Syntactic Theory and English Syntax (Autumn 2009), Geoff Pullum and Nik Gisborne

NOTES ON FUSED RELATIVES

The fused relative is dealt with in less than two pages (191-192) in SIEG, but takes up pages 1068-1079 of CGEL, with further detailed discussion later in a paper by Payne, Huddleston & Pullum (2007).

Fused relative constructions are not relative clauses; they are noun phrases containing structure that reflects that of relative clauses. The meaning is like that of a noun phrase containing a relative clause modifying a head noun of generic meaning, but the syntax is such that a single word at the beginning, usually the word what or some wh-word ending in -ever, does the work of the Determiner and the Head and the relative phrase:

ORDINARY NP	WITH RELATIVE CLAUSE		FUSED	RELATIVE
the thing which	Frankenstein created		what	Frankenstein created

Since what is perfectly OK as an interrogative pronoun, it is often the case that fused relative NPs look exactly the same:

I don't know

what

he created.

(interrogative)

I plan to destroy

what

he created.

(fused relative)

There are, however, good tests you can use to tell the difference. For now, one very simple one: else can only be suffixed to a one-word wh-phrase if it is interrogative, so we have:

I don't know

what else

he created.

(interrogative)

*I plan to destroy

what else

he created.

(fused relative)

That very clear grammaticality difference indicates the interrogativity of the first case of what he created and the fused relative status of the second.

CGEL treats fused relatives as involving fusion of two functions: (i) Head of the whole NP, and (ii) Prenucleus of the relative clause within the NP. These functions are fused in the sense that one word serves in both roles: what is both the Head of an NP and the Prenucleus of a relative clause at one and the same time, and the string of words what Frankenstein created in his laboratory is both an NP and a Clause.

It's bit difficult to give a sense of the tree structure with a rough text approximation (and it's impossible with a tree-drawing programme, because this isn't a tree!), but here's a shot at the structure for the slightly shorter fused relative NP what she wrote:

              NP
              |
             Head:
            Nominal
              | \
              |  \
              |   \
              |  Modifier:
              |  Clause[REL]
              |     / \
              |    /   \
              |   /     \
              |  /       \
              | /         \
              |/           \
       Head-Prenucleus:   Nucleus:
              NP           Clause
              |             / \
             Head:         /   \
              N           /     \
              |      Subject: Predicate:
              |         NP       VP
              |         |        |
              |       Head:     Head:
              |         N        V
              |         |        |
             what      she     wrote

Distinguishing relative from interrogative clauses

Certain expressions beginning with wh-words, specifically what, are ambiguous between fused relatives and open interrogative content clauses. Thus what Albert bought could be either: Alice didn't know what Albert bought has an open interrogative content clause complement of know; Alice didn't wash what Albert bought has a fused relative object of wash.

C. L. Baker (1970) proposes the following tests for distinguishing relative clauses from interrogative clauses.

1. A verb like know takes an interrogative clause complement, but a verb like wash does not. And most of the wh-words in English cannot begin a fused relative. So when we substitute for something like what Albert bought a clause beginning with some other wh-word, we get ungrammatical results as complements of wash:

   Alice didn't know why Albert bought it.
   *Alice didn't wash why Albert bought it.

   Alice didn't know whose dog Albert bought.
   *Alice didn't wash whose dog Albert bought.

   Alice didn't know which one Albert liked.
   *Alice didn't wash which one Albert liked.

   And we also get a clear contrast, of course, in the case of clauses introduced by whether:

   Alice didn't know whether Albert bought it.
   *Alice didn't wash whether Albert bought it.

2. An open interrogative can begin with a wh-word that binds the copular complement position of a cleft clause (the underlined position in It was this that I wanted); but a fused relative cannot:

   Alice didn't know what it was that Albert bought.
   *Alice didn't wash what it was that Albert bought.

3. A particularly simple test (already used at the beginning of this lecture) is that an open interrogative can have else after a one-word wh-phrase, but a fused relative cannot:

   Alice didn't know what else Albert bought.
   *Alice didn't wash what else Albert bought.

4. There are multiple-wh interrogatives but not multiple-wh relatives:

   Alice didn't know what was given to whom.
   *Alice didn't wash what was given to whom.

