Ellision of intervocalic voiced stops

[Cathbad]

Is this a widespread phonetic process? I'm asking because it seems to be happening quite a lot in certain Slovene speech styles, e.g.

da bom > daum [da.Om] or [daO_^m] or [dau_^m] (ellision of b)
pogledal > pogleu [po"gle.u[/i] (ellision of d)

What seems strange is that it's limited to /b d/. It also seems to occur in post-stress position, in words that carry some sort of primary stress (ie. it would never be ellided in a word like gradil [gra"diu_^], which is understandable since it is the onset of a stressed syllable). Oh, and there's no glottal stop or anything that replaces it, just a hiatus.

[Echobeats]

Slovene is da bomb!

[gsandi]

Is this a widespread phonetic process? I'm asking because it seems to be happening quite a lot in certain Slovene speech styles,

It's quite a widespread phenomenon in fact. It is also correlated with the parallel voicing of intervocalic voiceless stops, but it doesn't have to be.

Examples:

1. Intervocalic /d g/ tend to disappearance in the Western Roamnce languages. In French this is virtually universal: nuda > nue, plaga > plaie, augustu > août. It is also almost universal in Spanish and Portuguese, with occasional restoration because of the influence of Classical Latin. Exs: Spanish caer (from cadere), dedo (from digitu).

2. Intervocalic b disappeared in Romanian: caballu > cal

3. Contemporary Dutch tends to delete intervocalic d: goede > goeie. g became G or x ages ago, therefore could not participate in this change.

4. Intervocalic g disappeared in Old English, through the stage G. Many examples of silent <gh> are examples of this.

5. In Irish, intervocalic d and g often disappear, although occasionally a weakly articulated /G/ is pronounced. The original consonant is retained in the spelling <dh>, <gh>. This is part of the process of lenition, and affects even initial consonants, when they follow a word formerly ending in a vowel in a close syntactic relationship. /b/ may be also affected by disappearance (spelled <bh>), but is retained as /v/ or /w/ in many (most?) cases.

[Cathbad]

EB: haha.

gsandi: Yes - thanks! It seemed strange to me because I only ever heard of fricatization of voiced stops (as in Modern Greek and Spanish), but not of complete ellision. And I should have thought of août, really

[gsandi]

EB: haha.

gsandi: Yes - thanks! It seemed strange to me because I only ever heard of fricatization of voiced stops (as in Modern Greek and Spanish), but not of complete ellision. And I should have thought of août, really

The interesting thing is that the -g- in Spanish agosto (1) was retained from Latin, or (2) was reintroduced from the Latin form, which learned people (or at least the local priest) would have been familiar with.

(I'll have to see what Menéndez Pidal says, after I get home tonight!)

But it can't be an entirely learned borrowing, because then I would expect **Augusto.

[Etherman]

IIRC it happened in Albanian prehistory.

[Boşkoventi]

4. Intervocalic g disappeared in Old English, through the stage G. Many examples of silent <gh> are examples of this.

Are you sure about this? As far as I know, medial g became y or w, depending on the neighboring vowels (e.g. day < OE dæg, dawn < OE dagung), and silent "gh" represents OE /x/, which was usually written "h", and corresponds to Modern German "ch" (e.g. "night" < niht, cf. German "Nacht").

[linguoboy]

5. In Irish, intervocalic d and g often disappear, although occasionally a weakly articulated /G/ is pronounced. The original consonant is retained in the spelling <dh>, <gh>. This is part of the process of lenition, and affects even initial consonants, when they follow a word formerly ending in a vowel in a close syntactic relationship. /b/ may be also affected by disappearance (spelled <bh>), but is retained as /v/ or /w/ in many (most?) cases.

Is the [G] pronunciation a Scottish-Gaelic thing? Because I can't think of any Irish words with word-internal intervocalic [G]. /G/ is either deleted entirely or participates in diphthong formation.

Whether intervocalic /v(')/ is retained or deleted or merges with the nucleus depends crucially on the quality and length of the vowels involved. Munster is more prone than other dialects to delete the slender form, e.g. duibhe "blacker" > /'d_Gi:/, and even broad /v/ disappears in some words, e.g. diabhal > /'d_ji:@l_G/.

[Travis B.]

4. Intervocalic g disappeared in Old English, through the stage G. Many examples of silent <gh> are examples of this.

Are you sure about this? As far as I know, medial g became y or w, depending on the neighboring vowels (e.g. day < OE dæg, dawn < OE dagung), and silent "gh" represents OE /x/, which was usually written "h", and corresponds to Modern German "ch" (e.g. "night" < niht, cf. German "Nacht").

