Among the vowels in the conversation studied are 53 tokens of the // class, and 28 tokens of the // class. Measurements of the formant frequencies at chosen nucleus locations are presented in Figure . Some preliminary discussion is necessary to interpret this chart.
Herold (1990) notes that the phonological environment of these vowels is highly skewed: // occurs much more often than // before /l/ while // occurs much more often than //before /p/. Since /l/ has a strong backing effect on preceding vowels, (see below: Effects of Adjacent Consonants), and the phonetic realization of // is generally phonetically backer than that of //, confusion is possible between the effect of vowel identity and the effect of the phonological environment. Thus we also display the following consonant along with the tokens of the // class (Wells' lexical sets, THOUGHT and CLOTH). Since // tokens (Wells' LOT set) are much more common, and almost never precede /l/, we display them here with just a ``.'', and without the following consonant.
The measurements of // tokens are displayed with the symbol ``c'', plus the following consonant. As may be seen, // tokens are generally higher and backer than // tokens. This difference is consistent with the view that the merger has not taken place: the difference could be due to the underlying phonological distinction between the two lexical classes.
However, notice that following consonants for the raised and backed tokens are few: /l, n, /. Since /l/ is known to have a backing-and-raising effect on preceding vowel nuclei (again, see below in Effects of Adjacent Consonants), since nasals are known to frequently have a raising effect on preceding vowels,9.29, and since the raised tokens before /n/ are in the clitic word, on, it remains possible that these differences may be attributable to other allophonic or coarticulatory effects. Since the remaining THOUGHT-class tokens, with following /s, z, t, / occur well within the main body of the LOT distribution, it is likely that the difference of distributions of the two lexical classes is not due to an underlying lexical distinction.
It would appear that more data from more speakers, targeted towards answering this particular question would be useful in making certain whether or not // and // are merged in Chicano English, However, the evidence here, though superficially supporting the hypothesis that they two classes are separate for this speaker, on a closer examination do not support that hypothesis. It would appear that this is a Low-Back Merged dialect of English.9.30