Miglio, Viola (UC Santa Barbara)
Mexican & Chicano Spanish intonation: Differences related to Information Structure
Intonational patterns are constrained by the architecture of our vocal tract, but such constraints can be manipulated by the speakers and their speech communities, so as to become grammaticalized. Even if it is based on phonetic and acoustic correlates (F0, duration, intensity), intonation is thus part of the phonological system of languages (Gussenhoven 2002), and as such it is subject to variation and change (Guy 1986). Information structure functions in a similar way: all languages have a way of encoding 'new' information - information the speaker thinks the hearer does not know, and 'given' (or 'old') information - information the speaker thinks the hearer already knows (Clark & Haviland 1977, Prince 1981). The encoding of such differences, however, may well be language‐specific. Since there are few cross-linguistic studies considering the importance of information status in intonation (see however Prieto & Roseano 2010), we set out to test two varieties of Spanish: monolingual Mexican Spanish and Chicano Spanish (as spoken by English-Spanish bilingual speakers from California).
Since highly-energetic beginnings are associated with new topics and low-‐ energy beginnings with continuations (Gussenhoven 2002), we expected a correlation between 'new' information & high pitch/intensity. The opposite would occur at the end of a sentence: high endings point to continuation, low endings to finality & end-of-turn (Penfield 1989). Our hypothesis was that there could be measurable differences in how given-new information was encoded by bilinguals, in comparison with monolingual Mexican speakers.
The subjects were recorded in a semi-directed interview setting, which we considered would yield more trustworthy results than reading sentences aloud. As in Röhr & Baumann (2010), our preliminary results reveal that relatively higher pitch frequencies signal new information, in comparison to given information, and that items encoding new information also exhibit proportionally longer stressed and unstressed vowels, to those encoding given information. We do, however, find cross‐ dialectal variation between monolingual Mexican speakers and Chicanos, in that the intonation contours of Mexican Spanish speakers are flatter for given information (L‐L) and rising (L‐H) for new information, whereas in the case of Chicano speakers they are H‐L for both given and new information, but new information is marked by higher H pitch, longer vowel duration and more wavy intermediary peaks.
Our findings seem to corroborate Gussenhoven's hypothesis (2002) that some aspects of intonation seem to be shared cross‐linguistically (vowel length and high pitch for new info), whereas others are language‐specific (pitch contours).
Muller Levis, Greta (Iowa State University)
Levis, John (Iowa State University)
Benner, Susan (Iowa State University)
Contrastive Stress can be learned – But can it be taught at lower levels?
Sentence stress is perhaps the most important English suprasegmental. It promotes listener comprehension (L. Hahn, 2004), calls attention to new information (Halliday, 1967), is the only suprasegmental considered essential for ELF communication (Jenkins, 2002) and seems to be quite learnable in research studies examining both production and perception (Muller Levis & Levis, 2011; Muller Levis, Levis & Benner 2014; Pennington & Ellis, 2000). These last two studies examined a very specialized use of sentence stress, contrastive stress (CS), an iconic use of the voice that calls listeners’ attention to how a word is related to other words in a semantic category (Bolinger, 1986). CS can call attention to words that are not normally emphasized, as in (1), and to normally emphasized words with more extreme pitch movement, greater syllable length, and increased intensity, as in (2). In both cases, CS signals narrow focus (Ladd, 1996) as opposed to broad focus, the default use of sentence stress.
Did SHE say that?
I want to LEAVE.
CS tends to be taught primarily to advanced students. It also is usually taught simply using rules for which words to emphasize in a text that students read aloud. But several other things also need to be taught:
the degree of pitch movement to signal CS
destressing of any syllables after the CS
lexico-grammatical structures to express contrasts (Muller Levis, 2012)
the need for student control over these elements in free speech
In this paper, we report on a study in which we taught CS to two classes of intermediate students in an Intensive English program and compared their improvement to a control group. We discuss our reconsideration of teaching materials and our use of nonverbal resources including photos of signs, pictures of places and objects, weather forecast charts, games and numbers. Topics included weather, math sentences, prepositions of place, expressions of size, colors and sequential listing devices. At post-test, students showed significantly better performance. Our success suggests that the iconic use of CS is both teachable and learnable at lower levels of proficiency.
