Coordination and Case Assignment in English

Following a long line of research, we examine how pronominal case is assigned in Standard Present-Day English, with special reference to conjoined structures. In general, nominative is (canonically) assigned to subjects and accusative to objects, as in (1).

(1) She poisoned him against me [COCA]

However, pronouns can sometimes get a non-canonical case--- for example, it is possible to have conjoined accusative pronouns functioning as subject (cf. (3)). In that respect, we distinguish three varieties of English:

I. A PRESCRIPTIVELY CORRECT VARIETY that seems to have the same case assignment rules for conjoined and non- conjoined structures, i.e. all conjoined subject pronouns get nominative case, while conjoined object pronouns consistently get accusative case, as we illustrate in (2) for nominative subject pronouns (cf. Huddleston and Pullum et al.(2002: 462)).

(2) He and I tried to make conversation [BNC]

II. A COLLOQUIAL VARIETY that allows morphologically accusative pronouns in subject function. Our research shows that both pronouns can be accusative (as in (3)); in addition, we can get a combination of a nominative and an accusative pronoun (cf. (4)), or an accusative and a nominative pronoun (cf. (5)).

(3) How him and me painted the porch. [COCA] ]

[ PRED 'paint<SUBJ,OBJ>'
  SUBJ [ { [ PRED 'him'
             CASE Acc ]
           [ PRED 'me'
             CASE Acc ] } ]
  OBJ  [ PRED 'porch' ] ]

(4) She and us are going to be good friends [Huddleston and Pullum et al. (2002: 462)]

(5) Me and he are having a heated discussion about this [Source: http://www.digitalspy.co.uk/forums/showthread.php?t=287273&page=162b (Accessed on 13 April 2010)]

III. A HYPERCORRECT VARIETY that allows morphologically nominative forms to be used in object functions, as in (6). Other possible patterns feature a non-canonical conjunct in either first or second position (cf. (7) and (8)).

(6) he was carrying the shield and afforded all the protection to both he and I [BNC].

(7) They've awarded he and his brother certificates of merit [Huddleston and Pullum et al. (2002: 463)]

(8) The present was supposed to represent Helen and I... [Huddleston and Pullum et al. (2002: 463)]

Apart from conjoined structures, we also observe a crucial difference in the behaviour of clauses like (9) vs. (10).

(9) We/*Us all left.

[ PRED 'leave<SUBJ>'
 SUBJ [ PRED 'we'
        MOD { all } ] ]

(10)We/Us two left.

[ PRED 'leave<SUBJ>'
 SUBJ [ SPEC { us }
        PRED 'two' ] ]

We believe that this different behaviour holds the key to understanding the mechanism of case assignment in English. The non-canonical accusative is ungrammatical in (9) because all is a floating quantifier and the pronoun is the head of the phrase functioning as subject. It therefore must be assigned canonical case. In (10), however, the availability of non-canonical accusative case as an option is attributed to the dependent status of the pronoun --- it behaves like a determiner in the relevant phrase, and since it is not the head, it is free to get default/non-canonical case (cf. We/*Us/*The all left. vs. We/Us/The two left.).

An influential strand of research within Minimalism has tried to account for case assignment to conjoined pronouns by postulating a conjunction phrase (ConjP), with the conjunction as head, the first conjunct as a specifier and the second conjunct as a complement (cf. Johannessen (1998)). Johannessen (1998) claims that nominal ConjPs are sometimes regarded as "impure" NPs and are therefore assigned no case at all. In such circumstances, since the accusative is the default in what we term the colloquial variety of English, we get two accusative subjects in (3). Alternatively, the ConjP can be assigned case in the standard way, and case is then passed on only to its specifier (i.e. the first conjunct), because of specifier-head agreement (cf. (4)). No specifier-head agreement exists between the head and the complement, and the second conjunct is therefore expected to get default case, which it does in (4). However, this account already runs into problems, the most important being that sentences like (5) have a specifier which has received the default case and a complement which has been assigned canonical case; Johannessen's (1998) model predicts that such a situation is impossible. Later and more comprehensive work in the same spirit (e.g. Schütze (2001) and Quinn (2005)) also fails to account for such examples.

