Solution to Last Month's Mystery Spectrogram

Solution for September 2008

"Nothing beats a fudgey brownie."

So true, so true.

[n], IPA 116
Lower-case N
So I don't know what that noisy stuff is at the beginning. Maybe me wheezing into the microphone. The real nasal starts with the real voicing at about 75 msec. Note the nice zero at about 750 Hz and the nice clear pole at 1500 Hz. Ignore the murmur at 1000 Hz, which makes this look like a bilabial. Here's a generalization for you--if there seems to be both an alveolar pole and a labial pole, and the labial pole is weaker, assume it's an alveolar (unless the transitions are against you, which they're not here). Let's see if that holds up the next time this happens.

[ʌ], IPA 314
Turned V
Vowel. Low-to-mid, from the F1 at about 700. That's a little high of absolute mid, but probably close to mid enough (considering my range is from about 350 to about 950, maybe 700 is really mid afterall). F2 is just low of central, so this is a vaguely backish middish vowel.

[θ], IPA 130
Theta
Fricative. That's the easy part. It's sort of broad band and tilted to the high frequencies. So this might look sibilant. I've certainly produced sibilants that light before. But notice how strong the syllable (the preceding vowel) is, and even the following vowel. This is a particularly weak sibililant, if it's a sibilant, considering it's apparently stress-adjacent and fairly early in the utterance. So probably not sibilant. Unfiltered, so that suggests anterior (in the mouth, as in labiodental or interdental). Voiceless, by the way. Transitions are ocnsistent with something non-labial, I guess, so there you go.

[ɨ], IPA 317
Barred I
So here's a reduced vowel. Really. It's a little loud, but trust me. You can tell because the F2 is all transition, with nothing to suggest a separated target. So this is reduced. Trust me. And we transcribe it as barred-i due to the proximity of F2 to F3, following Keating et al (1994). But check those transitions in F2 and F3 approaching the following segment.

[ŋ], IPA 119
Eng
Fully voiced and high amplitude in the voicing bar, but right above that the F1 (or whatever you want to call it) is distinctly weaker than the previous vowel. There's also a pole or something just below 1000 Hz, and a stronger, broader one between 2000-3000 Hz. Note how the F2 and F3 transitions in the preceding vowel point a) together and b) to this general area. That transition pattern is pretty classic 'velar pinch' (although in the front part of the space) so this is a (fronted) velar. THe flat resonances, the absence of energy in the high frequencies and the general 'sudden'ness of the energy loss indicates a nasal.

[b], IPA 102
Lower-case B
Well, what we have here is a 'gap', an absence of any energy in the spectrogram. Indicating a plosive. Now, when we say 'no energy', we exclude the obvious energy down in the very lowest frequencies, i.e. where the voicing bar would be if there were voicing. Which in this case, there is. You can tell because of the nice even pulsing striations. So this is a voiced plosive. As to its place, the F2 and F3 transitions are rising sharply into the following vowel, i.e. the 'point' down into the closure, which is classically a bilabial cue.

[i], IPA 301
Lower-case I
So here's arguably the loudest vowel in the spectrogram. Fairly long, compared to the other vowels, this must be stressed. The F1 is low, indicating a high vowel. The F2 is outrageously high, up near 2250 or 2300 Hz, which is characteristic only of [i].

[t], IPA 103
Lower-case T
Anoterh gap. The transitions into it are falling (F2) and flat (F3), and the fall isn't 'sharp' the way the bilabial transition would be. It's 'pointing' at about 1700 Hz, rather than lower. So this looks sort of alveolar.

[s], IPA 132
Lower-case S
There is a segment here (probably we could tell that because of the noise is fairly consistent through the duration of the sound, roughly 75 msec around the 800 msec mark). If you thought it was just aspiration/VOT and maybe some noise associated with the release, the clues that you're wrong are that the 'release' moment isn't noticeably stronger and the noise doesn't die off from there, if you follow. For the most part, this is a very wide, stong band, centered in the highest frequencies, and so must be [s].

[ə], IPA 322
Schwa
Weak vowel. F1 and F3 are all transitional, which tells us this is reduced as well as unstressed. F2 is closer to F1, so following Keating et al (1994), I transcribe it as schwa.

[f], IPA 128
Lower-case F
Anoter fricative. It looks a lot like the previous one (the [s]) but is much, much weaker. Except the transitions in the surrounding vowels are all wrong. They all basically point downward, labial-like. So this this is a (strong) [f], rather than a weak [s].

[ʌ], IPA 314
Turned V
A lovely clean, clear vowel. F2 and F3 are pretty transitional, by which I mean 'straight', but the F1 suggest there's a 'target'--there's sort of a steady-state, followed by a curvy offset. So the F1, when it's flat, is sort of mid--just a hair bove 500 Hz. F2, though moving, is a little below central, and certainly on average is vaguely back. F3 is up where it doesn't tell us much, although it's rising. Don't ask me why. Well I have an idea, but I don't want to talk about it.

[d], IPA 104
Lower-case D
There's another, shortish gap, about 50 msec approaching the 1200 msec mark. The F3 in the previous vowel is clearly pointing up, which is consistent with alveolar, and the F2 transition also is heading toward that 1700 Hz mark.

[ʒ], IPA 135
Yogh
Fricative. Loud, i.e. high amplitude. And high frequency. Sibilant. Note that it seems to be centered in the visible frequencies rather than off the to, and dies off sharply below the F2 region. Post-alveolar.

[i], IPA 301
Lower-case I
Another high vowel with a very high F2.

[b], IPA 102
Lower-case B
Voiced plosive again, and look how sharply down past 1700 Hz the F2 transition into it is. Must be bilabial.

[ɹ], IPA 151
Turned R
So as the voicing begins just after 1400 Hz there's a weakness in the signal up to around 1500 or 1525 msec. But the resnances aren't flat, and there aren't any lower-frequency 'zeroes' characteristic of a nasal. So this looks like an approximant. F1 is just low of 'mid' and rises, F2 rises sharply to about 1500 Hz at the 'onset' of the vowel, and F3 has a (short) flat spot at about 1700 Hz. Just a coincidence, that frequency. But it's a decidedly 'low' F3, down in the 'territory' we associate with a (not particularly high) F2. So that's just 'low'. Low F3 = American [r].

[aɪ], IPA 304 + 321
Lower-case A + Upsilon
What we have here, folks, is a diphthong. When the 'vowel' starts, i.e. that moment when the energy above F3 kicks on, the F1 is quite high, indicating a fairly low starting vowel. At the same moment, the F2 is abotu 1500 Hz, or so. F3 we'll ignore since it's basically still 'recovering' from the previous sound. As the vowel smoves on, the F1 lowers, so the vowel moves from lower to higher, and the F2 also lowers, so it moves from centralish to backish. Or roundish, as the case may be.

[n], IPA 116
Lower-case N
Another nasal here. See how flat it is? See the zeroes? Can't be bilabial, with the F2 in the previous vowel isn't pointing down into the closure. While the F3 maybe droppng a little, there's no real degree of 'pinch' here. So probably alveolar. If the lower poles where clearer and less ambiguous, I'd say something about them. But they aren't, so I won't.

[i], IPA 301
Lower-case I
So weak and trailing off, look at that F2. What have we learned about F2s way up above 2000 Hz?