5. With some awkwardness, you can affix an integrated that-relative to an interrogative wh-phrase, but with a relative wh-phrase the results are hopelessly bad:

   Alice didn't know what that was new and different she could possibly come up with.
   *Alice didn't wash what that was new and different she could possibly come up with.

6. Replacing an interrogative content clause by a noun phrase explicitly denoting the answer to the relevant question preserves sense, but with fused relatives the sense is totally destroyed:

   Alice didn't know the answer to the question of what Albert bought.
   *Alice didn't wash the answer to the question of what Albert bought.
   

An additional test that Baker did not include is provided by a construction first noted in an obscure paper by Pullum (1973) and later much more fully and insightfully discussed in a paper by Kay and Fillmore (1997). The test hinges on the fact that the construction illustrated in What are you doing reading my mail? or What's this doing in here? is limited to interrogative constructions. So, sure enough, we find:

Alice didn't know what Albert was doing reading her mail.
*Alice didn't wash what Albert was doing reading her mail.

*Alice didn't like what Albert was doing reading her mail.

This is ungrammatical, even though Alice didn't like what Albert was doing is fine (since the last four words can be a perfectly good fused relative).

That makes a total of seven good tests for interrogativity of subordinate clauses. More can be based on the divergent uses of specific wh-words. There are slight discrepancies between which wh-words can be used as relative words and which can be used as interrogative words:

	open interrogative	integrated wh-relative	supplementary wh-relative
who	yes	yes	yes
whom	yes	yes	yes
human whose	yes*	yes	yes*
non-human whose*	no	yes	yes*
what	yes	no	no
which	yes	yes	yes
where	yes	yes	yes
when	yes	yes	yes
why	yes	yes	no
how	yes	no	no

*Note: whose must not have its head elided in relative clauses. So Whose is this one? (interrogative) is fine, but ?*I gave it back to the girl whose it was (integrated relative) is rather bad, and so is *I gave it back to Mary, whose it was.

Crucial examples:

Non-human whose in open interrogative:	*I asked whose hinges squeaked. (whose = which doors’)
Non-human whose in integrated wh-relative:	Oil any door whose hinges squeak.
Non-human whose in supplementary wh-relative:	Oil that door, whose hinges squeak.
What in open interrogative:	I didn't know what he was doing.
What in integrated wh-relative:	*I had no knowledge of the things what he was doing.
What in supplementary wh-relative:	*I had no knowledge of the money-laundering scheme, what he had not told me about.
Why in open interrogative:	I didn't know why he was yelling.
Why in integrated wh-relative:	I had no knowledge of the reason why he was yelling.
Why in supplementary wh-relative:	*I had no knowledge of the actual reason, why he had been yelling.
How in open interrogative:	I didn't know how he would respond.
How in integrated wh-relative:	*I had no knowledge of the way how he would respond.
How in supplementary wh-relative:	*I had no knowledge of that particular way, how they told me later he usually responded.

At the very least, the following tests are clear and useful:

1. If it begins with what, then it is not an integrated or supplementary wh-relative clause (it could be a fused relative or an exclamative or an interrogative).
2. If it begins with how, then it is not an integrated or supplementary wh-relative clause (it could be an exclamative or an interrogative, or possibly a fused relative).

All in all, then, distinguishing interrogative from relative clauses, is something that anyone who has taken this course should be capable of doing to a high degree of accuracy.

References

Baker, C. L. (1970) "Notes on the description of English questions: the role of an abstract question morpheme", Foundations of Language 6: 197-219; reprinted in Pieter A. M. Seuren, ed., Semantic Syntax, Oxford Readings in Philosophy, Oxford University Press, 1974.

Kay, Paul and Charles Fillmore, "Grammatical Constructions and Linguistic Generalizations: the What's X Doing Y? Construction", Language 75: 1-33.

Payne, John; Rodney Huddleston; and Geoffrey K. Pullum (2007) "Fusion of functions: The syntax of once, twice, and thrice", Journal of Linguistics 43: 565-603.

Pullum, Geoffrey K. (1973) "What's this sentence doing showing up in English?", York Papers in Linguistics 3: 113-115