Langston, Psychology of Language, Notes 7 -- Syntax
 
I.  Goals:
A.  Introduction.
B.  Formal grammars.
C.  Parsing strategies.
 
II.  Introduction.  We've discussed decoding processes (getting from sounds and letters to words), but we still haven't done anything with the stuff we've decoded.  I want to emphasize a couple of basic themes:
A.  Language things can be described by a grammar (a set of elements and rules for combining those elements).  As an exercise, we should review the elements and rules we know about so far:
1.  Perceiving speech.
2.  Reading.
3.  Meanings.
B.  Language is frequently understood in spite of a lot of ambiguity.  During comprehension, something has to be done to resolve ambiguity.  You can take two approaches:
1.  Brute force:  Represent every possibility.  For lexical access, this seems to be what happens.  If someone reads "He dug with the spade," "Shovel" and "ace" are equally activated.  The wrong meaning is quickly suppressed.  There are good reasons for lexical access to work this way, and it seems to be the only process that uses brute force.
2.  Immediacy assumption:  Make a choice, try to go on, if you get stuck, reevaluate.  As we start to get into syntax, you'll see why this has to happen.  There are too many possibilities to maintain all of them.
As an exercise, let's rehearse some of the ambiguities we've seen so far:
1.  Perceiving speech.
2.  Reading.
3.  Meanings.
C.  Now we come to the point where things get exciting.  Language usually comes at us in sentences, and that's our next level of analysis.  What is the meaning of a sentence?  There are three parts:  Syntax (the rules for grammar); Semantics (the rules for combining units of meaning); and Pragmatics (extra-linguistic knowledge that helps you interpret the content of a message).  We'll address each in turn.
To start grammar, I want you to try a couple of exercises that will get at our basic themes:
1.  Write down the meaning of "of."  It should be hard because it represents an empty syntactic category, and not a word.
2.  What is the meaning of this sentence:  "The boy saw the statue in the park with the telescope."  Could the statue have the telescope?  Could the boy be in the park?  This is a kind of syntactic ambiguity.
 
Top
 
III.  Formal grammars.
A.  Word string grammars (finite state grammars):  Early attempts to model sentences treated them as a string of associations.  If you have the sentence "The boy saw the statue," "the" is the stimulus for "boy," "boy" is the stimulus for "saw,"...  If a speaker is processing language, the initial input is the stimulus for the first word, which, when spoken, becomes the stimulus for the next word, ...
These ideas get tested with "sentences" of nonsense words.  If you make people memorize "Vrom frug trag wolx pret," and then ask for associations (by a type of free association task), you get a pattern like this:
 
Cue Report
vrom 
frug 
trag
frug 
trag 
wolx...
 
It looks like people have a chain of associations.  This is essentially the behaviorist approach to language.
Chomsky had three things to say in response to this:
1.  Long distance dependencies are problematic.  A long distance dependency is when something later in a sentence depends on something earlier.  For example, verbs agree in number with nouns.  If I say "The dogs that walked in the grass pee on trees," I have to hold in mind that plural "dogs" take the verb form "pee" and not "pees" for five words.  Other forms of this are sentences like "If...then..." and "Either...or..."  In order to know how to close them, you have to remember how you opened them.
2.  Sentences have an underlying structure that can't be represented in a string of words.  If you have people memorize a sentence like "Pale children eat cold bread," you get an entirely different pattern of association:
 
Cue Report
pale 
children 
eat 
cold 
bread
children 
pale 
cold or bread 
bread 
cold
 
Why?  "Pale children" is not a pair of words.  It's a noun phrase.  The two words produce each other as associates because they're part of the same thing.  To get a representation of a sentence, you need to use (at least) a phrase structure grammar.  That's our next proposal.
3.  Something can be a sentence with no associations.  "Colorless green ideas sleep furiously" still works as a sentence, even though you probably have no association between colorless and green.
B.  Phrase structure grammars (a.k.a. surface structure grammars):  Generate a structure called a phrase marker when parsing (analyzing the grammatical structure of the sentence).  It proceeds in the order that the words occur in the sentence, and the process is to successively group words into larger and larger units (to reflect the hierarchical structure of the sentence).  For example:
 
 (1)  The television shows the boring program
 
The phrase marker is a labeled tree diagram that illustrates the structure of the sentence.  The phrase marker is a result of the parsing.
 
