involvement, and cannot be decomposed into smaller
units (they are thus first articulation units).
If such devices were to convey the totality of human
experience, instead than as actually happens, a fairly
limited number of messages, the first articulation units of
their codes would need a tremendous increase in number,
with ensuing problems for storage facilities.
As to a structural semantics, a variety of views is present:
firstly, Bloomfield declared as impossible the scientifical
study of meanings: european scholars seem more trustful
about possible structural models for lexicon. In the late
fifties, Hjelmseev claimed that minimal meaningful units
could be utterly decomposed into smaller units, resemling
decomposition into phonems: for instance, “car” may be
divided into small meaningful units as “vehicle”,+ “motor
drive”+”four wheels”, and so on.
The grammar of a language is a scheme able to specify
sentences allowed in the language and to show the rules
for merging words into propositions.
In programs aimed to processing of natural language, the
asset their meaning and draw an appropriate response.
One of main contributions in this field was owed to
Chomsky (1950), with this theory of formal languages. A
formal language is said to be a set (possibly, infinite) of
finite length strings, made with a finite vocabulary of
symbols.
The grammar of a formal language has the following
basic concepts:
■ syntactic categories, like “sentence” and “noun
phrase”: they are said nonterminal symbols, or
variables
■ terminal symbols (that is, the words of a
common language) which are assembled in the
strings called sentences (A language is a subset in the
general set of all possible strings which may
assembled in all possible ways.)
■ the laws of re-writing, or productions, which
state relations among a number of strings
■ the start symbol.
The set of strings of terminal symbols that may derive
from this symbol using the sequences of productions
rules, is said the grammar generated language. 5
5. Linguistics and AI
Chomsky found four types of grammar, numbered from 0
to 3. The zero-type grammar is the simplest one, having
no restraint in re-writing laws. For further grammar the
form of re-writing laws becomes more and more strict.
For example, the grammar 3 (also said “context-free
grammar”) can have only one single non-terminal symbol
(on its left side) for every production. A context free
grammar produces statements like ab, aabb, aaabbb.
For instance:
<SEN TENCE> -> <NOUN PHRASE><VERB
PHRASE>
<NOUN PHRASE> -> the <NOUN>
<NOUN> -troad networks
<NOUN> —»railway networks
<VERB PHRASE> ->grow
Chomsky (Syntactic structure, 1957) devised a theory of
language to which refers transformational generative
grammar:”a statement is the surface expression of a
deeper structure, that is the real meaning of the
statement”.
The deep structure may suffer a number of formal
changes (in the order of nouns, in suffixes,.etc.), as it
comes to surfaces, although holding its substantial
meaning.
A new grammar of a language (like english) should be
generative, that is a finite length statement able to:
• take into account the infinity of possible statements
in the said language
• allot to any statement are structural description that
may capture the knowledge of language for a typical
individual
Chomsky depicted for the grammar a 3-parts
organization:
■ the first part should have a phrasal structure
which produces sets of morphemes for simple
declarative statements, each holding a derivative tree
■ a sequence of transformation rules should re
assembles sentences, adding or deleting morphemes
until the entire variety of possible sentences
represented
■ a sequence of morpho-phonemic rules should, at
the end, map all representations of sentences with a
set of phonemes.
The feature of formal definitions of languages, from the
viewpoints of computational linguistics and of processing
of natural languages, is that the structure of statements is
passed trough an algorithm of parsing, in order to be
analyzed.
Suppose that the phrase structure grammar has produced
the following derivation tree.
SENTENCE
NOUN PHRASE VERB PHRASE
NP-SINGULAR VERB NOUN PHRASE
DET NOUN AUX V NP-PLURAL
DET NOUN
The wood TENSE
the built area
surrounds