Sound Change Quickie Thread

[bradrn]

IPA vowels are defined by the position of the tongue - mid-close, high front, low back, etc etc.

This is the claim, yes. But in practice it’s false, and it’s been known to be false for ~40 years. You can consult Ladefoged and Maddieson’s Sounds of the World’s Languages for the details — there is no metric of tongue position which matches up well to the IPA vowel chart, whereas a plot of F1 vs F2 reproduces it almost perfectly.

(Or, if you don’t feel like digging up that book, read Geoff Lindsey’s article on the same topic.)

Change in frequencies is due to change in vocal tract shape, no? Our tongues don't "draw out" perfect triangles or trapezoids when filling up the vowel space, but still lowering the tongue increases F1, and fronting it increases F2.
That source as well states that IPA vowel chart is based on tongue positions, even though they don't form a perfect geometric shape.

There is some connection between tongue position and the vowel chart, yes. But the connection to acoustics is far stronger, and much more widely-used. When people want to describe a vowel system rigorously, they don’t even bother attempting to describe the tongue shape — they just show a formant plot, assuming it shows the vowel space. (Sometimes they even superimpose a vowel chart directly on the plot.)

By contrast, descriptions of consonants don’t usually bother talking about acoustics. They just describe the articulation straightforwardly.

[Travis B.]

travis i triple dog dare you to post clear audio recordings of you speaking complete sentences in your natural voice to prove that any of your ridiculous look-at-me transcriptions of how you supposedly pronounce things are anything close to accurate. [ʁˤʷʁ̩ˤːɰ ɰiːʑʁ̩ˤ(ː)] my ass

Well here is my attempt to record myself speaking more naturally in an unscripted fashion. It's still a bit careful (e.g. from listening it over I didn't really vocalize the /l/ either time I pronounced leisure but I didn't feel like redoing it again), but you should be able to get a good picture of how I speak normally from it.

[Travis B.]

Aight, I'll bite. What's the point of transcribing a vowel as [ʁ̩ˤ] instead of [ɜ]? You might as well transcribe [a i u] as [ʕ̩ j̩ w̩] or sth else to that effect.

[ɜ] is basically my DRESS (even though in careful speech my DRESS is a bit front-of-center). [ʁ̩ˤ] sounds nothing like my DRESS and is much closer to [ɯ] as bradrn mentions except I cram it further back in my throat (e.g. it is close to my vocalized /l/ except that the latter is not as far back and isn't nearly as pharyngealized).

And while we're at it, what's the exact supposed subtle difference between [p t k] and [b̥ d̥ g̥]?

I use [b̥ d̥ g̥] for convenience's sake when transcribing initial /b d g/ as they may be voiceless, very weakly voiced, or voiced depending on phonetic environment and carefulness, and I don't feel like transcribing the same word multiple times if I can help it. For instance, in everyday speech these are typically voiceless, but when speaking on the phone these are commonly voiced for me.

[Travis B.]

Aight, I'll bite. What's the point of transcribing a vowel as [ʁ̩ˤ] instead of [ɜ]?

Because [ɜ] and [ʁ̩ˤ] are not the same thing. The former is a vowel between [ɐ] and [ə], while the latter is a pharyngealised uvular fricative (or more likely an approximant). Consulting data from Ladefoged & Maddieson (1996), it appears that the latter sound is actually closer in articulation to [ɯ], though I suspect it’s quite different both acoustically and articulatorily.

For me I mean an approximant, never a fricative. Fricative [ʁ] reminds me of French.

And while we're at it, what's the exact supposed subtle difference between [p t k] and [b̥ d̥ g̥]?

I believe some sources use the latter notation to show slack voice. In my opinion it’s a confusing convention.

As I mentioned, I just use it as a convention to mark consonants which are typically either tenuis or weakly voiced, but which can become fully-voiced in careful speech.

[Travis B.]

This is the claim, yes. But in practice it’s false, and it’s been known to be false for ~40 years. You can consult Ladefoged and Maddieson’s Sounds of the World’s Languages for the details — there is no metric of tongue position which matches up well to the IPA vowel chart, whereas a plot of F1 vs F2 reproduces it almost perfectly.

(Or, if you don’t feel like digging up that book, read Geoff Lindsey’s article on the same topic.)

Change in frequencies is due to change in vocal tract shape, no? Our tongues don't "draw out" perfect triangles or trapezoids when filling up the vowel space, but still lowering the tongue increases F1, and fronting it increases F2.
That source as well states that IPA vowel chart is based on tongue positions, even though they don't form a perfect geometric shape.

There is some connection between tongue position and the vowel chart, yes. But the connection to acoustics is far stronger, and much more widely-used. When people want to describe a vowel system rigorously, they don’t even bother attempting to describe the tongue shape — they just show a formant plot, assuming it shows the vowel space. (Sometimes they even superimpose a vowel chart directly on the plot.)

