The 2006 Smiley Award Winner: Kalusa

The 2006 Smiley Award is presented to Kalusa, an unplanned constructed language begun by Gary Shannon, and fleshed out by many more. Congratulations to Gary, and the entire Kalusa community!

Smiley Award 2006

The Origin of Kalusa

Ever since I've known Gary via Conlang, he's spent his time coming up with kooky projects that involve language and web technology, somehow. For example, there was the movement to replace fonts for pictographic language resulting in Piktok. There was the Assertion Based Language Experiment; the conlang by mutation idea; the 30 minute conlang challenge; a listing of the only six verbs you'll ever need; the entertaining SOALOA; and even a system for using the Roman alphabet as a syllabary. I was always intrigued by Gary's ideas and projects, and couldn't wait to see what he came up with next. (Update: Gary has now documented partially or in full each of these language projects on his main page, which can be found by clicking here.)

One post of Gary's discussed an accidental conlang that he and a friend of his came up with. He expressed a desire to have a "collaborative conlang created by a group of people but without any 'planning' whatsoever". The result was the conlang game Madjal. Madjal was a language anyone could add to, but it had some definitive grammar rules (e.g., only two word imperatives), and a back story. It generated some interest, but didn't quite catch on the way I think Gary was hoping it would.

Enter Kalusa!

The Kalusa Language

The Kalusa language entered this world rather modestly. It began on May 22nd, 2006, with four sentences with four English translations posted on the Kalusa homepage:

  1. Ma vito es John. "I see John."
  2. Ira vito es palu. "He sees the cat."
  3. Ira vito es teku kia ruba. "He sees the red book."
  4. Ma vito es da ruba. "I see the red one."

Accompanying these sentences was no grammar and no grammatical information of any kind. Instead, as with Madjal, Gary opened Kalusa up to any and all to modify and expand. The difference was that this time, there were some who took up the challengemore than just some. Quite a few, in fact. And in less than a week, the original corpus of four sentences had ballooned to over a thousand.

Kalusa as a Collaborative Effort

The power of Kalusa lies in its community of contributors. Creating a language from the ground up is difficult, as many of us know. Further, it takes quite awhile. What's more, one spends a lot of time figuring out grammar, and writing down grammatical rules, be the phonological, morphological or syntactic. And, then perhaps, after a week of toil, the language will consist of a more or less sizable grammar with few words and even fewer example sentences (and those that exist will probably include many of the same words, e.g. Kamakawi with "hug", "woman" and "fish", in various orders). It's a grattifying, but painstakingly slow process that can take a lifetime.

The nice thing about Kalusa is that there is no grammar work, because it's essentially done for you. If you refresh the Kalusa page every five or so minutes, there'll be new sentences that employ features of the grammar created thus far. All the visitor has to do is look at it, figure out the structure (which may be different than what the author intended), and decide whether it's acceptable or not. And then, to propose a sentence, all one needs to do is take what one has learned and reapply with a tweak here and therea new noun, a new verb, perhaps a passive version of an active sentence, etc. The great part is that if a visitor wants to create a bit of grammar, they can, but if they don't feel like it, they can contribute to the corpus, and either bolster or diminish the status of current grammatical patterns.

Further, what's best about having an army of contributors is you simply don't have to worry about aspects of grammar that don't interest you. For example, one of my small contributions to Kalusa was a method for employing modals and a way to handle biclausal structures (e.g., "I should eat" and "I want you to read this book"). One thing I didn't want to even deal with was conjunctions. Were this me creating a language on my own, I probably would have left the conjunctions until last, and would've ended up doing something hacky, or rather English-like. Since it was Kalusa, I simply sat back and watched the conjunctions develop. And develop they didin an interesting way, too, no less (e.g., the word for "and", ib, seems to have also been taken up as the comitative marker, serves as a kind of emphatic particle in places, and even works like the word "too" in certain sentences [and what's more: it was originally the word for "with", whose meaning was quickly extended to cover the function of the conjunction "and". I'd been using it so much as "and" that I totally forgot!]). In this way, the grammar is self-perpetuating. And, since there's a team working on the language, the grammar is also self-cleaning. If anyone misuses a particle or misconstrues the meaning of a word, Kalusans are ready to correct the error (or incoperate the innovation, as the voting public sees fit).

