Fricativ loss

[Tropylium⁺]

There are examples for the loss (usually with at least a possible [h] stage) of pretty much any voiceless fricativ:
• ɸ h ∅ (Tungusic)
• f (? ɸ) h ∅ (Spanish)
• θ h ∅ (IIRC Scottish Gaelic)
• ɬ ∅ (Zan languages)
• s h ∅ (Greek, Iranian, etc.)
• ʃ (? x ) h ∅ (Votic)
• x h ∅ (dialectally, English)
• ħ (? h) ∅ (Akkadian)

I'm seeking to fill the series. Does anyone have examples for:
• ʃ (without loss of /x/ or /s/)
• ɕ, ç (in contrast to a non-palatal shibilant /ʃ/ or /ʂ/)
• χ (without loss of /x/)
• sʲ, xʷ, or any fricativ with secondary articulation (without loss of the corresponding plain fric.)
• any cases of loss, without loss of /h/

[Xephyr]

I'm seeking to fill the series. Does anyone have examples for:
...
• ʃ (without loss of /x/ or /s/)
...

Unless I'm majorly brainfarting here, Spanish lost /ʃ/ when it became /x/.

[linguoboy]

• θ h ∅ (IIRC Scottish Gaelic)

I don't think that's accurate. /h/ from /θ/ is lost in some Irish varieties (e.g. Cois Fhairrge), but only word-internally and in coda position. I don't know of any examples of unconditional loss of /θ/.

[Astraios]

I believe Irish oíche shows /ç/ > Ø, as it's pronounced (by two singers, at least) ['i:].


EDIT: But I don't know how widespread that is...

[Tropylium⁺]

I'm seeking to fill the series. Does anyone have examples for:
...
• ʃ (without loss of /x/ or /s/)
...

Unless I'm majorly brainfarting here, Spanish lost /ʃ/ when it became /x/.

It lost /ʃ/ as a phoneme, yes, but I mean a development ʃ ∅ specifically.

[Xephyr]

I'm seeking to fill the series. Does anyone have examples for:
...
• ʃ (without loss of /x/ or /s/)
...

Unless I'm majorly brainfarting here, Spanish lost /ʃ/ when it became /x/.

It lost /ʃ/ as a phoneme, yes, but I mean a development ʃ ∅ specifically.

Oohhhh.. that's what you're looking for?

[linguoboy]

I believe Irish oíche shows /ç/ > Ø, as it's pronounced (by two singers, at least) ['i:].

Again, in the dialects in which this deletion takes place[*], the loss occurs only word-internally and in coda position. In initial position (e.g a chéile), it is retained.

[*] I assume one of the two singers you have in mind is Enya, a native-speaker from Gaoth Dhobhair.

[Astraios]

Again, in the dialects in which this deletion takes place[*], the loss occurs only word-internally and in coda position. In initial position (e.g a chéile), it is retained.

[*] I assume one of the two singers you have in mind is Enya, a native-speaker from Gaoth Dhobhair.

Ah, OK. Still, if the OP would accept Spanish initial f > h, this is relevant.

And yes, she is one. The other is Áine Ní Dhroighneáin.

[Tropylium⁺]

Again, in the dialects in which this deletion takes place[*], the loss occurs only word-internally and in coda position. In initial position (e.g a chéile), it is retained.

[*] I assume one of the two singers you have in mind is Enya, a native-speaker from Gaoth Dhobhair.

Ah, OK. Still, if the OP would accept Spanish initial f > h, this is relevant.

I'm accepting Spanish since the resulting †/h/ disappeared universally. It doesn't sound like it's necessarily a similar case here.

/h/-loss is so well attested that just examples of lenition to /h/ would suffice, really.

[linguoboy]

Ah, OK. Still, if the OP would accept Spanish initial f > h, this is relevant.

How so? Old Spanish /h/ is lost in all Spanish dialects outside of some Andalusian varieties.

Ironically, /f/ is the one voiceless fricative which is deleted entirely in lenition position in Gaelic.

[Astraios]

Well, I was thinking that since f didn't > h everywhere in Spanish, and /ç/ doesn't > Ø everywhere in Irish, they both work.

[Whimemsz]

Well, I was thinking that since f didn't > h everywhere in Spanish, and /ç/ doesn't > Ø everywhere in Irish, they both work.