 
What's the grammar?  It's a series of rewrite rules.  You have a unit on the left that's rewritten into the units on the right.  For our grammar (what we need to parse the sentence above) the rules are:
 
P1:  S -> NP VP
P2:  VP -> V NP
P3:  NP -> Det (Adj)* N
L1:  N -> {television, program}
L2:  Det -> {the}
L3:  V -> {shows}
L4:  Adj -> {boring}

The rules can be used to parse and to generate.
C.  Transformational grammars:  There are some constructions you can't handle with phrase structure grammars.  For example, consider the sentences below:

 (2)  John phoned up the woman.
 (3)  John phoned the woman up.

Both sentences have the same verb ("phone-up").  But, the "up" is not always adjacent to the "phone."  This phenomenon is called particle movement.  You can't parse or generate these sentences with a simple phrase-structure grammar.  Problems like this were part of the motivation for transformational grammar's development.  Chomsky is also responsible for this insight.
What is transformational grammar?  Add some concepts to our phrase-structure grammar (both technical and philosophical):
1.  The notion of a deep structure:  Sentences have two levels of analysis:  Surface structure and deep structure.  The surface structure is the sentence that's produced.  The deep structure is an intermediate stage in the production of a sentence.  It's got the words and a basic grammar.
2.  Transformations:  You get from the deep structure to the surface structure by passing it through a set of transformations (hence the name transformational grammar).  These transformations allow you to map a deep structure onto many possible surface structures.
3.  Expand the left side of the rewrite rules:  Transformation rules can have more than one element on the left side.  This is a technical point, but without it you couldn't do transformations.
Why do we need to talk about deep structure?  It explains two otherwise difficult problems:
1.  Two sentences with the same surface structure can have very different meanings.  Consider:

 (4a and 4b)  Flying planes can be dangerous.

This can mean either "planes that are flying can be dangerous" or "the act of flying a plane can be dangerous."  If you allow this surface structure to be produced by two entirely different deep structures, it's no problem.
2.  Two sentences with very different surface structures can have the same meaning.  Consider:

 (5)  Arlene is playing the tuba.
 (6)  The tuba is being played by Arlene.

These both mean the same thing, but how can a phrase structure grammar represent that?  With a deep structure (and transformations) it's easy.
Let's get into this with Chomsky's (1957) Toy Transformational Grammar.  His model has four basic stages:
 

 
Phrase structure rules allow you to construct basic trees (what we've already seen).  Then lexical insertion rules put on the words.  That makes a deep structure.  Then, you do some transformations.  Finally, you go through a pronunciation stage and you have the surface structure, or the final sentence.
The rules (different for each stage):
a.  Phrase structure rules:
P1:  S -> NP VP
P2:  NP -> Det N
P3:  VP -> Aux V (NP)
P4:  Aux -> C (M) (have en) (be ing)
b.  Lexical insertion rules:
L1:  Det -> {a, an, the}
L2:  M -> {could, would, should, can, must, ...}
L3:  C -> {ø, -s (singular subject), -past (abstract past marker), -ing (progressive), -en (past participle), ...}
L4:  N -> {cookie, boy, ...}
L5:  V -> {steal, ...}
c.  Transformation rules:
Obligatory
T1:  Affix (C) V -> V Affix (affix hopping rule)
Optional
T2:  NP1 Aux V NP2 -> NP2 Aux be en V by NP1 (active -> passive transformation)
d.  Morpho-phonological rules:
M1:  steal Æ /stil/
M2:  be Æ /bi/ etc.
How does this work?  I've done an example for "The boy steals the cookie."  (Tree given in class)
 
Produce the sentence "The boy steals the cookie."  (Tree given in class)
 
Here's a harder one because it involves the passive.  Produce "The cookie is stolen by the boy."  (Tree given in class)
 
You try "The cookie could have been being stolen by the boy" as an exercise.
Psychological evidence for transformations.  Early studies rewrote sentences into transformed versions.  For what follows, the base sentence is "the man was enjoying the sunshine."
 