Munro, Murray (Simon Fraser University)
Derwing, Tracey (University of Alberta)
Windows of Opportunity for L2 Fluency and Pronunciation Development
As learners proceduralize their phonological, lexical, and grammatical knowledge of the L2, their ability to produce extemporaneous speech with fewer hesitations, false starts and pauses may improve. Such learning may be a natural consequence of ongoing L2 experience, though several research studies also point to the benefits of instruction focusing on fluency. While many investigations have considered the cognitive underpinnings of oral fluency, relatively little is known about long-term L2 fluency development because few studies have probed oral learning trajectories over extended time periods. In addition, knowledge about the relationship between long-term fluency and pronunciation development is very limited. As part of a 10-year study comparing two groups of Mandarin and Slavic language ESL learners (N = 25 per group at the outset), we collected extemporaneous speech samples at several times. All learners were adult immigrants of low oral proficiency at the beginning of the study, and all were enrolled in similar ESL classes. Speech materials were submitted to perceptual evaluations of fluency, comprehensibility, and accent; in addition, temporal-acoustic measures of fluency were made. While some aspects of pronunciation learning appeared to be constrained by a window of maximal opportunity during the first year of residence in an English-speaking environment, listeners’ judgments, as well as measures of pausing, speech rate, and hesitation phenomena indicated that, for some learners, significant positive changes in fluency continued long after completion of ESL instruction. However, fluency improvement was mitigated by some of the same social and individual factors that also influenced the L2 speakers’ pronunciation development.
Murphy, John (Georgia State University)
Baker, Amanda (University of Wollangong)
Where We’ve Been and Where We’re Going: 1850-2014 and beyond
This paper’s review and analysis of 150 years in the teaching of L2 pronunciation is organized around four waves of instructional innovations. The first wave began in the 1850s and prioritized imitative-intuitive ways of teaching. The second wave (1880s-early 1900s) witnessed the formation of the International Phonetic Association and introduced analytic-linguistic instructional practices. For much of the 20th century these first two waves vied for teachers’ attention while specialists defined and illustrated the primary characteristics of L2 phonology, English language phonology in particular. By the mid-1980s a third wave emerged which introduced teachers to more communicative means of teaching pronunciation. This third wave was spearheaded by instructional methodologists and resulted in publication of three genres of communicatively-oriented, pronunciation-centered resource materials (classroom textbooks, activity recipe collections, and teacher training texts). It not until the mid-1990s, however, that clusters of empirical researchers began to explore research questions designed to support pronunciation teaching, a defining characteristic of the field’s contemporary fourth wave. While the presenters devote most of their attention to implications for pronunciation teaching characteristic of the third and fourth waves, in a final section, they predict a probable 5th wave likely to emerge in the coming decade. This anticipated fifth wave of pronunciation teaching will be characterized by: (1) a continued synthesis of the first four waves; (2) a sociocultural turn in classroom teaching and teacher development practices; along with (3) specialist explorations of science/research, values/beliefs, and apprenticeship models of both classroom teaching and teacher preparation practices.
Offerman, Heather (Purdue University)
Olson, Daniel (Purdue University)
The Effect of Visual Feedback on VOT Productions by L2 Learners of Spanish
Research suggests that pronunciation instruction should be developed and taught in the L2 classroom (Derwing & Munro, 2005; Elliott, 1997) to facilitate production of intelligible and comprehensible utterances in the L2 (Derwing & Munro, 1997). Although accentedness does not always create intelligibility issues, it can be the catalyst to negative NS perceptions of L2 learners’ speech (Derwing & Munro, 2009). One distinctive marker of accent among NSs of English learning Spanish is the duration of aspiration values for the voiceless plosives /p, t, k/ (Lord, 2005). English plosives are realized as [pʰ] [tʰ] [kʰ] the onset position (Hualde, 2011) with long-lag aspiration duration values (Lisker & Abramson, 1964), while durations in Spanish are short-lag (Lisker & Abramson, 1964).
