When reading about rare or antique languages I often came across statements like "the most widespread / widely accepted theory is that it is an isolated language". You can read that for example in wikipedia for Basque, Korean or maybe Albanian. The reasons given are often like: "there are no clear evidences for the contrary" My question is, if it is really a valid zero-hypotheses to assume a language isolate? To me it seems that such a reasoning is rather subject to scientific "fashions" and that at the moment its quite fashionable to claim isolates, rather then relation. Any educated opinion on that?
It's not at all clear what the term "language isolate" means: one could think it means that the language is not related to any other known language. People mistakenly say "not related" when they mean "I can't show that it is related", in other words people sometimes confuse lack of evidence and knowledge of non-relationship. As long as you understand "language isolate" to mean "insufficient evidence to justify a claim of relatedness", then it's obviously the only valid null hypothesis, until one gains evidence to support a claim of relationship.
A more complete argument would read: We compared Basque to the Indogermanic, Semitic, Finno-Ugric, and the Caucasian language families as well as to the Etrucsan language and found no provable connections.
A "famous" language like Basque was probably also compared to Dravidian, Athabascan, Sino-Tibetan and Na-Dene (with no provable connection), but somewhere comparison always stops. Usually, "isolated" languages are compared to the geographically plausible neighbours, not to all known language families and isolate languages of the world.