The brain science
The underlying model
During our ongoing research and development, as we were disambiguating sentences,
it became clear that there is a single underlying model for human language.
To comprehend a sentence, we need to use the underlying brain model to intersect
meaning. Languages are beautiful in their generous use of embedded clues to meaning.
Noun phrases readily convert to clauses to provide meaning. It is awesome how sentences
can embed multiple clauses to allow a single sentence to convey the equivalent meaning
of many sentences. Brevity is wit. The main criticism of the current linguistic
science has to be in making the already complex science even more complex. By building
more complexity on the model even when it isn't producing results makes the fall
back to statistics even more compelling.
Part of speech = brain sense
Brains evolved in conjunction with senses. Senses identify the world and include
vision, touch, taste, balance, movement (using muscles, which are tracked in position)
and so on. Language describes all these things and also includes time. So what is
needed to improve on our current linguistic model of the world?
A problem with current linguistics
At its most basic, language models the real world, but linguistics is too vague
and too complicated. Today I can say that all verbs are nouns. Why? Consider this
construct. 'I ran' is a sentence that has a noun subject and uses the verb 'run'.
Now consider the sentence "I had a run". OK. Now it is a sentence with the same
subject noun, a new auxiliary verb 'had' and a noun phrase 'a run'. Surely the noun
phrase is composed of exactly the same verb? Now our dictionary needs to define
run as a verb and a noun - but isn't it the same thing? What about these examples:
- I ran. I had a run.
- I slept. I had a sleep.
- I am chewing. I am having a chew.
- I fought. I made a fight. I had a fight. It was a fight I had.
- John fought Sue with a knife agressively. John's fight with Sue with a knife was agressive.
While not all verbs work this way, for those that do it is preferable to use the
single verb definition instead of creating one for a noun. As it represents an action,
we track the definition as a verb definition.
Of course another way to convert a verb is to add ‘ing’. ‘The running of the company’
is difficult. Or just ‘The running is difficult.’ Running is being used as a noun
here. But the meaning is the same as the verb, isn't it?
Or add ‘er’. The runner is tall. Runner is a noun, converted from a verb – someone
who runs or more fully, someone who does run.
Although our dictionary is currently comprised of this duplication storing the diffently
derived forms, the future dictionary doesn’t need to, which will improve its accuracy.
Our patterns are intended to reflect the experience that generates them. Learning
'running' is an action based on experience - so would be a stored verb. The same
applies for eating, drinking and attacking. Things like dogs, cats and tables are
physical and so would be stored as nouns. Even though we can 'table' a report (a
verb). How something happens - quickly, slowly, with emotion - are stored as 'how'
not an adverb. Things like 'on' like she 'hit on' the waiter is stored as a verb
phrase, not an adverb because it is experienced with the verb. Things like 'on'
is stored as a preposition, because it is experienced as a common identifier for
when and where, but the phrase itself isn't a prepositional phrase, but part of
the 'when' or 'where' phrase rather than an adverbial phrase, for instance. In short,
to learn a language automatically we need a brain location to store it and the current
term 'adverb' is too ambiguous to use.
As you may notice, where there is no obvious, unambiguous part of speech to use,
we rely on the actual words in a language that form questions: "how", "when", "where"
and "why", for example.
Core theory: Nouns rule (they make clauses work)
The underlying brain model is the noun. Nouns, not verbs as is commonly asserted,
are the key to languages because it is the key to comprehension. Although the shortest
English sentence is a single verb in imperative form, like 'leave', there is an
implied noun that provides meaning. Otherwise when the king yells 'leave!', he could
potentially get up and go! As that never happens, we know he includes a hidden (pro)noun
in his command: "YOU leave", not "I leave" or "She leaveS".
Of course that is simplistic, but core to comprehension. There is more to language
than the noun, but how and what the rest of the model is becomes a result of investigation
into comprehension. As you use the intelligent dictionary, you will quickly notice
that clauses are everywhere. A clause is a noun with its association. Its association
is the linkset - the set that provides meaning to all words.
Examples of brain model - clauses
The easiest way to explain something is to see it in action. That's because our
brain links things automatically based on experience. Consider the following equivalent
phrases and clauses:
- The tall cat... the cat is tall.
- The sleeping cat... the cat is sleeping.
- The tired cat... the cat is tired.
- The very fat old Italian cat... the cat is very fat. the cat is old. the cat is
Italian.
- The black and white cat... the cat is black and white.
- the cat sleeping on the mat... the cat is sleeping on the mat
- the cat hit by the feather... the cat was hit by a feather
- the cat biting the dog... the cat is biting the dog
Clearly, rolling clauses into noun phrases is a quick way to identify something
and then use it. Our languages even let us embed meaning inside a phrase.
That is the cat the dog bit.
This becomes two sentences:
That is the cat. The dog bit the cat.
This construct is very efficient. Simply by using a word at the start of a phrase,
we keep the word current, but still add the relevant information to help clarify
its detail. It is much harder to say the two sentences than one because there is
more implied information in the one sentence than the two. For clarity, we would
probably say something more like: "That is the cat. The dog bit the cat I just mentioned."
which leads to the restructured: "That is the cat. The dog bit the cat. I had just
mentioned the cat."
The brain model
When you see a matched sentence that includes embedded clauses - whether clauses
themselves, noun phrases, verb phrases or passive phrases - Project Turing intersects
and displays valid matches where they exist, or the entire range of possibilities
where the dictionary doesn't contain an improved, reduced set. And as our brain
is understanding a complicated word, the types of clauses are complex enough to
model the things we need to talk about.