Determining the Productivity of Resultatives: A Reply to Goldberg and Jackendoff

1. Introduction

Goldberg and Jackendoff (2004) present an analysis of English resultative constructions which incorporates corpus data presented in Boas 2003. While Goldberg and Jackendoff (henceforth G&J) offer an interesting account, their article has a number of problems that I believe need clarification in order to avoid any misconceptions that readers of Language might arrive at about my own work. The goal of this discussion note is to discuss a number of empirical as well as theoretical issues that were incorrectly cited, misrepresented, or left out by G&J. To set the stage for the remainder of this discussion note, I first correct a number of inaccurate citations by G&J, which may lead to confusion when one attempts to locate examples in Boas 2000b and 2003, and then sort out a few factual inaccuracies that occur in G&J’s article. Finally, in the remainder of the note I examine some theoretical claims made by G&J, comparing and contrasting them with the results presented in Boas 2003, which, in my view, were not adequately represented by G&J.

2. Incorrect references and citations

The reference listed by G&J (p. 565) as Boas 2000, ‘Resultatives at the crossroads between the lexicon and syntax’, is referred to at various points throughout their article. However, none of the data or results attributed to the paper listed by G&J as Boas 2000 are found in this paper, possibly leading to confusion in a reader’s search for them.¹ In fact, most of the data cited by G&J actually occur in a different work. To avoid further confusion, I from now on refer to the ‘Resultatives at the crossroads between the lexicon and syntax’ paper as Boas 2000a and to my dissertation Resultative constructions in English and German (where most of the data to which G&J refer are to be found) as Boas 2000b.

The reference listed by G&J (p. 565) as Boas 2003 appears with the incorrect title, namely Resultative constructions in English and German. The correct title of Boas 2003, which appeared as a substantially revised version of my dissertation (Boas 2000b), is A constructional approach to resultatives.² Having clarified the status of the different titles of my works, I now address some factual inaccuracies in order to correct any misconceptions one might have about Boas 2000b, 2003 as a result of reading G&J’s article.

3. Factual inaccuracies

In addressing the issue of the productivity of resultative constructions, G&J (pp. 558–63) use corpus data presented in Boas 2003. While referring to the corpus of more than 6,000 resultative sentences collected from the British National Corpus (BNC), G&J make incorrect claims about the size of the BNC, which—as we see in §4—has important ramifications for their analysis as well as their presentation of the account developed in Boas 2003. For example, on p. 559 the authors write: ‘In Boas’s (2003) search of the 10-million-word British National Corpus, . . . ’. A similar statement occurs on p. 562, where G&J point out that ‘there are many verbs that occur only once in the 10-million-word corpus’ and ‘a particular verb appears only once in a 10-million-word corpus’. The problem with these statements is that they are incorrect. In Boas 2000b:21 and Boas 2003:1 the correct size of the BNC is stated as 100 million words.

Another issue relating to the Boas 2003 data cited by G&J concerns inaccurate statements about the source of the corpus data. For example, they claim: ‘In Boas’s (2003) search of the 10-million-word British National Corpus, some adjectives such as wet, sleepy, brown, and dirty appear as RPs only with lexical resultative verbs’ (p. 559).³ While Boas 2003 includes corpus data about the distribution of wet, sleepy, and dirty, it does not include corpus data on brown. I do not know how this data could have been attributed to me since a discussion of brown does not occur in my work.^4...