Lambda the Ultimate

There are linguistic theories under which language "structure" is not distinct from the lexicon. In Ronald Langacker's "cognitive grammar," for example, there is no separate "grammar" and "lexicon," there is only an inventory of symbols, or "constructions," where a symbol is defined as a structure which has a semantic and phonological pole.

It's more involved than I can go into in a short comment, but in this system the phonological pole can be incompletely specified to the point where it can be used to describe structures which include no specific words, only arrangements of potential words. Grammatical constructions would be symbols of this type -- symbols with very very abstract phonological poles. (Their semantic poles would generally be very abstract too.)

The grammar of a language, to Langacker, is nothing but the complete inventory of its constructions, and there are no "rules" apart from constructions.

Charles Fillmore and Adele Goldberg (yes, that Adele Goldberg) have done some work on a very similar theory, which they call "Construction Grammar." Adele Goldberg, appropriately enough, suggests an object-orientation-style inheritance structure which relates different constructions.

George Lakoff has done a lot of work popularizing similar linguistic theories.

All of these theorists would be considered rebels against classic Chomskian orthodoxy, which is so terribly hostile to the Sapir-Whorf hypothesis, though as far as I know only Lakoff has engaged in defense of qualified Sapir-Whorfian claims (in his book _Women, Fire, and Dangerous Things_ there is a chapter about a number of possible versions of the hypothesis, some of which he thinks are defensible).

I'm assuming "that Adele Goldberg" refers to the Dr. Goldberg of Smalltalk fame. For what it's worth, I'm pretty sure this is actually a different Adele Goldberg. Her extensive CV contains no mention of Smalltalk, and she looks strikingly different from Smalltalk's Adele Goldberg.

This is pretty confusing, especially since they both seem to have spent time at Xerox PARC. I just happen to remember the linguist Goldberg from my few linguistics classes at UIUC...

Aymara has `trivalent logic' http://www.aymara.org

jan

I take it you have never heard of hair colouring. ;-)

To my eye, both photos are the same person a few years apart with different hair colour and style.

I hope your never asked to examine a police line-up. ;-)

The thing I find most interesting about programming languages is that they are almost entirely dedicated to structure, as opposed to primitives. Most programming languages have at most a couple hundred primitives (variables, functions, pointers, addition, etc) which can be used to build any program. But just try to get by in the world with only a couple hundred words.

Natural languages are more like the first versions of BASIC: a special keyword for every operation of interest, all of which are built into the language itself instead of being defined in a library.

Some natural languages seem to be more composable than others. For example Greek, with its prefixes, roots, and suffixes. Or Chinese, where characters are frequently composed out of simpler characters. I've occasionally wondered whether this is simply because they had less outside influence than English, with its mix of Anglo-Saxon, German, and French.

Are "pure" languages (Greek, Chinese) usually more orthogonal and consistent than "mongrel" languages (English)? Or am I merely generalizing from too small a sample set?

The designer(s) of Loglan realized that for it to achieve even a modest amount of success, it would have to be relatively easy to learn. One aspect of that is their system for creating new words from their basic set of primitives, in a somewhat orthogonal way. While I didn't find that very helpful in learning or guessing the meaning of a new combination-word, I did find it to help somewhat as a mnemonic, to remember those words once learned. Another aspect of that focus on ease of learning was the way they selected the primitives; that was IMHO somewhat less successful.

What I found to be the killer (aside from my lack of time to pursue it, and its status as little more than a very interesting mental/linguistic exercise) was the excess of precision. There's a difference between ambiguity and vagueness, and I think humans require support for some vagueness in their speech.

I toyed once with constructing a logical spoken language and orthography, and was interested in this idea of minimizing "primitives". My general approach was to enforce a means of simplifying and orthogonalizing the basics, and simultaneously coordinating this with the phonetics of the language. The resulting system was designed to phonetically designate various aspects of the sense of a word as tense or case using a suffix.

The interesting part was that I organized the orthogonalized "dimensions" of a word's sense to match some notion of phonetical dimension and coordinate. If you look at the International Phonetic Alphabet's organization, you see some structure that can be employed relatively intuitively. So my informal result was that a lot of meaning could be rattled off in a few syllables (which could be redundantly used for emphasis or confirmation), and that the listener had some immediate intuitive means of deciphering it from its sound.

I never did this as part of a formal research program, though, but I might pick it up again and polish it enough to publish some guide to the idea given enough interest (a minimal number of details are on my home page). There were some aims that I had with this system that I never fully explored. Oh, and I never have placed much stock in predicate logic itself for natural language. Dynamic predicate logic (as in the Dynamo programming language) in the least is more suitable for framing and interpreting thought processes and descriptions and stories.

Languages that have mostly or all native speakers are generally more complex. Languages like English that have had many non-native speakers over the years are less complex than say, a language spoken only by a small tribe for thousands of years. Simpler still are pidgin languages that are spoken primarilily by nonnative speakers (such as Tok Pisin). For more on this a good book is "The Power of Babel" by John McWhorter.

But just try to get by in the world with only a couple hundred words. See: http://ebtx.com/lang/eminfrm.htm for a language that attempts this.

In English, common words like "I", "me", "cat", "eat" are short, monosyllabic and easy to say, as compared to infrequently used words such as "antidisestablishmentarianism", which are difficult to say, but are so rare that it doesn't matter.

I looked as a Lojban dictionary and it looks like all the words are 5 characters in length, regardless or how frequently they're used.

I think that a good synthetic language would try to maximize information bandwidth by simplifying common words.

Yes, that occurred to me, although they still look different to me. In any case, I really think the documentary evidence indicates that they're different people. PhD's from different places in different fields, completely different careers, and so on. If it's really just one person, she's apparently living two lives and has taken great pains to keep them separate...

Apart from physical resemblance (or lack thereof), do you have any other reason to believe they're the same person?

I hope your never asked to examine a police line-up. ;-)

Well, if I were, it looks like I'd be more likely to let a guilty person walk than to identify an innocent, which, in my opinion, is a better state of affairs anyway... ;-)

It looks like yours is the better policy; the link supplied by Florian Hars suggests they are NOT the same person.

I wonder if they are related though... it seems odd that share the same name as well as the same jaw line and nose. ;-)