How does an original proto language produce its daughter languages?

Question

I am trying to understand the principles how a proto language produces it daughter languages, do they proliferate from dialects of the same proto language or do they proliferate from dialects of other daughter languages?

E.g. Are English, German, Dutch and Icelandic all originally dialects of Proto-German or are they dialects of another daughter language. Does this actually happen where a daughter becomes the proto for another set of languages? (like Italian seems to descend from Latin but Italian itself seems to be a proto language for later Italian languages?

Answer 1

A proto-language is a hypothesis - it's a theory about the history of a language family. A proto-language is a model of the closest common ancestor of the daughter languages, but which is not directly attested itself (ie, we don't have anything written in it.) So neither Latin nor Italian are proto-languages, because we have direct written evidence of both. Proto-Germanic and Proto-Indo-European are models of languages we think were spoken at one time, but which were never written down.

user6726's answer is a good explanation of the tree model of linguistic families. Many language families can be effectively explained through the tree model. But you should also be aware that there are some changes in languages which cannot be accounted for through simple descent as in the tree model. Some changes are introduced from neighbouring languages; these are called areal features. The wave model of linguistic families goes further and says that most changes spread as areal features rather than strictly through descent in a splitting family tree.

Answer 2

The theory is that there is a community, whose members speak "a language" (one language). They go about life, roaming the plains of whatever, and their children learn that language. As long as they remain a coherent community whose members are in contact with one another, everybody speaks the same language. The language might change a bit, but it changes for everybody. But sometimes one group of people head off somewhere else, and have little to no contact with their cousins. Languages change all the time, but when you have a single community whose members are in contact, the change tends to be in the same direction. Once you put a mountain or other obstacle between two communities, the speech of one group can become substantially different from the speech of the other group. Initially, that might be called (by linguists) a difference of "dialect", but with enough time and change, you have different languages. The "proto-language" is the language as it was spoken by a single community, and the "descendant languages" that are used to reconstruct the proto-language are the languages that you actually encounter.

Obviously, contact and social influence are not an all or nothing proposition. Suppose that humans shipped 10,000 people to colonize a planet in another galaxy. Even if everybody used English, not everybody in this community would speak the same dialect of English. Some would be from California, some from New York, some from South Africa (and so on), and some would be second-language speakers from France or Mongolia. One or two Mongolian speakers would probably not have a substantial impact on the resulting language 2,000 years after, but if the Mongolian community were 1/3 of the population, there is a fair change that that would have an influence on later languages. There would already exist dialects of English, and distinctions could be passed on depending on whether your ancestors were from the Mongolian-influenced dialect vs. the California-influenced dialect. Hence the assumption that "Proto-Indo-European" or some other language was originally completely homogenous is likely to be false.