When speaking, talkers modulate the signal they produce to balance the conflicting goals of conveying meaning and speaking fluently. How talkers manage this modulation is responsive to information content (e.g. focus prosody, predictability) as well as sociolinguistic factors (e.g. gender, dialect). It is clear that many global phonetic characteristics change consistently with this modulation (e.g. speaking rate, vowel dispersion/reduction,) which may affect how easily the listener can understand the message. A second question is whether talkers also modulate the precision of phonetic contrasts so that they are more/less clearly conveyed to the listener. This talk will investigate whether and under what circumstances phonetic contrasts are enhanced by talkers and provide evidence that modulation may not be as precisely targeted as has been assumed. I will then turn to the issue of individual differences between talkers and argue that many of the differences between talkers can be captured by where they fit on the spectrum of more or less clear articulation. Together these results can reduce the complexity of both the production and perception computations required by talkers and listeners.