Practical Grammar 4

Verbs and VPs

Next, we want to associate verbs and VPs with reasonable f-structures. As with NPs, we do this in two steps:

a. We add appropriate features to the lexical entries of verbs, and
b. we add the correct annotation to the VP-rules, so that each node in the VP tree gets the desired f-structure.

Features of verbs

Let us look at the following sentences:

(1) She is happy.
(2) They are happy.
(3) He was happy.

Exercise 4.1

State which feature the constrast between (1) and (3) motivates. What are the feature's possible values?

Check your answer

The feature is TENSE and its possible values are pres, past, future.

Exercise 4.2

State which feature the constrast between (1) and (2) motivates. What are the feature's possible values?

Check your answer

The feature NUMBER (NUM), whose possible values are sg and pl.

Exercise 4.3

1. Go to https://xlfg.labri.fr/ and log in.
   

Then you locate our workshop:

1. Choose Workshops|All Workshops.
2. Look for the workshop "Syntax 1" for the current semester.
3. In the column "Action", click the button "Consult the workshop".

Now you find a project of the workshop:

1. On the right side of the screen is the list of projects of our workshop.
2. Look for "Grammar 3" and click on "Create your own version".
3. Now click on "Edit your version".

• Add appropriate NUM features to all the nouns in the lexicon.
• Add appropriate TENSE and NUM features to all the verbs in the lexicon.

Annotating the rule VP -> V NP

Let us now turn to the rule that combines a V and an NP into a VP:

VP -> V NP;

In the previous unit, we formulated the following annotated c-structure rule for combining a D and an N into an NP:

1. NP -> D N
2. {
3.  ↑=↓1;
4.  ↑=↓2;
5. }

Reminder:

1. ↑=↓1; in the rule above means that the NP and its first daugher share an f-structure.
2. ↑=↓2; in the rule above means that the NP and its second daugher share an f-structure.
3. ↑=↓1; and ↑=↓2; means that there is a single f-structure that the NP and its two daughers share.
4. As a result if the feature values of D and N are compatible, then the f-structure of the NP will be the unification of the f-structures of D and N.
   

Exercise 4.4

• In Grammar 3, give the VP rule the same annotation as the NP rule above.
• Annotate the two NP rules as well.
• Add 'broke the bottle' to the test sentence list and mark it as grammatical.
• Parse.
• Inspect the f-structure of the VP: is this the correct result?
  

• Now add 'broke the bottles' to the test sentence list and mark it as grammatical.
• Parse.
• Inspect the result. Is it correct?

If not, what seems to be the problem? What is the difference between the two cases?

Check your answer

a. 'broke the bottle': the verb and the NP are both singular.
b. 'broke the bottles': the verb is singular, but the NP is plural.
c. The annotations ↑=↓1; and =↓2; require that there is a single f-structure that the VP and its two daughers share.
d. But that is impossible, since the NUM feature cannot simultaneously have the values sing and pl.
e. Therefore, the string 'broke the bottles' has no well-formed f-structure and the grammar does not accept the string as grammatical. This, of course, is a wrong prediction!

The source of the problem

Recall why we annotated the rule NP -> D N with ↑=↓1; and ↑=↓2; . We wanted to capture the following facts:

(1) this bottle
(2) *those bottle
(3) those bottles
(4) *those bottle

(1)-(4) show that D and N agree in number. By giving Ds and Ns number features in their lexical entries and requiring the f-structure of the NP to be the unification of the f-structures of the D and N, we predict that the NP will have a well-formed f-structure only if the values for the feature num of D and N are compatible.

Now it should be clear why it was a mistake to give the VP rule the same annotation as the NP rule: this predicts that when V and NP have different values for NUM, the VP cannot have a well-formed f-structure. This is the wrong result because while V must agree with its subject, it does not have to agree with its object, as shown by the grammaticality of

(5) broke the bottles

where V is singular, but its object is plural! For this reason, the annotation of the VP rule must differ from the annotation of the NP rule in that the VP rule does not unify the f-structures of the two daughters of the VP.

Grammatical functions

The considerations above lead us to the following task:

Both V and NP should contribute to the f-structure of the VP, but the information in the f-structure of the VP that comes from V should should not be unified with the information that comes from the NP.

Here is the solution. We change the annotation of the VP rule as follows:

1. VP -> V NP
2. {
3.  ↑=↓1;
4.  (↑ OBJ) =↓2;
5. }

• Do this in your grammar.
• Parse and inspect the f-structures of

(6) broke the bottle
(7) broke the bottles

Exercise 4.5

• Add the S rule back to the grammar.
• Annotate it as follows:

1. S -> NP VP
2. {
3.  (↑ SUBJ) =↓1;
4.  ↑=↓2;
5. }

• In the Declarations editor, change the start symbol to "S".
• Add the test sentence "Lilly breaks the bottles".
• Parse.
• Inspect the result and interpret it.

Valence

Exercise 4.6

• Add the following test sentences to your grammar, telling the program that they are ungrammatical:

(1) *John [disappeared Fred].
(2) *Lilly [saw].

• Add the words in (1)-(2) to the lexicon, if necessary.
• Parse each sentence.
  
• Does the grammar make the right prediction?
  
• If not, formulate in grammatical terms what the problem seems to be.

The PRED feature

Valence is the representation of the knowledge speakers have about what other kinds of constituents a word needs to combine with. You will remember from traditional grammar the distinction between intransitive and transitive verbs. These are just names for those verbs, respectively, which do not need a direct object (i.e. the verb disappear) and those which do (like see).

In Lexical-Functional Grammar, the valence of a word is expressed by the value of its PRED(icate) feature:

(3)
John disappeared.
[PRED   'DISAPPEAR<SUBJ>']

(4)
Joe saw Fred.
[PRED   'SEE<SUBJ, OBJ>']

Exercise 4.7

• Open the project Grammer 4
• Add annotated grammar rules for S, VP, and NP to parse sentences (1)-(2). Do not change the lexicon yet!
• Add (1)-(4) as test sentences with their grammaticality status.
• Parse.
• Add the correct PRED feature to the lexical entries of the verbs.
• Parse.

Homework

Read the following pages in the textbook:
section 2.2: p. 13-17 
section 2.3: p. 22-25
section 2.3.2: everything

Note: You will probably not understand everything in these pages! Do your best on the basis of what we've done in class so far.

   Main page  Week 2  Week 3  Week 4  Week 5  Week 6  Week 7  Week 8  Week 9