English vowel systems and lexical sets

[Travis B.]

Okay, /d/ does weird things to preceding vowels for me. For instance, /æ ɛ ʌ ʊ/ seem to be realized as [æː ɛː ɐː ɵː] before /d/ for me even though they are typically realized as [ɛː ɜː ʌː ʊː] before lenis consonants otherwise* (and [ɛ ɜ ʌ ʊ] before fortis consonants). And yes, from checking more closely, matter and latter do not have the same vowel quality as madder and ladder for me.

* with some exceptions, such as /æ/ before nasals, where the dialect here has [ɛ̆ə̯̆]~[ɛə̯] (something I did not grow up having but which I picked up inconsistently as an adult), and before coda /g/, where I often have [eə̯].

[bradrn]

Meanwhile, I had a go at plotting my diphthongs (based on https://joeystanley.com/blog/geomtextpath/):

How are you exporting and then storing the data for each diphthong from Praat to use it in R?

Go to spectogram, zoom in on dipthong, Formants>‘Extract visible formant contour’, go to new formant object, Tabulate>List, then File>Save as. Then repeat for next diphthong. It’s annoying but workable when I don’t have too many words.

[Travis B.]

I had trouble picking out sane formants with Praat, where it would often produce obvious garbage F2 values. As a result, I suspect that simply naively exporting formants from Praat may not produce valid contours in some cases.

[bradrn]

I had trouble picking out sane formants with Praat, where it would often produce obvious garbage F2 values. As a result, I suspect that simply naively exporting formants from Praat may not produce valid contours in some cases.

Well, by eye they looked reasonable enough. There are certainly some artefacts but I think the results are basically valid.

[Travis B.]

My daughter also seems to have fronting of GOOSE before /d/, because she pronounces food as [fʉːt] or even [fyːt] where I have [fuːt].

[Travis B.]

From checking, I have different vowel qualities in putting [ˈpʰʊɾɘ̃(ː)ŋ]~[ˈpʰʊɘ̯ŋ] and pudding [ˈpʰɵːɾɘ̃(ː)ŋ]~[ˈpʰɵːɘ̯ŋ].

[Raphael]

From checking, I have different vowel qualities in putting [ˈpʰʊɾɘ̃(ː)ŋ]~[ˈpʰʊɘ̯ŋ] and pudding [ˈpʰɵːɾɘ̃(ː)ŋ]~[ˈpʰɵːɘ̯ŋ].

Are there people who don't?

[Travis B.]

From checking, I have different vowel qualities in putting [ˈpʰʊɾɘ̃(ː)ŋ]~[ˈpʰʊɘ̯ŋ] and pudding [ˈpʰɵːɾɘ̃(ː)ŋ]~[ˈpʰɵːɘ̯ŋ].

Are there people who don't?

Historically both words have /ʊ/ -- and I have never seen described the vowel coloring I apparently get from /d/ after a vowel (hence why I never suspected it until I measured my formants).

[Raphael]

Historically both words have /ʊ/ -- and I have never seen described the vowel coloring I apparently get from /d/ after a vowel (hence why I never suspected it until I measured my formants).

Thank you. Looks like I misremembered how pudding is pronounced.

[bradrn]

I noticed yesterday that the first syllable of water seems to have a unique vowel. In quality it seems the same as NORTH/FORCE/THOUGHT, but it’s short rather than long. Does anyone else have this? Are there any other words where this vowel occurs?

[Travis B.]

I noticed yesterday that the first syllable of water seems to have a unique vowel. In quality it seems the same as NORTH/FORCE/THOUGHT, but it’s short rather than long. Does anyone else have this? Are there any other words where this vowel occurs?

Are you sure this isn't plain old vowel length allophony? Even most English varieties with phonemic vowel length have vowel length allophony on top of it (even though the details may differ, e.g. while vowel length allophony here is not sensitive to syllable boundaries, vowel length allophony in SSBE is syllable boundary-sensitive such that vowels are shortened only if the fortis obstruent falls in the coda of the syllable containing the affected vowel).

[bradrn]

I noticed yesterday that the first syllable of water seems to have a unique vowel. In quality it seems the same as NORTH/FORCE/THOUGHT, but it’s short rather than long. Does anyone else have this? Are there any other words where this vowel occurs?

Are you sure this isn't plain old vowel length allophony?

I think it isn’t, because I just realised it forms a minimal pair with warder [ˈwoːɾɐ].

[Darren]

I noticed yesterday that the first syllable of water seems to have a unique vowel. In quality it seems the same as NORTH/FORCE/THOUGHT, but it’s short rather than long. Does anyone else have this? Are there any other words where this vowel occurs?

Are you sure this isn't plain old vowel length allophony? Even most English varieties with phonemic vowel length have vowel length allophony on top of it (even though the details may differ, e.g. while vowel length allophony here is not sensitive to syllable boundaries, vowel length allophony in SSBE is syllable boundary-sensitive such that vowels are shortened only if the fortis obstruent falls in the coda of the syllable containing the affected vowel).

I will Praat my vowels when I have time, but I think the AusEng system is simply that coda glottalised obstruents shorten the preceding vowel slightly; so technically there are four lengths [pʰʊ̆ʔt pʰʊd pʰʊːʔt pʰʊːˑd]. This is very surface-level and the shortened long vowels (e.g. port, wart) never encroach on the plain short vowels (pud, would).

[anteallach]

If I measure the length of time between the end of the fricative noise for the /s/ and the release burst for the stop I get almost exactly the same time for sit as for Sid. However, the sound dies away sooner and the F2 in particular weakens earlier in sit, suggesting that in fact the vowel is a bit shorter and the closure of the stop a little longer with a /t/. (The release burst is also stronger with /t/ than with /d/.)

The difference is considerably larger for seat vs. seed; the time between the end of the fricative noise and the release burst is clearly and consistently longer in the latter, by about 0.04ms (about 0.26 and 0.30). Indeed seat isn't generally much longer than sit, but seed is much longer than Sid.

For PRICE there is another difference: in side there is a lengthy initial phase which really looks like a monophthong, similar to my TRAP but a little further back and maybe not quite so open; there is then the expected glide but it is quite late in the vowel. In site the initial phase is much shorter with a very obvious glide, although the starting point is similar (so I don't have much sign of "Canadian raising" for this vowel).

[jcb]

Meanwhile, I had a go at plotting my diphthongs (based on https://joeystanley.com/blog/geomtextpath/):

How are you exporting and then storing the data for each diphthong from Praat to use it in R?

Go to spectogram, zoom in on dipthong, Formants>‘Extract visible formant contour’, go to new formant object, Tabulate>List, then File>Save as. Then repeat for next diphthong. It’s annoying but workable when I don’t have too many words.

So, I extracted the formants, cleaned them up a little bit, plotted them with R, and got this cool image:

Figuring out how to smooth the lines in R like the later examples showed was too complicated, so I didn't do it.