To properly view the phonetic symbols in the text below, you must have installed either SILDoulos IPA93 or Lucida Sans Unicode. If you are desperate to see phonetic symbols in SIL Sophia or SIL Manuscript, or some other kind of Unicode, drop me a line.

Solution for April 2003

"An example of a spectrogram"

The backstory: Towards the end of March, I got an e-mail from Dr. Stephen Linsday at UVic asking assistance with a figure for an intro psych textbook. He wanted an example of a spectrogram to embed in the discussion of the ear having to decode information encoded as frequency and amplitude. After tossing out a few subversive ideas ("Psychology sucks. Study linguistics instead" would have been too long anyway), I offered a version of this phrase, which now seems to have met the standards of the publisher. Woo hoo.

Ash, IPA 325,
[Q], [æ]
I guess I should have transcribed this as a schwa, but since I was sort of over-articulating, I didn't do nearly the amount of reduction as I usually do in this spectrogram. So this has moderately high F1, so the vowel is lowish. And the F2 is just a little high of neutral, so this is frontish. Lowish and frontish is probably /ae/.

Lower-case N, IPA 116,
[n], [n]
Looking back, I should have transcribed this as a nasal flap, but it didn't seem so short when I was doing the figure--it's not that much shorter than the /z/ coming up, or the /kt/ sequence coming up later than that. But this definitely has the quality of a flap--a short interruption in amplitude. The nasal quality is suggested by the presence of the resonances up above (regular flaps are basically super-short plosives, so they'd have gaps above the voicing bar), separated by zeroes. Must be alveolar if it's a flap. And nasal to boot.

Barred I, IPA 317,
[], [ɨ]
You'd think I'd hit the first syllable of a content word with a more distinct articulation, but I think I got trapped by my own rhythm. Having not reduced the preceding syllable, and having to stress the following syllable, I must have 'deaccented' or something, this syllable. Cuz this don't look like no [E]. It's short and unstressed (compare the amplitude of F2 in the preceding and following vowels--this vowel just ain't as loud), so I transcribed it accordingly.

Lower-case G, IPA 110,
[g], [g]
Looks voiced to me. Looks sort of like a gap, though it's mushy (and we know by now my velar closures tend to be mushy--I think it's my outsized uvula). The transitions into it are pretty typically (front) velar, with F3 falling just a touch and F2 rising (i.e. velar pinch). That F2 transition can't really be anything else. So it's velar and voiced (in case you weren't sure, it looks like it is). Doesn't look nasal, so that pretty much limits the options.

Lower-case Z, IPA 133,
[z], [z]
This is vaguely voiced, but it does get lost in the noise. We've got seriously broad-band (covering a lot of frequencies) noise here, getting weaker in the lower frequencies. It doesn't look classically /s/ like, in that it isn't clear the noise gets stronger in the very high frequencies (by which I mean frequencies off the top of the spectrogram). So beyond marking this as a voiced fricative, I'm not sure there's a lot of other information. The F2 transition into the next vowel suggests coronal (note the apparent rise from 1600 or 1700 Hz to a peak at just about 400 msec), but that's not much to go on. Still, there's nothing obviously non-/s/-like about it either. So it's a [z].

Ash, IPA 325,
[Q], [æ]
Well, there's a nice long vowel for you. This vowel starts just high of mid and goes very, very low (F1 rises). The F2 starts quite front--well, once it's done transitioning from the preceding--and then falls a bit, which is centralization. So technically, this starts at about [e] and falls to [a]. Well, that's not really a likely combination in English. So if you think this is two things, keep them. But if you don't, split the difference, and you get something around IPA [ae]. Lo and behold.

Lower-case M, Lower-case P, IPA 114, 101
[mp], [mp]
I hope somebody is going to study these nasal-plosive sequences really soon, cuz I Just Don't Get It. I'm not 100% sure about this segmentation, so I'm going to discuss these two at once. From about 500 msec to about, well 530 msec, there's some evidence of voicing. Then somewhere around then thre's cahnge. This (sort of) corresponds to the moment when there's some kind of transient aup above. This could well be the labial release, or it could be some kind of wild clunk, i.e. a transient associated with the sudden closure or opening of the velopharyngeal port, or some wad of something randomly flying around and hitting something. Euw, but work with me. Following that, the voicing loses energy and dies rapidly, as it would during a regular closure. So that's our evidence that there's two things going on here--a bit of 'open' voicing, followed by a bit of 'dying voicing', with a clunk of some kind in between. Unless the clunk is the release, in which case I don't know what's going on. So anyway, the apparent transition into this bit suggest labial (F2 and F3 falling), and notice they both rise slightly into the following vowel. So we have something vaguely voiced and possibly sonorant followed by something mostly voiceless and pretty definitely obstruent, both labial. Moving on.