By contrast, descriptions of consonants don’t usually bother talking about acoustics. They just describe the articulation straightforwardly.

A good example of this is SSBE [ɒ], i.e. LOT or CLOTH, which typically unrounded and rather sulcalized, in contrast with NAE [ɒ], i.e. THOUGHT or CLOTH in many very many varieties, which is typically weakly rounded rather than sulcalized, despite being transcribed with the same symbol and sounding similar.

[Emily]

thank you for posting the audio. i probably didn't need to be so aggressive, apologies. that being said, your /r/ didn't sound uvular, your initial /l/ sounded like a pretty clear /l/, and your /ʒ/ certainly didn't sound like /ʑ/

[bradrn]

travis i triple dog dare you to post clear audio recordings of you speaking complete sentences in your natural voice to prove that any of your ridiculous look-at-me transcriptions of how you supposedly pronounce things are anything close to accurate. [ʁˤʷʁ̩ˤːɰ ɰiːʑʁ̩ˤ(ː)] my ass

Well here is my attempt to record myself speaking more naturally in an unscripted fashion. It's still a bit careful (e.g. from listening it over I didn't really vocalize the /l/ either time I pronounced leisure but I didn't feel like redoing it again), but you should be able to get a good picture of how I speak normally from it.

This definitely sounds odd to me compared to the American dialects I usually hear. On the other hand, it’s mostly not that difficult to understand.

thank you for posting the audio. i probably didn't need to be so aggressive, apologies. that being said, your /r/ didn't sound uvular, your initial /l/ sounded like a pretty clear /l/, and your /ʒ/ certainly didn't sound like /ʑ/

I can buy this /r/ being articulated with a uvular constriction. Similarly, the first /l/ in long (right at the beginning) sounds definitely velar, something like [ʟ~ɰ~w] (not sure which but most probably [w]). Not sure about /ʒ/.

[Darren]

I tend to agree with most of what Travis says. The pronunciation of "particularly" at 2:32 has very obvious backing of /r/ and /l/ in it, same in "upload" near the end. It's very noticeable in late unstressed syllables like -ly, but not so obvious in most word onsets (still every once in a while a word stands out to me). And the /ʒ/ in "leisure" sounds a lot sharper than my /ʒ/ so I would be tempted to transcribe it [ʑ].

What stands out to me the most is intervocalic lenition with words like "anyways" and "dialect" and "syllable" where -n- and -l- seem to disappear almost entirely. Also when you said "can't" and "can" in isolation I parsed them as "cat" and "cad" at first, since vowel nasalisation is very marginal in AusEng.

[Travis B.]

thank you for posting the audio. i probably didn't need to be so aggressive, apologies. that being said, your /r/ didn't sound uvular, your initial /l/ sounded like a pretty clear /l/, and your /ʒ/ certainly didn't sound like /ʑ/

I can buy this /r/ being articulated with a uvular constriction. Similarly, the first /l/ in long (right at the beginning) sounds definitely velar, something like [ʟ~ɰ~w] (not sure which but most probably [w]). Not sure about /ʒ/.

My /r/'s don't sound "uvular" in that they don't sound like French [ʁ]'s - they have no frication and are commonly quite vowel-like.

My /l/'s, when they are "clear", are velar laterals, and I articulate then quite differently from, say, German [l] (for instance if my tongue touches the roof of my mouth at all it is in the dorsal region).

About /ʒ/, like all shibilants for me, has both palatoalveolar and alveolopalatal allophones. In the case of leisure the constriction is in the palatal region while in the case of, say, vision it is in the postalveolar region.

[bradrn]

My /l/'s, when they are "clear", are velar laterals, and I articulate then quite differently from, say, German [l] (for instance if my tongue touches the roof of my mouth at all it is in the dorsal region).

To me the term ‘clear /l/’ refers to an alveolar articulation, and ‘dark /l/’ refers to a velar (or velarised) articulation. From this description I would have no problem saying that your /l/ is always dark — as, indeed, is the case in Australian English.

[Travis B.]

My /l/'s, when they are "clear", are velar laterals, and I articulate then quite differently from, say, German [l] (for instance if my tongue touches the roof of my mouth at all it is in the dorsal region).

To me the term ‘clear /l/’ refers to an alveolar articulation, and ‘dark /l/’ refers to a velar (or velarised) articulation. From this description I would have no problem saying that your /l/ is always dark — as, indeed, is the case in Australian English.

That is the terminology I am used to too ─ and "clear" /l/'s are not a native speech sound for me; I had to teach myself to pronounce them when learning basic German.

[bradrn]

My /l/'s, when they are "clear", are velar laterals, and I articulate then quite differently from, say, German [l] (for instance if my tongue touches the roof of my mouth at all it is in the dorsal region).

