Discussion

For the minimal pairs, the words with [l] consistently had substantially closer $F_1$ and $F_2$ values than those with [w]. In the table below, we have the minimum average difference for words with no [l] and the maximum for words with an [l], by speaker.

Speaker	Min [w] diff	Max [l] diff
David German	578	125
Venger Jamison	545	157
Michael Stone	449	263
David Chudzicki	589	186

This data suggests that the cutoff for hearing an [l] instead of a [w] is somewhere between 450 Hz and 250 Hz. Nearly every recorded sample from the minimal pairs section classified properly under this criteria, with the exception of about five distributed words which are likely errors.

We can then apply this filter to the test words to see which speakers use [l] and which use [w]. The table below has `x' for each word that had an apparent [l], `-' for those that don't, andand `?' for those that are in the middle.

Speaker	poke	Polk	polka	yoke	yolk	folk	moment	both
German	-	x	x	-	x	x	-	-
Jamison	-	x	?	-	x	x	-	?
Stone	-	?	-	-	?	?	-	-
Chudzicki	-	-	-	-	-	-	-	x

Both German and Jamison appear to be putting [l] in words that are spelled with l while Chudzicki doesn't. Stone is somewhere in between. Because this was a formal setting and people were reading, I would expect this [l] inclusion frequency to be higher than normal. Yolk and yoke are often listed as homophones, and there are on-line guides telling people when to write one and when the other. This suggests that there are large numbers of people who no longer make this distinction. English orthography is nice in some ways because it preserves distinctions that people do not always make, such as wh/w or, in this case, ol/o. While I would want to do further testing, it seems likely that a sound change has turned /l/ to /w/ in the environment /o_k/, and that this change has not get gone to completion.

As for both, it is clear that Chudzicki pronounces it with an [ol] in place of an [ow], but it is not clear why. It may have to do with some sort of hypercorrection, either on his part or the part of someone he heard as a child. He is not the only one that does it, and bolth is more common a spelling than you would expect as a typo.³ Now that the phonetic distinction between both and bolth is clear, it would be interesting to do further testing to find out where it is common and where not.