Kalusa as a Competition

Why is contributing to Kalusa so fun and so dreadfully addictive? With the ability to create sentences as well as the ability to vote in favor of or against proposed sentences, one has a vested interest in following Kalusa's progress. Here's a brief overview of some of the battles that have been fought:

  • Andru vs. Ezikize: The word for "man" sparked quite a bit of controversy early on. The original word for "man" was andru, probably derived from Greek andros. Somewhat later on, though, several words referring to humans appeared which all ended in -ze: ukivinze, "girl"; kalemaze, "woman"; aritaze, "boy"; and ezikize, "man". And thus, the battle began. One visitor would post a sentence using andru, and the next would post the same sentence using ezikize, and each would work to vote the other's sentence down. Sentence after sentence with either andru or ezikize would appear, and visitors were forced to take sides, or try to divert the interest of the Kalusa community. The tide turned when a poster tried to resolve the conflict by defining andru as "old man" and ezikize as "man". Soon there were a few other sentences trying to propose alternate definitions for andru. Soon, though, it became clear that ezikize would win the day, and indeed it did. There are no longer any sentences in the Kalusa corpus that contain the word andru. Nevertheless, it fought with valor, and its courage will not be forgotten!

  • Ma vs. Ngo/Nga: The very first word of Kalusa, according to the corpus records, was ma, which was understood to mean "I"that is, a first person singular pronoun. It remained as such, until one day, somewhere around sentence 800, an enterprising individual decided to introduce the gendered first person singular pronouns ngo, "I (masculine)" and nga, "I (feminine)". This was an interesting addition, as Kalusa had never previously shown any gender distinction in its pronominal system (though it does distinguish animacy in its third person pronouns: ira "s/he" vs. ema "it"). The introduction of these pronouns divided the community into three parts: those who rejected the new pronouns, those who were in favor of the new pronouns, and those who sought a compromise, where ma was the neutral first person singular pronoun. Sentences started appearing left and right. Some included all three pronouns in their sentences; some just one or the other. Votes were cast; hearts were broken; empires crumbled. Eventually, the issue was "decided" by a vote. 89.7% of those voting decided in favor of using only ma, with the remaining 10.3% voting in favor of ngo and nga. Nevertheless, both ngo and nga live on in the corpus, and one still sees a sentence using one or the other or both appear every now and then. Only time will tell if ma has really won the day.

  • Es vs. Ku vs. Null: When it comes to case marking, there can be no compromise! And if there's any one issue that has more bitterly divided the Kalusa community than any other, it is the issue of how to deal with the direct object of a transitive verb. From the first four sentences, we have evidence of an accusative marker of a kind in es. Nevertheless, ever since Kalusa's inception, there have been those that have used it, and those that haven't. Those who do use it vote down sentences that fail to use it; those that don't use it consistently don't seem to distinguish between a sentence with or without es. And in the midst of this, a new marker appeared: ku. First introduced as a word meaning roughly "about", ku came to be used as an accusative marker for verbs of experience (e.g., "see", "know", etc.). This bewildered members of the es faction, who countered each ku sentence with a corresponding sentence with es. In the meantime, those that didn't use es at all, continued to use sentences without either, confounding both the es and ku groups. Soon, though, the form ku began to lose its specialized meaning, and began to be used to mark the direct objects of all transitive verbs, experience verbs or active. Who will win this battle? It's nearly impossible to tell. All three types of sentence exist, and still continue to be proposed. Further, there are quite a few existing sentences of each type in the corpus. It will be interesting to see how the development of the accusative in Kalusa progresses.

