Class notes from July 27

One of the preconditions for the development of distinctively nasalized vowels exists in English.

"demonstration experiment"

Phonetic base for phonology:
The case of labial-velars, [kp, gb, etc]
Chomsky and Halle(1968): determined by phonological evidence, not phonetic evidence
[m,n,ng,w] cavity length (of anti-resonance): m>n>ng>w
w->devoiced/_t e.g. 'lieutenant'

Intensity: inherent intensity + the way of sounds modified when it passes through a resonant cavity.

# Phonological phenomena are determined by basic physical phonetic principles, not abstract taxonomic caracter.

Ohala 1990 is a "demonstration" experiment.
The "acoustic modulation" theory of phonotactics shows the goodness of phonetic knowledge in phonological study.
"Feature geometry" (Clements 1985) wouldn't work for the labia-velars and assimilation test, and the fact that [velar] place influence [voice].
Anderson (LI 1981 "Why phonology isn't "natural") <-> we are making the area of ignorance smaller and smaller. More fundamenta folly is that: the sound pattern and its change are led by grammars, not by the listeners.
between ~ betweenity....so this phonological rule is not very productive.
"it is going to necessary to discover conditions on theory cnstructions...."
(Chomsky 1967, Verbatim discussion in Brain mechanism underlying speech and language)
Just by examining surface phonology, it is hard to get what's inside speaer's brain.
The most important attitude in science is doubt, and eExperiments are the way to resolve doubt.
rationalism vs. empiricism: In my opinion, generative phonology is more like logic, and phonetics is more of science. I am strongly in favor of phonetics and acoustic phonetics, but to me it seems that there exists mental abstract representation in human mind. So, logic is done with abstract features (like numbers and symbols in math), and science is done in domain of phonetics and acoustics. Of course the two domains must not be separated, and we need a fine method to combine the two areas together.

Applicable answer

Applicable answer

Q1. The airflow change in the sequence of sounds [hih....]: When pronouncing [h] in front of the vowel [i], OQ is bigger in the portion of [h] than in the portion of the vowel, and so is the airflow. Since there is more constriction at the vocal folds when pronouncing a vowel, not just OQ but also the amount of the air goes through the vocal folds is less in case of vowels than in case of [h]. The vocal folds are more open when pronouncing [h] than when pronouncing a vowel, so it's like a breathy phonation.

Q2. OQ change in 'who' (time 0.72-0.82 sec): OQ is small at the beginning, increases toward the middle, then decreases again.

Q3. In the [h] of "who", EGG shows that there is no vocal folds vibrating, but strangely, acoustic effects are observed, even though it's damped. Also interestingly, we see the oral airflow is oscillating. The acoustic effects seems to come from the oral airflow fluctuations. The reason I guess: after vocal folds stop vibrating, a small amount of airflow comes out of the glottis, which is still slightly vibrating. Acoutstic effects come from the weak vibration of the remnant airflow.

Q4. In the [z] of "has", we see the phenomenon which is just the opposite of what we observed in Q3. There are glottal oscillations, but no acoustic effects.

Q5. Looking more closely at the OQ, we see that at the transition of one vowel to the other, the OQ decreases, which evidences presence of glottal constriction.

Q6. The initial consonants [p], [k], [t] of "applicable", "keep", "to" have big non-oscillating airflow, which suggests that the consonants are aspirated without vocal folds vibation. The amount of airflow is smaller in [k] of "keep" than [p] of "applicable", maybe because the place of articulation is closer to the microphone in case of [p] than [k]. Measurements of VOTs are as follows:

VOT
[p] of 'applicable' 64ms
[k] of 'keep' 57ms
[t] of 'to' 19 ms

Q7. In fact, the final stop in "keep" is not actually a final stop. It seems that [i] in "keep" is devoiced, and [p] has only a weak burst, parsed like the onset to "it". Anyway, we can see a short period of glottal vibration at the end of "it", rather than burst of [t] release: because the airflow is very small. However, it is a bit strange that where glottis is open, the airflow is small.

Damp Skunk

Q1. 'damp skunk' without /s/ sounds like 'damp gunk' (or 'damp kunk'), I am not quite sure.

Q2. /s/+'damp' sounds like 'stamp'. Actual articulation is voiced, but after /s/ I hear it as a voiceless. I hear what I'm used to, not what is actually produced.

Q3. By swapping stop burst, 'damp skunk' sounds like 'daŋk skump'. Again, this is the evidence for default perception. Positional assimilation alone cannot explain this kind of misperception.

