Interactions between groups and individuals result in the diffusion of linguistic variation and therefore less linguistic differentiation between populations, just as isolation of populations leads to more linguistic differentiation. But our understanding of what interaction and isolation mean is limited by the social and linguistic group and the types of evidence that are considered when establishing theories on language variation and change. In this study we use evidence from history, oral tradition, ethnography, and acoustic analysis of archival language recordings to understand the diffusion of a sound change in the Tlingit language. We focus on the voiced velar approximant /ɰ/, termed ‘gamma’ in Tlingit language research. This sound provides an excellent testing ground as it is known to be variety-specific and its distribution over time and area provides information about how, when and with whom it changed. ‘Gamma’ is acoustically analyzed with manual measurements using multiple acoustic measures (F1, F2, intensity, and harmonic-to-noise ratio) from archival data for one speaker whose variety had retained ‘gamma’ and who produces it in their recordings and one speaker whose dialect had already lost gamma. This acoustic analysis provides phonetic evidence relating to its eventual loss in all varieties and helps us to better understand the acoustic properties of the voiced velar approximant given that it is understudied and cross-linguistic rare. We find that gamma is most approximant-like in all the acoustic properties, but it is also variable and heavily influenced by the surrounding segments. This contributes to the potential for it to change in an attempt to better maintain phonological contrasts. Gamma’s geographical diffusion seems to be well explained through geographical proximity, but we find other evidence supports an irregular diffusion of gamma loss through geographical space and that geographic isolation is only partially relevant for those areas where gamma is retained the longest. Geographically closer regions are likely to have more linguistically similar inventories and ‘gamma’ is constrained to particular peripheral areas, as predicted by an geographically radial outward diffusion of the loss of gamma. Evidence points to one area as the origin of the change and then its spread by movement within and across Tlingit communities following the Juneau and Klondike gold rushes. Therefore, despite contemporary appearances, it is likely that gamma loss diffused irregularly through population movements. Why then is gamma retained at only the three peripheral areas? We find that geographic isolation may explain two of the areas where gamma is retained the longest, but this does not seem to be the only explanation. We suggest that cultural conservatism and complex social identities from intensive contact with other societies and languages is also relevant for the retention of gamma. In one area geographic isolation offers no explanation and instead language contact may justify retention via similar phoneme inventories. Through multiple types of evidence, we are able to provide a more nuanced picture of the distribution of gamma and the diffusion of linguistic change across Tlingit varieties. This study provides a springboard for other work that integrates evidence from multiple fields in order to better understand language variation and change.