Panlingua, by Chaumont Devin, May 9, 1998.

Chapter 5, Deductions.

Because this is a somewhat unrefined compilation of ideas, for now I will call it "The Jungle." Welcome to the jungle!

Many of the most important discoveries in the hard sciences are made by deduction. For example, no one has ever seen an atom. Atoms are known only by deduction, and yet nearly everything we touch today is in some way related to a detailed knowledge of atoms. Yet in linguistics it would seem there might be some unwritten law, namely that nothing can be determined by deduction. This approach leads to endless discoveries about the most trivial of details about syntax, which can be observed directly, and virtually nothing about anything else.

I do not subscribe to this rule, therefore most of my work employs deduction. I employ deduction because no one has ever devised any other means to examine the detailed workings of the human brain. Fortunately for us, deduction is a very powerful tool. Here are some of the assumptions, observations, and deductions upon which my theory of Panlingua is based:

If Panlingua is the universal subsurface language common to all of mankind and surface language is the summation of all spoken languages and if surface language is free to assume various forms and if the meanings of surface forms are represented in Panlingua then any Panlingua feature will probably appear in surface language sometime.

If Panlingua is subsurface language and spoken and written languages are surface language and parsing is translation from surface to subsurface language and generation is translation from subsurface to surface language then translational processes are continually at work in the human mind.

If all biological organisms evolve towards maximum efficiency over time then these translating components must likewise improve over time

If races have been separated long enough to change physically yet infants of any race can readily acquire the language of any other then Panlingua itself cannot be changing over time

If maximum efficiency means minimum translation and Panlingua doesn't change over time then surface language must become more like Panlingua over time

If some feature is true of Panlingua and surface language becomes more like Panlingua over time then that feature will evidence itself in surface language if it can

If surface language becomes more like Panlingua over time and a feature is common in surface language then that feature probably reflects some feature of Panlingua

If surface languages are spoken and written languages and Panlingua's features can be deduced by examining many surface languages and it is important that the details of Panlingua be deduced then it may be critical that all surface languages be preserved. If Panlingua does not change over time and surface languages are free to take any form and only a few surface languages survive

and these are employed to deduce the features of Panlingua then the probability of success will be small

If the human linguistic apparatus uses Panlingua and the human linguistic apparatus works and no other linguistic apparatus works then the only linguistic devices that work use Panlingua

If lower animals use subhuman Panlingua and lower animals think poorly and higher animals think better and higher animals show signs of linguistic understanding then Panlingua must have evolved

If animals using advanced Panlingua think better then thinking must be a function of Panlingua and thinking improves as human Panlingua is approached

If animals using advanced Panlingua have better linguistic ability then linguistic ability must be a function of Panlingua and linguistic ability improves as human Panlingua is approached

If the only linguistic apparatuses that work use Panlingua and computers are linguistic devices and linguistic ability improves as human Panlingua is approached then the better a computer can perform linguistically the closer it has come to using human Panlingua

If better linguistic performance means approaching human Panlingua then computational linguistic systems approach human Panlingua as they improve

If computational systems improve as they approach Panlingua then all refinements in computational linguistics converge on Panlingua

If computational linguistic systems improve steadily and this improvements means approaching Panlingua then it is not necessary to conserve surface languages to deduce Panlingua and people can learn Panlingua by studying machines

If many surface languages are available and traits common to many must belong to Panlingua and no machine has yet learned a language and it doesn't look like any machine will do so very soon then to learn Panlingua from surface languages may be faster than from machines

If a state is the way something is and brains can model the way various things are in various states then a brain can model a thing in some state just by knowing what thing and what state

If a brain can model various things in various states and language can represent various things and the states they are in and such a representation can reflect reality then a brain can use language to model reality in this way

If each set of things and states can be called a frame and moving through a set of frames can reflect reality over time and changing the states of some things makes a next frame and language can represent these changes of state then a brain can use language to update its model through time

If without exception all languages encode the maintenance or assumption of various states by various things and nothing less or more then Panlingua must also encode the maintenance or assumption of various states by various things and nothing less or more

And if a brain can use this information to form its model of reality then this must be what brains do with Panlingua all the time

If the words of languages all depend on other words and word order changes through surface languages without changing meaning and Panlingua does not require word order to exist then Panlingua must be shaped like a tree

If Panlingua structures are shaped like trees and any linguistic sequencing must be retained then Panlingua must be shaped like a BINARY tree

If Panlingua is shaped like a binary tree and one regent word can have many dependent words meaning one parent node of the tree can have many daughter nodes then Panlingua must be shaped like a Tinkertoy structure in the syntactic plane

If the sticks of this Tinkertoy structure ar synlinks and atoms connected by horizontal synlinks are siblings and siblings have the same parent node and synlink type encodes word dependency type then the types of these horizontal synlinks are the types of the synlinks linking the atoms on their right-hand sides to their regent words (parent nodes)

If Panlingua structures are like trees and a noun is a word representing a thing and thoughts are things and entire thoughts can replace nouns in all surface languages then entire thoughts can probably replace nouns in Panlingua also

If a verb is a word representing the maintenance or assumption of some state and every thought has but one and only one verb and its dependents and this is the same definition as that for "clause" and all subordinate clauses are treated as nouns from higher clauses and the heads of all subordinate clauses are verbs and a verb cannot simultaneously be a noun then a noun will be required to serve as regent to the verb of every subordinate clause in Panlingua

If the purpose of part-of-speech classifications is to determinewhat synlinks and lexlinks are possible for a certain symbol and if this information is contained in the lexicon and the purpose of the lexicon is only for communication with the outside world and the lexicon is not necessary to the internal functions of language then part-of-speech is not a component of the ontology or Panlingua

If each word in a sentence has but one and only one meaning and each word in a sentence has but one and only one syntactic role then each word in a sentence must have but one and only one lexlink and but one and only one synlink

If a word need nothing more than a synlink and a lexlink to mean what it means then the meaning of a word is completely defined by its synlink and its lexlink

If a word need nothing more than a synlink and a lexlink to mean what it means then a word requires no symbol or sequential position in a line of text

If both a synlink and a lexlink are required for meaning then meaning cannot be divorced from syntactic structure

If both a synlink and a lexlink are required for meaning then the syntactic/semantic dichotomy is a fiction of modern linguistics

Etc.