0% found this document useful (0 votes)
204 views

Parsing

This document discusses several key concepts in sentence comprehension, including: 1. Surface structure vs deep structure - how the actual words differ from the underlying meaning. 2. Parsing - the process of analyzing a sentence's syntactic structure by assigning words to linguistic categories like nouns and verbs. 3. Parsing strategies like late closure and minimal attachment that aim to reduce working memory load by attaching new words to the current constituent being parsed.
Copyright
© © All Rights Reserved
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
204 views

Parsing

This document discusses several key concepts in sentence comprehension, including: 1. Surface structure vs deep structure - how the actual words differ from the underlying meaning. 2. Parsing - the process of analyzing a sentence's syntactic structure by assigning words to linguistic categories like nouns and verbs. 3. Parsing strategies like late closure and minimal attachment that aim to reduce working memory load by attaching new words to the current constituent being parsed.
Copyright
© © All Rights Reserved
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
You are on page 1/ 26

Dr.

Francisco Perlas Dumanig


Sentence Comprehension and
Memory
Surface Structure vs. Deep Structure
Surface structure represented by the words you hear spoken or
read.

Deep structure represented by the meaning underlies in an
utterance.

Examples:
The boy threw the ball.
The ball was thrown by the boy.
Flying planes can be dangerous.
Competence vs. Performance
The way people produce language is not
equivalent to their knowledge of language.


Competence the speaker knows about the
structure of the language(Chomsky 1957, 1965).

Performance an explanation of how we can
understand speech.
Syntactic structure of sentences

To understand a sentence, the listener or reader must determine
its syntactic structure. The assignment of words in a sentence to
their relevant linguistic categories is called parsing a sentence.

Examples
1. The boy threw the ball.
2. They are eating apples.
Parsing
is the process of assigning elements of surface structure to linguistic
categories.

Due to limitations in processing resources, we begin to parse sentences
as we see or hear each word in a sentence.

We use syntactic, semantic, and pragmatic knowledge to comprehend
sentences.

An ongoing debate is whether we use these forms of knowledge
simultaneously or whether we process syntactic information first.

We hear thousands of sentences every day and respond to many,
perhaps most, with barely any notice of their structure.

We often forget the exact words a person uses to convey a message, but
some sentences linger in our memories for years.
Comprehending a sentence involves attention to syntactic, semantic, and
pragmatic factors.

Consider a simple active declarative sentence:
The actor thanked the audience.

At the syntactic level, we identify the constituent or phrase structure of the
sentence; that is, we identify the actor as a noun phrase (NP), thanked as a
verb (V), and the audience as another NP.

At the semantic level, we identify the semantic or thematic roles played by
various words in the sentence:
Actor is the agent
audience the recipient of the action.

At the pragmatic level, we probably have some knowledge about the real-
world circumstances in which this sentence would make sense. It might, for
instance, describe the end of a play after an actor has taken a bow.
Questions to ponder

Do we use our syntactic, semantic, and pragmatic knowledge
simultaneously when we comprehend a sentence?

Do certain factors take priority at various stages of the
comprehension process?

What kinds of cognitive processes are involved when a sentence,
unlike this simple declarative one, is complex enough to be a burden
for working memory?

IMMEDIATE PROCESSING OF SENTENCES
Steps

The first step in the process of understanding a sentence is to
assign elements of its surface structure to linguistic categories, a
procedure known as parsing.

The result of parsing is an internal representation of the linguistic
relationships within a sentence, usually in the form of a tree
structure or phrase marker.
S S

NP NP

det
det N
The
The
actor
S
NP VP
det N V

The actor thanked


S

NP VP

det N V NP

The actor thanked the


S

NP VP

det N V NP


The actor thanked det N

the audience Figure 6.1 Five
stages in the parsing of a sentence.

We may think of parsing as a form of problem solving or decision making in
the sense that we are making decisions (although not necessarily in a
conscious manner) about where to place incoming words into the phrase
marker we are building.

Just and Carpenter (1980) suggest that we make these decisions immediately
as we encounter a word, a principle they call the immediacy principle.