Previous research suggests that the use of visual feedback aids L2 learners in improving various features of pronunciation (deBot, 1983; Saito, 2007), including segmental features in Spanish (Olson & Offerman, 2013). The present study demonstrates that the use of visual feedback treatments aids NSs of American English in producing more target-like realizations of /p, t, k/ in Spanish. A total of three treatments were carried out, which included showing the Experimental group spectrograms of NSs’ and NNSs’ Spanish productions with guided teacher instruction. Generalizability from a controlled task, tokens embedded in a carrier phrase, to less controlled tasks, such as tokens within novel sentences, continuous speech (story task), and spontaneous speech (picture task), were measured for the Experimental and Control groups.
Results conclude that the Experimental group improved their aspiration reduction significantly for each elicitation task from the Pre-test to the Post-test, as well as in comparison to the Pre-test and Post-test results of the Control group. Of particular interest, participants were able to generalize more target-like productions into a spontaneous speech task, or the picture task.
Pak, Chin-Sook (Ball State University)
Kuriscak, Lisa (Ball State University)
The effects of peer tutoring on speech production and perception outcomes for learners of Spanish
The need for and benefits of pronunciation instruction have been discussed in recent literature with calls for more research on the effectiveness of various approaches and methods for explicit pronunciation instruction (Couper, 2011; Dewing & Munro, 2005; Lord, 2005, 2008; Morin, 2007). This study focuses on upper-level Spanish language students enrolled in a Spanish Phonetics course and the effectiveness of a service project (tutoring sessions for lower-level language students) as a pedagogical tool for improving knowledge and production of target-like speech. The tutoring project (“Pronunciation Clinic”) provided students in this Phonetics class with opportunities to analyze speech, to explore strategies for teaching pronunciation (as many students in the class are Spanish Teaching majors), and to support a peer struggling to improve his/her Spanish conversation skills. In doing so, the semester-long project ultimately aimed to increase the awareness of students' own pronunciation abilities and improve their production. Based on survey and speech-analysis data collected from 16 students who completed pre- and post-questionnaires, pre- and post-audio recordings of their own speech, and reflection journals, this study examines the effect of such a pedagogical tool on students' production and perception as follows: 1) metalinguistic awareness of their own pronunciation (i.e., their ability to identify in their own speech both positive traits and traits that require improvement); 2) changes over time in their attitudes toward Spanish, including their confidence and their motivation to continue to study and improve in their abilities; and 3) changes over time in their pronunciation (i.e., in their production of specific vowels and consonants). In sum, this paper describes the challenges and benefits of this kind of pedagogical innovation for teaching Spanish Phonetics (both in terms of pronunciation outcomes as well as attitudinal measures), offers guidelines for those who may want to apply this approach, and suggests directions for future research.
Qian, Mandy (Iowa State University)
Automatizing the principled identification and extraction of minimal pairs
Minimal pairs, in their highlighting of phonetic contrasts, are used in many areas of language
education such as clinical phonological treatment (Elbert, Rockman, & Saltzman, 1980; Weiner,
1981), pronunciation teaching, research, and materials development (Levis & Cortes, 2008). In
investigations and practices involving minimal pairs, what pairs to select is an important decision. Practitioners and researchers in general tend to let learner needs inform their choices so that words are not selected randomly but in a principled way with attention given to word frequency and positional variation of the phoneme. This process on the one hand improves the suitability of the selected words for learners with specific proficiency levels and linguistic backgrounds. On the other hand, it can get laborious especially if approached manually — as is frequently the case.
