The role of spoken vocabulary knowledge in second language speaking proficiency

ABSTRACT

Research has suggested the important role of vocabulary knowledge in second language (L2) speaking proficiency. However, earlier studies tended to disregard the congruence in test format between assessing vocabulary knowledge and speaking skills with the former predominantly measured in written format. The current study measured vocabulary knowledge in spoken format to university students speaking English as an L2, and investigated whether spoken vocabulary knowledge predicts speaking proficiency. Forty-six university learners completed written and spoken forms of productive vocabulary test (i.e., Lex30) as well as a story narrative task. Elicited speech samples were rated in terms of four aspects of L2 speaking proficiency (fluency, vocabulary, pronunciation, grammar), and the rating scores were compared with productive vocabulary scores. Results showed a significant correlation between the spoken and written vocabulary scores with a closer examination of the data indicating a gap between the two forms. Results of the vocabulary-speaking link indicated that spoken vocabulary knowledge was associated with all but one of the L2 speech ratings, while written vocabulary knowledge was not related to any of the rating scores. The current study provided methodological and practical implications with respect to the central role of modality in vocabulary testing.

KEYWORDS:

• Vocabulary knowledge
• vocabulary testing
• productive vocabulary
• speaking proficiency
• modality

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Although alternative updated word lists could be selected, this study used JACET 8000 (2003) given the accumulated evidence suggesting that this word list works for measuring productive vocabulary knowledge elicited through the Lex30 task (Fitzpatrick and Clenton 2010; Fitzpatrick and Meara 2004) and for the purpose of exploring the vocabulary-speaking relationship (Uchihara and Saito 2019). Future research might consider other word lists such as Nation’s (2012) BNC/COCA word-family lists.

2 The current study used different cue items from different wordlists: items from Nation (1984) for W_Lex30 and items from JACET (2003) for S_Lex30. Fitzpatrick and Clenton (2010: 542) reported (a) a significant correlation (r = .692, p < .001) between the two parallel forms of the test, (b) no significant difference for the mean scores (t = 0.81, p = .425), and (c) no significant difference for the variance statistics (p = .277). Despite the relatively smaller correlation for the purpose of establishing ideal test equivalence, the significant correlation together with the results of the mean and variance statistics suggested that ‘the Lex30 tests meet at least threshold criteria for equivalence’.

3 Ideally, both subjective and objective measures were used. However, given our wider focus in measuring speaking proficiency, it was not feasible to adopt objective measures to quantify all temporal, lexical, grammatical, and phonological features of L2 speech. Future research should narrow the scope of speaking proficiency and measure not only perceived proficiency using human rating but also linguistic features in greater detail such as lexical use (e.g. sophistication, diversity, density, and accuracy; Zaytseva et al. 2021), phonological accuracy (e.g. acoustic measures of segmental and prosodic features; Saito and Plonsky 2019), and temporal properties (e.g. articulation rate, location of pauses; Suzuki et al. 2021).