International Journal of Learner Corpus Research - Volume 5, Issue 2, 2019

This paper introduces a new corpus resource for language learning research, the Trinity Lancaster Corpus (TLC), which contains 4.2 million words of interaction between L1 and L2 speakers of English. The corpus includes spoken production from over 2,000 L2 speakers from different linguistic and cultural backgrounds at different levels of proficiency engaged in two to four tasks. The paper provides a description of the TLC and places it in the context of current learner corpus development and research. The discussion of practical decisions taken in the construction of the TLC also enables a critical reflection on current methodological issues in corpus construction.

The present study investigates the use of filled pauses in the Trinity Lancaster Corpus Sample by 1,244 learners/second-language speakers of English at different proficiency levels (viz. B1-C2) from six different mother-tongue backgrounds, including speech communities in which English is spoken as a Foreign language (Italy, Mexico, Russia and Spain) and two speech communities where English is spoken as a second language (China/Hong Kong, India). Apart from the speakers’ country of origin and their proficiency levels, other learning context variables gathered from the learner profiles of the corpus sample are also taken into consideration. The study revealed that, while the number of filled pauses is indeed significantly different across proficiency levels, other learning context variables, such as the country of origin, the age of acquisition or the examiners’ experience turn out to have a much stronger effect on the learners’/speakers’ use of filled pauses. These findings are discussed with regard to their implications for learner corpus research in general, as well as the relevance of including such performance phenomena in current descriptor scales such as the Common European Framework of Reference for Languages in particular.

Light verb constructions (LVCs), that is, combinations like take a walk or make a choice, are often claimed to be problematic for non-native speakers of English. In this paper, spoken data from the Trinity Lancaster Corpus are used to explore the use of these constructions across different sections of the corpus, representing different proficiency levels (from lower intermediate to upper advanced) as well as different types of acquisitional contexts, namely English as a Foreign Language (EFL) and English as a Second Language (ESL). The results of the study reveal a tendency towards an increased frequency of LVCs, as well as more complex and abstract uses, as we move from an intermediate to an advanced level and from an EFL to an ESL context. For the EFL speakers, this development corresponds to a better approximation to native English. For the ESL speakers, on the other hand, LVCs seem to have become ‘constructional teddy bears’, used more often than by the native speakers themselves.

Based on datasets of L1 Italian and Spanish learner language culled from the Trinity Lancaster Corpus Sample, this paper investigates how verb-argument constructions (VACs) develop in the spoken English of L2 learners across proficiency levels. In addition to proficiency and L1 effects, we focus on the potential influence of native English usage on learner VAC production. Insights into learners’ productive knowledge of five target VACs and the verbs used in those VACs are gained through (1) comparisons of normalized entropy scores for verbs in VACs; (2) correlation analyses comparing for each VAC the verbs produced by groups of learners and by native English speakers; and (3) regression analyses comparing learner verb-VAC associations against indices of VAC usage, including verb-VAC frequency, VAC-verb association strength and contingency. Results indicate that, across L1 backgrounds, more proficient learners are more productive in their VAC use and closer to patterns in L1 English usage than less proficient learners.

This study investigates the use of lexical backchannels in the discourse of L2 English users sitting Trinity College London’s Graded Examinations in Spoken English (GESE). It is based on the Trinity Lancaster Corpus Sample and explores the language produced during the Discussion, Conversation and Interactive tasks of the language examinations by L2 English users from Chinese, Indian and Italian linguistic backgrounds, whose proficiency ranges from the B2 to C2 levels (i.e. high intermediate, advanced, expert) of the CEFR. The findings suggest that the L2 users with an Italian background and to a lesser extent those with a Chinese background often supported their examiners’ turns with items conveying uncertainty, while those with an Indian background with items of certainty. Furthermore, the L1 Chinese speakers used lexical backchannels the most, especially those expressing surprise or request for confirmation, while the speakers from India used them the least. Implications for the assessment of oral proficiency are discussed.

Our research examines the use of three stance adverbs of certainty (actually, really and obviously) across B1, B2 and C1 levels in the Trinity Lancaster Corpus (TLC). Particularly, we examined the occurrence of these adverbs in the subset of Spanish L1 speakers from Mexico and Spain. Really, actually and obviously were found to display a distinctive frequency of use across different proficiency levels and the different speaking tasks analysed. Dialogic tasks favoured a more frequent use of really and actually, while obviously was hardly used. Qualitative analyses of the pragmatic functions of really and actually revealed that there is an increase in the use of meanings to express hedging in really and factualness in actually across the proficiency levels. Our research confirms the finding in Gablasova et al. (2017) that the type of speaking task conditions speakers’ repertoire of linguistic devices, although we argue that this conditioning operates on different levels.