Wordlists

The Wordlists strand (Capel Reference Capel2009) focuses on graded vocabulary development for learners. The current lists of entries for British English and American English cover A1–B2-level words and phrases, providing definitions, native speaker examples, frequent collocates, and learner examples from the CLC. In the Wordlists, the levels A1–B2 have been assigned at sense level, since it is evident from the CLC that different senses are acquired at vastly different levels of proficiency (indicating that words may also be criterial for CEFR proficiency levels).

The methodology employed in the creation of the Wordlists includes the use of data from Cambridge University Press dictionaries (Cambridge Advanced Learner's Dictionary and Cambridge Learner's Dictionary), taking account of frequency in large international corpora of English and also considering current classroom materials. The sources used to decide on a level include:

• • Cambridge Learner Corpus

• • Spoken and written native speaker corpora

• • Cambridge ESOL Vocabulary Lists (KET & PET)

• • Breakthrough (Trim n.d.), Waystage (van Ek & Trim Reference van Ek and Trim1990b), Threshold (van Ek Reference van Ek1975a, Reference van Ekb; van Ek & Trim Reference van Ek and Trim1990a) and Vantage (van Ek & Trim Reference van Ek and Trim2001)

• • Cambridge Readers wordlists

• • Wordlists in English language teaching coursebooks and vocabulary skills books

• • Cambridge English Lexicon (Hindmarsh Reference Hindmarsh1980)

This research has already provided a number of interesting insights. For example, it has shown that the most frequent senses are not always the first taught. Furthermore, evidence from the CLC sometimes indicates earlier production than is represented in some existing vocabulary lists. Strong first language (L1) influences are detected, especially at A1/A2 levels. The acquisition of affixes is shown not to be straightforward, and L1 frequency seems to be a useful guide here. In addition, common phrases and collocations appear to merit separate treatment.

Criterial features of English grammar, morpho-syntax and usage

This strand is being directed by Professor John Hawkins at RCEAL. It focuses on sets of lexical, semantic and morpho-syntactic features that appear to distinguish the CEFR levels of English and is already providing useful data on the transfer effects of different L1s (Hawkins & Buttery Reference Hawkins and Buttery2008).

Four types of criterial features have been identified (Hawkins & Buttery Reference Hawkins and Buttery2009): (i) positive grammatical properties (i.e. correct properties of English) that are acquired at a certain second language (L2) level and generally persist at higher levels, (ii) negative grammatical properties, (iii) positive usage distribution, and (iv) negative usage distribution.

For instance, a property which is acquired at B2 may differentiate B2, C1 and C2 from A1, A2 and B1, and will be criterial for the former set of levels. Such a property is, for example, a new verb co-occurrence (Williams Reference Williams2007), appearing at B1: NP–V–NP–NP (e.g. She asked him his name), which will be criterial for all the levels from B1 upwards. With regard to the usage distribution, the positive will match native speaker uses and the negative will not. For example, the relative distribution of subject relatives (the professor who wrote the book) to object relatives (the book that the professor wrote) approximates that of L1 English speakers’ usage at C2 but not at lower levels and hence this distribution is criterial for C2 level. A negative usage distribution has been detected in the lexical frequency skewing for common verbs of English (e.g. know, see, think) whereby they are overrepresented in the learner corpora compared to the native speaker usage and the frequency skewing in the learner usage declines with increasing proficiency.

The initial hypotheses and search for criterial features were based on the Threshold series (T-series) by van Ek & Trim (Reference van Ek1975a, Reference van Ekb, Reference van Ek and Trim1990a, Reference van Ek and Trimb, Reference van Ek and Trim2001). A preliminary corpus study has indicated tendencies to confirm or challenge what occurs at each level as defined in the T-series. Initial results reveal several features for which the T-series adequately predicts the level of occurrence. For example, the use of reflexive emphatic pronouns appears to distinguish A2 from B1, as does the indefinite article a/an for frequency (e.g. once a day) as well as structures such as V + -ing (e.g. walking stick) and V + -ed/-en (e.g. broken promises). Some other features apparently differ from the predicted level assigned in the T-series. For example, complementising verb and complementising adjective structures do not appear at the same time but rather at B1 and B2 respectively. It may be that the complementising Verb + Vinf (e.g. forget/remember to lock the door) is acquired earlier than complementising Adj + Vinf (e.g. likely/expected to arrive) because the latter kind of modification is rarer across the L1s or less frequent in L2. Other factors may have a role to play, namely the specifics of the task.

