Monthly Archives: November 2010

What makes a language complex?

There is a thread on the CBB forum concerning the most complex conlang and it was wuite strange to me what many people understood as complexity. Complexity, IMNSCO does not only mean a grammar with many categories, as many people seem to think. So here a few thoughts on what makes a language complex:

  • Strange phones/phonemes: Yes, Hindi and Tagalog might consider a certain sound common, but this does not mean that for me as a German speaker, it is. I am guilty of that sort of complexity for at least one British-English speakers about Rejistanian: Apparently, /h/ and /x/ were indistinguishable for him.
  • FUBAR phonotactics: “Sandhi, bloody sandhi” ūüėČ
  • Strange use of aspiration: This belongs to the previous point, but then, it is a more specific difficulty
  • A misleading orthography. Apparently, I am guilty of this as well: “r” for /x/ makes sense given the history of the sound or the fact that [R] is a valid allophone. However, Englishspeakers do not expect this. The fact that I write /s`/ as “x” (rejistanian, quuxlang) or “z” (Kenshuite He Mo Gie, Tsali) is probably misleading at first, but not on the level of English orthography’s oddity
  • Odd clusters of consonants or vowels: Someone mentioned to me that he prefers Volap√ľk over Esperanto because it does not have clusters like “-nstr-” which are hard to pronounce for people not used to them. From that time, I liked Volap√ľk better. While certain sounds are very easy to pronounce in isolation, in certain combinations, they are hell. I created Jasabag√©’a to eliminate difficult clusters, because I know how very guilty I am of that one: /zt/, /ss`/, /wk/ and /lt/ in initial positions of words in various conlangs
  • Tones: From a European perspective, tones are rather difficult and I assume that this is also true if the conlang has nother pattern than your L1. If your L1 has ‘only’ low and high tones and you want to speak a conlang with 9 different tones, there will be an issue. At least I assume that.
  • Dissimilarity: Try to learn these quuxlang words: “hibalama” means “woman and one of her children”, “luZatso” means “father, mother and girl”, “kae” means “human carries object” (all from quuxlang). And now remember these words: “lana” means “child”, “jile” means fast, “c√≠m√≠” means twelve (all from jasabag√©’a). And now remember “rapida” means fast, “birdo” means bird and “modifi” means to edit (all from Esperanto). I think it is safe to assume* that the Esperanto terms are easiest to remember. But even if you use a priori vocabular, if the concepts differ too much, this constitutes complexity.
  • References: It is an issue in natlangs that certain words mean something because they refer to something related. Some examples here: “Iraq is another vietnam”, “Tempo” (literally: speed) for paper tissues because the most popular brand is called like that, and “‘xikila” for “to qualify using 2 different routes” because the team which first ‘xikila-ed into the rejistanian national soccer cup was Xikila**. As the last example showed, even conlangs can and probably will develop these.
  • Irregular grammar: The grammar does not have to have many rules, but if these are really strange and have many exceptions, the language is hard to learn. Esperanto has for example 4 different ways in which an adjective can agree with the noun, Tsali has 2. Despite that, Esperanto is easier because when you know how to decline “granda” (big) (granda domo, grandan domon, grandaj domoj, grandajn domojn), you also know how to decline “nova” (new) and bela (beautiful), while Tsali does not tell you that: mazdzu’ feri, mazdzu’tik feri but elat feri, elati feri and fadi feri, fadiu’ feri.
  • Different grammar: Let’s assume that you had to learn Esperanto in one week. In terms of grammar, it would definitely be doable. Now let’s assume you had to learn rejistanian in one week, it seems as easy if there were not these darn subjunctive moods, which in this form do not exist in English, the week passes and you realize that you still have no idea when exactly to use ‘lanja, when ‘meshi, etc. In general, a grammar which makes distinctions which your natlang does not make is more difficult. As such, even 2 different evidental forms are harder to learn 3 tenses which have an equivalent in your language.
  • Complex grammar: Now we’re at the point you wanted to be. Yes, a language with no tenses is easier than one with 10 tenses***.
  • Pragmatic differences: Here, you can find things like the very indirect, ‘spiral-shaped’ way in which Chinese say things, the many pitfalls of politeness in German, the fact that according to the Language Construction Kit book, poles consider the imperative for less rude than English speakers…

BTW: this must be a record: 860 words about complexity and no hate against Toki Pona. ūüėČ

* yes, “assume makes an ass out of u(sic!) and me”, but bear with me. It might make sense nontheless.
** a fourth league team can qualify for he cup by winning either the league or the “nantical” (ie: the nanti-wide tournament for teams in the leagues below the 3rd one and non-ASR teams). Xikila did both in the first year of the cup.
*** “Yes, of course there are ten of these, this is why they are called tenses!”


Untranslatables and Badly-translatables

Sometimes, there are things, which have one term in one language and require an entire sentence in another one. This is not only true for terms which are very nation-/culture-specific (ie: tropical nations having no word for iceberg), but also for terms, which you would expect the language to have. For example, according to several native speakers who are interested in football (British usage) or soccer (for our American readers), English lacks a term for “Tabellenletzter”. Despite claiming to be the country which invented the sport and despite having a relegation system, this simple term does not exist. Thus there is no Tabellenletzter in the premier league but a “club which is on the last spot in the league table”. I am not sure why this is. It is not as if England does not have struggle against relegation. However, maybe this is related to the culture. If a culture treats failing as, well, a failure and has a generally optimistic outlook on life, there might be a reason for a lack of terms which detail failure. Or maybe this is too Sapir-Whorfian. I do not even know how I could falsify such a hypothesis in a language.

This brings me to the topic of conlangs. I have written about how to create vocabulary, but how do you prevent creating it by accident? One idea is to have an expression which is used instead and document this one. Kenshuite He Mo Gie however uses something slightly different. Kenshuite He Mo Gie’s file is more a diary of the language than a reference of it. As such, when I want to document a glaring hole, which should be there, I am happy to write “TERM intentionally left blank” or something as direct to tell me to stay away from directly translating this thing. It probably is a good idea to have a private section of your dictionary with these terms.

NaNoWriMo and conlanging

NaNoWriMo is upon us and I have to admit that I am one of the participants (at 5k words atm and writing in English). But I have thought about conlanging in respect to it. What kind of conlang would be best to win the NaNoWriMo with? Here are some ideas of mine:
1) It needs to have a well-developped grammar
2) It needs to have well-developped pragmatics, you do not want to think about how citizens of your conculture greet themselves, whether and how they use names, this kind of things
3) It needs to have a sufficient amount of root words. This does not mean that it needs to have a dictionary of 23k words like Kankonian supposedly* has, it means that there need to be sufficient ways to say what you want to say in your conlang either by words of by constructions.

And now fo some nice-to-have stuff:
1) It needs to be entered with a normal QWERTZ/Y keyboard, or whatever you use. Point is: no one wants to constantly remember which key is which character or to learn the 10-finger system for your conscript, or to enter diacritic characters from a character table.
2) It needs to be verbose. Well, yes, 50k words need to be written somehow.
3) It needs to use distinct words instead of affixes.
4) It needs not to compound but instead form terms.

This leads to a scary thought: Toki Pona is a language, which I tend to bash, but it might be the best language for the NaNoWriMo.

* I have no reason to doubt Khemehekis’ words but sheer incredulity. However, if the inventor reads this posting and takes offense, please provide evidence. Not because I am a paranoid and untrusting person, but because this seems to be an extraordinary claim, requiring extraordinary evidence.