And, indeed, there have been other skurmishes involving word order, lexical items, phonology, and heresy (look up the word perai in the Kalusa corpus). I've no doubt there will be more. One thing, however, is very interesting. Even if one disagrees with a particular lexical item, or way of doing things, one must still learn the lexical item or construction. Why? Because the very sentences we vote down or up constitute the entirety of the Kalusa language. Whether or not one agrees with it, if a large enough segment of the Kalusa community are going to use construction y, well then, you'd better learn construction y, like it or not! For if it makes sense, and at least a minority of Kalusans like it, the construction will stay, and will be used thereafter. Thus, a distinction can be drawn between proposals that are disliked, and proposals that are incomprehensible. In this way, one can draw a distinction between standard Kalusa, non-standard Kalusa, and ungrammatical Kalusa. And what's fascinating about that is that these distinctions can be made even though the grammar of Kalusa has never been written down in any way shape or form. All that exists are the sentences and the community of users.

Kalusa and Language Evolution

One of the interesting things about Kalusa is that, in a way, it's a naturally evolving language, similar in a number of ways to a pidgin. A pidgin arises in a contact situation where a number of languages are spoken by members of a diverse community, and a need to communicate arises. Usually, no one has the time to learn everyone else's language (especially if there's four or five of them), and no one can take time to learn a single language. As a result, a pidgin emerges: a compromise language that borrows a lot of words from one of the languages and the grammar from a number of the others. It's necessarily simpler than any one of the languages that influenced it, and tends not to have a fixed grammatical structure. Instead, there are competing structures and competing lexical items. Over time, these languages settle down and become what are known as creoles, which function just as other natural languages do. The one thing about pidgin languages that has remained a mystery is what they look like at their inception. Most of the data that exists on the world's pidgins is extremely unreliable, and is usually nothing more than a partial document of a fragment of the pidgin after it had been spoken for more than fifty years. To this day, the elementary stages of just about all of the world's pidgins remain a mystery.

From a pidgins and creoles perspective, Kalusa is interesting because it models the elementary stages of a pidgin. In several ways, it's very different, of course. The users of Kalusa do not need to use the language to accomplish some other task (e.g., bartering, doing work, etc.), and most of the users are somewhat familiar with foreign languages and some elements of linguistic theory. On the plus side, though, there is pretty much no input. All we had was four sentences, and from that the entire language spawned. And, sure enough, the language's growth has mirrored the growth of a pidgin in several rather unsurprising ways. For example, there has been a constant competition between vying structures and lexical items, as detailed above. The grammar hasn't been fixed, by any means, but has been in a state of flux. All this one might expect, not only based on linguistic observations of previous pidgins and creoles, but also based on pure common sense. That is, if no one knows how to speak language X, why would one expect anything else other than confusion? Indeed, that the language would develop a structure seems like it would be surprising. And yet, it does. Some theorists have supposed that this is due to an innate ability in humans to create grammar. Derek Bickerton has, in fact, made his career discussing his Bio Program, which he proposed to explain why all English-based creoles look alike. According to such theories, it might be the case that Kalusa would look like all other English-based creoles, and would follow the principles of Universal Grammar. The evolution of the language itself, however, points to a different analysis.

As I've watched the language grow, it seems to me that evolution of Kalusa has been determined by four factors: innovation, augmentation, analogy, and misanalysis. Innovation is simple enough to explain. New structures and new words have to come from somewhere. When there is no basis for the creation of a new structure, one must be created ex nihilo. And this has been done. UG theorists claim that in this situation, one falls back on UG principles. This hasn't been the case. For example, the first biclausal structure was created as follows. Someone created the sentence:

Jesus ira dun wepan. "Jesus wept."

Based on previous sentences, we could figure out that ira was a third person pronoun (used resumptively here), dun indicated that the action happened in the past, and wepan was "weep". To create the sentence, "I want to weep", I decided to do something off the wall:

Ma wepan, ma ziresh. "I want to weep."

This is almost like a topic-comment structure. There are several bizarre things about this sentence. First, all Kalusa sentences up to that point had been pretty much head-initial (SVO word order, NG, NA, PN, etc.). UG principles would dictate that the most logical structure for a sentence like "I want to weep" would look something like English, with "want" coming first, and "weep" coming second. I felt like putting "want" last, though, so I did. Further, I used no particle to connect the two clauses, and forced there to be two instances of the first person pronoun. This is definitely not what one would expect.

