Palatalization(?) help

[Nmmali]

I want my conlang to have palatalization in front of /i/ and /ɪ/, but in the voiced stops only (and the ones that are not labial, because, yeah).
Like: /di/ > /d͡ʒi/ and /gi/ > /ʒi/.

So first off, I can't really find any examples of languages that do this without doing the same thing in the voiceless versions. Romance does /ki/ > /si/ or /tʃi/, Japanese does /ti/ > /tʃi/. I saw something about Arabic doing /g/ > /ʒ/, universally, but it's not really the same thing, all though I will probably doing that specifically anyway because I don't like g.

Also, I want to keep that distinction between /d͡ʒ/ the affricate and /ʒ/. So I'm saying for now that the /g/ > /ʒ/ is a much older sound change and /dʲ/ > /d͡ʒ/ is more recent. So that way, a sister language (if I bother making up one), would have no /g/ as well, but still allow /di/ (which, indeed, should be that sister language's word for "no").


But anyway, would y'all think that is implausible, to have /dʲi/ to /d͡ʒi/, but no change in /tʲi/?

Thank you.

Edit: that sister language as of now will probably not have the same vowel reductions, so the word for no would be /di/.

[Zekoslav]

There's inscriptional evidence that in Latin gi and di palatalised before ci and ti. However, strictly speaking this was not palatalisation but iotation: this happened before a vowel when these sequences were pronounced [gj], [dj], [kj] and [tj], and the voiced clusters assimilated to [jː] and/or [d͡ʒː]* while the unvoiced clusters remained.

*My (and some expert's) interpretation, which is based on the varying outcomes in modern Western Romance languages, is that originally there was free variation between [jː] and [d͡ʒː] similar to modern Spanish.

Palatalisation of /g/ but not /k/ could be explained by /g/ becoming /ɣ/ unconditionally (a very plausible change), which then becomes /ʒ/ when palatalised and something else when not palatalised, if you don't like the sound. From cross-linguistic evidence we can see that /ɣ/ is more susceptible to palatalisation that /k/ (e.g. in Old English).

As for palatalisation of/di/ but not /ti/, I don't know if it's as plausible, but weird asymmetries in sound change happen all the time. /g/ to /ʒ/ but /d/ to /d͡ʒ/ actually happens in my native language, in Brazilian Portuguese and probably elsewhere. In both cases it's, as you've guessed, due to different timing of the change.

P. S.: As for labials, yeah... they are hard to palatalise because their articulation doesn't involve the tongue.

[sangi39]

There's inscriptional evidence that in Latin gi and di palatalised before ci and ti. However, strictly speaking this was not palatalisation but iotation: this happened before a vowel when these sequences were pronounced [gj], [dj], [kj] and [tj], and the voiced clusters assimilated to [jː] and/or [d͡ʒː]* while the unvoiced clusters remained.

*My (and some expert's) interpretation, which is based on the varying outcomes in modern Western Romance languages, is that originally there was free variation between [jː] and [d͡ʒː] similar to modern Spanish.

Palatalisation of /g/ but not /k/ could be explained by /g/ becoming /ɣ/ unconditionally (a very plausible change), which then becomes /ʒ/ when palatalised and something else when not palatalised, if you don't like the sound. From cross-linguistic evidence we can see that /ɣ/ is more susceptible to palatalisation that /k/ (e.g. in Old English).

As for palatalisation of/di/ but not /ti/, I don't know if it's as plausible, but weird asymmetries in sound change happen all the time. /g/ to /ʒ/ but /d/ to /d͡ʒ/ actually happens in my native language, in Brazilian Portuguese and probably elsewhere. In both cases it's, as you've guessed, due to different timing of the change.