Technically speaking, early Old English had an unpalatalized phoneme /ɣ/ and a palatalized phoneme /ʝ/ (both realized as plosives when geminated), along with the etymologically separate /j/, for what is conventionally written as g. At some point, /ʝ/ became /j/ while /jː/ and /ʝː/ became /dʒ/*; the point is not quite clear, but there are sound changes that distinguish /j/ from West Germanic /j/ from /j/ from West Germanic /ɡ/ which likely took place after palatalization. Later on during the Old English period, /ɣ/ when prevocalic or geminated became /ɡ/*, and in other environments became /j/ or /w/ depending on the surrounding vowels. Soon thereafter, around the start of the Middle English period, sequences of a vowel followed by /j/ or /w/ became diphthongs, which were simplified through mergers over the course of the Middle English period.

* with geminateness being retained, only being lost far later during the Early New English period

[Soap]

Many modern Spanish dialects have gone further and elided the voiceless stops which turned into voiced stops when the original voiced stops went silent. So you hear -á and -áo for -ada/-ado, giving us words like "hoosegow" as a loan from Spanish juzgado. Like Slovene, I believe it doesnt happen if the stop is at the beginning of a stressed syllable, even if it's intervocalic.

<s>There's also a few irregulars, like "sé" where we would expect *sabo, from Latin sapere. Or is that a suppletive?</s> Never mind, it's irregular even in Italian which never voiced the intervocalic stops.

"wheel" was once hweogol, unless the synonym {hweol}, also attested in OE, came from a separate PIE root.

[linguoboy]

There's also a few irregulars, like "sé" where we would expect *sabo, from Latin sapere.

If anything, we'd expect *sepo since the Latin 1S.PRS was SAPIO and the parallel form CAPIO yielded quepo. Quoth Penny "The form sé is best explained as analogical imitation of he."

[gsandi]

There's also a few irregulars, like "sé" where we would expect *sabo, from Latin sapere.

If anything, we'd expect *sepo since the Latin 1S.PRS was SAPIO and the parallel form CAPIO yielded quepo. Quoth Penny "The form sé is best explained as analogical imitation of he."

It may well be so, but this is going back to Proto-Western-Romance (PWR), where I reconstruct *sai (analogical to *ai 'I have' and *Bai 'I go'). Hence Old French sai, Portuguese sei, Catalan sé, Old Provençal sai, Romanche (Sursilvan) sai, Piemontese sai, Old Padovan sè.

I consider these forms among the most powerful arguments for the reality of PWR as a language that actually existed. They correlate strongly with some phonetic changes (voicing of intervocalic voiceless consonants, é > í under the influence of a following i, k'l > L) and morphological changes (retention of plural in -s [shared with Sardinian], , imperfect of -ere/-ire verbs loses the -B-, conditional based on infinitive + imperfect of havere).

[gsandi]

5. In Irish, intervocalic d and g often disappear, although occasionally a weakly articulated /G/ is pronounced. The original consonant is retained in the spelling <dh>, <gh>. This is part of the process of lenition, and affects even initial consonants, when they follow a word formerly ending in a vowel in a close syntactic relationship. /b/ may be also affected by disappearance (spelled <bh>), but is retained as /v/ or /w/ in many (most?) cases.

Is the [G] pronunciation a Scottish-Gaelic thing? Because I can't think of any Irish words with word-internal intervocalic [G]. /G/ is either deleted entirely or participates in diphthong formation.

Thanks for the correction.

Irish is not a language I have studied, although I am fascinated by its complex phonetics/morphology.

Having followed up on this, you are right when it comes to intervocalic examples of <dh> and <gh>. They were vocalized in early Modern Irish ("The Celtic Languages" (ed. by Martin Ball), p. 110), as in: laghad /L@id/ 'paucity', cadhan /k@in/ 'barnacle goose', the latter an everyday word in the Gaeltacht, I am sure.

dh/gh = /G/ initially in lenition, however, in broad environments.

[AnTeallach]

5. In Irish, intervocalic d and g often disappear, although occasionally a weakly articulated /G/ is pronounced. The original consonant is retained in the spelling <dh>, <gh>. This is part of the process of lenition, and affects even initial consonants, when they follow a word formerly ending in a vowel in a close syntactic relationship. /b/ may be also affected by disappearance (spelled <bh>), but is retained as /v/ or /w/ in many (most?) cases.

Is the [G] pronunciation a Scottish-Gaelic thing? Because I can't think of any Irish words with word-internal intervocalic [G]. /G/ is either deleted entirely or participates in diphthong formation.

Thanks for the correction.

Irish is not a language I have studied, although I am fascinated by its complex phonetics/morphology.

