Depending on which font(s) you have installed, one or the other symbol in paragraph headings may not display correctly. To properly view the phonetic symbols on this page, you must have one of the following fonts installed on your system.

All are available freeware. I'm committed to keeping my recommendations to a) freeware fonts with b) decent looking IPA compliant symbols. If anybody has any recommendations for freeware Unicode fonts with good looking IPA support, I'll test them out and add them to the style sheet.

Solution for October 2003


"They're made of reclaimed wood."

Trivium: It was going to be 'they're made from reclaimed barnboard', but I couldn't get a "fr" I was happy with, and I decided 'barn board" would just be too much rhoticity for anyone but me to be interested in.

Eth, IPA 131,
[D], [ð]
Starting just a bit after 100 msec, there's some fairly significant voicing going on, but without a lot in the way of resonance. So there's something voiced going on here. I probably should have transcribed it with a Raising Sign, since it looks pretty much like a stopped one of these, whatever it is, but I didn't. Maybe I saw more frication on the original. Oh well. Clearly voiced, probably obstruent, not at all fricative, so either this is a plosive or a very weak (in the sense of having no noise--in other words, "strong" in an absolute consonant/fortition kind of way) fricative. The F3 is a little ambiguous, depending on how you interpret the transition at 200 msec, but isn't overwhelmingly bilabial or velar looking. The F2 transition around 200 msec definitely points away from velar and doesn't really point 'low' enough to indicate bilabial. So this is probably coronal. So if it isn't Eth (or [d]), I don't know what else it could be. I've seen high F4s associated with dentals, but not consistently enough to be sure if that's "always" true.

Epsilon + Rhoticity Sign, IPA 303 + 419,
[EÕ], [ɛ˞]
Following convention, I've transcribed a lax vowel before an /r/ (not to give anything away, but there you go), as well as with rhoticity (ditto). And if you want to compare it with "true" [eI] diphthongs coming up later (ditto ditto), you'll see that the F1 is in fact just a little bit higher (indicating a slightly lower vowel, although still not what I'd really want to call 'mid'), and the F2 is definitely a little lower (though that may have to do with the rhoticity, i.e. the F3 pushing down into the F2 space), which may indicate something less front. So you pick. Still a very front vowel, not at all low, with a mid-ish rather than a high-ish F1. And r-colo(u)red.

Turned R, IPA 151,
[¨], [ɹ]
I was convinced that the F3 (and F1) leveled off a little here, while the F2 kept on sinking, so warranting a distinct segment, but now I'm not so sure. But this F3 approaching 2000 Hz or below is usually a pretty good indicator of a North American English /r/ floating around somewhere.

Lower-case M, IPA 114,
[m], [m]
Well, here's a pretty good nasal, for those of you who were clamo(u)ring for one. It's clearly of lesser amplitude than the vowels on either side, has good sharp amplitude boundaries with very sharp transitions on either side. Good healthy zeroes between most of the formants, and in particular a band of distinctly low energy (i.e. a zero) below 1000 Hz. All those zeroes indicate side resonance which is usually a good indicator of nasality, especially combined with the transitional information. The pole (weak as it is) above the low zero is about 1000 Hz, which is typical of my bilabial [m]s.

Lower-case E + Small Capital I, IPA 302 + 319,
[eI], [eɪ]
Well, the F1 here is just low of neutral, so let's call this a vaguely high or higher-mid vowel in traditional terms. Even abstracting away from the absurd F2 transition, there's indications that there's a less-front onset before the very front position the F2 assumes around 450 msec. Which is about as much of a distinct [e] you're ever going to see. Acoustically, these always struck me as more like mid-onglided [i]s than [e] with an offglide. Clearly if there's anything you want to call 'steady', it's the 'glide' portion, and not the nucleus. But that's I fight I need to have with Hillenbrand inter alia once I have my new data properly measured.

Fish-hook R, IPA 124,
[R], [ɾ]
Ah, good ol' flap. This is a very (very) short interruption to the resonance on either side, like a tiny plosive--the closest thing you'll ever see to a concrete demonstration of one-mouth/two-mouth theory in spectrograms. This gap is even short for a flap, looking like it's around 20 msec long or so. Apparently fully voiced throughout, there's nonetheless some disturbance to the airflow or something following the 'release', such that it looks vaguely aspirated (though not long enough to count as aspirated in the "chinchilla/Japanese quail" sense, even if it weren't clearly voiced). The F2 and F3 transitions into and out of it are helpful, pointing fairly clearly to alveolar (no pinch, with an F2 locus around 1750 Hz or so), even though we don't usually see amazing alveolar transitions with flap (they just happen too quickly and ballistically for everything to transition correctly). Maybe this is really a short /d/. But more on the notion of short [d]s later.

