Presentation

Diagnostic Language Assessment for L2 Pronunciation: A Worked Example

---

Abstract

Although pronunciation teaching has come back into fashion, pronunciation assessment remains mostly focused on evaluating global qualities(Isaacs & Trofimovich, 2017; Kang & Ginther, 2017). This assessment approach offers little to learners and teachers who desire finer-grained information about their pronunciation abilities to help guide their learning activities. Diagnostic language assessment, however, can be used to identify learner strengths and weaknesses to guide learning. While diagnostic assessment holds great promises for informing learning, there are relatively few examples of truly diagnostic tests, and this paucity is especiallystark for pronunciation. This article describes the design of a diagnostic pronunciation test for Korean as a second language, which includes tasks targeting both production and perception of Korean phonemes. From this test, learners receive information on how well they perceive and produce Korean sounds, which can raise awareness of difficulties and steer learners toward suitable learning activities. In addition to describing test and task specifications, key principles for pronunciation diagnosis are reviewed. Finally, an approach to interpreting and validating diagnostic test scores is discussed.INTRODUCTION While second language (L2) pronunciation instruction is back in vogue and perhaps even enjoying a renaissance (Levis, 2019), pronunciation assessment, in particular, has not evolved to serve the purposes of teaching and learning. Most contemporary work on pronunciation assessment falls under the umbrella of proficiency assessment, which focuses primarily on evaluating global qualities of pronunciation skills, often in the context of assessing pronunciation as a subcomponent of speaking skills (Isaacs & Trofimovich, 2017; Kang & Ginther; 2017). Such assessments are certainly useful, but they do little to address the questions most relevant to teaching individual learners, like “What specific pronunciation difficulties does this student actually have?” Unfortunately, types of assessments more relevant to pronunciation learning and teaching receive little attention, and in some ways, have not shown much development from Lado’s (1961) recommendations (e.g., assessing individual L2 segments, using pictures to elicit sounds), many of which, to be fair, are still useful. This lack of development is especially unfortunate given experts’ calls to diagnose learner needs and individualize instruction (e.g., Levis, 2018).One promising avenue for assessment that is more attuned to pronunciation teaching and learning is diagnostic language assessment (DLA; Alderson, 2005; Alderson et al., 2015). In DLA, the focus is on identifying the specific strengths and weaknesses of individual learners rather than placing them on a continuum of general ability, as is the case with typical proficiency assessment. The overarching goal in DLA is to promote subsequent learning: By identifying student weaknesses, teachers and learners know what to prioritize in their teaching and learning activities. Whereas proficiency assessments (and most forms of language assessment) are designed in reference to a description of a target ability (e.g., “speaking”), DLAs go beyond this and IsbellDiagnostic language assessment for L2 pronunciationPronunciation in Second Language Learning and Teaching 11128incorporate what is known about language development into the design of assessments. In other words, DLA reveals information about the targeted skill/knowledge as well as provides insights on developmental progress. DLA is well-positioned to meet the recommendations of experts in pedagogy, such as Levis (2018, p. 239), who proposes that “teachers will take note of each student’s challenges either through diagnosing the student’s speech or through noticing difficulties during class.” When diagnosing learners’ pronunciation difficulties, intelligibility is the most suitable criterion for evaluation in instructionally-relevant assessment: Intelligibility is the primary goal of communication, and intelligible pronunciation, rather than native-like, is a goal within reach of L2 learners and thereby aimed for in instruction (Levis, 2005). It is now intelligibility, rather than nativeness, that constitutes descriptions of pronunciation ability in proficiency frameworks like the Common European Framework of Reference (CEFR, Council of Europe, 2017), with L2 learners attaining more intelligible pronunciation as their global proficiency increases. While such a trend is likely to exist across learners, individual deviation from that trend is likely. For example, some learners at lower levels of overall proficiency may exhibit highly intelligible pronunciation, and some otherwise high-proficiency learners may still struggle in terms of pronunciation. From a diagnostic perspective, these individual differences are of primary concern. In this paper, I exemplify DLA through a test I developed,the Korean Pronunciation Diagnostic (KPD), that supports learners in developing moreintelligible pronunciation. In what follows, I discuss some key principles for pronunciation diagnosis, describe the design of the KPD, and lay out an approach to investigating its validity. This latter part is presented in brief due to space limitations but included due to its importance in assessment: Making a test is only the beginning of assessing learners, and valid use of test scores should always be kept in mind. Some Key Principles for Diagnosing PronunciationIt is desirable for a pronunciation diagnostic to yield information about strengths and weaknesses in word-level pronunciation phenomena (e.g., segments, lexical stress; see Levis, 2018). While discourse-level phenomena, such as suprasegmentals, do have an impact on intelligibility, empirical research on intelligibility suggests that segmental features have considerable effects, too (Kang et al., 2018; Loewen & Isbell, 2017). Thus, at least as a starting point, a focus on segmental pronunciation features in diagnosis is justifiable. In the case of Korean, an initial focus on segmental features is further supported by its phonological characteristics. Korean is a syllable-timed language without lexical stress (e.g., English), lexical tone (e.g., Chinese), or other word-level suprasegmental features used to distinguish lexical items, resulting in functional loads entirely dependent on consonant and vowel relations (Oh et al., 2015). At the same time, Korean’s phoneme inventory is average in size and its syllable structure is fairly restrictive, resulting in considerable neighborhood density in the Korean lexicon (i.e., number of words differentiated by a single segment; Holliday et al., 2017). In other words, intelligible production of Korean segments is especially crucial for a speaker’s intended words to be accurately identified by an interlocutor.In terms of pronunciation learning, the roles of perception of speech sounds and awareness of one’s own pronunciation are particularly relevant to DLA. A key finding of pronunciation development is the link between perception and production: Accuracy in production is linked to accuracy in 125% 138 / 379

How to Cite:

Isbell, D. R., (2019) “Diagnostic Language Assessment for L2 Pronunciation: A Worked Example”, Pronunciation in Second Language Learning and Teaching Proceedings 11(1).