According to this view, when we first see or hear a word, we access its
meaning from permanent memory, identify its likely referent, and fit it into the
syntactic structure of the sentence.

The alternative to immediate processing is to take a wait-and-see
approach: to postpone interpreting a word or phrase until it is clearer where a
sentence is going. However, considerable evidence for immediacy principle is
available.

The primary reason that we use immediate processing is that the number of
decisions involved in understanding even a single sentence can be quite large
and thus can overload our cognitive resources.
Suppose we heard sentence:
John bought the flower for Susan.

This sentence is syntactically ambiguous. It might mean that John bought
the flower to give to Susan or that John bought a flower as a favor for Susan,
who intended to give it to another person.

This ambiguity is encountered when we hear the word for.

Flower has more than one interpretation also. It could mean flower or
flour (remember, the sentence was heard)

Although immediacy of processing reduces memory load, it may lead to
errors in parsing.

For example, consider sentence fragment :
The florist sent the flowers

The sentence looks like a simple declarative sentence in which the florist is
the subject and sent the flowers is the main verb phrase. But suppose it
continues as:

...was very pleased.

Although it at first appears to be ungrammatical, in fact this is a grammatical
sentence with an embedded relative clause (a clause that modifies a noun).

One of the reasons that the sentence is difficult to comprehend is that the
embedded clause is a reduced relative clause; it is not signaled with a
relative pronoun.

The florist who was sent the flowers was very pleased.

Another reason is that declarative sentences are more familiar than relative
clauses, so we are more likely to place our bets on that outcome.

If we took a wait-and-see approach, we would not be surprised by the
continuation. But we are surprised, so it appears that we immediately
interpret the fragment in.
Parsing Strategies
Late Closure Strategy- this strategy states that, wherever possible, we
prefer to attach new items to the current constituent (Frazier, 1987; Frazier
& Fodor, 1978; Kimball, 1973). A primary motivation for this strategy is that
it reduces the burden on working memory during parsing (Frazier, 1987).

Example:
Tom said that Bill had taken the cleaning out yesterday.

Here the adverb yesterday may be attached to the main clause (Tom
said) or the subsequent subordinate clause (Bill had taken). Frazier
and Fodor (1978) argue that we tend to prefer the latter interpretation.

Jessie put the book Kathy was reading in the library

The prepositional phrase in the library could modify either the verb put or
the verb reading. We tend to prefer attaching the prepositional phrase to
the latter verb (Frazier & Fodor, 1978)

Since Jay always jogs a mile seems like a very short distance to him.

The ambiguity in this sentence is a little artificial because it lacks a comma
after jogs. Nonetheless, the participants eye fixations were interesting.

Frazier and Rayner found that fixation times on the last few words were
longer than on the earlier ones, implying that readers had misinterpreted the
term a mile and had to make some later adjustments.
Minimal Attachment Strategy states that we prefer attaching new items into the
phrase marker being constructed using the fewest syntactic nodes consistent with
the rules of the language (Frazier, 1987; Frazier & Fodor, 1978).

For example:
Ernie kissed Marcie and her sister
A sentence fragment could be interpreted as either a noun phrase conjunction
(that is, both Marcie and her sister were recipients of a kiss) or as the beginning
of a new noun phrase.

According to minimal attachment, we prefer the former interpretation (Frazier,
1987)

Frazier and Rayners (1982) study cited earlier also found evidence for the
minimal attachment strategy. For example:
The city council argued the mayors position forcefully.
The city council argued the mayors position was incorrect.

The first sentence is consistent with minimal attachment in that the adverb
forcefully is attached to the current constituent, the VP. In contrast, the second
sentence is a complement construction that requires building a new constituent.
Frazier and Rayner found that reading times were faster for the first sentence
than for the second
Figure 6.2 Tree diagrams for (a) The city council argued the mayors position forcefully and (b)
The city council argued the mayors position was incorrect. (Based on Making and Correcting
Errors During Sentence Comprehension: Eye Movements in the Analysis of Structurally Ambigous
Sentences, by L. Frazier and K. Rayner, 1982, Cognitive Psychology, 14, p. 181, Academic
Press.)