This strand also comprises a search for semantic features that characterise learners’ progress in the strategies used to convey meaning in L2. For example, a sub-corpus was created by Filipović (Reference Filipović2009) based on 20 verbs that have the potential to collocate actively and to be used in numerous syntactic environments whereby meanings of different quality (e.g. from physical to figurative) and complexity are conveyed as the direct result of a learner's progress (e.g. catch, drag, fall, hit, stagger, stand, throw). Verbs with different frequencies were chosen to check how quickly different meanings can be mastered once the item enters the system and what kind of meanings are learned first. For example, stagger enters relatively late (at C2 level) but it is immediately used in both physical and figurative meanings (e.g. staggering from the heavy blow).

A number of general observations can be made at this stage about the development of meaning-conveying strategies in learners across the levels. The A levels are characterised by the prevailing of the physical sense of verbs almost exclusively, apart from few meaning shifts such as catch transport or fall in love. The B levels exhibit figurative and more complex, yet extremely transparent, meaning, e.g. a bright light hit my face or catch every little detail. At the C levels, more complex and creative semantic modifications begin to appear as well as fully figurative and metaphorical uses that successfully mimic L1 language habits (e.g. throw yourself into learning English, take it as a friendly and useful suggestion).

Further and statistically precise analysis based on larger samples will establish the exact ratios and relationships of these semantic features, as well as determine how decisively criterial they are for the respective levels at which they seem to be fully acquired (see also Hendriks Reference Hendriks2008; Hendriks & Saville Reference Hendriks and Saville2009).

Functional progression at C1 and C2

This strand is being led by the University of Bedfordshire Centre for Research in English Language Learning and Assessment (CRELLA). An early phase of research linked definitions of the language function construct with the role of language functions and their linguistic realisations in language models, in frameworks of language proficiency levels and in language teaching and testing syllabuses and materials. A review of the historical background to language functions and related concepts was conducted (Hawkey Reference Hawkey2007) and a report was produced entitled Functional progression at the C level of the CEFR (Green Reference Green2007; Hawkey & Green Reference Hawkey and Green2008). This also built on the work of van Ek & Trim (op. cit.) and described developments of enhanced functional descriptors for English at the C levels of the CEFR. The study extracted claims about the functions addressed at the C levels using data from CEFR lists of function words and meta-documents such as test specifications, proficiency scales, teaching syllabuses and curricula, proficiency frameworks, coursebook outlines and materials in current use around the world. In a second phase of research, these claims were tested against the perceptions of teachers, the content of teaching and testing materials, and against learner data in the CLC (Green Reference Green2009). Initial analysis suggests that sets of communicative functions such as arguing, criticising and evaluating; editing, refining and structuring; defining, defending and justifying; abstracting, interpreting and integrating are salient at the C levels, together with increasing sensitivity to the relationship between choices of linguistic form and contexts for use. However, the T-Level categories of imparting and seeking information, expressing and finding out attitudes, socializing, deciding on and managing the course of action, assuring and repairing communication and structuring discourse may also be capable of encompassing the range of functions identified at the C level in the materials analysed by the researchers. This categorisation does not, in itself, appear to discriminate clearly the heuristic, imaginative and pragmatic opportunities (Bachman & Palmer Reference Bachman and Palmer1996) that the C level seems to open up for learners.

These studies suggest that additional categories of function may be required or that a different approach to presenting them should be found in order to capture progression at the C levels.