topics: |  linguistics | syntax | information-theory

Bampton lectures in America no.28 [Columbia U, 1986].

Describes a formal theory of language structure where entities are defined
by their frequency of occurrence, rather than by phonetic or semantic
properties. 

Sufficient regularities [in syntax] have not been found... In each language
there are some relations that can be called grammatical, but a satisfactory
general definition is lacking.  Furthermore, grammatical relations are
unique to natural language, and if we can describe language only in such
terms we will be unable to compare language to anything else, not even with
such close relatives as gesture on the one hand and mathematics on the
other. [p.1-2]

Finally, the elements on which grammatical relations hold are not
adequately defined.  The one type of element that is precisely established
is the set of phonemes, the charcacteristic sounds of language.

In general, the investigation of a field, and the defining of its entities,
is carried out in a metalanguage of this field, a language of broader
informational capacity than the given field.  This is clearly so in
mathematics and logic, where the precision as to what is in the field enables
us to recognize that statements said about the field are not in it...  But NL
has no external metalanguage.  [p.2]

Procedures for discovering the elements of language

A. Phonology:
	it is possible to determine the phonemic distinctions in a language
	by a behavioral test that does not involve the specific meaning of
	words ... test with S,H both speakers of the lg:
		S : utters two words (e.g. sea, see; or hard, heart)
		H : says if these were repeated words or different
	phonemes = most economical collection of distinctions arising in diff
		contexts- e.g. the ph of pin vs the p of spin

B. Word boundaries
	check the n-grams - e.g. in sentence "if he comes call me",
	the probability of "ifh" is lower than other phrases.

C. Sentence boundaries
	similar to above, but with words, and with a more complex stochastic
	process.  at some points the probabiliity of the next word returns to
	the situation at the start of the sentence.

the phoneme sequence property is less interesting - cannot be tied to how
meaning arises; but the word sequence relation is of
decisive importance for the structure and meaning of sentences.

Ellipsis:
   Given two sentences connected by "and" - e.g.
		I knocked and I entered
		John came over and John introduced Mary
		   etc.
    suggests that the "least grammar" that accounts for John V and W goes
    via John V and John W plus the zeroing of the element in the second
    that appears n the same position as in the first. 8