English diphthongs are represented as bisegmental (e.g., /aɪ/) and assumed to have two acoustic targets (onset, offset). These phones are standardly represented with variable symbology (e.g., /ai/, /aj/), and previous work indeed reports variability in diphthong offsets. To investigate whether variation can be explained by dialectal and/or individual differences, we examined diphthongs (/aɪ, aʊ/) produced by 41 speakers of American English from Ohio (Midland, Northern regions) and Louisiana (Southern region).
Comparison of offset spectral estimates to nearby monophthongs indicate that the /aɪ/ offset was relatively consistently acoustically close to [ɪ], but the /aʊ/ offset was highly variable for both groups. While some of these findings, such as the degree of dynamic spectral change in diphthongs, can be explained by dialectal differences, it is also possible that the diphthongs have different underlying structure (e.g., more clearly biphasic onset-offset for /aɪ/ than for /aʊ/).