A Reader in Sociophonetics

84 Rebecca Roeder

Technology has advanced considerably since Stevens and House pub-
lished their article in 1963. For the current study, measurement of vowel
tokens in a nasal environment was done through careful identi¿ cation of the
nasal formant, followed by individual readings for F1 and F2 in Praat.
Production does not match perception in this situation, however. Evi-
dence shows that listeners average the ¿ rst oral and nasal formants when per-
ceiving a vowel (Beddor and Hawkins 1990, Stevens 1998). As pointed out
by Plichta (2004), this means that nasalized /æ/ will be perceived as higher
than oral /æ/. Plichta (2004) discovered that nasalization of vowels, even in an
oral environment, is a common feature of NCS speech, but perhaps it is most
salient perceptually in a pre-nasal context. If so, this provides an impetus
for the genesis of the NCS, in that pre-nasal /æ/ was perceived as raised and
subsequently produced as raised by children learning the dialect as a ¿ rst
language, creating the ¿ rst step in the chain shift.

3.4 Preceding Manner and Voice

With regard to preceding manner and voice, the following four environments
were tested: voiced and voiceless stops, voiceless fricatives, and liquids. The
words used in each environment are listed in Table 3.5. Results show that
preceding voiced stops are correlated with a signi¿ cantly lower F1 for /ae/ in
four of the sixteen speakers. Only one of these four is in Group A, supporting
previous hypotheses that raising after voiced stops is a universal tendency.

Table 3.5 Preceding Manner and Voice Features and Words—Excluded (single-

tons): zap (voiced fricative), jazz (affricate)

Manner/Voice Words

Voiced Stop banker, bath, dad, gamble, gang, badge

Voiceless Stop cabin, cash, pal, past, pat, tab

Voiceless Fricative Saginaw, Sam, thank

Liquid black, brag, Lansing, laugh, plant, rack, rag

Nasal mash, mattress, nap

/h/ or Vowel Initial apple, ask, has, have

Although preceding manner and voice were not tested separately in the previ-
ous sociolinguistic studies mentioned, these ¿ ndings coincide with Stevens
and House (1963) and Hillenbrand et al. (2001). Since Stevens and House