- Brief response to Müller
Stefan Müller (2018) criticizes my article, ‘The lexicalist hypothesis: Both wrong and superfluous’ (Bruening 2018), on several grounds. I go through each of his criticisms here and show that none weaken the arguments I made. The point of the original paper is undiminished, and the lexicalist hypothesis should be rejected as empirically false and explanatorily inadequate.
2. Psycholinguistics and processing
Müller (§2) criticizes a phrasal analysis of a nominalization like God’s declaration of them to be wrong on the grounds of directionality. Because many phrase structure analyses build syntactic structures from the bottom up, he says, we would be required to have the VP declare them to be wrong before we get the nominalization declaration of them to be wrong. Since humans process language incrementally from left to right, according to Müller this account is therefore unworkable.1
However, the only justification that Müller gives for his claim is a reference to Fodor et al. 1974. This work is often cited for the conclusion that the derivational theory of complexity is incorrect and thus there is no evidence for a transformational approach to syntax. This is a mischaracterization of the psycholinguistics literature of the 1960s and 1970s. The literature of the time actually concluded very little, as did Fodor and colleagues (1974) themselves. For an in-depth discussion of that literature and what it did and did not show, see Phillips 1996:Ch. 5. Note also that there are many different views regarding possible connections between the grammar and the parser; in many of them, there is no contradiction at all between a model of syntax that builds structure bottom-to-top and a parser that parses structures left-to-right. See Phillips & Lewis 2013 for a detailed discussion of the different views and the issues (and evidence) involved.
Rather than discuss these issues, I would like to discuss a comparison case. Compare the nominalization at issue, God’s declaration of them to be wrong, with the corresponding gerund, his declaring them to be wrong. The syntactic literature has universally concluded that gerunds include a full VP structure, because they can have VP elements like negation, auxiliaries, accusative case, and adverbs, as in his not having declared them so vehemently to be wrong (was viewed as a tacit endorsement). In a left-to-right processing model, the parser, upon encountering declaring in his declaring them to be wrong, will have to anticipate and begin building a full VP, even though declaring is category N.2 This is simply going to be necessary, in any model, if parsing takes place left-to-right (as it clearly does). Now, if the noun declaring in his declaring them to be wrong can cause the parser to build a full VP structure, why can’t the noun declaration in his declaration of them to be wrong? If the syntactic facts justify VP structure in a nominalization like declaration, as I and others have argued (Bruening 2017b and references [End Page e67] there), then this is what the parser must do. Given that the parser must be able to do it for a gerund, there is no reason to think that it cannot do it for a nominalization. Whatever view one adopts on the relation between the grammar and the parser, the model one proposes is going to have to allow something of category N to trigger the construction of VP structure. While some models might be ruled out if they cannot do this, there are many more that are compatible with left-to-right processing, even if they have a bottom-up syntax (again, see Phillips & Lewis 2013).3
3. Phrases inside derivation
In §2.1 of my article, I argued that the lexicalist hypothesis is incorrect in its assertion that phrases cannot form the input to derivational morphology. I gave several examples of such cases, including phrases occurring inside compounds. Müller (§3) asserts that these are not a problem for the lexicalist hypothesis, because they are quotations. He repeats the quotation analysis from Wiese (1996), which I addressed at length in my article. According to...