Negative:  "The man was not enjoying the sunshine."
Passive:  "The sunshine was being enjoyed by the man."
Question:  "Was the man enjoying the sunshine?"
Negative + Passive:  "The sunshine was not being enjoyed by the man."
Negative + Question:  "Was the man not enjoying the sunshine?"
Passive + Question:  "Was the sunshine being enjoyed by the man?"
Negative + Passive + Question:  "Was the sunshine not being enjoyed by the man?"
 
I have you read lots of these and measure reading time.  The more transformations you have to undo, the longer it should take.  That happens.  Note the problem with unconfounding reading time from the number of words in the sentence.
D.  Semantic grammars:  The PSG and transformational grammars parse sentences with empty syntactic categories.  For example, NP doesn't mean anything, it's just a marker to hold a piece of information.  These syntax models have some problems:
1.  Not very elegant.  The computations can be pretty complicated.
2.  Overly powerful.  Why this set of transformations?  Why not transformations to go from "The girl tied her shoe" to "Shoe by tied is the girl"?  There's no good reason to explain the particular set of transformations that people seem to use.
3.  They ignore meaning.  Chomsky's sentence "They are cooking apples" isn't ambiguous in a story about a boy asking a grocer why he's selling ugly, bruised apples.  Sentences always come in a context that can help you understand them.
Semantic grammars are different in spirit.  The syntactic representation of a sentence is based on the meaning of the sentence.  For example, consider Fillmore's (1968) case-role grammar.  Cases and roles are the parts each element of the sentence plays in conveying the meaning.  You have roles like agent and patient, and cases like location and time.  The verb is the organizing unit.  Everything else in the sentence is related to the verb.  Consider a parse of:
 
 (7)  Father carved the turkey at the Thanksgiving dinner table with his new carving knife.
 
 
The nodes here have meaning.  You build these structures as you read, and the meaning is in the structure.  You can do things purely syntactic grammars can't.  For example, consider:
 
 (8)  John strikes me as pompous.
 (9)  I regard John as pompous.
 
If you analyze these two syntactically, it's hard to see the relationship between the "me" in (8) and the "I" in (9).  But, there is a relationship.  Both are the experiencer of the action.  One problem is the number of cases.
E.  Something to keep in mind:  Competence vs. performance.  People have pushed each of these to generate and comprehend language (I've used a modified PSG to read and write scripts for Friday the 13th movies).  But, that's only demonstrating competence:  being capable of solving the problem.  It doesn't address performance:  What people actually do.  The question of what people actually do hasn't been determined.  We might agree that some form of syntactic analysis has to take place, and we might agree that these grammars can achieve that, but that isn't saying we agree on performance.
 
Top
 
IV.  Parsing strategies.  When people are reading text, how do they parse on the fly?
A.  Some constraints on parsing:  People have a very limited capacity working memory.  This means that any processes we propose have to fit in that capacity.  The problem of trying to do it with limited resources will be the driving force behind the strategies.  They're both ways to minimize working memory load.  One way to minimize load is to make the immediacy assumption:  When people encounter ambiguity they make a decision right away.  This can cause problems if the decision is wrong, but it saves capacity in the meantime.  Consider a seemingly unambiguous sentence like:
 
 (10)  John bought the flower for Susan.
 
It could be that John's giving it to Susan, but he could also be buying it for her as a favor.  The idea is that you choose one right away.  Why?  Combinatorial explosion.  If you have just four ambiguities in a sentence with two options each, you're maintaining 16 possible parses by the end.  Your capacity is 7±2 items, you do the math.  To see what happens when you have to hold all of the information in a sentence in memory during processing, try to read:
 
 (11)  The plumber the doctor the nurse met called ate the cheese.
 
