zompist bboard

Not really about actual conlangs in history, but as a hypothetical question. Assume we happen upon a previously-undiscovered language with extensive documents in said language. Though it's not apparent, this hypothetical langauge is a constructed language. Through available tools, is there any way one could demonstrate this conclusively, or even come up with a strong hypothesis?

In theory, no. I see no obstacle in the way of any degree of naturalism.

In practice, I can think of two telltales, though they're subtle and would just be clues, not proofs.

1) Most conlangers take shortcuts, and this might be noticeable. E.g., no borrowings at all, or borrowings from just one language; absolutely regular sound change; absolutely regular morphology.

2) More subtly, a conlanger can only imitate things she's aware of. E.g. a pre-1800s conlang might include rule-less sound changes; a pre-1960s conlang might have unlikely syntax because the conlanger knew nothing about generative grammar. Though there are few universals, there are some near-universals and a conlang might violate a bunch of them.

(Note that naturalistic conlangs are a fairly recent thing.)

zompist wrote:(Note that naturalistic conlangs are a fairly recent thing.)

Because linguistics has progressed since Zamenhof, and so has taste.

Well, the changes in taste aren't always progress. Zamenhof brought in a fashion for a posteriori auxlangs, which were produced in depressing quantity. Many of them are arguably improvements on Esperanto, but the whole problem with that approach is that people could never agree on the criteria to use.

I think Volapük is underappreciated. It's a little weirder than Esperanto, which I think is a positive. I don't know if Shm Jay still keeps up with it...

Do some of the Australian langs that were made up because of speech taboos (e.g. Damin, and a few sign languages) count as conlangs? Most aboriginal signs are just cyphers, but Damin is fairly divergent.

Hypothesis: it is a conlang if this ratio is high or was high in the past:

deliberately invented words by living speakers = C
total lexicon

The classic conlang has a C of 100% (where some single person or group makes all the words up out of his own head (quenya, klingon). Presumably a natlang has a low ratio, but not zero, because of language change. Borrowings from other languages do not count unless they change meaning, but slang terms (mec, yob, jazz) do. If somebody takes a natlang and deliberately applies sound changes to derive a daughter language, the derived words count as inventions, and the result is a conlang. Sound changes as a result of natural language drift do not count, becuase no speaker intends to change her mother tongue out of recognition. Pidgins (which consist entirely of borrowings) do not count, and therefore creoles do not.

Corollary: A language evolving naturally has a low C throughout the process. A poet like Shakespeare may coin hundreds of words and raise the numerator, but he is increasing the total lexicon at the same time, so C does not increase all that much.

I suppose if large numbers of taboo words are lost from Damin at the same time as large numbers of words are invented, and it was developed over a short period (i.e. over a single lifetime) it might be approaching a threshold value of C above which a language can be considered to be a conlang.

I suppose if large numbers of taboo words are lost from Damin at the same time as large numbers of words are invented, and it was developed over a short period (i.e. over a single lifetime) it might be approaching a threshold value of C above which a language can be considered to be a conlang.

Damin, according to it's speakers, was invented by the gods. Considering it exists to circumnavigate a taboo, not against certain words, but against speech itself, it probably fits your definition of conlang.

Damin is probably categorisable as a group conlang, conllaboolang, maybe (constructed-collaborated-taboo-language) ?

I deliberately didn't include a term for "number of people inventing words" so that group conlangs would qualify.

At what point might a conlang like Esperanto become a natlang?

treegod wrote:At what point might a conlang like Esperanto become a natlang?

Two to five hundred years is I assume a good enough gap in time.

Elector Dark wrote:

treegod wrote:At what point might a conlang like Esperanto become a natlang?

Two to five hundred years is I assume a good enough gap in time.

125 down, anywhere between 75 and 375 to go.

Does modern, spoken Esperanto have any irregularities (or well-defined sound changes) yet?

> Does modern, spoken Esperanto have any irregularities (or well-defined sound changes) yet?

