June 13th, 2021
The problem of using language for interactive storytelling keeps nagging at me. On the one hand, I have powerful evidence in my failure with Deikto that my language-based approaches are, so far, failures. But there remains a deep-seated force that refuses to accept failure.
I had a physics professor who taught me one of the fundamental rules of good thinking:
"Always drill down to the essence of the problem."
This rule is fundamental to my thinking process. No matter how I look at it, the essence of the problem of interactive storytelling is language. Indeed, the essense of the problem of human-computer interaction is language. It is inevitable that interaction between humans and computers will someday be carried out through language. You want proof? OK, here it is:
So, if the essence of the problem of interactive storytelling is language, then there’s not much point in working on anything else, is there? If you want to get to the top of the mountain, there just isn’t any value in working on a cure for cancer or earning a lot of money — you might as well start walking, even if it’s a very long way to the top.
Conlangs
One promising field is that of conlangs: constructed languages. There are zillions of these things; it seems that every linguistics student and everybody else who isn’t a linguistics student feels qualified to build a conglang. I have often wondered if any of these conlangs might provide the basis for a language for interactive storytelling. We can readily dismiss a large number of conlangs for a variety of reasons. For example, we can eliminate from consideration those conlangs such as Esperanto that are intended to be general-purpose languages. For our initial efforts, we absolutely MUST take a minimalist approach. We can also dismiss the many conlangs created for works of fiction, such as the Klingon language for Star Trek. A few languages remain that might provide us with a base for further work.
BlissSymbolics
This is a graphical language. It uses symbols to represent words. The symbols are not entirely arbitrary; there is a rational basis for each symbol, although the meaning is not immediately obvious to a beginner. For example, here is the word for “movie theater”:
The first symbol, meaning “house” or “building”, is a classifier; it specifies that whatever follows it is some sort of building. The second symbol, a square, represents a camera, and the third symbol, an arrow, indicates that the camera is a motion camera — that is, a movie. Thus, this symbol means “house of movies” — a theater. It’s all quite reasonable but it won’t be easy to use.
Toki Pona
This is a clever language that is meant to be minimalist in style. Its vocabulary amounts to only 123 words. It attempts to reduce the concepts of expression to their absolute fundamentals. Thus, one word in Toki Pona would be considered to represent a number of words in English. For example, the Toki Pona word anpa means both floor and defeat, as both are really lowest cases. This kind of ambiguity has interesting philosophical implications, but probably renders Toki Pona useless as a basis for a language of interaction in interactive storytelling.
Pared-down versions of English
There have been many attempts to boil the richness of English down to a simple, easy-to-learn language — to murder Shakespeare. Some of these, such as Basic English, are almost fraudulent in the way they simplify English in the most complicated manner imaginable. Special English, used by the Voice of America, pares down the English vocabulary, but is still much too large for our purposes.
Roget’s Thesaurus
I have long felt that Roget’s Thesaurus provides us with a classification system that might be usable for our purposes. It has a little more than a thousand categories, each of which could become a word. Here are some examples of its entries:
#111: Transience
#222: Weaving
#333: Cold
#444: Visibility
#555: Manifestation
#666: Consumption
#777: Permission
#888: Hope
#999: Illegality
Each category includes nouns, verbs, adjectives, and adverbs, so we’d need some kind of classifier attached to each glyph to denote which word class it belongs to. For example:
Indo-European roots
This idea has been bouncing around inside my head for some time now. We have been able to derive around 450 Proto-Indo-European roots; these are words that show up as cognates in many different languages in the Indo-European language family. For a word to show up as one of these roots, it has to be a heavily-used word. Thus, this set of roots comprises an empirically-derived condensation of language to its most important components. You can see a list of these roots here. Clicking on one entry will give you the translation of the root. The only weakness is that these words represent the lifestyle of people living about 6,000 years ago. For example, we know the Proto-Indo-European word for “horse”: *hekwos. That’s because horses were important to those people 6,000 years ago. Nowadays, not so much. Perhaps, however, we could use hekwos to mean car.
Most of these roots suffer from the same ambiguity that we have with Roget’s Thesaurus; we would again need classifiers to denote whether we’re using a word as a noun, a verb, an adjective, or an adverb.
Determinatives
A scheme used in a number of written languages uses ‘determinatives’ to specify exactly what a word means. For example, both Chinese and ancient Egyptian logographic systems use special determinatives to clarify the meanings of their logograms. Here’s an Egyptian example:
The first symbol means “good, nice, pretty”. The second symbol means “bounteous, productive thing” and the third symbol is a determinative meaning that this good productive thing is a young woman. So this word means “nubile young woman”.
Here’s a Chinese example of a determinative:
The black word on the right would normally mean “horse”. But the red determinative on the left means “female”, which changes the meaning of the word to “mother”. I don’t know why the Chinese consider a mother to be a female horse, but as a speaker of English, I am in no position to make fun of irrational aspects of other languages.
For our purposes, however, we needn’t rely on such written hacks to apply determinatives; we can use a number of tricks to specify the type or class of a word. We can surround it with a geometrical shape, as I did above with the word “hope”. We can use color in several ways, as either a background color or a foreground color. I realize that graphic designers frown upon the use of color for this purpose, as it causes problems for color-blind users. I’m ruthless about this; only about 8% of people are color-blind. If I can solve 90% of my problem by jettisoning 8% of my audience, I’m willing to make that trade-off. Damn the torpedoes, full speed ahead.
