- On the Absence of Superiority and Weak Crossover Effects in Yoruba
For more than thirty years, generative grammarians have been interested in accounting for the acceptability of (1) in contrast to (2). Such examples are usually cited to show the effect of the Superiority Condition—which in contemporary terms requires an attractor to attract the closest phrase to check its feature—in syntax (Kuno and Robinson 1972, Chomsky 1973, 1995).
(1) Who do you think t bought what?
(2) *What do you think who bought t?
Many researchers, including Chierchia (1991), Chomsky (1995), Hornstein (1995, 2001), Huang (1995), Dayal (1996), Wiltschko (1998), Bošković (1999), and Barker and Shan (2004), have made diverse proposals about how to account for the unacceptability of examples such as (2). Most of these authors assume that there is a superiority effect in all languages.1 However, the contrast is not found in Yoruba and several other Niger-Congo languages, as shown in (3) and (4) (Manfredi and Oyelaran 2000; see also Adesola 2005).
(3) Ta ni o rò pé ó t ra kíni?
who be you think that 3SG buy what
'Who do you think bought what?'
(4) Kí ni o rò pé taní rà t?
what be you think that who buy
'What do you think that who bought?'(* in English)
The acceptability of (4) in Yoruba raises a fundamental question about whether Superiority is really a universal condition of syntax. [End Page 309]
A related issue that has also received much attention for decades in generative grammar is how to account for the unacceptability of examples such as (5) and (6).
(5) *Whoj does hisj mother like tj?
(6) *Whatj did you give itsj owner tj?
The unacceptability of such examples has been attributed to the Weak Crossover (WCO) Condition (Postal 1971, Chomsky 1977, Koopman and Sportiche 1982, Safir 1984, 2004).
(7) Weak Crossover Condition2
A variable cannot be the antecedent of a pronoun to its left.(Chomsky 1977:201)
This condition also seems to be relaxed in Yoruba and several other Niger-Congo languages. For example, (8) is acceptable in Yoruba on the bound reading.
(8) Tak ni ìyá tk?
who be mother his like
'Whok did hisk mother like?'
It is therefore legitimate to investigate why the WCO effect seems to be systematically absent from Yoruba.
The goal of this squib is to provide a unified theoretical account for the absence of superiority and WCO effects from Yoruba. The Yoruba facts support Hornstein's (2001) claim that superiority is an instance of the WCO effects. I also show that the structure of wh-questions in Yoruba is such that it neutralizes WCO effects.
The rest of this squib is divided into three main sections. In section 1, I make a case for analyzing superiority as WCO. In section 2, I investigate the structure of wh-questions in Yoruba. In section 3, I discuss how the structure of wh-questions in Yoruba neutralizes WCO effects.
1 Superiority as a Weak Crossover Effect
One of the central claims of the Minimalist Program is that the derivations that involve the least effort are the best. Empirical support for the claim is seen in the fact that many languages show a contrast between the (a) and (b) examples in (9)-(11). The grammatical (a) examples involve the shortest move, in obedience to the Minimal Link Condition (Chomsky 1995).
a. Who bought what?
b. *What did who buy t?
a. Whoi did you give ti whatk?
b. *Whatk did you give whoi tk? [End Page 310]
a. Whoi did you persuade ti to buy whatk?
b. *Whatk did you persuade whoi to buy tk?
In the (a) examples, the closest wh-phrase is attracted to check the uninterpretable wh-feature of the complementizer. The (b) examples are excluded because the closest eligible phrase is not the one that is attracted to check the uninterpretable feature of the complementizer. However, sentences that correspond to these examples are acceptable in Yoruba. For example, the sentences in (12...