Having followed up on this, you are right when it comes to intervocalic examples of <dh> and <gh>. They were vocalized in early Modern Irish ("The Celtic Languages" (ed. by Martin Ball), p. 110), as in: laghad /L@id/ 'paucity', cadhan /k@in/ 'barnacle goose', the latter an everyday word in the Gaeltacht, I am sure.

dh/gh = /G/ initially in lenition, however, in broad environments.

The Akerbeltz website (for Scottish Gaelic) does give a couple of examples of [G] between vowels - one is modhail [mOGal] - but says it's "rare".

[Boşkoventi]

4. Intervocalic g disappeared in Old English, through the stage G. Many examples of silent <gh> are examples of this.

Are you sure about this? As far as I know, medial g became y or w, depending on the neighboring vowels (e.g. day < OE dæg, dawn < OE dagung), and silent "gh" represents OE /x/, which was usually written "h", and corresponds to Modern German "ch" (e.g. "night" < niht, cf. German "Nacht").

Technically speaking, early Old English had an unpalatalized phoneme /ɣ/ and a palatalized phoneme /ʝ/ (both realized as plosives when geminated), along with the etymologically separate /j/, for what is conventionally written as g. At some point, /ʝ/ became /j/ while /jː/ and /ʝː/ became /dʒ/*; the point is not quite clear, but there are sound changes that distinguish /j/ from West Germanic /j/ from /j/ from West Germanic /ɡ/ which likely took place after palatalization. Later on during the Old English period, /ɣ/ when prevocalic or geminated became /ɡ/*, and in other environments became /j/ or /w/ depending on the surrounding vowels. Soon thereafter, around the start of the Middle English period, sequences of a vowel followed by /j/ or /w/ became diphthongs, which were simplified through mergers over the course of the Middle English period.

* with geminateness being retained, only being lost far later during the Early New English period

Thanks for the history lesson, but can you name any words where OE intervocalic /g/ (or /ɣ/ if you like) disappeared rather than vocalising, or where Modern English "gh" represents OE /g/?

But, incidentally, wouldn't [ʝ] have been an allophone of /ɣ/? Words like dæg and dagung, suggest this, with the vowels determining whether each word has [ɣ] or [ʝ]. At least that's how I've always understood it.

"wheel" was once hweogol, unless the synonym {hweol}, also attested in OE, came from a separate PIE root.

Hweol and hweogol seem to have been variant forms of the same word. The OED2 even gives the "g" as optional in the Proto-Germanic form ("χwe(g)ula-", also "χweχula-", both from *qʷeqʷlo-, i.e. *kʷekʷlo-).

[Travis B.]

Thanks for the history lesson, but can you name any words where OE intervocalic /g/ (or /ɣ/ if you like) disappeared rather than vocalising, or where Modern English "gh" represents OE /g/?

The typical pattern was to vocalize, but I do think I have seen places where OE /ɣ/ simply disappeared, but I cannot think of them offhand aside from your example of hweogol versus hweol.

But, incidentally, wouldn't [ʝ] have been an allophone of /ɣ/? Words like dæg and dagung, suggest this, with the vowels determining whether each word has [ɣ] or [ʝ]. At least that's how I've always understood it.

Palatalization in OE is generally considered to be phonemic, generally agreeing with adjacent vowels but having discrepancies that make it hard to chalk up to mere allophony throughout the whole of OE times.

Another note is that the merger of OE /ʝ/ with OE /j/ long preceded the vocalization of other unpalatalized and unhardened cases of /ɣ/ to either /j/ or /w/ as conditioned by adjacent vowels. The primary evidence that OE /ʝ/ remained a separate phoneme from OE /j/ for any period of time after initial palatalization, which took place around the beginning of OE times, is there are some sound changes that occurred later in OE that applied to only West Germanic /j/ proper and which ignored OE /ʝ/, such as the elision of /j/ but not /ʝ/ after consonants.

[Niedokonany]

In colloquial Polish there's a tendency to elide voiced labials intervocalically in an unstressed syllable; I certainly do that a lot when I don't particularly strive to pronounce things clearly; although I believe /w/ is the most affected, /v b m/ are as well.

chciała > ["xt_s\a.a] "she wanted"
nie mam > ["JE.am] "I don't have"
mama > ["ma.a] "mum"
kupiłem > [ku"pi.Em] "I bought"
trzeba > ["t_s_-E.a] "one must"
czasami > [t_s_-a"sa.i] "sometimes"
nie wiem > [Je:m] "I don't know" (which "dangerously" approaches nie jem "I don't eat" :P)
żeby ["z_-E.1] "in order to (purpose clause marker)"

(I haven't spectrogrammized myself on this, but I suspect that e.g. I sometimes collapse the resulting [E.a] into [{:])

Additionally, /g/ may dissappear from the genitive ending -ego in rapid speech, so tego > ["tE.O].