This talk is an informal presentation of my two main research interests. The first part relates to the degree to which continuous phonetic characteristics might be used and accessible by phonological computational system. In Hungarian vowel harmony, transparent vowels [i] and [e] in front and back harmony words have been assumed to be phonetically the same irrespective of the harmony domain in which they occur. In Slovak, yer and non-yer vowels [e] and [o] have also been assumed to be phonetically identical. Hence, in both cases, phonological systems require a contrast that is assumed to be neutralized phonetically. In both cases, however, phonetic evidence casts doubts. In the second part I will briefly touch upon the prosodic correlates of turn-taking behavior suggested in the literature and potential use of filled pauses in this behavior. I will try to formulate some hypotheses that I plan to test in the corpora available to me.