Study Guides (390,000)
US (220,000)
UCLA (3,000)
LING (40)
LING 1 (20)
Green (4)

LING 1 Lecture Notes - Lecture 4: Phrase Structure Rules, PronounExam

Course Code
Study Guide

This preview shows page 1. to view the full 5 pages of the document.
LING 1 (Fall 2015) Supplementary reading for Week 4
Toru Minamimoto
Syntax – the study of how sentences are made up from words
§1: Facts and theories
It is a fact that I will visit my parents tomorrow is a good English sentence. It is also a fact that *I visit my parents
tomorrow will is not a good English sentence. It is, then, also a fact that native speakers of English can make this kind
of judgements, telling apart descriptively grammatical English sentences from utterances that are descriptively
ungrammatical. Native speakers of English can make this kind of judgement on any sequence of words that they have
never heard before. Therefore their brain must have some sort of computational system that allows them to make such
Linguists want to know exactly what this system is, so we build hypothetical models (set of hypothetical
rules) that can imitate what native speakers have in their brain. If our hypothetical model can imitate native speakers
and make the same judgements as native speakers would make (that is, if it can generate all potential sentences that
are acceptable in English and no sentence that is unacceptable in English), then we can say that our model is probably
close to what native speakers have in their brain. This is, of course, a very distant goal, even if we limit our target to
one single language, English.
Toward this goal some hypotheses have been proposed. The finite state grammar is one attempt, but linguists
now know that it can’t be a model of human language (though it may have useful applications elsewhere). The phrase
structure grammar appears to be more promising.
§2: Finite state grammars and phrase structure grammars
A finite state grammar focuses on the relation between immediately neighboring words: in short, it is made up of
rules stating which words may follow which words.
But it is now known that finite state grammars can not represent how actual sentences are built. Here is one
example sentence that a finite state grammar cannot deal with: in the sentence The man who designed all these
buildings is a German, the singular form of the verb is is determined by the singular noun man, appearing far away
from the verb is. Since a finite state grammar contains rules only for two immediately neighboring words, it cannot
deal with this kind of relationship between distant words. (I picked an example I thought easy to grasp. To be precise,
this specific example may be analyzed with a finite state grammar, in a very unelegant, uneconomical and unlikely
way that would devote much resource to this problem alone.)
More promising is the phrase structure grammar. It focuses on the grouping of words, without paying attention to
which words may end up neighboring each other. It is capable of dealing with long-distance dependencies, and
furthermore, we know that grouping of words – constituency – is important in English sentences (and in all human
languages, as far as we know). Therefore the phrase structure grammar seems to be a good model for how sentences
are built.
You're Reading a Preview

Unlock to view full version

Only page 1 are available for preview. Some parts have been intentionally blurred.

§3: Constituency
Sentences are not mere sequences of words. Some words within a sentence are closely tied together in a way other
words are not.
Example sentence: I will visit my parents tomorrow.
It is intuitive that, in this sentence, the two words my and parents are tied closely together, while the two words
parents and tomorrow are not. But what about the two words will and visit ? It is not entirely obvious.
A unit made up of words closely tied together is called a constituent. Each word is a constituent; the whole
sentence is one big constituent. Which words form constituents and which don’t is sometimes obvious (like my
parents above), sometimes not (like will meet above – in fact will meet is not a constituent). Therefore we need some
objective tests that confirm constituency. There are many different tests, but in LING 1 we will learn three of them:
- Stand-alone answer to a wh-question
- Forward movement
- Replacement with a pronoun
These tests must be done on native speakers. All these tests depend on the native speakers ability to distinguish
acceptable (descriptively grammatical) expressions from impossible (descriptively ungrammatical) expressions.
VERY IMPORTANT – The logic behind these constituency tests is, “if one group of words passes a constituency test,
then that group is a constituent” or, equivalently, in order for one group of words to pass a constituency test, it needs
to be a constituent”. It follows from this that, if one test says that a group of words is a constituent, then we have
evidence that it’s a constituent. If one test does not say that it’s a constituent, then that group of words may or may
not be a constituent – we just don’t have evidence that it is. (This means that, in principle, a group of words may be a
constituent even if all the three tests fail. In fact, it is difficult to prove that a group of words is not a constituent.)
Some other important things to note:
Constituents are made of neighboring words. For example, in a sentence I will meet my parents tomorrow, it
is not allowed to think that the two words will and tomorrow may form a constituent (even though it’s true
that these two words both express futurity), because some words come in between
This does not mean that the two words will and tomorrow in the sentence above don’t belong to one
constituent. In fact there are big constituents that contain both will and tomorrow, like will meet my parents
tomorrow – the point here is that there’s no constituent that includes will and tomorrow to the exclusion of all
the other words.
The point above is not limited to discontinuous words. When we think of constituency, it is important to
always consider which group of words forms a constituent to the exclusion of all the others. For example, in
the sentence I will meet my parents tomorrow, there are constituents that contain both meet and my, like meet
my parents, meet my parents tomorrow, etc. Crucially, however, there is no constituent that contains meet and
my and nothing else.
Constituents can contain constituents. In the example above, the constituent will meet my parents tomorrow
contains (among others) a constituent meet my parents, and this latter contains (among others) a constituent
my parents, etc. etc. This multiple-layered structure may be illustrated by a tree or by the bracket notation.
You're Reading a Preview

Unlock to view full version