So -Inversion Revisited

I will concern myself in this squib mainly with the derivation of the construction with initial so in the following sentences:

(1)

1. a. John can speak French, and so can Mary.

2. b. Frank adores dogs, and so does his wife.

I claim that there are arguments against the so-called subject-auxiliary (or operator) inversion (or I-to-C movement) analysis, which Quirk et al. (1972, 1985) and Huddleston and Pullum (2002) employ in the description of the construction in (1) on the assumption that initial so is an adverb. I also argue that initial so is a verb phrase proform.

1 Problems with the I-to-C Movement Analysis

First of all, consider the following examples given by Quirk et al. (1985:882) (irrelevant details are omitted or modified in (2)):

(2)

1. a. You asked him to leave, and so did I.

2. b. The corn is ripening, and so are the apples.

3. c. You've spilled coffee on the table, and so have I.

4. d. The corn isn't ripening, {and neither/nor} are the apples.

Quirk et al. (1985) claim that the sentences with initial so in (2a-c) are derived via the application of subject-operator inversion (or I-to-C movement) in the same way as the sentence with neither or nor in (2d) is. However, there are counterexamples to the claim. Consider the following examples.¹ ((3a) is taken from the British National Corpus, (3b-e) were pointed out to me by Karen Courtenay (pers. comm.), and (3f) is from Dorgeloh 1997:89.)

(3)

1. a. If the necessity for the agreement is overlooked, then so will be the necessity of the child's having been trained to react to certain things in a certain way for it to mean anything by 'It is red'. (Godfrey Vessy, Inner and Outer, 1991)

2. b. In the museum field, the equity is different, and so must be their approach.

3. c. The results of education are long term and far reaching and so must be our commitment.

4. d. "The Queen's servants are quite accustomed to the Queen and her headaches. " And, she does not add, so must be His Majesty.

5. e. She is virginal, and so must be her priestesses.

6. f. But the main elements have been retained and so, more importantly, has been the standard. (Times, 15 Dec. 92, 35)

It is clear in (3) that I-to-C movement cannot account for the position of be in the right conjuncts. If I-to-C movement alone had applied, the resulting sentences would have had be after the subject, not before.² Consider, further, these sentences:

(4)

1. a. Bill must be a genius and so must be Ann.

2. b. *Bill mustn't be a gossip and neither must be Ann.

3. c. Bill mustn't be a gossip and neither must Ann be.

While (4b) is unacceptable, (4a) is acceptable to many speakers. Note also that (4c) is acceptable, which shows that (2d) does indeed involve I-to-C movement. These observations make it clear that the construction under discussion cannot be derived only via I-to-C movement.³

2 The Alternative

I have shown that the pattern under discussion cannot be accounted for by assuming that it arises only through I-to-C movement. Here I would like to propose an alternative analysis in which VP-preposing (or topicalization), subject postposing, and obligatory proverbalization as well as I-to-C movement are responsible for the derivation of the relevant construction, in a theory-neutral (or framework-independent) way, using a description that makes use of rule-based systems.

The first process is VP-preposing, illustrated in (5). ((5) is taken from Quirk et al. 1985.)

(5) They have promised to finish the work, and finish it they will.

In (5), the VP finish it is moved to sentence-initial position, and the auxiliary element will is left behined. That is, the relevant derivation of (5) is (6).

(6) VP-preposing

. . . [CP[VP finish it]i [IP he [I' will ti]]]

VP-preposing first triggers I-to-C movement (see (7bii)). It also triggers...