Grammaticality Judgment Tests

Bialystok (1979) had three groups of English speaking learners of French (i.e.

317 participants) participate in her study. The instruments used in the study were 24 isolated French sentences, in which 6 sentences were grammatically correct and 18 sentences contained grammatical errors on categories such as adjectives, indirect or direct object pronoun, and verb formation. The participants were assigned to one of the three conditions: (a) they listened to the sentence and judged if it was grammatical or not (b) they identified the affected parts of speech in the ungrammatical sentences (c) they matched one of the nine rules provided in the test to explain the error violated by each ungrammatical sentence. All the participants were given spontaneous and delay time conditions. The results showed that implicit knowledge was employed to judge the grammaticality of the sentences while explicit knowledge was essential in further analysis of the incorrect sentences.

Sorace (1985) conducted a study to investigate the development of metalinguistic knowledge and the relationship between knowing and using a language with two groups of students of Italian (i.e. nine beginners and eight intermediate students). The participants were asked to complete three types of test: a written Judgment Test (JT), which was meant to tap into metalinguistic knowledge, an oral Picture Description Task (PDT), which represented a non-communicative condition and an informal conversation with the interviewer, which served as a communicative

26

occasion. The JT included ungrammatical sentences with errors of six grammatical structures in Italian and the participants were required to (1) write in English if the sentence was correct or not (2) make corrections of the errors (3) formulate the statement of the rules to explain the errors. A developmental pattern of metalinguistic knowledge emerged in the study that formulation of the rules was the most difficult task for them to achieve, appearing at the last stage of the developmental pattern.

As stated in the previous section, Green and Hecht (1992) investigated the relationship between implicit knowledge and explicit knowledge. As for their measurement of implicit and explicit knowledge, they adopted a grammaticality explanation test, in which the participants were asked to correct twelve sentences and then provide explanations of the rules. The results showed that the participants’ ability to correct the errors (i.e. implicit knowledge) exceeded their ability to provide rule explanations for the errors (i.e. explicit knowledge).

In addition, there were studies focusing on particular aspects of explicit knowledge. Han & Ellis (1998) invited 48 adult learners with intermediate-plus level from a university intensive English program to their study. Their study aimed to explore ways of measuring implicit and explicit knowledge and examined the roles of implicit and explicit knowledge in general language proficiency. Scores were gained from the following instruments which focused on learners’ knowledge of verb complements: a timed oral production test, a timed grammaticality judgment test, a delayed grammaticality judgment test, the TOEFL, the SLEP (Secondary Level English Proficiency Test) and an interview aimed to tap into participants’

metalanguage. The results showed that there was a clear distinction between these measures that executed with time limit (for implicit knowledge) and these that did not (for explicit knowledge). All measures were found to correlate with scores on the SLEP but only the delayed grammaticality test was discovered to correlate with the

27

scores on the TOEFL.

Hu (2002) invited 64 Chinese learners of English to participate in his study.

What differed from the previous studies was that the aspect studied was the prototypicality of six English structures. The participants’ metalinguistic knowledge about the six structures was measured through a verbalization task and a judgment task was implemented for the participants to judge the prototypicality of the uses. As for their attention to form, it was operationalized by two production tasks (spontaneous writing tasks and error correction tasks) and by time limit. It was found that participants’ displayed greater accuracy with more prototypical ones in writing and error-correction tasks and a significant interaction existed between prototypicality and attention to form.

All the aforementioned studies investigated learners’ explicit knowledge by taking advantage of the grammaticality judgment tasks to measure participants’

explicit knowledge. According to Ellis (2004), participants were usually required to perform the three operations as elicitation of their explicit knowledge: “(a) identification of the ungrammatical sentences (b) correction of the errors and (c) provision of rules” (p.249). Green & Hecht’s (1992) study did not include (a) since all the sentences in the study were ungrammatical. To overcome the problem of potentially inaccurate and imprecise verbalization in (c), Bialystok (1979) developed a test for learners’ receptive knowledge of rules as learners could choose one reasonable rule from a list of nine rules. Several studies (e.g., Green & Hecht, 1992;

Hu, 2002) required participants to verbalize the rules and most studies examined a range of grammatical features except that Han & Ellis (1998) focused on the use of verb complements and Hu (2002) investigated the prototypicality of six English structures.

These studies “attempted to investigate explicit knowledge as conscious

28

awareness” (Ellis, 2004, p.249) but only Sorace (1985) operationalized the Judgment Test (JT) as a measure of metalinguistic knowledge. As a matter of fact, in terms of the content of the Judgment Test, it is more like a measurement of explicit knowledge than of metalinguistic knowledge. Other studies using the grammaticality judgment test claimed that learners’ capacity to judge the grammaticality of a sentence could reflect their access to explicit knowledge but could not ensure their use of metalinguistic knowledge. Ellis (2004) also maintained that grammaticality judgment tests could be a way of measuring explicit knowledge but obtaining a measure of metalanguage required verbal reports. Perhaps that is why Ellis (2005) designed an untimed GJT and a metalinguistic knowledge test as measures of explicit knowledge but separated GJT from metalinguistic knowledge test.