full-fledged pidgin

Question 1

What does “skunk” sound like with /s/ gone? How do you account for this
transformation?

/skʌŋk/ with the /s/ removed now sounds like [gʌŋk]. This is because the so-called `voiced’ velar plosive in (at least American) English is actually a voiceless, unaspirated velar plosive with a lower fundamental frequency.

Question 2

What does “damp” sound like with the /s/ in front of it? How do you account for
this transformation?

/s/ + /damp/ now sounds fairly convincingly like [stæmp] (convincingly enough to fool my linguist roommates). I suspect two possible explanations for this phenomenon:

1. English phonotactics prevent the listener from hearing /d/ even though it is acoustically distinct from /t/.
2. The expected /d/ and /t/ sounds are, as described above, more similar than they are dissimilar and the listener uses context to distinguish the two.

These are surprisingly difficult to tease apart with the data I have available. One approach might be just to collect a few hundred /d/ and /t/ productions in various contexts, measure them, and then quantify the similarity or dissimilarity of the two sounds for the speaker. As it is, though, I don’t even have one example of a /t/ for this speaker. Another approach might be to record a speaker saying the minimal pair /tæmp/ and /dæmp/ (in a frame like ’say /tæmp/ again’ and with some distractor words so he doesn’t artificially enunciate them). Again, though, I feel I should be able to make this distinction with the available data.

A third approach might be to move some other sound in front of the /dæmp/ to see if another sound (e.g. /æ/ also causes the /d/ to sound like a [t]. I did this and it does not, for me, make /d/ sound like [t], but this also isn’t a real word so perhaps I’m not listening to it as speech.

I think there’s a clue to this puzzle in the fact that moving the /s/ too close to the /d/ (anywhere from 0 to about 40ms in my informal testing) fails to cause the changed perception — /d/ still sounds like /d/. This /d/ does not differ from a /t/ in terms of voicing and the solution isn’t one or the other of my two stated possibilities but the unification of the two. This perceptual shift is the same as /kʌŋk/ –> /gʌŋk/ in question 1. With the context removed (or, in this case, added) we here the /k/ as a /g/ because, for all intents and purposes, that context is what’s different between the two sounds.

Question 3

Swap the stops. What do the words sound like now? How do you account for these
transformations? Remember, you did nothing to interchange the nasals in these words.

I suspect that it is now supposed to sound like [stæŋk'gʌmp], but I get something more like [stæŋk'gʌŋmp]. /stæmk/ definitely sounds like [stæŋk] to me, but I can still clearly hear what sounds like an /ŋ/ before the [mp] in `gump’. This clearly, I think, is interference from English phonotactics. There’s nothing /ŋ/-like about this /m/ and there’s nothing /m/ like about this /ŋ/. Incidentally, if I remove one of the pulses from the nasal at the end of `gump’ the /ŋ/ disappears and all I can hear is the illusory /m/. This is incredibly cool.

Question 4

Is it still “damp”?
No, but now it is [stæŋk] (I think that logically follows).

Question 5

When you delete the final burst of “damp” what does the word sound like? Where
did the velar nasal go?

Now it sounds just like /dæmp/ with an unreleased /p/ (I can’t get the upper corner/unreleased diacritic to work). Since there are no longer any signs of the /k^h/ release burst there are is no phonotactic motivation to perceive the /m/ as an engma.

Question 6

Can you get it to sound like “damn”? What does it sound like? Why shouldn’t this
procedure succeed in getting “damn”?

No, I can’t get it to sound like `damn’. Three (related reasons): (1) `damn’ has a longer /m/ consonant in it than `damp’ does. (2) There is too much coarticulatory information from the /p/ closure on this /m/. The way to get rid of it would be to shorten the consonant (see also reason #1). (3) One might expect that we could just chop off the end of the existing /m/, extend the remaining consonant by a dozen or so milliseconds, and have `damn’. Unfortunately, there’s also a lower pitch in the /m/ of `damn’ than in the /m/ of `damp’. I believe this is related to the duration of the segment — in the shorter /m/ there’s just no time to set up a standing wave in the oral cavity so the /m/ in `damp’ is nasal but not as bilabial as the /m/ in `damn’.

Question 7

Can you get it to sound like “gun”? What does it sound like? Why shouldn’t this
procedure succeed in getting “gun”?

No, I can’t get this to sound like “gun” either. I think the explanation is the same: the portion of the vocal tract resonating for the /n/ in `skunk’ is less complex than in the longer /n/ of `gun’.