Methods and Materials

I recorded subjects on a word-list containing words in three categories. The first consisted of words that I believed were minimal pairs in American English. From these I hoped to get an idea of how [l] and [w] contrasted after [o]. They were,

[l]	[w]
bowl	bow
bolt	boat
knoll	know
gold	goad
soul	sew
mold	mode

The second category was smaller, and consisted of the words on which I expected variation. They were,

both	folk	yoke	yolk	moment	poke	Polk	polka

The final category was filler words, included both to keep the subjects from getting suspicious and to prevent them from getting used to the vowel. All these words were then shuffled by a computer program that randomly arranged them into a list, with each word repeated three times. The list was then broken up into seven word segments and arrainged in columns on a page.

The subjects were instructed to read each block of seven words as if they were a sentence, pausing briefly after each block. This experimental paradigm was somewhat different from previous ones we'd used, lying somewhere between a list reading and a carrier sentence.¹. I used this paradigm because I could get lots of examples in mostly natural positions without tremendously long recordings that would have annoyed my subjects and taken a long time to analyze.

For this paper I use data from four subjects. They were,

Name	Age	Gender	From
David German	20	male	St. Louis, Missouri
Venger Jamison	19	male	New Jersey
Michael Stone	21	male	Maryland
David Chudzicki	20	male	Upstate New York

While I could have varied age and gender, the sample is small and I didn't want additional, possibly confounding factors.

We recorded the subjects reading the lists with a Plaintronics DSP-400 noise canceling microphone and used the software program Praat to calculate formant values. For each word we recorded the values for the first and second formants at the point in the vowel at ther point where they were closest together. The charts below show the three repetitions of each word with their first two formant values ($F_1$ and $F_2$), in addition to both their difference and the average of their difference.