At this point, the theorist can say that I was consciously creating this sentence, and so it doesn't count as spontaneous creation. Fair enough. But one would also predict that such a bizarre structure would fall out of use. Quite the opposite turned out to be true. Many others have created sentences using ziresh in just this way, and the construction itself has spawned other modal-like words that work in exactly the same way. For example:

Feni da eyani, ma muresh. "I hope everything is all right."

It seems that what's happening is not that users are seeing a sentence, finding that it violates UG principles, and then proposing alternates that are in accord with UG principles. Rather, once a pattern is proposed, users can use that pattern to generalize to other patterns. This is where analogy comes into play. Users use the patterns that have been presented to create similar forms, and generate lexical items to be used in those patterns, whether the patterns are expected or unexpected. The only thing required is that they be a recognizable pattern that users agree on. Once the patterns are settled on, they can be augmented in various ways to build up the grammar and vocabulary required for Kalusa to continue to grow.

One interesting aspect of the growth of Kalusa has been the role of misanalysis. With over 1,000 sentences and who knows how many words, it's hard to tell what sentences are instantiations of established patterns, and which are non-standard. Further, without interlinears and a grammar, every sentence is open to multiple interpretations. For example, the da used in the fourth sentence that Gary proposed was apparently intended to be a nominalizing particle. Few if any seemed to pick up on this, though. Instead, da has been misanalyzed three times. First, it was misanalyzed as a definite article. This misanalysis didn't catch on. Then, however, it was misanalyzed as a copula which takes a noun for a subject and an adjective for an object. This use became so widely accepted that it essentially took over, and now da's primary function is to act as such a copula. That wasn't the end of it, though. Da has also been adopted as complementizer in sentences like, "The man that I saw". I don't know if this usage will survive, but sentences using da in just this way exist still in the corpus.

All in all, watching the evolution of Kalusa has been informative, and entertaining. It may not be the quintessential pidginization experiment, but I think it has shown us something interesting about the evolution of language.

How Kalusa Has Made Me Smile

Kalusa is probably the single most entertaining language creation game that exists. Translation relays are fantastic, but they require a lot of effort, and take a lot of time. You can't just plop down for an hour and decide to do a translation relay. You can, however, sit down for an hour and play in the Kalusa sandbox. And since there are a ton of other people working on the language, there's always something to do. Indeed, I fought in the battle of Ezikize, and have marveled at the ingenuity of the Kalusa community. I know the story of A'Tuin, and have been amused by the antics of perai sam (the heretics!). I love watching my ideas get adopted and augmented, such as with this killer sentence:

Za kape es ma, za dun agada, ma qaru.
/you love ACC. I, you PAST stop, I fear/
"I fear that you have stopped loving me."

And I love seeing really cool innovations, such as this one:

Za bogi es palu biti niq, ma bogi es palu noq. "You have a mere kitten, I have a real cat."

Most of all, it's fun to sit down at the computer every few hours to discover that the Kalusa language has changed in some significant way. In my opinion, it's been an incredibly successful collaborative language experiment thus far, and I hope to see it grow far into the future. And so, I'm happy to be able to award Kalusa the very first Smiley. Ma amuz es Kalusa! (Or Ma amuz ku Kalusa, or Ma amuz Kalusawhatever turns out to be grammatical.)

[Note: Gary is no longer hosting the Kalusa homepage, and so the project is, sadly, defunct. Nevertheless, it remains one of the most intriguing language creation experiments I've ever seen. Perhaps one day the time will be right to give it another try.]

Back to the Smiley Award Main Page

This page was last modified on Monday, January 17, 2011.
This website was last modified on .
This page can be viewed normally, as a milk or dark chocolate bar, in sleek black and white, or in many other ways!
All languages, fonts, pictures, and other materials copyright © 2003- David J. Peterson.

free counters