It did, though? The only exception is before resonants, including /w/ in the sequence /we/ derived from VL */ɔ/ (so it's retained in fuerte and frente). Otherwise, /f/ only occurs in modern Spanish in loanwords, either from Latin itself or from other languages.

[Whimemsz]

[EDIT: this was a stupid post because I didn't reread the OP before making it, sorry!]

[Astraios]

It did, though? The only exception is before resonants, including /w/ in the sequence /we/ derived from VL */ɔ/ (so it's retained in fuerte and frente). Otherwise, /f/ only occurs in modern Spanish in loanwords, either from Latin itself or from other languages.

That's what I meant - /ç/ disappears in whatever Irish dialects it is, but only word-internally and in coda-position. I guess it's not the same then, because the Irish thing is only dialectal.

[alice]

Manx lost all word-final fricatives irrespective of voice, but I don't know if that's a useful answer. Although it does put me in mind of Manx cats for some reason.

[Niedokonany]

Polish has had positional

s > s_j > s\ > j, z > z_j > z\ > j

and you could imagine the j being subsequently merged with a vowel or something, though it hasn't happened yet

[Soap]

I don't think it's proper to say "Spanish lost all /f/". Almost all of the /f/ in Latin was word-initial or occurred in compounds. I can think of at least one word which probably isn't a loan in which the /f/ was retaiend medially: infierno "hell". I'm assuming it's not a loan because it shows the e > ie change which happened quite early, and because it's logical that a word pertaining to religion that is such a basic concept would not be lost and subsequently re-loaned from Latin.

There are probably lots of other examples; that's just the first one I thought of.

[linguoboy]

I can think of at least one word which probably isn't a loan in which the /f/ was retaiend medially: infierno "hell". I'm assuming it's not a loan because it shows the e > ie change which happened quite early, and because it's logical that a word pertaining to religion that is such a basic concept would not be lost and subsequently re-loaned from Latin.

That's not a safe assumption given how many semicultismos there are in Spanish, particularly in the area of religious life. In fact, it's because Spanish was under continuous influence from liturgical Latin that these emerged. I mean, is "Hell" a more basic concept pertaining to religion than "God"? And yet in Dios we see retention of final /s/--a change so early it's posited for the Vulgar Latin stage!

The fully popular development here is (1) loss of /n/ before fricatives (cf. MENSA > mesa) followed by (2) /f/ > /h/. E.g., CONFUNDERE > cohonder. So if infierno wasn't influenced by liturgical Latin, the form we would expect is *ehierno.

[Whimemsz]

I can think of at least one word which probably isn't a loan in which the /f/ was retaiend medially: infierno "hell". I'm assuming it's not a loan because it shows the e > ie change which happened quite early, and because it's logical that a word pertaining to religion that is such a basic concept would not be lost and subsequently re-loaned from Latin.

That's not a safe assumption given how many semicultismos there are in Spanish, particularly in the area of religious life. In fact, it's because Spanish was under continuous influence from liturgical Latin that these emerged. I mean, is "Hell" a more basic concept pertaining to religion than "God"? And yet in Dios we see retention of final /s/--a change so early it's posited for the Vulgar Latin stage!

The fully popular development here is (1) loss of /n/ before fricatives (cf. MENSA > mesa) followed by (2) /f/ > /h/. E.g., CONFUNDERE > cohonder. So if infierno wasn't influenced by liturgical Latin, the form we would expect is *ehierno.

This is exactly what Ralph Penny says in his History of the Spanish Language: Infierno is a semi-learned form; it's directly inherited from Latin, but its shape has been influenced by the constant use of Latin in liturgical contexts.

[Hakaku]

ɸ → h → ∅ (Tungusic, Japonic)

I'm not quite sure that this is true for Japanese, rather, /ɸ/ became an approximant /w/ in intervocalic position, and was subsequently deleted everywhere except before the vowel /a/ (though a few Southern dialects fall exception to this rule). This change took place before /ɸ/ came to be pronounced [h] /h/.

