Solution to Last Month's Mystery Spectrogram

Solution for February 2007

labeled spectrogram
"A bell adorns the gate."

So, this month, or rather last month, we continue the 'return to basics' trend, with a bunch of voiced plosives. Check out those transitions. The goal here was to get each voiced plosive with a preceding schwa, so you could see the transitions. Well, I tried.

[ə], IPA 322
Schwa
Well, it's kind of short, and the F2 is closer to the F1, so this is a schwa. The F1 falls from about 700 to about 500 Hz, F2 is steady about 1300 Hz and F3 is a little high, around 2600 or 2700 Hz. Starts with a 'hard pulse', but not really a glottal stop. Vaguely midish, and not amazing backish. Hmm. Well, it was supposed to be a schwa, perhaps opened a little by a) overarticulation and b) utterance initialness.

[b], IPA 102
Lower-case B
Just over 100 msec of gap, with evidence of relatively strong voicing for the first, I don't know, two-thirds or so. That's as voiced as a plosive this long is ever going to get, due to aerodynamics. Fairly sharp burst, by which I mean there's a very narrow vertical line accompanying the onset of voicing. It's much 'sharper' than my plosve releases usually are. Note that it doesn't have much in the way of high frequencies, and the Transitions all trend down into it, and up out of it. F4 especially. Can only be a bilabial.

[ɛ], IPA 303
Epsilon
Well, everything is mjoving, so we're going to have to do our best. F1 starts at about 500, and rises a little. Owing to the lowness of the transition, I'd say that overall this vowel looks like it's basically a little more open (higher F1) than mid, so we're talking a midish, to lower-midish kind of vowel. F2 is about as neutral as you can get without actually trying, and trends down (but that's clearly in transition to the following minimum. F3 is rising from neutral. So I'd say from 300 msec to about, well 400 msec, we've got a vaguely mid-to-lower-mid, not very front but not at all back vowel. Which brings us to the [æ - ɛ] part of the space. And for me [æ] is usually going to be longer. But whatever. Somewhere down there.

[ɫ], IPA 209
Tilde L (Dark L)
SO from 425-500 msec or so there's this region of full voicing, resonance, with a funny zero or something between F2 and F3. So we're dealing with something basically open and sonorant, but with less overall energy than a vowel. But continuous formants with the surroundig vowels, so this isn't a nasal. And anyway, there's much too much F1 for it to be a decent nasal. So this is an oral approximant. F1 is, wlel, a little lower than mid, so it's sort of higher-mid, but not at all high. F2 is about as low as it gets for me, aroudn 1000 Hz. F3 is definitely higher than anywhere else in the utterance, although it's a little fuzzy. So the low F2 indicates something very back, think velarized, and the raised F3 is usually a good indicator of laterality. Therefore, dark /l/. An /r/ would have a low F3, and a /w/ would have a lower F1 and an unraised F3.

[ə], IPA 322
Schwa
another short, inconsequential vowel. Trust me.

[d], IPA 104
Lower-case D
But the transitiongs. F4 is doing something weird, but it basically straight. F3 may be trending down just a little, but since it was so high it can hardly do anything else. And not at all 'dropping like a rock' like you might get with a labial. F2 is not trending down at all, which leaves us alveolar or velar as possibilities, and there's just no way the F3 and F2 are 'pinching' together, even if you believe the F3 is trending down. So probably this is an alveolar. Longish closure duration, and quite strong voicing for the first three-quarters or so, so definitely voiced.

By the way, if you're wondering, a voiceless plosive can show some voicing energy during the closure, especially in codas. But typically the voicing is weaker, dies off quickly (this and the previous [b] the voicing seems sort of 'steady'--cf the coming plosive), and usually doesn't last into the second half of the closure. Just a rule of thumb.