We propose that case assignment in English is regulated by the following principles: SUBJ pronouns MUST get nominative case if they are governed by a finite verb. Otherwise, they can get default accusative case in the colloquial variety. Similarly, an OBJ pronoun that is governed by a finite verb or a preposition will have to get canonical accusative case. If no such government relation exists, pronouns are free to get default nominative case in the hyper-correct variety. A governed pronoun will appear as a governed grammatical function of the governing PRED, not as a member of such a function or as a non-governed function.

All the pronouns are governed in (1) and therefore must get canonical case. This is taken care of by the standard constraints in (11) that a case-assigning predicate like poison will impose (cf. Butt (2006: 145 ff)).

(11) PRED 'poison<SUBJ, OBJ, OBL>, (^ SUBJ CASE) = Nom, (^ OBJ CASE) = Acc

On the other hand, if the relevant case assigner does not govern the pronoun, the pronoun is free to get non- canonical/default case. Such conditions obtain when the pronoun is a dependent and not the head, as in (10) but not (9), and in conjoined phrases, as neither of the conjuncts is the head (cf. (3)). Variation is expected in subject complements (It is I/me), because a subject complement (PREDLINK) is neither a subject nor an object. We also predict that there will be variation in the absence of a case-assigning FINITE predicate, as in so called Mad Magazine Sentences or in the following examples:

(12) Who's there? Me.

(13) We were in Greville's office, I/me sitting in his swivel chair behind the vast expanse of desk, Annette sorting... [Huddleston and Pullum et al. (2002: 460); bold face and accusative pronoun mine]

Topicalisation and left dislocation are another context where variation is expected (e.g. Me, on the other hand, am not so patient [Grano (2006: 6)]). We might therefore assume positional case assignment, with displaced constituents remaining free to get default case. Using the annotations on c-structure trees, we can assign nominative to subjects appearing in the specifier of IP, and accusative to objects appearing as sisters of V or P (cf. Quinn (2005)). Anything that does not appear in its canonical phrase- structure position will thus be free to get default case. Positional case assignment would supersede the case specifications in (11) and make them redundant. Alternatively, we can keep the specifications in (11) and say that topicalised constituents appear as the value of TOPIC at f-structure; then, we need to use the restriction operator in the functional control equation that connects the topic and the relevant grammatical function in order to exclude case. Either of these explanations can be unproblematically extended to relative and interrogative pronouns which are displaced and associated with TOPIC or FOCUS.

On our account, case-assigning predicates in English will either impose the standard requirements presented in (11) or there will be case requirements associated with certain c-structure positions. For conjoined phrases, we assume that case is always a non-distributive feature in English so the case requirements will not reach any of the conjuncts (cf. Schwartz (1985), Parker et al. (1988), Schütze (2001), Sadock (2005)). The actual morphological shape a pronoun will take in any given context will be modelled with additional constraints attached to individual lexical entries. In the colloquial variety for example, a nominative form like he will be restricted to appearing in subject functions only (cf. (14a)), whereas an accusative form like him will be left unrestricted and will be free to appear anywhere where case requirements (positional or imposed by a PRED) do not reach and a default form is possible (cf. (14b)). So, an unrestricted accusative pronoun in the colloquial variety is allowed wherever accusative is assigned or where no case specifications are imposed. The lexical-entry constraints will be reversed in the hypercorrect variety, whereas the prescriptively 'correct' variety will have restrictions on both nominative and accusative pronouns.

(14) Partial lexical entries for a nominative and an accusative pronoun in the colloquial variety (cf. Butt (2006: 145 ff)):

a.  he: (^ CASE) = NOM
        (SUBJ ε* (SPEC) ^)

b.  him: (^ CASE) = ACC

We summarise the lexical-entry constraints for the three varieties in (15). These lexical-entry constraints ((14), (15)) can only work together with the specifications in (11) (or their positional counterparts). We cannot do without the case requirements in (11), because they prevent us from over-generating, which would happen if we only relied on the lexical-entry constraints --- for instance, the interaction of the two types of constraints rules out *Him painted the porch.

(15) Lexical-entry constraints in the three registers of English

The model presented here has a lot in common with the work of Schwartz (1985), Emonds (1985/6), Parker et al. (1988), Schütze (2001), Peterson (2004), and Sadock (2005), but arguably we offer a more comprehensive and satisfactory analysis that does not suffer from some of the problems besetting previous work. At the same time, we only state when the assignment of non- canonical/default case is possible, without being able to predict, say, the frequencies of the different patterns or the preference for certain pronouns to appear in first or second position, as these may be due to extra-syntactic constraints that remain for future research.