This presentation introduces a system written in Perl that automates minimal-pair selection. With variables such as L1 background, word frequency and syllable environment controlled, the
system can identify, select, and extract minimal pairs automatically from the Carnegie Mellon
University Pronouncing Dictionary, a public domain dictionary used as a standard reference for
over 15 years. The selections follow Swan and Smith’s (2001) phonology guide. This guide was
chosen as the theoretical framework because it values learner-centeredness and recognizes that
students speaking different mother tongues struggle with distinct pronunciation errors. Using the system, minimal pairs are respectively generated for English learners with different mother
languages (e.g. Dutch, German, Spanish, Arabic, Chinese, and Korean). These minimal pairs are
directly applicable to classroom and research use. Minimal pairs can also be easily generated for
additional learner groups by the system if error lists for their native languages similar to those in
Swan and Smith (2001) are input.
The beauty of this system is twofold. In practice, it is laborsaving. Pedagogically, it facilitates
conscious minimal-pair selections that cater to diverse learner needs, which ultimately promotes learner motivation and outcomes (Nunan, 1988; Rodgers, 1969).
Elbert, M., Rockman, B., & Saltzman, D. (1980). Contrasts: The use of minimal pairs in
articulation training: clinician manual. Austin, Texas: Exceptional Resources.
Levis, J., & Cortes, V. (2008). Minimal pairs in spoken corpora: Implications for pronunciation
assessment and teaching. In C. A. Chapelle, Y.-R. Chung, & J. Xu (Eds.), Towards
adaptive CALL: Natural language processing for diagnostic language assessment, 197-
208. Ames, IA: Iowa State University.
Nunan, D. (1988). The learner-centred curriculum. Cambridge: Cambridge University Press.
Rogers, C. R. (1969). Freedom to learn: A view of what education might become. Columbus:
Charles E. Merrill.
Swan, M., & Smith, B. (2001). Learner English: A teacher's guide to interference and other
problems. Cambridge: Cambridge University Press.
Richards, Monica (Iowa State University)
A Comprehensive Analysis of the Word Stress of New General Service List Vocabulary
Michael West’s (1954) General Service List of English Words (GSL) has grounded L2 English vocabulary acquisition materials (e.g. Coxhead's 2001 Academic Word List) for decades, but an updated corpus-based list of general English vocabulary has long been needed. Browne, Culligan & Phillips' (2013) New General Service List (NGSL) fulfills this need with its list of approximately 3000 vocabulary tokens second language learners of English are most likely to encounter in a wide variety of academic and nonacademic contexts. This presentation provides a comprehensive analysis of word stress patterns identifiable in the NGSL, demonstrating that 97% of NGSL vocabulary falls into eight word stress pattern categories. That is, because research has repeatedly demonstrated that L2 English intelligibility is substantially impacted by word stress for both L1 and L2 English listeners (e.g. Field, 2005; Isaacs & Trofimovich, 2012; Zielinski, 2008) and because, as McCrocklin (2012) points out, several studies have indicated that word stress is indeed learnable by L2 English speakers, this presentation provides a comprehensive account of the patterns most important for L2 English teachers/materials developers/learners to target.
Roccamo, Ashley (Pennsylvania State University)
What to Target in Second Language German Pronunciation Instruction:
Findings from the Classroom
In recent years, research has suggested that target-like pronunciation in a second language (L2) does not seem to be a feasible goal for the majority of adult learners. As such, the main goal of L2 pronunciation instruction has shifted in the direction of increasing intelligibility and a learner’s ability to effectively communicate in their L2 (Levis, 2005). This is often done by targeting the segmental or prosodic errors that most hinder a learner’s ability to make themselves understood. A large body of previous research has worked to identify the most detrimental errors in L2 English; research has investigated what is most salient to native speakers of English when listening to accented speech (e.g., Saito & Lyster, 2011) and determined the value of functional load in segmental instruction (Brown, 1991; Munro & Derwing, 2006). Yet little has been done to identify the most valuable segments or prosodic features to train in other world languages. This paper presents a preliminary foray into this area, combining standard native speaker ratings with acoustic analyses to provide an introductory investigation into which aspects of German pronunciation are most important for comprehensible and less-accented speech.