The problem is you can't decide on a structure until very late in the sentence, meaning you're holding it all in memory.  If I complicate the sentence a bit but reduce memory load, it gets more comprehensible:
 
 (12)  The plumber that the doctor that the nurse met called ate the cheese.
 
Or, make it even longer but reduce memory load even more:
 
 (13)  The nurse met the doctor that called the plumber that ate the cheese.
 
Now that we've looked at the constraints supplied by working memory capacity and the immediacy assumption, let's look at the strategies.
B.  Parsing strategies.  There are two problems.  The first is getting the clauses, the second is hooking them up.
1.  Getting the clauses (NPs, VPs, PPs, etc.).
a.  Constituent strategy:  When you hit a function word, start a new constituent.  Some examples:
Det:  Start NP.
Prep:  Start PP.
Aux:  Start VP.
b.  Content-word strategy:  Once you have a constituent going, look for content words that fit in that constituent.  An example:
Det:  Look for Adj or N to follow.
 
c.  Noun-verb-noun:  Overall strategy for the sentence.  First noun is agent, verb is action, second noun is patient.  Apply this model to all sentences as a first try.  Why?  It gets a lot of them correct.  So, you might as well make your first guess something that's usually right.  We know people do this because of garden-path sentences (sentences that lead you down a path to the wrong interpretation).  Example:

 (14)  The editor authors the newspaper hired liked laughed.
 
You want authors to be a verb, but when you find out it isn't, you have to go back and recompute.
d.  Clausal:  Make a representation of each clause, then discard the surface features.  Evidence:
 
 (15)  Now that artists are working in oil prints are rare.  (863 ms)
vs.
 (16)  Now that artists are working longer hours oil prints are rare.  (794 ms)
 
In 15, "oil" is not in the last clause, in 16 it is.  The access time for "oil" after reading the sentence is in parentheses.  When it's not in the current clause, it takes longer, as if you've discarded it.
2.    Once you get the clauses, how do you hook them up?  More strategies:
a.  Late closure:  The basic strategy is to attach new information under the current node.  Consider the parse for:
 
 (17)  Tom said Bill ate the cake yesterday.
 
(We'll need some new rules in our PSG to pull it off, I'm skipping those to produce final phrase markers.)
 
Late Closure and Not Late Closure trees go here.
 
According to late closure, "yesterday" modifies when Bill ate the cake, not when Tom said it (is that how you interpreted the sentence?)  It could modify when Tom said it, but that would require going up a level in the tree to the main VP and modifying the decision you made about it (that it hasn't got an adverb).  That's a huge memory burden (once you've parsed the first part of the sentence, you probably threw out that part of the tree to make room).  So, late closure eases memory load by attaching where you're working without backtracking.
Evidence:  Have people read things like:

 (18)  Since J. always jogs a mile seems like a very short distance to him.

With eye-tracking equipment, you can see people slow down on "seems" to rearrange the parse because they initially attach "a mile" to jogs when they shouldn't.
b.  Minimal attachment:   Make a phrase marker with the fewest nodes.  It reduces load by minimizing the size of the trees produced.  Consider:
 
 (19)  Ernie kissed Marcie and Joan...
 
Minimal Attachment and Not Minimal Attachment trees go here.
 
The minimal attachment tree has 11 nodes vs. 13 for the other.  It's also less complex.  The idea is that if you can keep the whole tree in working memory (you don't have to throw out parts to make room), then you can parse more efficiently.
Evidence:  Consider:
 
 (20)  The city council argued the mayor's position forcefully.
 (21)  The city council argued the mayor's position was incorrect.
 
In (21) minimal attachment encourages you to make the wrong tree and you have to recompute.
3.  Note that these are strategies.  Both help you meet the goal of keeping your burden as small as possible.  It doesn't mean this is all you do or that these necessarily compete during processing.
 
Top
 


Psychology of Language Notes 7
Will Langston

 Back to Langston's Psychology of Language Page