This study explores how articulation recovery might be accomplished in the absence of clear acoustic output consequences. Based on perception data from Tashlhiyt Berber utterance-initial voiceless singleton and geminate stops (e.g. tut vs. ttut), we show that auditory information alone is not sufficient for native listeners to elicit the standard perception performance expected from native listeners on a native contrast. Implications of the results on the general issue of the nature of speech targets are briefly discussed.