This paper presents the results of an analysis of the speech of forty-four L2 learners of German who received pronunciation training as a complement to their elementary and intermediate level classes. Three areas of German pronunciation were chosen for training over a period of six weeks, including lexical stress assignment, voiceless palatal and velar fricatives ([ç] and [x]), and the fricative and vocalized /r/. Learners completed a word-reading task consisting of 75 German words both before and after training. Speech samples were subjected to four native German speakers for ratings of comprehensibility and accentedness, and acoustic analyses of these speech samples were conducted by two linguists to determine phonetic accuracy.
Results of these analyses will begin to inform researchers and instructors as to which aspects of German pronunciation are most susceptible to improvement after training, as well as which aspects are most influential to L2 comprehensibility and accentedness as perceived by native speakers of German. Ultimately, an analysis of this type can help inform decisions about the focus of German pronunciation lessons for the highest chance of helping L2 learners of German improve their pronunciation and become more comprehensible.
Rojczyk, Arkadius (University of Silesia)
Schwartz, Geoffrey (Adam Mickiewicz University)
Balas, Anna (Adam Mickiewicz University)
The production of word boundary C#V sequences in English by Polish learners
While a great bulk research in second-language speech has concentrated on vowels and consonants as constituting a segmental composition of words, considerably fewer studies have looked into the acquisition of temporal and spectral properties that are exploited to signal word boundaries in the target language. In the case of C#V sequences of Polish learners, this task is complicated by the details of realization of target language voicing contrasts. When the final consonant is voiced, English shows a kind of liaison processes, by which find out may be indistinguishable from fine doubt. When the final consonant is voiceless, the liaison process is not so absolute; the /k/ in like old is not typically produced with aspiration as it is in lie cold. In both cases, however, glottalization of word-initial vowels, typical in Polish, blocks syllabic reorganization of words separated by a word boundary. Moreover, in find out, glottalization reinforces the context for a feature typical of Polish-accented English: final devoicing.
In the current study we investigated the realization of English C#V sequences in the production of advanced Polish learners. The purpose of the study is to find if, and to what extent, Polish advanced learners of English are able to attain native-like realizations of word- boundaries by suppressing the tendency to produce glottalization on initial vowels. The material included two-word phrases such as stand on, tried out placed in carrier sentences. The measured parameters were glottalization of the word-initial vowels, duration of the preceding consonant, duration of periodicity of the consonant. These measures were compared to the participants' default native realization of similar sequences in Polish in order to see not only if their productions in English approached those typical for native speakers but also to see if they diverged from their default realizations in L1.
Sakai, Mari (Georgetown University)
Expanding the vowel space: A corpus investigation of native speakers of Greek, Japanese, and Spanish
Greek, Japanese, and Spanish all have a five-vowel inventory. Adult native speakers (NSs) of these languages often have a difficult time learning English, because their first language (L1) vowel space must expand to accommodate twelve English monophthongs. Many studies have investigated the difficulty NSs of Spanish face when trying to perceive and produce the English vowel contrasts that do not exist in their native language (e.g., Cenoz & Garcia Lecumberri, 1999; Iverson & Evans, 2009), but very few have looked at the vowel productions of NSs of other languages with a similar vowel system (e.g., Lengeris & Hazan, 2010). By observing the second language (L2) vowel productions across multiple language groups, we can better understand how the vowel space is constrained by the quantity of L1 vowel categories.
The current investigation used an audio corpus (the Speech Accent Archive) to measure the /i/ and /I/ realizations of NSs of Greek, Japanese, and Spanish to determine if the Greek and Japanese speakers pattern in the same way as the NSs of Spanish. A Hidden Markov Model, called the Penn Forced Aligner, and PRAAT scripts were used to parse and extract vowel information, and all tokens of /i/ and /ɪ/ were normalized using the Bark Difference Metric. Preliminary analyses of individual and group means show that the Spanish speakers did not have separate categories for these vowels, as expected, but the Greek and Japanese speakers did, although they were far from English NS targets. Results are discussed in light of the Speech Learning Model (Flege, 1995) and age of onset. This study adds valuable information to theoretical questions about non-native phonemic category formation.