Those who learned Esperanto intentionally as a medium of international communication are somewhat hostile to irregularities or sound changes. IMHO they will never deviate far from the grammar, pronunciation and core vocabulary that Zamenhof designed.

I don't know what the native speakers of Esperanto are up to. In pre-internet, pre-Skype days they might not have had many chances to converse with other native speakers outside of their own families. Now at least in theory they could have worldwide conversations and create some sort of mutant dialect.

dhokarena56 wrote:Does modern, spoken Esperanto have any irregularities (or well-defined sound changes) yet?

There's a very nice article here about, among other things, the changes Esperanto has undergone, which I found originally through Zompist's blog some weeks back.

Highlights include: the accusative is doomed; there's a whole new range of periphrastic tenses; it's developing a new mediopassive voice and a new subjunctive mood; and the vowel endings are getting re-purposed and oozing into dimensions never intended for them. Most of these changes to Esperanto are mainly in literary Esperanto, not so much natively-spoken Esperanto. Likewise, regular sound changes don't yet seem to be in evidence - native speakers are just not in regular enough contact with each other.


Nightwriter: what people think they are doing with their language and what they are actually doing are, generally, two quite separate questions. People having all the above changes probably mostly still think they're adhering well to Zamenhof.

I suppose it might go the way of Latin, dialects will diverge and languages will emerge. Then there will be those that learn "Classical Esperanto" in the same way people learn Latin, existing alongside its spawn. Zamenhof's work will always be around to learn from.

ol bofosh wrote:I suppose it might go the way of Latin, dialects will diverge and languages will emerge. Then there will be those that learn "Classical Esperanto" in the same way people learn Latin, existing alongside its spawn. Zamenhof's work will always be around to learn from.

Given the absence of sound change, it sounds more to me like the divergence between Classical Latin and post-Classical Latin (preserved to this day as an international auxiliary language). Still mutually intelligible, but obviously not the same beast.

One possible candicate could be the so-called Voynich Manuscript.
It is a medieval book, written in a scribe that hasn't been decoded by anyone.
We are not enterily certain what the book contains : one of the guesses is "a tract on alchemy".

http://en.wikipedia.org/wiki/Voynich_manuscript

The best hypothesis on the Voynich manuscript is that it was produced using a Cardan grille. The text doesn't have the frequency characteristics of a language (most of which would be noticeable even if it were a cipher).

zompist wrote:The best hypothesis on the Voynich manuscript is that it was produced using a Cardan grille. The text doesn't have the frequency characteristics of a language (most of which would be noticeable even if it were a cipher).

I have recently come across an interesting blogpost by computational linguist Jürgen Hermes (in German) which suggests that the Voynich text may have been produced using an elaborate substitution cipher which replaces each letter with a full word, with a different word (taken from a special kind of dictionary) each time the letter occurs. The method in question has been described in Johannes Trithemius' "Polygraphia" (1508), originally producing fake Latin text (which is grammatical but nonsensical, achieved by taking into account the part-of-speech of the cipher words, cf. this post). The method was apparently copied by a few other 16th century cryptographers and adapted to produce fake text in other languages, and Trithemius himself adapted it (possibly in a slightly modified form) to producing text that looks like words from a taxonomic conlang. Here's an excerpt from Trithemius' codepage for the latter; the image also appears in the first article I linked above. Anyway, according to Hermes, text produced with these ciphers has letter frequencies and distributions that are very similar to the odd statistics of the Voynich manuscript.

Elector Dark wrote:Damin is probably categorisable as a group conlang, conllaboolang, maybe (constructed-collaborated-taboo-language) ?

I wonder if it was some sort of language game-out-of-control process that created it. Boontling was created in a way that resembles chanspeak.

One other weird thing about Damin is it is the only click language in the world not in Africa, and clicks are posited as being some of the most ancient phones in any language so...maybe some sort of contact scenario?

Damin was constructed to be deliberately weird as hell.

Damin is also extinct.

Do we know for sure it was a conlang? I thought the idea that it was conlanged was just speculative.

It was constructed by the elders of whatever tribe to deliberately be weird.

According to who? Ken Hale?