Lower-case L + Mid Tilde + Syllabicity Mark, IPA 209 (155+428) + 431,
[l`], [ɫ̩]
The only real evidence that there's something gon gon here is the funny discontinuity in the F3 range here--that and the apparent zero in the F4 (while the F4 is low) that suddenly kicks off after about the 630 msec mark. Note the little hump in F3 at around then that indicates that this is an /l/. Not much to go on, but there you go.

Open O, IPA 306,
[], [ɔ]
I don't know where this vowel came from, but there it is. It's quite definitely mid (look at that F1). It's quite definitely back (look at that F2) and probably round. Ick. But there it is.

Lower-case V, IPA 129,
[v], [v]
There's something here. It's definitely less sonorous than either vowel around it, and there's a bit of noise between 1500 and 2500 Hz or so. So it's a fricative and voiced. There's not a lot of frication, and the voicing bar suggests something rather open (or approximant), so that lets out any of the 'strong' fricatives. Probably /v/ or Eth. Figure out which later.

Schwa, IPA 322,
[], [ə]
Finally, a reduced vowel that looks like a reduced vowel.

Lower-case S, IPA 132,
[s], [s]
Finally, an [s] that looks like an [s]. Extremely broad-band, very high amplitude, and higest amplituded in the high (as in above 6000 Hz, which you can't see but you can infer) frequencies.

Lower-case P, IPA 101,
[p], [p]
Well, there's gap from 875 msec or so to 950 msec or so. That's about all I can say. There's no real information into or out of it, but there it is. I'm very concerned that the first little bit of noise, or whatever that is upon release looks like it's even with the F1 in the following vowel, but just above the F2 and very possibly just below the F3. Looks a little like velar pinch. But that would be wrong. So it's voiceless, it's plosive, it looks slightly velar, but it's not.

Epsilon IPA 303,
[E], [ɛ]
Well, this vowel is a little low of mid, i.e. its F1 is a little above 500 Hz. It's a little bit front, juging from the F2, but it doesn't move at all the way both previous Ash-es move, so this is definitely a monophtnong. So lower-mid, or open-mid, depending on how you were trained, and front. It's worht noticing the F3 is falling slightly through the duration of the vowel.

Lower-case K, IPA 109,
[k], [k]
See, cuz there's *actually* velar pinch at the end of this vowel, which could conceivably be a clue that whatever is going on at the beginning of this vowel, it's not as velar looking as it might have been. So anyway, the onset of this gap is pretty definitely velar.

Lower-case T, IPA 103,
[t], [t]
On the other hand, there's no way that the offset of this gap looks velar. The F2 falls, the F3 falls, which looks coronal. Also, the release noise is sibilant and /s/-like, suggesting alveolar as well. So it turns out this gap represents the closure of two different plosives, and this one (the second one) is alveolar. And voiceless.

Turned R, IPA 151,
[], [ɹ]
The F3 is already quite low, and getting lower, which might lead you to just /r/. I hear something other than a syllabic /r/ here tho, although I can't really explain why. But the low F3 is the giveaway here.

Barred O + Rhoticity Sign, IPA 323 + 419,
[P], [ɵ˞]
Okay, I don't know what's going on here. Barred-O is sort of the round, higher-mid version of schwa, and this is definitely r-colo(u)red (low F3). I marked it as round a) because it seems to be round and b) I wanted to capture the F2 and F3 *lowering* through here, although that may be coarticulatory with the /r/ in the next syllable.

Lower-case G, IPA 110,
[g], [g]
Okay, my velar plosives just aren't very stoppy most of the time. But there's some indication, at least in the preceding falling F2/F3 is pinching, and the burst/noise in the middle of this gap (just ahead of 1300 msec) is centered in the F2-F3 range, which is typical of velars (for the same reason that velar pinch happens--discuss). So this is velar. It's probably not as fully voiced as I thought it was when I prepared the figure, but there you go.

Turned R, IPA 151,
[], [ɹ]
The /r/ is definitely there in the onset, in the sense that the F3 is about as low as it gets toward the [g]. Low F3.

Ash, IPA 325,
[Q], [æ]
Well, the F1 is fuzzy, but there's no reason for the voicing bar to suddenly go from 0 up to 1000 Hz if there weren't a high first resonance to support it, so this is a high F1, and therefore a low vowel. If you look at the F2 maximum, around 1750 Hz starting around 1400 msec, that's sort of the same range as the F2 in the two previous Ash-es, and just a hair higher than the preceeding Epsilon. So this is *really* front. and very low.

Lower-case M, IPA 114,
[m], [m]
Well, there's something sonorant that starts abruptly at about 1550 msec. SO it's voiced and sonorant, and it's probably nasal a) because the F1 of the preceding vowel is so fuzzy, b) because the nasal zeroes seem to kick in during the preceding vowel, and c) because if it were an approximant it would probably be more 'continuous' with the preceding vowel. So it's probably nasal. ANd judging from the sudden fall in F2 at the end of the preceding vowel (and sort of the F3) suggest bilabial more than anything else. ALso, there's just a hint of something at the left edge of this segment at around 1500 Hz, which is sort of where my bilabial nasal pole ends up.