(55) A Comparative Typological Profile of Language Isolates

     Language isolates as a group have received relatively little attention in the field of comparative typology. This study compares three samples (1."true" isolates, 2. semi-isolates, and 3. non-isolates) and asks... [ view full abstract ]

     Language isolates as a group have received relatively little attention in the field of comparative typology. This study compares three samples (1."true" isolates, 2. semi-isolates, and 3. non-isolates) and asks whether the first two exhibit tendencies that distinguish them from the latter. All information is drawn from the most recent edition of the online World Atlas of Language Structures (WALS). I create sample 3 by removing isolates from the random 100-language sample on WALS's website. The first sample consists of all languages who are the sole member of their language family and whose name is identical to their family's. The second consists of one language from all families with three or fewer members and all languages who are the only members of their family but whose name is different from that family's. I do not stratify the first two genealogically, since they are by definition already stratified. Sample 3 is left as it was genealogically stratified by WALS, and I stratify all samples using Dryer's (1989) large linguistic areas. I establish the frequency with which a variety of morphological and phonological features appear in each sample. If significant differences appear between the samples, I propose that the explanation must stem either from or be caused by the isolation. That is, either a) certain features are less likely to diverge or spread, causing isolation, or b) change caused by language contact operates differently when languages are unrelated to each other, resulting in unique features. I predict, based on the scant existing literature on this topic, that minor differences will show up. However, if they don't, other routes must be explored to explain why some are isolates and some are not.