S

NP VP

V NP
adv
(a) The city council argued the mayors position forcefully.
S

NP VP

V S
NP VP
(b) The city council argued the mayors position was incorrect.




Modular Versus Interactive Models
The parsing strategies identified by Frazier are consistent with the
modular approach to language comprehension in which comprehension
as a whole is the result of many different modules, each devoted to a
particular aspect of comprehension (Fodor, 1983).

In this view, parsing is performed initially by a syntactic module that is not
influenced by higher-order contextual variables such as the meaning of
the sentence or by general world knowledge.

Frazier (1987, 1995), for example, claims that parsing is executed by a
syntactic module, and these contextual factors influence comprehension
at a later stage.

An alternative view is that syntax and semantics interact during the
comprehension process (Britt, Perfetti, Garrod & Rayner, 1992; Crain &
Steedman, 1985;Taraban & McClelland, 1988; Tyler & Marslen-Wilson,
1977)
One type of interactive view is called the constraint-based model
(MacDonald, Pearlmutter, & Seidenberg, 1994; McClelland, 1987;
Trueswell, Tanenhaus, &Garnsey, 1994).

In this model, we simultaneously use all available information in our
initial parsing of sentence- syntactic, lexical, discourse, as well as
nonlinguistic, contextual information (Tanenhaus, Spivey- Knowlton,
Eberhard & Sedivy, 1995).

The florist sent the flowers was very pleased.
The florist sent the flowers to the elderly widow.
The florist who was sent the flowers was very pleased.

this sentence fragment may be parsed in one of two ways. The
parsing favored by the minimal attachment principle is that sent is the
verb (MV), as in sentence (12). This interpretation is a reduced
relative clause (13). This ambiguity occurs because English permits
the reduction or deletion of relative clauses such s who was.

Rayner, Carlson and Frazier (1983) examined whether the plausibility of
real-world events influenced the immediate parsing of sentences.

In sentence (14), the interpretation that the performer received the flowers is
considerably more plausible:

The performer sent the flowers was very pleased.
Parsing, the process of assigning elements of the surface structure of a
sentence to linguistic categories, is the first step in understanding a sentence.
As a result of processing limitations, we begin to analyze sentence structure
as soon as we see or hear the first words.

Two theories of parsing have been discussed. The modular approach
suggests that the words of a sentence activate syntactic processing strategies
that are used to organize the words into a phrase marker.

These strategies indicate that we prefer to attach incoming words to the most
recent constituent as opposed to attaching them to earlier constituents or
developing new ones.
Although the strategies are generally useful, they sometimes lead to
errors and subsequent reanalyses of syntactic structure.

The interactive approach emphasizes that we use all available
information including lexical, discourse, and contextual factors. Whereas
the modular approach insists that syntactically based strategies are
used first, with lexical and discourse factors coming in later, the
interactive model asserts that we simultaneously use all available
information to parse sentences.

Current research supports the role of lexical and contextual factors in
parsing, but the role of discourse factors is less evident.

Models of Sentence Parsing
Garden Path Model of Sentence Processing
The parser makes only one initial syntactic analysis of a word
sequence. This initial parse is made on the basis of several rules
and parsing principles.

There are two important principles in garden path model: late
closure principle and minimal attachment principle


Garden path model
Late closure principle focuses on the way in which listeners (or
readers) might determine when they have reached a major
clause boundary.

In doing this one might attempt to close a clause boundary either
at the earliest point possible or to hold off until the latest point
possible.

Examples:
Because Jay always jogs a mile. (clause boundary comes after
a mile late closure)
Because Jay always jogs a mile, this seems like a short distance.
(clause boundary comes after jogs - early closure)

Minimal attachment
This principle states that listeners or readers
attempt to interpret sentences in terms of the
simplest syntactic structure that is consistent with
the input. This is done by using the fewest
phrase structure nodes possible.

Example
Because Jay always jogs, a mil seems to be a short
distance to him.

You might also like