A language isolate, in the absolute sense, is a natural language with no demonstrable genealogical (or "genetic") relationship with other languages; that is, one that has not been demonstrated to descend from an ancestor common with any other language. They are in effect language families consisting of a single language.
Most of the languages spoken in Europe belong to one single language family: Indo-European. Basque is the sole surviving non-Indo-European language in Western Europe, it is classified as a language isolate. Besides Indo-European, there are to be found languages of four other families in Europe; the Uralic family and the Altaic stock are represented, and we have to add two language families in the Caucasian area, namely South Caucasian and North Caucasian.
As a matter of fact, Basque has been compared with a lot of different languages (and language families) all over the world. For many people it is not easy to accept that some language(s) should escape classification. And there are always some enthusiasts who are not willing to acknowledge that some dozens of „correspondences“ can be found
between any two languages of the world, so they try to point out imaginary „macro-families“ or „super-stocks“. A serious research of the relations of Basque will start in South-Western Europe. Here we find first two areas of investigation: Basque and Aquitanian, and Basque and Iberian. Then we may ask: are there traces of extinct languages - which could have been relatives of Basque - to be found in modern languages of Western Europe? It seems to be quite possible that some words of an ancient language akin to modern Basque could have survived in Western Romance.
In their 1999 paper “Adaptation of Man to the Mountains: Revising the Mogollon Concept,” scholars David Gregory (Preservation Archaeologist at Archaeology Southwest; now deceased) and David Wilcox (Senior Curator of Anthropology, Museum of Northern Arizona; now retired) developed a provocative hypothesis that attempted to account for how the Zuni language could have become a linguistic isolate, unrelated to other languages known in the new world. They proposed that Zunian emerged as a distinct language between 6,000 and 7,000 years ago in the Mogollon highlands of the central Southwest.
It is necessary to distinguish language isolates from unclassified languages, languages so poorly known that they cannot be classified, though sometimes listed as isolates. An unclassified language is one for which there is not enough data (documentation/attestation) to know whether it has relatives – these languages lack sufficient data for them to be compared meaningfully with other languages and therefore their possible kinship remains unknown. Isolated languages are not grouped in larger genetic classifications because for them there do exist data and comparisons of these data with other languages do not reveal linguistic kinship.
In the most common view, an isolate is a language which has no relatives, that is, that has no demonstrable genetic relationship with any other language. It is a language which has not been shown to be the descendent of any ancestral language which has other descendants (daughters). Thus, language isolates are in effect language families with only one member. The best known and most cited linguistic isolates are Basque, Burushaski, and Ainu.
It is the disagreement that still continues with regard to placement of the Korean language that begs to be resolved. Altaic? Isolate? A “mini”-family perhaps comprising Korean and Japanese? Linguists may argue, however, that the debate is over given that linguistic textbooks tend to routinely place Korean (and Japanese for that matter) as an isolate. However, several guide books I have come across about Korea still place Korean within the Ural-Altaic family when discussing the language.
Some linguists still recognize the inclusion of Korean in the Altaic family (and by extension, recognise the Altaic family as a legitimate family). The majority view it seems though is that Korean is a “language isolate” (or isolated language). This term has various meanings, but in an absolute sense it would mean that the Korean language is unrelated to any other language, living or dead. It is on its own. An orphan, as it were.
The pre-Indo-Aryan language the Bhils probably spoke is lost, and we don't know its genetic affiliation. West central India is almost entirely Indo-Aryan speaking now. Presumably other linguistic families were more strongly represented in these areas in earlier times. There are other - fairly large - groups in central India, the Baiga for one, who now speak a variety of the local `Hindi dialect', but who probably had their own, non-Indo-Aryan, language earlier. That `Old Bhili' was related to `Old Nihali' - that there was an ancient Nihali-Bhili family - is a plausible surmise (this was suggested by Koppers and by Shafer and accepted by a number of others), but as yet there is no linguistic evidence for it, and I have seen no strong claims based on ethnographic materials to support the case.
The interest in Nihali, such as it is, in certain narrow academic quarters, lies in the fact - possible fact - that it is (in interesting ways) no language at all but a `so-called' or seeming language, and/or that it is a mystery, a lost - possibly `paleolithic' - language (something like the Tasaday of the Philippines, what Tasaday was purported to be but without the heavy public relations flak that surrounded it). It is, perhaps, the only remnant of an ancient - pre-Munda, pre-Dravidian, pre-Indo-Aryan language family, with no living relatives, but perhaps a sister language of the language the Bhils spoke before they lost their own language and it was supplanted by the various Indo-Aryan `Bhilis'. Nihali has been noticed by historical linguists for the very high percentage of borrowed vocabulary, and the variety of (proposed) sources for that borrowing, and the `suspiciously simplified' syntax of the language.
If one were to tell a biologist that one should not believe in mammals because no one has ever reconstructed Proto-Mammal and then explained how this animal evolved through all the intermediate mammals before arriving at the current array of 4,006 species, the biologist would laugh and think it's a joke. Yet if you tell a traditional historical linguist the same thing regarding, say, Amerind [a proposed higher language family], he will nod solemnly and say "of course".
One of the most exciting developments in the past two decades has been the identification of a language family, now called Dene-Caucasian, that includes several of these supposed [language] isolates. The six branches of this family -- Basque, Caucasian, Burushaski, Ket (Yeniseian), Sino-Tibetan, and Na-Dene -- [have been shown].