Q4. It sounds like 'daŋk'.

Q5. 'dam', or 'damp' without release. I cannot decide which it is, maybe because I already know this word.

Q6. The steady state portion until the formant transition part sounds like 'dae'. When the formant transition part is included, it sounds like 'dae' followed by some labial consonant. I hear it as dam(p), maybe because I know the word already. I wonder what a naive listener would classify them: It could be dap, dab, dam, I guess, becauseony acoustic information is articulatory movement to some kind of labial sound, or formant trasition.

Q7. From the beginning point of 'kunk', I increased the length of the selection I hear. It changes from 'kə' to 'kəŋ'. Why we don't get 'gun' from 'kunk' is that, I guess, it's partly because the place of k closure before /ŋ/ is different from /n/; the former will be backer than the latter. And the difference between the place of articulation of /k/ will make change in perception. Or maybe not?...

How do we perceive, by signal or inference?

Class notes from July 17th.

[1] Signal vs. Inference
Acoustics often does not include the whole set of features. Ambiguity is inherent in acoustics.
“Inferential power of listeners”:
Stevens's view: how you get features is out of signal only. He is reluctant to refer to listeners’ inference. Recognition models (e.g. merge model of recognition, (Denis Noris) assume that signals carry everything. However, even if nasal sound is missing, listeners parse nasals. Listeners perceive words despite of undershooting of articulation ("Phonetic inferencing"). In my opinion, recognition model cannot include inferential abaility of listeners on its first stage. Recognition based on signal (only) and recognition by inference seems like the events that take place in two different modules. 'Recognition' is based on signal, a purely physical level. 'Inference' is based on knowledge, everything a person knows about the world.

Chilin: "There is an implicit way, a weighted model. If there is not enough information, context overrides. This is a linguists’ view of engineering."

[2] The Aerodynamics of nasality

Professor Keith Johnson demonstrated the equipments that measure oral/nasal pressures.
General purpose transducer amplifier has four channels.

1)? labial: oral air pressure goes up.
2) nasal air pressure
3) oral air pressure
4) acoustic channel

[3] We took a look at professor John Ohala’s record, using the equipments that measure oral/nasal pressures and DAQalzer. The recorded words are:
amba
damnation
pap’a (ejective) ….clipped
baba (implosive) ..minus pressure

(The transducer amplifier couldn’t capture ejectives’ pressure.)

A low vowel lowers soft palate, changes the volume of nasal cavity and resulting nasal flow.
(not the source of slight difference of F1 for nasalized vowels from non-nasal vowels)

notes from the previous class [2]

This is a note from the previous class by John Ohala.

[1] Historically, consonants with high airflow (such as s and h) often leads to nasalization. Experiment: /pas/ (1) Lengthen the steady state of /a/ (2) cut off /s/, and you will perceive nasalization.

[2] A place-assimilation joke: Do you know where to find mangos?; wherever a woman goes.

[3] Geminates promote voicelessness.

[4] Voicelessness and pitch
S.Kammu -> N.Kammu
kla:ng klá:ng
gla:ng klà:ng

* Similarly in Korean, vowels following tense stops usually have higher pitch than those following lax stops.

[5] Which controls F0, larynx or lungs?

notes from the previous class [1]

This is a note from the previous classes by professor John Ohala.

[1] Velars tend to be devoiced more easily than labials. Labials have higher compliance than other consonants.

[2] Default perception and source of noise
"slit" sounds like "split" by adding silent in between s and l. This is probably because /p/ is a consonant with the weakest burst, since it has no downstram resonating cavity. On the other hand, apicals have a small cavity through which airjet goes and thereby noise is produced. Another source of noise: when airjet hits the teeth, which is the case of apicals.

[3] Articulatory assimilation vs. homorganical percept?
Replacing 'an' in 'anta' with 'am' in 'ampa' result in a sound perceived like 'anta'. Even if what was articulated is [am], people hear it as [an] in the environment of [t]. This could be evidence for the hypothesis that people perceive homorganically. This also can be related to 'default percept'. Ear mishears, and brain misinterprets. Maybe this is because our ears and brain know what is going on inside our vocal tract? I think both take part in the process of production and perception of human speech, or the speech chain. They are not independent activities.

[4] tiny click in hymnal
When hymnal is whispered, there is a very tiny click sound, which is produced by a small chamber created by gestural transition from /m/ to /n/. This can be heard only when it's whispered because it's very quiet.

[5] F0 modulation
In fact, the difference between cone and co-own is difference in F0.