We report a study investigating the acoustic basis of intelligibility variation using a speech-analysis resynthesis paradigm termed hybridization (Kain et al., 2008) to blend acoustic properties of sentences produced by male speakers with known differences in baseline intelligibility. Speakers were identified on the basis of previously reported transcription intelligibility scores for Harvard Sentences (Stipancic et al., 2016). Speakers were selected from an existing database to represent a range of intelligibility scores amongst a pool of 28 male speakers. One speaker with PD and a median intelligibility score was identified as the base speaker. Three speakers with PD having the lowest, 25th percentile, 75th percentile intelligibility scores and one neurotypical speaker having the highest intelligibility score were identified as the donor speakers. Using the hybridization paradigm, acoustic properties including the sentence level energy envelope, F0 envelope, segment durations, short-term spectra, were individually or in combination extracted from sentences produced by each donor speaker, and donated to the same sentence produced by the base speaker to form hybrid sentences. Transcription scores of hybrid and original sentences obtained from 520 crowd-sourced listeners were the primary outcome. Statistical analyses indicated that Spectrum, Duration+Spectrum, and Intonation+Energy+Duration hybrid variants were the strongest predictors of intelligibility variation. Singular acoustic properties of duration, intonation, or energy did not account for intelligibility variation. These results indicate that 1) between-speaker hybridization may produce intelligibility improvements of dysarthric speech, and 2) both segmental and suprasegmental properties of the acoustic signal mediate intelligibility losses associated with dysarthria.