Three experiments examined the time course of phonological encoding in speech production. A new methodology is introduced in which subjects are required to monitor their internal speech production for prespecified target segments and syllables. Experiment 1 demonstrated that word initial target segments are monitored significantly faster than second syllable initial target segments. The addition of a concurrent articulation task (Experiment lb) had a limited effect on performance, excluding the possibility that subjects are monitoring a subvocal articulation of the carrier word. Moreover, no relationship was observed between the pattern of monitoring latencies and the timing of the targets in subjects' overt speech. Subjects are not, therefore, monitoring an internal phonetic representation of the carrier word. Experiment 2 used the production monitoring task to replicate the syllable monitoring effect observed in speech perception experiments: responses to targets were faster when they corresponded to the initial syllable of the carrier word than when they did not. We conclude that subjects are monitoring their internal generation of a syllabified phonological representation. Experiment 3 provides more detailed evidence concerning the time course of the generation of this representation by comparing monitoring latencies to targets within, as well as between, syllables. Some amendments to current models of phonological encoding are suggested in light of these results. © 1995 Academic Press, inc.Most current models of speech production propose that articulation is preceded by the generation of an abstract representation of the form of the target word or utterance (Dell, 1986(Dell, , 1988Garrett, 1975;Levelt, 1989;Levelt & Wheeldon, 1994;Shattuck-Hufnagel, 1979. Indeed, the existence of an internal abstract speech code is a basic assumption within psycholinguistic theory. In speech recognition, a store of abstract lexical representations has been postulated to accommodate the fact that no two tokens of a given word form are acoustically identical (Lahiri & Marslen-Wilson, 1991). Similarly, in speechThe authors thank Ger Desserjer and Jouke Riepkema for their help with the running and analysis of the experiments reported, and Vincent Evers for carrying out speech lab measurements. Many thanks also to Antje Meyer, Pienie Zwitserlood, and an anonymous reviewer for their comments on an earlier version of this article. Address correspondence and reprint requests to Linda Wheeldon, School of Psychology, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK. production the same word, in different contexts, can be spoken with very different segmental and syllable structure, intonation, duration, and amplitude. This generalization has long been captured in linguistic theory by the proposal that every linguistic item has a unique phonological representation which encodes only that information which can distinguish among words in the language. Phonological representations are categorical in nature and consist of discre...