Linguistic Security, Ideology, and Vowel Perception 261the vowel that matched the speaker’s /æ/, and only two of the respondents
chose the actual token representing the speaker’s /a/. Instead, most respon-
dents chose tokens that were produced from formants that matched those
found in Peterson and Barney (1952)—vowels that are considered “standard.”
Even more intriguing, the remainder of the respondents chose tokens that I
have labeled “hyper-standard.” This label refers to the fact that these tokens
are farther from the actual token then even the standard token—farther back
and lower in the case of /æ/, and farther back in the case of /a/.
In other words, the respondents did not choose the NCS token as the one
that best-matched the vowel of a fellow Michigan speaker, but rather chose
the one that conformed to their notion of a standard. While it did not conform
to what the speaker produced, it is consistent with the notion that Michigan
speakers are SAE speakers. This gives us a picture of speakers who believe
that they are speaking a standard variety of a language, but who do not in
fact ¿ nd actual features in their own dialect to be standard. I suggest that
their notions of standardness come not from the acoustic evidence they are
presented with every day as they talk to their cohorts, but from the fact that
their sense of linguistic security allows them to ¿ lter out evidence incongru-
ent with their beliefs.
In other words, their ideological beliefs of themselves as standard speak-
ers causes them to misidentify vowel tokens in their own speech, if they are
being explicitly asked about those tokens. I suggest that this is a direct result
of the linguistic security that these speakers display. In short, speakers in
regions characterized by high degrees of linguistic security (i.e., speakers
who believe that their regional dialect is “standard”) have a great deal at stake
in their self-stereotype, and this causes them to ¿ lter out information (in this
case, acoustic evidence) that runs counter to their own beliefs. Giddens (2000)
offers a suggestion of precisely why these speakers would do this:
[T]he plethora of available information is reduced via routinised attitudes
which exclude, or reinterpret, potentially disturbing knowledge. From a neg-
ative point of view, such closure might be regarded as prejudice, the refusal
seriously to entertain views and ideas divergent from those an individual
already holds; yet, from another angle, avoidance of dissonance forms part
of the protective cocoon which helps maintain ontological security. (187)Thus, (routinized) language attitudes can inhibit the acquisition of accurate
knowledge of language variation, if there is a cost to such acquisition. In the
case of Anglo Detroiters, that cost is the surrender of their identity as speak-
ers of Standard American English, a high cost indeed.