Schwa, IPA 322,
[«], [ə]
Short(ish) vowel, nothing but transition, and probably r-colo(u)red to boot. Oops. F2 slightly front, so this is probably better transcribed as an r-colo(u)red barred-i. But the point is that this vowel is all transition, with nothing to indicate a robust 'target'. So call it reduced (or schwa) and move on.

Lower-case V, IPA 129,
[v], [v]
Well, for once in our lives (actually, I'm usually careful about these things in spectrograms), a truly fricative fricative. This fricative is even voiced (judging by the striations in the low frequencies). In the upper frequencies (above 1000 Hz) there's some very broad-band energy, and except for it apparently being strongest in the F2-F3 range (the F3 range is low for, um, other reasons, so this is basically just the F2 range), there's not a lot of spectral 'shaping' to the noise (note that F4 in the surrounding vowels is kind of dead in the fricatives). The energy in the noise is not what I'd call outrageously strong, either. So this probably isn't a sibilant. (If it were a sibilant, it might well be Esh, or rather Yogh, but the zero you see below the F2 range is a little low for those.) So if this isn't a sibilant, it must be a voiced [h], an Eth, or a [v]. It's probably not an /h/, which would have more obvious-looking resonances. So we're left with one of the others. Unfortunately, due to the F3 frequency, there's not a lot to tell us that it's [v] and not Eth. The transitions are equivocal, and the F4 isn't telling us much.

Turned R, IPA 151,
[¨], [ɹ]
Well, there's a canonical /r/ for you. F3 well below 2000 Hz, well-defined F1 and F2 below it. There's even some energy in F4, sort of. But the frequency of the F3 is the giveaway. I've placed the boundary between the /r/ and the following vowel a little arbitrarily, but it seems like the F3 starts to move for real at about 725 msec. The F2 (which is also moving throughout, but more obviously than the F3) I think changes slope about that same moment, so I figure that's where the boundary might be, if we have to put it somewhere.

Lower-case I, IPA 301,
[i], [i]
Well, abstracting away from the necessary transition from the preceding /r/, we've got something that's heading to higher than mid (I think the F1 goes from lower-than-where-it's-been to even lower than that--but not by much, I admit), and extremely front (high F2). When the F2 gets above 2000 Hz, I always assume it's an /i/. Even very strong /j/ offglides don't often get up that high. On the other hand, it would be easy to be distracted by the onset F2 frequency, which is a little depressed due to the preceding /r/, into thinking this was an /ei/ or something. But it's not.

Lower-case K + Right Superscript H, IPA 109 + 404,
[kH], [kʰ]
Well, in addition to the F2 screaming up to 22O0 Hz or wherever, the F3 doesn't seem to be racing out of its way, such that the two look like they'd collide. Or 'pinch', depending on your point of view. Hence this is probably a (very front) velar. It's apparently voiceless, and there's a fair amount of noise following the release. (I take the release to be that transient at about 850 msec. If that's the case, then the actual VOT here is close to 100 msec, which is fairly long, even for a velar. But nonetheless, falls well to the aspirated side of things.

Lower-case L + Mid Tilde + Under-Ring, IPA 155 + 428 + 402,
[lò8], [ɫ̥]
Well, this isn't the darkest of /l/s and I probably shouldn't have used the mid-tilde for velarization here. The voicelessness is due to the prolonged aspiration of the preceding /k/, but it is typical for the second member of this kind of cluster to be decidedly voiceless, extending the aspiration well through the 'duration' of the second segment. If you believe in segments. There's a change of some kind at about 890 msec or somewhere, probably not a closure moment, but I don't know what it is. But there's also a change in the trajectories of all the formants at or near the moment when voicing finally kicks in, suggesting a target independent of the aspiration. The F2 is rising out of it, but the F3 and F4 are both raised during that last bit of aspiration, so /l/ is probably the best guess.

Lower-case E + Small Capital I, IPA 302 + 319,
[eI], [eɪ]
This one is a little more canonical. The F1 seems to be pretty mid throughout, the F2 rising from vaguely neutral to quite front (before dropping sharply into the next segment. So mid and front-moving. Not a lot of choices.