P. S.: As for labials, yeah... they are hard to palatalise because their articulation doesn't involve the tongue.

I suppose you could have a sort of "stepped" sound change, e.g. [tj kj dj gj] > [tj kj dj ɟʝ] > [tj kj ɟʝ ʝ] > [tj kj dʒ ʒ], perhaps? The assimilation still only affects the voiced clusters, but it affects [gj] first, which then further lenites before it affects [dj].

Either that or something like [tj kj dj gj] > [tj kj ɹj~ɾj ɰj] > [tj kj rj j] > [tj kj ʒ j] > [tj kj dʒ ʒ]? The voiced plosives become fricatives and then approximants first, and I don't think at that point it's unreasonable for [rj] to become a "palatal" without a similar change affecting [tj]. The only issue I could think with this particular route is that I'd expect to see [d] and [g] shift to fricatives/approximants in other environments as well.

[Nmmali]

Thank you both for your help. As for [g], for now I'm going to have an unconditional shift to [ɣ]. This, in turn, will drop altogether word initially, become [j] elsewhere, and develop into [ʒ] when lenited left of [i:] and [ i ], the latter transforming into modern [ɪ].

The only issue I could think with this particular route is that I'd expect to see [d] and [g] shift to fricatives/approximants in other environments as well.

What other environments, for instance?
The only thing that comes to mind is the lenition of [d] in the syllable coda, and presently the phonotactics do not allow it the coda anyway, except the cluster [nd], and, well, yeah. [nd] is about as hard-core as the coda gets, so [nd͡ʒ] is maybe just... no.
Not doing it before [ɛ e:] either. In fact, the only place I can see it happening any further might be [dr] in the onset, and that could easily become [d͡ʒ], but I'm not sure if I want to make that commitment yet. Off the top of my head I can think of [drɑn], and I'm not sure if I want another homophone like [d͡ʒɑn], the masculine inanimate article that procedes vowel onsets, or [d͡ʒɑ],which is the normal form of that article and the second person feminine oblique pronoun. Of course, that wouldn't be intolerable, just that I use my number system every day, and it's habit, and I'm too lazy to change it. But I might. I could always do the retroflex thing instead and have [ ɖ ] in that instance. That's a thought. It's already done for historic *[sr] into [ʂ]. I would much sooner prefer that to [d͡ʒ] in that context.
But anyway, thanks again.

[sangi39]

Thank you both for your help. As for [g], for now I'm going to have an unconditional shift to [ɣ]. This, in turn, will drop altogether word initially, become [j] elsewhere, and develop into [ʒ] when lenited left of [i:] and [ i ], the latter transforming into modern [ɪ].

The only issue I could think with this particular route is that I'd expect to see [d] and [g] shift to fricatives/approximants in other environments as well.

What other environments, for instance?
The only thing that comes to mind is the lenition of [d] in the syllable coda, and presently the phonotactics do not allow it the coda anyway, except the cluster [nd], and, well, yeah. [nd] is about as hard-core as the coda gets, so [nd͡ʒ] is maybe just... no.
Not doing it before [ɛ e:] either. In fact, the only place I can see it happening any further might be [dr] in the onset, and that could easily become [d͡ʒ], but I'm not sure if I want to make that commitment yet. Off the top of my head I can think of [drɑn], and I'm not sure if I want another homophone like [d͡ʒɑn], the masculine inanimate article that procedes vowel onsets, or [d͡ʒɑ],which is the normal form of that article and the second person feminine oblique pronoun. Of course, that wouldn't be intolerable, just that I use my number system every day, and it's habit, and I'm too lazy to change it. But I might. I could always do the retroflex thing instead and have [ ɖ ] in that instance. That's a thought. It's already done for historic *[sr] into [ʂ]. I would much sooner prefer that to [d͡ʒ] in that context.
But anyway, thanks again.

I think, by "other environments" I might have meant "between vowels", but the that sort of lenition could probably be unconditional as well. I think I was assuming you wanted to have /d/ and /g/ remain voiced plosives, appearing as [dʒ] and [ʒ] respectively before /ɪ/ and /i:/.