, whence it spread to the a-stems as well (IIRC, the old ending -as is still attested in some Old Czech toponyms).*1) Perhaps there are a few cases where morphological change made an allophone phonetic, or contributed to its phoneticisation. I could imagine that, if morphology started using "+ feature X" as a marker, that could make this feature (or inversely, "- feature") spread to a part of the phoneme system that originally wasn't contratsing that feature - fictive example, language uses voicing as a plural marker, and as a result extends the consonant system from /p b t d k/ to /p b t d k g/, creating a voiced counterpart to /k/ that didn't exist previously. But I don't know any real life examples for such a development.

*1) NB that, due to the development *ks > *kx > x, /x/ already wasn't limited distributionally to position after /i/, /u/, /r/, /k/ at that point any more, that's why it can be argued as well that the phonemic status of /x/ made the morphology change possible, and that at most the morphology change made a phoneme more frequent that already existed.

[gach]

The patterns with which many Finnic and Saamic languages have lost their word final nasals might count as examples of grammar affecting sound change, though not in the sense of solidifying new phonemes. For example Estonian has lost the final nasals of its genitive-accusative (*-n < *-n/m) and illative (*-sen) case endings as well as nasal final noun stems (if there are any that still preserve an original final nasal please tell me). Thus we have

jõgi "river" < *joki
jõe "river.GEN/ACC" < *joke-n < *joke-n/m
jõkke "river.ILL" < *joke-sen

süda "heart" ~ Fi. sydän
seeme "seed" ~ Fi. siemen
naine "woman" ~ Fi. nainen
eriline "different" ~ Fi. erilainen

But when you look at the conjugation of verbs you immediately notice that the 1st person singular has preserved its original nasal ending completely intact (-n < *-n < *-m), as you can see from the partial paradigm

annan "give-SG1" < *anta-n
annad "give-SG2" < *anta-t
annab "give-SG3" < *anta-pa

This seems unexpected if you have been looking at the nominal inflection. It might be that the symmetry of the conjugation paradigm has motivated the preserving of this ending. Again, if anyone knows more about this change I'd be happy to hear.

You find a similar pattern in North Saami where again the genitive, accusative and illative have lost their final nasals while on verbs the 1st person singular ending is alive and doing well despite being phonetically no different than the other word final nasals.

Just to stay on the safe side, if anyone is left wondering, I'd like to add a further point. Although highly erosive phonological processes will certainly have an effect on the morphology of the language, they don't have to change the typology of the grammar into any direction, such as towards higher fusionality or use of analytic constructions. For example, the three cases I've been talking here (genitive, accusative and lative/illative) also merged in Proto Permic due to heavy phonological erosion. The cases didn't loose their independent identities, however, but acquired new secondary suffixes from existing elements in the language. The new case suffixes are both equally bound and agglutinating than their original counterparts.

[jal]

are there any cases of grammatical changes leading to changes in the phonology

Though I can't think of real-life examples, as some of the other commenters already mentioned more-or-less in between their other comments, new allophones may come to life if certain sounds, otherwise not normally adjacent, would find themselves juxtaposed because of word order changes, affixation or whatnot. Later, these allophones may develop into full phonemes.


JAL

[Nortaneous]

fictive example, language uses voicing as a plural marker, and as a result extends the consonant system from /p b t d k/ to /p b t d k g/, creating a voiced counterpart to /k/ that didn't exist previously. But I don't know any real life examples for such a development.

In Nias, initial consonant mutation is the only source of /v/.

[hwhatting]

fictive example, language uses voicing as a plural marker, and as a result extends the consonant system from /p b t d k/ to /p b t d k g/, creating a voiced counterpart to /k/ that didn't exist previously. But I don't know any real life examples for such a development.

In Nias, initial consonant mutation is the only source of /v/.

Do you have more information? If this is a case of a grammticalized sandhi phenomenon, this would be just another case where phonological developments create an alternation that became used in morphology. Only if the /v/ was created in analogy to other initial mutations (e.g, /t/ > /T/, /d/ > /D/, /k/ > /x/, /g/ > /G/, while originally labials were not affected and only became so analogically), one can really speak about morphology driving phonology.