Tlön, Uqbar, Orbis Tertius

    Tlön, Uqbar, Orbis Tertius is a 1940 short story by Jorge Borges, translated into English in 1961 and appearing in the collection Labyrinths.

    In modern terms, it describes the discovery of a secret society of sci-fi engaged in a multi-generational worldbuilding project--creating an encyclopedia of the world of Tlön. This is essentially the "explain a film plot badly" summary, but a proper summary is very philosophical and literary, and if you want that you can go read the Wikipedia page or something. I'm just here for the linguistic references!

There are no nouns in Tlön's conjectural Ursprache, from which the "present" languages and the dialects are derived: there are impersonal verbs, modified by monosyllabic suffixes (or prefixes) with an adverbial value. For example: there is no word corresponding to the word "moon," but there is a verb which in English would be "to moon" or "to moonate." "The moon rose above the river" is hlör u fang axaxaxas mlö, or literally: "upward behind the onstreaming it mooned."

    This isn't actually particularly odd. Many indigenous North American languages--especially the Salish family--are famous for having a heavy preference for verbs, and deriving nouns from relative clauses or participle-like constructions. Native American languages are not particularly famous for monosyllabic roots, but there's no particular reason those features should not be combined. Such a language could easily turn up in nature, and I would not be slightest bit surprised if someone discovered a language exactly like that somewhere in Papua New Guinea! Borges clearly did not generate a complete Tlönian conlang for this short work, but there is some translated Tlönian there which we might as well try to analyze, just for fun. I don't know what the original text looked like, but at least in English translation, there is not a one-to-one correspondence between Tlönian words and English words, which is a plus. Many times, translations between arbitrary will end up with the same number of words just by chance, but by not matching up in number, we know that the words cannot match up one-to-one in sense, which gives us an excuse to think up different ways that information could be organized in the Tlönian sentence. Obviously, with such little data, it's impossible to settle on one obviously correct answer, but I like to think that Tlönian uses something like a relational noun construction (where the "noun" is of course actually a verb), and has reduplication for extended actions, leading to hlör = upward; u fang = at what-is-behind; axaxaxas  = multiply-reduplicated form of ax, "to flow", maybe with an adverbial suffix as for "onward", "towards a goal"; and mlö "(it) is-the-moon". Incidentally, "Axaxaxas mlö" is also the title of one of the books mentioned in Borges more famous story The Library of Babel.

    We get one other word of Tlönian, though, which seems very mugh like a noun: hrönir (singular hrön), referring to duplicate instances of things which are lost once and then found multiple times. Maybe it's actually a verb meaning "to be found multiple times" and -ir isn't so much a plural as a pluractional or something. About other parts of Tlön, we are told that

In [languages] of the northern hemisphere [..] the prime unit is not the verb, but the monosyllabic adjective. The noun is formed by an accumulation of adjectives. They do not say "moon," but rather "round airy-light on dark" or "pale-orangeof-the-sky" or any other such combination. In the example selected the mass of adjectives refers to a real object, but this is purely fortuitous.

    This, too, is actually not so strange after all. Per Topics in Warlpiri Grammar by David Nash, Warlpiri does not formally distinguish nouns from adjectives, and can string them together in any order to pinpoint a more precise concept which is the intersection of all the provided descriptors. What would be strange is if, rather than merely focusing on "adjectives", northern Tlönian in fact only had words for semantic attributes and not entities; but Borge himself seems to have had a hard time conceiving of that, given that he includes "sky" in the provided glosses. The philosophy of not having fixed words for specific objects, but just using contextually-relevant descriptions as needed, regardless of whether we think of any of those descriptors as "adjectives" or "nouns" is, however, strongly reminiscent to me of the communicative philosophy of Toki Pona. So even if this is a less naturalistic vision of language than the first version of Tlönian represents, it has at leat been shown to be emminently workable by a conlang community arising some 61 years after Borge posed this idea.

If you liked this post, please consider making a small donation!

The Linguistically Interesting Media Index

Some Thoughts on Iljena

    Iljena is an alien conlang by Pete Bleackley, also the author of Khangaþyagon which I reviewed previously.

    The key conceit of Iljena is that all words encode both a nominal root and a verbal root--and based on both the grammar notes and the dictionary, there are no other parts of speech. All verbs are monovalent, and you construct large propositions by chaining together noun-verbs that describe what each participant is doing. It's sort of like the disambiguation strategy sometimes employed in natlangs where a transitive clause that lacks distinctive subject and object marking (like two neuter nouns in a positive-polarity Russian sentence, or nouns with equal animacy in a direct-inverse language) can be split in two with an antipassive clause and a passive clause--i.e., instead of "Bob saw Bill", "Bob saw, Bill was seen". Except that Iljena doesn't have a passive construction, it just has enough different verb roots to cover all the necessary meanings, whether one is an agent or patient or instrument or whatever in any particular scene.

    With the lack of any other parts of speech, however, it is unclear how boundaries between clausal constituents are determined, how attachment ambiguities might be resolved, or how references to events-as-things are made, and the only ordering constraint is that 

Word order is used to convey the flow of the action between the participants, and to bring together closely related participants.

    However, David Gil has shown us that you don't really need to formalize all that grammatical machinery all of the time, and the corpus of Conlang Relay texts in Iljena, which have been translated reasonably faithfully by following relay participants, demonstrates that it does work well enough. Pete's own documentation notes that Iljena could be considered a "verbless" language, based on the idea that verb roots could instead be interpreted as noun cases (which is one of the possible solutions to verblessness I discussed in my own article How To Not Verb), but he (and the fictional Leyen people who speak Iljena) prefer to think of the relevant open lexical class as verb roots, rather than case morphology--and I tend to agree. The complete lack of function words makes Iljena a decidedly non-human language, but that's fine--it's not supposed to be!

    As noted, Iljena does seem to work just fine as it is, so I won't presume to suggest improvements--but I think it would be neat to see a language that takes the one verb--one noun approach and embeds it in a larger system of grammatical function words for eliminating structural ambiguities. And it would also be neat to see some more detailed analyses of the existing corpus texts, beyond simple interlinear glosses, that might be able to extract more empirical rules about Iljena grammatical structure.

Some Thoughts... Index