This study investigated whether lexical access is affected by inherent
acoustic variations that contribute to the identity of a phonetic fea
ture and ultimately a phonetic segment. Two experiments were conducted
to determine whether the magnitude of semantic (associative) priming
in a lexical decision task is influenced when the acoustic manifestati
on of the initial voiceless stop consonant (specifically, voice onset
time) of a prime word semantically related to a lexical decision targe
t was systematically manipulated (e.g., prime: ''king''; target: ''que
en''). Results showed no effect of the acoustic manipulations at the l
onger interstimulus interval (250 ms): however, at the shorter interst
imulus interval (50 ms), the magnitude of semantic facilitation decrea
sed as a function of the voice onset time manipulations. In addition,
there was a tendency toward slower lexical decision latencies when the
prime word had a real word voiced counterpart than when it did not. T
hese results suggest that activation levels of words in the lexicon ar
e graded, depending on the subphonetic shape of the input word. Result
s also suggest that words which are phonologically similar to the inte
nded word candidate are activated to some extent, whether the input pr
ovides a relatively poor phonetic representation of the intended word
or a good one.