[o], IPA 307
Lower-case O
So everything's moving again, but the F1 is basically straight, around 600-700 Hz. Theres a 'moment of interest' where the F2 reaches a minimum, at around 750 msec, and more or less at the same moment, F4 reaches a minimum and changes direction too. So I'm going to suggest that that 'moment' is important, and the fact the F3 is just sliding through it just means that it's sliding on to its big 'moment', a bit later. So at the 750msec moment, F1 is mid or a little lower than mid (although nowhere near as open as the end of the second vowel, before the dark /l/ kicks in), so basically pretty mid. The F2 is low, indicating something fairly back and round. The F3 we can ignore because it's on its way somewhere else. SO mid-ish and very back and round. Not that many possibilities in my voice (western American English).

[ɹ], IPA 151
Turned R
SO the enxt moment where anything 'happens' is at about 850 msec, whe F2 and F4 are moving, but F3 hits its minimum. F1 is still mid-ish, F2 is as high as it can be without hitting F3, and the F3 is down around 1700 Hz. When F3 gets that low, it's just gotta be an [ɹ]. Ignore the F4. It'll just confuse you.

[n], IPA 116
Lower-case N
Well, this is a nasal. Nice strong voicing, but F1 just goes away. There's a region of low energy (a 'zero') just above that, and there's just a little nergy around 1500 Hz, which is never there in a bilabial nasal. The transitions into the closure don't suggest a bilabial either. Nor do they 'pinch' as for a velar, so again we've probably got an alveolar on our hands.

[z], IPA 133
Lower-case Z
So around 1000 msec, the voicing continues but drastically loses energy. A the same time the upper resonances switch off, and there's a trace of noise in the very high frequencies. This is pretty typical of voiced [z]. The frequency and bandwidth of the noise is consistent with [s], but very, very weak, which is consistent with the attempt to maintain voicing.

[ð̝], IPA 131 + 429
Eth + Raising Sign<
There's a short gap following the end of the sibilant noise, apparently around 1075 msec or so, which is also when the voicing dies away. But the following bit of noise kicks on at about 1100 msec, has a different spectrum to it (it's got more stuff in the lower frequencies, and it's 'shaped' into resonances a little more), and the noise seems not to accompany a burst of any kind. So we've probably got another fricative, but this one 'raised', i.e. partially occluded. The 'voicing' I get from a) knowing the answer and b) how short both the gap and the frication phase are (underlyingly voicless fricatives tend to be longer overall, and louder). Can't be sibilant, which leaves the labiodental, the dental, and [h]. I guess I could make a case for the noise being continuous with the resonatnces of the voicing in the following vowel, but that's really only true of the F2. So probably not an [h]. But how you tell the difference between [v] and [ð] I'm not so sure. Let the lexicon sort it out.

[ə], IPA 322
Schwa
Now this is another short vowel. Technically, I should have transcribed it as a barred-i, following Keating et al. (1994), but I didn't. Sorry.

[ɡ], IPA 110
Lower-case G
ANother voiced gap. This one has weaker voicing that dies off, but it still has voicign for most of the closure duration, so in the balance, I'd say this was voiced. If the voiced closure duration were shorter, relative to the voiceless duration, I might say voiceless. But not aspirated. Ennyhoo, See how the F2 and F3 in the reduced vowel seem to be pointing together, and how in the following vowel the still seem to be together? That's velar pinch. So this has to be a [ɡ].

[e], IPA 302
Lower-case E
So now we have a vowel. F1 is a little low, but still basically mid-ish. F2 is very high, around 2100 Hz, and very flat. F3 is neutral, but if the F2 were any higher it would end up pushing the F3 out of the way. So we've got somethign mid, but very, very front. And not diphthongized, in case anyone was wondering.

[t], IPA 103
Lower-case T
And one last gap, from 1450 to 1550 msec, where it gets released into some noise which is much stronger in the higher frequencies than the lower. That is, it's suspiciously sibilant looking. Specifically like a short [s]. What plosive releases into something which could be [s]-like? That's right, [t].

Et voilá there you go!

So, how'd you do?