The author reports four experiments that examined phonological processes in
spoken word production. A form-preparation paradigm was applied to the que
stion of whether phonological features can be preplanned to facilitate spok
en word production. In Experiment 1, monosyllabic words were produced in se
ts different in form, or in sets sharing either the initial segment or init
ial segments differing only in voicing. Only shared initial segments yielde
d facilitation. A similar pattern of results was observed when the sets wer
e matched for the following vowel (Experiment 2), when words were produced
in response to pictured objects (Experiment 3), and when place of articulat
ion rather than voicing was manipulated (Experiment 4). The special status
of identity suggests that segments are planning units independent of their
features. The results are explained in terms of the WEAVER model of word-fo
rm encoding, in which a serial encoding of segments is followed by a parall
el activation of features. A WEAVER simulation of the experiments is presen
ted which supports these claims.