Lower-case M, IPA 114,
[m], [m]
The sudden change in the bandwidth of F1, which corresponds, more or less, to the bottom of the F2 drop (which begins at a moment of sudden loss of amplitude a few msecs earlier), tells us something is going on here. And the best guess is some kind of nasal, given the zero-ey quality of the upper frequencies. The F2 transitions, and the 1000 Hz or so pole suggest the bilabial nasal, although there's more energy at 1500 Hz (the /n/ range for me) than I'd normally like. But there may be a reason for that....

Lower-case D + Subscript Arch, IPA 104 + 432,
[d9], [d̯]
Okay, controversy. It's typical for combinations of nasal and voiced plosive to look like this. A stretch of nice normal nasal, followed (if your lucky) by a miniscule gap, and an obviously oral release. Which is what we have here. So I've decided that since this isn't a flap in the usual sense, but a plosive whose closure is hidden by (or coarticulated with, or something) the preceding nasal, it's just 'short'. So I used a breve. Okay, bad solution. I'll keep thinking. (I think the UCLAbet recommendations distinguish these with a nasal closure and an oral release, but the IPA doesn't have different symbols for closure and release phases.) Ennyhoo, since the preceding nasal is bilabial, this plosive is obviously bilabial, right? Oh, I'm sorry. This is actually a coronal release. Its coronality may be indicated by the 1500 Hz pole that doesn't seem to belong in the otherwise bilabial nasal, or the apparent F2 transition (has anyone done an acoustic study of nasal-to-nasal coarticulation?). That transition (as we approach the gap at 1100 msec, from that strongish harmonic at 850 Hz or so up to the 1500 Hz mark) may be the tongue blade rising behind the bilabial closure. Or maybe not. I don't know. The release noise is centered in the F2/F3 range, which makes it look velar rather than either bilabial or alveolar. I hope that has something to do with tongue body movement anticipating the upcoming [w], but hope may not get me too far. The release spectrum could conceivably be bilabial, and it definitely doesn't look classically alveolar. But the transitions coming out of it are definitely not bilabial. They look velar, actually. But I think they *could* be alveolar, shaped (that is, lowered) by the combination of the bilabial release beforehand and the rounding coming up. But I'm really treading water here, since the lip rounding clearly gets rounder until you get to about 1200 msec. So those of you who think this is a [g], go to the top of the class, but then try to makes sense of the whole spectrogram. Sometimes, life just doesn't work out the way we want it to.

Lower-case W, IPA 170,
[w], [w]
Well, ignoring the release of the preceding plosive, and sort of ignoring the amazing transition going on, there's a funny drop in amplitude about 25 msec after the release of the plosive, which continues to almost 1250 msec. Lower amplitude like this probably means a very close articulation, and therefore this is probably an approximant. (Nasal is another possibility, but the formants are too well defined, I think.) F1 is just sitting there about 400 Hz, but the F2 is as low as my F2 ever gets, down there around 800 Hz. Then there's this huge expanse of nothing until you get up to the F3, which is, well, slightly raised. What I can see of it, at least. So this is either another /l/, or I could take my cue and claim that the F4 seems to be a little low, and wonder if this isn't a [w]. Which would be right, but don't ask me how you should be able to tell for sure. I need to work on my approximants.

Upsilon, IPA 321,
[u], [ʊ]
Well, its a little higher than mid, judging from the F1, although it might be lowering ever so slightly between 1300 and 1400 msec. The F2 goes from incredibly low (back and round) to sort of nowhere. So we've got something that starts vaguely in the higher-backer-rounder part of the vowel space, and moves, if anywhere, towards schwa. Sounds like an Upsilon to me.

Lower-case D, IPA 104,
[d], [d]
Well, there's definitely about 100 msec of voicing starting just shy of 1400 Hz. So there's something there. It probably isn't an approximant, due to the sudden loss of upper frequencies. It probably isn't a nasal, which probably wouldn't kick on that abruptly. There'd be *some* anticipatory nasalization, wouldn't there? And there's absolutely no evidence of any higher resonance going on. Looks like a pretty good candidate for a plosive, although the voicing goes on an awfully long time. But if it's a plosive, it doesn't look velar (no pinch, to speak of) or bilabial (F2 is definitely heading up), and, lo and behold, doesn't it seem just to reach 1700 Hz before the closure hits? How's that for an example of locus equation theory? So this is probably alveolar. And voiced.