I was performing some Catford-style "experiments" with nasal consonants, and found that slight opening of the mouth or rounding/unrounding of the lips has no particular sonic effect on, for example, [n], [ɲ] or [ŋ]. That matches intuitively with my understanding of nasal consonants, since it doesn't seem possible for any changes in front of the point of articulation to have an effect on a sound with no air past the articulation point. However, whenever I try to produce [ɴ], rounding/unrounding and mouth movement have a relatively strong effect—one that sounds a lot like a shifting formant. I hear no fricative sounds, so I'm virtually certain that I'm not accidentally producing some sort of nasalized [ʁ]. I suppose that leaves the possibility that, depending on rounding, I'm accidentally producing a nasalized back vowel such as [o] or [ɤ] (apologies, I haven't figured out how to add the nasalization diacritic).

I looked at the section in Catford on uvular consonants, and although it does mention a certain "messiness" for the stops [q] and [ɢ] (which seems potentially significant to my problem), it describes the uvular nasal as unproblematic and easily found by producing [ŋ] and simply lowering and bringing back the tongue.

My main question is: is this just the nature of [ɴ], or should it be possible to produce it in a manner such that changes in the lips have no effect? Secondarily, if indeed an ideal [ɴ] is possible, do you have any advice on how to produce it properly?


I suggest recording and measuring your measurements (but be careful to not totally believe the numbers). I found that in all cases including [n] vs. [nʷ], there was some difference in formants, though hearing it in [n] vs. [nʷ] was very hard. The acoustic differences are subtle and hard to characterize except with [ɴ] vs. [ɴʷ] which is quite clear. Unfortunately, I can't compute the predicted resonances and anti-resonances from the assumed vocal tract, so all I can say is that tube length matters, and there are indeed striking quantal effects going from [ŋʷ] to [ɴʷ]. I don't know whether there is new tech or insight into the problem of nasal zeros and LPC analysis (LPC assumes an all-pole model); you can see the problem in the spectrogram, that there are formanty band in some of the nasals between the LPC-detected formants.

The best way to mimic the IPA ideal is to practice with expert recording. The UCLA phonetic lab has an IPA table recorded by Peter Ladefoged, and John Esling has one likewise. For recording and analyzing, I recommend a good head-mounted mic -- forget cell phone or laptop built-in mics. Also, MP3 compression (and LPC analysis) may defeat your efforts, so use WAV format. Praat is a free and good tool for inspecting sounds.

  • Thank you, and those links are very helpful. I've seen several sites like those, but these are especially comprehensive. So if I'm understanding you correctly, you're saying that I'm right that [ɴ] has substantially more variance with lip changes, but wrong that the other nasals have none? I'm not sure how to go about recording and measuring my sounds, I assume you're talking about using some sort of spectral analysis? Jan 5 '17 at 15:12
  • Also, do you know how a secondary articulation can affect the sound of a nasal? I understand that tube length matters, but, at least theoretically, isn't the oral cavity in front of the articulation completely excluded in a proper nasal? Jan 5 '17 at 15:15
  • 1
    I think it is explained by passive linkage to the front resonator, i.e. sympathetic vibration, which is why it is very low amplitude.
    – user6726
    Jan 5 '17 at 16:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.