Speculating about linguistics when you are not a linguist is irresponsible, so let’s get started!

To-day I’ve been thinking about mutual intelligibility. Two languages are mutually intelligible if their speakers can understand each other. For example, Thai and Lao are mostly mutually intelligible—Thai speakers can understand most of spoken Lao, and vice versa. English and Scots are partially mutually intelligible. When I listen to recordings of people speaking Scots, I can pick up many words and kind of understand what’s going on, even if several words are opaque to me. Not so with Frisian or Dutch or German. Although English is closely related to each of those languages through a common ancestor (which we might call Proto-West Germanic), they have diverged far enough that I can’t understand people when they are speaking Dutch.

Considering mutual intelligibility as a mathematical relation, I naturally ask myself whether it is reflexive, symmetric, and/or transitive.

Is it reflexive? Yes. Every language is mutually intelligible with itself, by definition (otherwise its speakers wouldn’t be able to communicate and it wouldn’t be a language).

Is it symmetric? Usually but not always. One way symmetry can fail is if one language is much more widely spoken. For example, Scots speakers can understand English more than English speakers can understand Scots, purely by mere exposure to the massive relative abundance of English-language media.

Is it transitive? If languages A and B are mutually intelligible and languages B and C are mutually intelligible, does that imply that languages A and C are mutually intelligible? Signs point to no. There exist dialect continua, progressions of languages across geographical areas such that neighbouring language varieties are mutually intelligible but the differences accumulate enough so that the languages at the ends of the continuum are not.

I like the idea of dialect continua. Theoretically, I could speak a message in language A to my language B-speaking friend Blair, and she could relay my message to her language C-speaking friend Cassie, and so forth, until it reaches its intended recipient Hana in language H after a gauntlet of interlinguistic Telephone. And nobody in the chain has to speak anything but her own language! Granted, in normal games of Telephone things get lost to comical effect, so perhaps I should not expect message integrity to be upheld.

I wonder if you could construct a dialect continuum between any two languages, say, French and Japanese. That is, construct a series of conlangs such that together they bridge the space between French and Japanese. The first, FJ-Bridge₁, would be a slightly Japonic kind of French that French speakers could understand, and the last, FJ-Bridge_N-1, would be a variety of French-flavoured Japanese whose speakers could understand normal Japanese. This would be a massive undertaking—each language from FJ-Bridge₁ to FJ-Bridge_N-1, inclusive, would have to be a fully-fledged language! Assuming that this can be done, what is the minimum number of languages N that we’d need to construct?

This N would be a metric of distance between two languages. Let’s call it the continuum distance. Intuitively, I’d expect N to correlate with “phylogenetic” language distance (i.e., how long ago did the two languages diverge from a common ancestor?). That is, I’d expect French-to-Spanish to require a shorter bridge than French-to-Norwegian, which would require a shorter bridge than French-to-Japanese. One advantage the continuum distance would have over the phylogenetic distance is that you can define it over unrelated languages. But I wonder how many “surprises” there’d be, where language A is more closely related to language B than to language C phylogenetically, but the A-to-C continuum bridge turns out to be shorter than the A-to-B continuum bridge.

Lastly, could you learn a language just by walking across the continuum bridge? If I’m a native speaker of French, and I want to learn Japanese, could I spend a week or so immersed in a FJ-Bridge₁ environment until I feel comfortable, and then move on to an FJ-Bridge₂ environment? And iterate N times, over which Qu’est-ce que c’est would morph continuously into Kore wa nan desu ka before my very own ears?