Articulation of a given speech sound is often contingent on the intention to produce subsequent sounds (co-articulation). Thus, intended acts affect the way current acts are executed. Here we show that the intention to subsequently repeat a short sentence, overtly or covertly, significantly modulated the articulatory musculature already during listening or reading (i.e., during the input phase). Young adults were instructed to read (whole sentences or word-by-word) or listen to recordings of sentences to be repeated afterwards. Surface electromyography (sEMG) recordings showed significant reductions in articulatory muscle activity, above the orbicularis-oris and the sternohyoid muscles, compared to baseline, during the input phase. These temporary reductions in EMG activity were contingent on the intention to subsequently repeat the input overtly or covertly, as well as on the input modality. Inhibition was stronger when an overt response was intended. When participants intended to repeat the sentences covertly, only auditory input but not reading resulted in significant reductions in sEMG activities during the input phase. Higher temporal resolution analyses also showed different patterns of activity modulations before the cue to respond, depending on the input modality and the intended response. Only when repetition was to be overt, a significant build-up of activity occurred after sentence presentation, before the cue to respond; this build-up was most pronounced when the sentence was heard or read wordby-word. Neurolinguistic models suggest that language perception and articulation interact; the current results suggest that the interaction begins already during the input phase, listening or reading, and reflects the intended responses.