Practical Grammar 7

Prepositional Phrases: explaining the complex annotation

The textbook contains a c-structure rule for VP like the following:

1. VP → V          PP
       ↑=↓   (↑ (↓ PCASE)) = ↓

which translates into the following xlfg rule:

1. VP → V PP
2. {
3. ↑=↓1;
4. (↑ (↓2 PCASE)) = ↓2;
5. }

Together with the lexical entries of the verbs donate and put and the prepositions to and on, rule (1) yields f-structures like the following:

   

Note that OBL-REC appears twice in the f-structure on the left, with two different functions:

1. OBL-REC is an attribute of the f-structure of donate.
2. OBL-REC is the value of the feature path: OBL-REC|PCASE.

The same is true for OBL-LOC in the f-structure on the right. This is the result of the annotation

(↑ (↓2 PCASE)) = ↓2;

in rule (1).

Explaining the meaning of (↑ (↓ PCASE)) = ↓

The annotation on the PP looks a lot scarier than it actually is! Let us look at its structure piece by piece. To do this, we will begin by looking at the annotations in the following simpler rule one more time:

1. VP → V DP
2. {
3. ↑=↓1;
4. (↑ OBJ) =↓2;
5. }

Remember that by definition

• ↑ is "the mother's f-structure" and
• ↓2 is "the f-structure of daughter 2"

So, in the tree licensed by the rule above ↑ is the VP's f-structure. Let us call that f-structure f_VP. And ↓2 refers to the DP's f-structure. Let us correspondingly call that f-structure f_DP. With that, the formula (↑ OBJ) =↓2 becomes

(fVP OBJ) = fDP

Given the following:

(fVP OBJ) = the value of the attribute OBJ in the f-structure of the VP

(f_VP OBJ) = f_DP means:

The value of the attribute OBJ in the f-structure of the VP (= fVP) is the f-structure of the DP (= fDP).

This translates into the following graphical representation:

[fVPOBJ   [fDP ]]

In other words, all the functional information associated with the DP daughter describes the OBJ of the VP.

With this understanding, let us now look at the rule

1. VP → V PP
2. {
3. ↑=↓1;
4. (↑ (↓2 PCASE)) = ↓2;
5. }

Replacing the up and down arrows in line 4. by contant names yields the following:

(fVP (fPP PCASE)) = fPP

We can restate this formula as follows:

(fVP X) = fPP and X = (fPP PCASE)

When we break down the formula like this, we get two expressions:

1.(f_VP X) = f_PP
2. X = (f_PP PCASE)

We know the meaning of the first formula. Let us represent it in graphical form:

1'. [_fVP X   [_fPP ]]

This means that the PP's f-structure is the value of the attribute X in the VP's f-structure. Line 2. above tells us what X is: it is the value of the attribute PCASE in the PP's f-structure. Graphically:

2'. [_fPP PCASE   X]

Now, we replace [_fPP ] in 1'. by line 2'. The result is:

1''. [_fVP X   [_fPP PCASE   X]]

And this is simple: it says that the PP's f-structure is the value of the attribute X in the VP's f-structure and that X is also the value of the attribute PCASE in the PP's f-structure.

Where does the value of PCASE come from?

The obvious question that arises now is "What determines X?" This question has a simple answer: the lexical entry of every preposition specifies its PCASE value. Here are some examples from the textbook:

to P
[PCASE OBL_REC]   (example 36b)

So, the entry of the preposition to says that it expresses the PCASE value "oblique recipient".

on P
[PCASE OBL_LOC]   (example 43c)

whereas the entry of the preposition on says that it expresses the PCASE value "oblique locative".

With this, we know what the value of X is in the formula below when the embedded f-structure is the f-structure of the preposition to:

 1''. [fVP X   [fPP PCASE   X]]

It is the PCASE value of to:

 1''. [fVP OBL_REC   [fPP PCASE  OBL_REC]]

Summary

(↑ (↓2 PCASE)) = ↓2 means that the f-structure of the second daughter in the mother's f-structure is the value of the attribute which bears the same name as the value of PCASE in the second daughter's f-structure. In other words: prepositions have a value for the feature PCASE and the f-structures of their PPs become the value of the attribute with the same name in the mother's f-structure.

Exercise 7.1

Add the lexical and grammatical tools to your version of Grammar 5 at https://xlfg.labri.fr/, to make it license the two sentences below.

(3) Oscar donated clothes to charity.
(4) Robin put food on the table.

Your grammar should yield the f-structures in (42) and (45) on page 57f of the textbook. [Note that your version of (42) should also contain information about the SUBJ of the sentence!]

Thematic Roles

So far, we have completely ignored semantics. But xlfg is capable of representing the assignment of thematic roles to the arguments of predicates. The textbook contains a list of thematic roles on page 13. Using these roles, the next exercise asks you to assign an argument structure to each predicate that governs grammatical functions. The notation is extremely simple: just put a period and a role name after each GF name, as follows:

[PRED 'pred<GF.Role>']

Here are some concrete examples:

• [PRED 'pred<SUBJ.Agent>']
• [PRED 'pred<SUBJ.Agent, OBJ.PATIENT>']
• [PRED 'pred<SUBJ.Agent, OBJ.PATIENT, OBL_LOC.LOCATION>']
• [PRED 'pred<SUBJ.THEME>']

Exercise 7.2

1. Go to https://xlfg.labri.fr/ and log in.
2. Open your version of Grammar 5.
3. Click on "Output Parameters"
4. Under "Output for Argument-Structure:" select "Draw the Argument-Structure as an acyclic graph"
5. Return to "Input"
6. Assign the thematic role THEME to the SUBJ of the verb disappeared.
7. Parse and look at the output for the sentence John disappeared.
8. Study the relationship between the f-structure and the Argument Structure in the output.
9. Now assign thematic roles to all the other governed grammatical functions in your lexicon and make sure you get the expected Argument Structure output.

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