To me the term ‘clear /l/’ refers to an alveolar articulation, and ‘dark /l/’ refers to a velar (or velarised) articulation. From this description I would have no problem saying that your /l/ is always dark — as, indeed, is the case in Australian English.

That is the terminology I am used to too ─ and "clear" /l/'s are not a native speech sound for me; I had to teach myself to pronounce them when learning basic German.

‘Clear’ [l] isn’t a native sound to me either — my /l/ is either [ɫ] or [u~ɯ]. However I did get a lot of early exposure to Israeli Hebrew, which does have clear [l], so that helps me in pronouncing it. (If I’m not concentrating I tend to say [ɫ] anyway… the difference isn’t big.)

[Travis B.]

To me the term ‘clear /l/’ refers to an alveolar articulation, and ‘dark /l/’ refers to a velar (or velarised) articulation. From this description I would have no problem saying that your /l/ is always dark — as, indeed, is the case in Australian English.

That is the terminology I am used to too ─ and "clear" /l/'s are not a native speech sound for me; I had to teach myself to pronounce them when learning basic German.

‘Clear’ [l] isn’t a native sound to me either — my /l/ is either [ɫ] or [u~ɯ]. However I did get a lot of early exposure to Israeli Hebrew, which does have clear [l], so that helps me in pronouncing it. (If I’m not concentrating I tend to say [ɫ] anyway… the difference isn’t big.)

Thing is, I find "clear" [l] easy to pronounce — actually, I find combining alveolar contact with velarization, i.e. [ɫ], harder than it because I find what is essentially double articulation of it difficult to enunciate.

[Gareth3]

I have a few fairly reasonable sound changes in my conlang:

u > y _C[front vowel]
[stop] > nasal _...[nasal] (nasal harmony)

The problem is that I don't want any of these changes to affect the root, only the suffixes. For example:

root pu
plural suffix ku
case suffixes ti n

The case-number pattern I want is:
This
pu...puku
puti pukyti
pun pungun
Not this
pu...puku
pyti pykyti
mun mungun
At the moment I'm just saying the last syllable of the root is always stressed, and the changes only work in and next to unstressed vowels. But I'm not sure if that restriction is legitimate. Should I add something to each root to block these changes?

[bradrn]

Reformatting your table for you:

This	Not this
pu puku	pu puku
puti pukyti	pyti pukyti
pun pungun	mun mungun

Honestly, I don’t think this needs any special justification. Analogy is a powerful force and could easily work to homogenise the roots (perhaps with one or two exceptions if you feel like it).

[Gareth3]

Thanks, that table was a mess. That's a good point about analogy, you're saying the speakers would think of the system as "root + suffix", and not apply the sound changes to the root? That's even more likely to work if there's plenty of roots that wouldn't change anyway, like "me".

[bradrn]

That's a good point about analogy, you're saying the speakers would think of the system as "root + suffix", and not apply the sound changes to the root?

Well, not quite… as I understand it, what happens is that the sound change happens as usual, but then over the next few generations speakers notice the common pattern of ‘unchanging root + suffix’ and generalise it to all roots. This sort of analogy (where speakers notice a pattern and generalise it) is a very strong force in language evolution, especially when it comes to making things less irregular.

(A similar example from English: the verb help used to have past tense holp, but in modern English the suffix -ed has been generalised, such that the standard past tense form is now helped. Similarly for heave, shave and a bunch of other verbs. But English verbs are complicated enough that analogy can work in the opposite direction too: e.g. bring sometimes gets past tense brang or brung, by analogy with verbs like wring and spring.)

[Gareth3]

Makes sense. Depending on how exactly the phonology ends up, most of the roots wouldn't be affected anyway.

[Darren]

(A similar example from English: the verb help used to have past tense holp, but in modern English the suffix -ed has been generalised, such that the standard past tense form is now helped. Similarly for heave, shave and a bunch of other verbs. But English verbs are complicated enough that analogy can work in the opposite direction too: e.g. bring sometimes gets past tense brang or brung, by analogy with verbs like wring and spring.)

I have heard that English verbs reached peak regularity in the Middle Ages and since then the trend of analogy has been more in favour of strong verbs, although I don't know where they got their figures from.

[bradrn]

(A similar example from English: the verb help used to have past tense holp, but in modern English the suffix -ed has been generalised, such that the standard past tense form is now helped. Similarly for heave, shave and a bunch of other verbs. But English verbs are complicated enough that analogy can work in the opposite direction too: e.g. bring sometimes gets past tense brang or brung, by analogy with verbs like wring and spring.)

I have heard that English verbs reached peak regularity in the Middle Ages and since then the trend of analogy has been more in favour of strong verbs, although I don't know where they got their figures from.

Huh, really? I’m skeptical.