/omopu/ → /omoɸu/ → /omo(w)u/ → /omou/
/omopi/ → /omoɸi/ → /omowi/ → /omoi/
/ɸa/ → /wa/ (→ southern /a/)
*/kapite/ → /kapute/ → /kaɸute/ → /ka(w)ute/ → /koote/ (Kansai)

The change /s/ (→ /z/) → /∅/ is also attested in Early Middle Japanese before the high vowel /i/, a change which was retained most prominently in true adjective endings.
/kasite/ (→ /kazite/) → /kaite/
/sasite/ (→ /sazite/) → /saite/
/kawajusi/ → /kawajui/ → /kawaii/

This reduction is also observed in the south, though it may or may not occur by way of /si/ [ɕi] → /hi/ [çi] → /i/.
/kagosima/ → /kagoima/ (/kagohima/ also occurs)

Some authors posit that /k/ was deleted in a number of endings by way of [ɣ] (via onbin).
/siroki/ → /sirogi/ [siroɣi] → /siroi/
/kakite/ → /kagite/ [kaɣite] → /kaite/ (some dialects)

[Tropylium⁺]

ɸ → h → ∅ (Tungusic, Japonic)

I'm not quite sure that this is true for Japanese, rather, /ɸ/ became an approximant /w/ in intervocalic position, and was subsequently deleted everywhere except before the vowel /a/ (though a few Southern dialects fall exception to this rule). This change took place before /ɸ/ came to be pronounced [h] /h/.

No, not Japanese itself. I was thinking of Ryukyuan here (and rechecking, I actually I got that the wrong way round: there's a variety which adds epenthetic /h/ before vowels)

Anyway, it's still an example of ɸ h.

[Hakaku]

Anyway, it's still an example of ɸ h.

Yes, it's an example of /ɸ/ → /h/, but not an example of /ɸ/ → /h/ → /∅/. Otherwise, you could make the claim about every single fricative, since stuff like /ʃ/ → /h/ are attested in a number of languages. In reality, however, such a subsequent change as /h/ → /∅/ isn't always admissible due to a number of phonological constraints.

[Tropylium⁺]

In reality, however, such a subsequent change as /h/ → /∅/ isn't always admissible due to a number of phonological constraints.

Can you elaborate?

[Zhen Lin]

The change /s/ (→ /z/) → /∅/ is also attested in Early Middle Japanese before the high vowel /i/, a change which was retained most prominently in true adjective endings.
/kasite/ (→ /kazite/) → /kaite/
/sasite/ (→ /sazite/) → /saite/
/kawajusi/ → /kawajui/ → /kawaii/

I'm not sure the last one counts. I'm pretty sure the -i endings in modern Japanese descend from the -ki endings. (Evidence: the -shiku-type adjectives became -shii, not plain -i.)

[Hakaku]

I'm not sure the last one counts. I'm pretty sure the -i endings in modern Japanese descend from the -ki endings. (Evidence: the -shiku-type adjectives became -shii, not plain -i.)

I've always been under the impression that both the terminal and attributive endings reduced to -i, and then merged. Could well be wrong though.

In reality, however, such a subsequent change as /h/ → /∅/ isn't always admissible due to a number of phonological constraints.

Can you elaborate?

Well, the glottal fricative is a phoneme like any other. Just because it has a certain tendency to disappear in certain languages, doesn't mean that this same change will happen in others. In fact, in many languages /h/ is a very stable sound. In others, it might be an emerging sound. And in others again, it might be a transitional change, and the end result could well entail its entire disappearance, lead to a different phoneme such as a glottal stop, or leave behind a trace such as aspiration or breathy vowels.

From a phonological point of view, the change /h/ → /∅/ can be discouraged if the resulting sequence violates phonotactic rules. For instance, its entire disappearance could lead to undesirable consonant clusters and vowel sequences, and it may also result in the resyllabification of entire syllables. Another issue is that the existence of /h/ might be tied down to a process of debuccalization, which means that it could very well correlate with a voiced counterpart /ɦ/. Should one disappear, there's a very strong chance the other disappears along with it. And their disappearance could bring about innumerable homophones. A third issue is that /h/ may not be what it seems and could, in fact, encompass multiple realizations, as is the case of Japanese: [h ç (ɕ) ɸ]. The question here is, does the entire phoneme disappear, or does only a certain realization of it disappear?

Now, it's not to say that it absolutely can't happen, but in a language where all of these issues are relevant, it's far less likely that the change /h/ → /∅/ will fully take place. Instead, you're more likely to see other phonemes or phonetic qualities take its place in order to fill the void; and in many other cases, the change will be incomplete and will only take place under restrictive circumstances. Perhaps here a better question to ask would be, under what conditions do all of the languages above exhibit h → ∅.