This study examines the effects of auditory priming on second-language (L2) speech production. Mandarin learners of English were presented with an English vowel as an auditory prime followed by an English target word containing either a tenseness congruent (e.g., prime: /i/ – target: “peach”) or incongruent (e.g. prime: /i/ – target: “pitch”) vowel. Pronunciation of the target vowel was measured in terms of duration and formant frequency, as well as intelligibility by native English listeners. Results show a more English-like formant frequency distribution and an increase in intelligibility of the /i/ and /ɪ/ productions in the congruent relative to incongruent condition, suggesting that auditory speech information can positively affect the pronunciation of difficult L2 speech contrasts.

This chapter provides a critical review of the research on L2 learners’ lexical stress production and perception conducted over the past three decades, which has sought to explain cross-linguistic variability in L2 learners’ ability to reach target-like generalizations in their stress placement and to encode stress lexically. The chapter begins with a discussion of generative approaches to the study of lexical stress in L2 learners, which focused on the influence of the native-language (L1) phonological grammar on L2 learners’ stress placement. These approaches were subsequently challenged by the seminal work of Susan Guion and colleagues, which examined the influence of statistical regularities on L2 learners’ (and native speakers’) stress placement in novel words. The chapter then discusses phonological approaches to L2 learners’ perception and processing of lexical stress, focusing on Peperkamp and Dupoux (2002)’s Stress Parameter Model and the predictions it made for the encoding of stress in lexical representations by listeners from different L1 backgrounds. These approaches were later refined in studies investigating the importance of phonetic cues to lexical contrasts in the L1 for determining whether L2 learners can perceive lexical stress. The chapter concludes with directions for future research on L2 lexical stress.

This chapter reviews research examining the acquisition of English /r/ and /l/ by native Japanese (NJ) speakers from the perspective of the revised Speech Learning Model. The research shows that the English liquids can be learned after the end of the so-called “critical period” for speech learning, but that the two liquids are learned in different ways. This derives from the fact that the English /r/ is perceived to be more dissimilar phonetically from the Japanese liquid, /R/, than English /l/ is. NJ speakers who have received a substantial amount of English input produced and perceived English /r/ with high levels of accuracy due to the formation of a new phonetic category for English /r/. The lower level of accuracy observed for English /l/ is attributed to the formation of a composite Japanese /R/-English /l/ category based on the Japanese /R/ and English /l/ productions to which Japanese-English bilinguals have been exposed. The SLM-r predicts that bilinguals will continue to produce and perceive English /l/ less accurately than English /r/, regardless of how much English input they have received, and that learning the English liquids will induce modifications in how NJ speakers will produce and perceive their native /R/.

The lack of appropriate methodology creates an obstacle to progress in L2 speech research. Specific suggestions are provided here regarding how to obtain samples of L1 and L2 speech production that are representative of bilinguals’ typical productions of L1 and L2 speech sounds. Recommendations are made regarding methods that might be used to assess the perception of L2 speech sounds, to determine when a new phonetic category has been formed, and to obtain more accurate estimates of percentage L1 and L2 use. Finally, a new technique is unveiled for measuring the quantity and quality of L2 input to which learners have been exposed. The new technique provides a way to determine, for the first time, what kinds of L2 input distributions promote the formation of new phonetic categories for L2 sounds.

Here we present the revised Speech learning model (SLM-r), an individual differences model which aims to account for how phonetic systems reorganize over the life span in response to the phonetic input received during naturalistic second language (L2) learning. We first review research leading to the formulation of Speech Learning Model, or SLM (Flege, 1995), before presenting a synthesis of that model and then its revision. The SLM-r proposes that the mechanisms and processes needed for native language (L1) acquisition remain accessible for use in L2 learning, without change or exception, across the life span. By hypothesis, the formation or non-formation of new phonetic categories for L2 sounds will depend on the precision of L1 categories at the time L2 learning begins, the perceived phonetic dissimilarity of an L2 sound from the closest L1 sound, and the quantity and quality of L2 input that has been received. According to the SLM-r, the phonetic categories making up the L1 and L2 phonetic subsystems interact with one another dynamically and are updated whenever the statistical properties of the input distributions defining L1, L2, and composite L1-L2 categories (diaphones) change.

In this study, segmental and prosodic properties of word-length stimuli were assessed together. Six talkers from 5 L1 backgrounds (American English, Hindi, Korean, Mandarin, and Spanish) were recorded reading English stop-initial trochaic words. The productions were played for 20 monolingual American English-speaking listeners rated the accentedness of each talker. For each token, the deviation from native English productions was determined for segmental (VOT, vowel quality) and three prosodic properties (ratios of duration, intensity, f0 across the two syllables). For each non-native language background, a linear mixed-effects regression model was created to predict accentedness ratings from the phonetic deviations, and the significance of each fixed effect was examined. In each model, the significant predictors included both segmental and prosodic properties. For Hindi and Spanish talkers, the single best predictor was segmental; however, for Korean and Mandarin talkers, the single best predictor was prosodic. Thus, even for short stimuli, both segmental and prosodic information must be considered in accounting for accentedness judgments. We conclude that listeners are sensitive to the different ways that foreign accent may be manifested across different non-native backgrounds.

This chapter considers the relation between production and perception of L2 tone in speakers of Kiên Giang Khmer who are fluent to varying degrees in Southern Vietnamese. In addition to directly comparing L2 to L1 performance in tonal production and perception, we explore how perception might be related to the internal organization of a speaker’s own production system by comparing distances between f0 curves to accuracy in a speeded AX discrimination task. Relative to native speakers, we found considerable individual variation among speakers of Kiên Giang Khmer with L2 knowledge of Vietnamese in the degree to which they approximated Vietnamese tonal targets. Production accuracy was most strongly related to age, while discrimination performance correlated best with education. In addition, we observed a weak correlation between the acoustic distance of a Khmer speaker’s production of tone T to the native Vietnamese production of T, and the ability to discriminate tone T from other tones. However, speakers who acoustically separated two tones in their own productions were also more accurate at discriminating those tones in perception, regardless of how well those productions approximated native speaker targets.

The bulk of the work on non-native speech has focused on average differences between L1 and L2 speakers. However, there is growing evidence that variability also plays an important role in distinguishing L1 from L2 speech. While some studies have demonstrated greater variability for non-native than native speech (e.g., Baese-Berk & Morrill, 2015; Wade et al., 2007), others have demonstrated that under some circumstances non-native speech maybe less variable and that variability in non-native speech may shift as a function of many factors, including task (Baese-Berk & Morrill, to appear; Baese-Berk, Morrill, & Bradlow, 2016) and L1-L2 pairing (Vaughn, Baese-Berk, & Idemaru, to appear). In the present study, we ask how variability manifests in L1 and L2 speech by speakers from a variety of language backgrounds. Specifically, we ask whether a speaker whose L1 speaking rate is highly variable is also highly variable in their L2. We also ask whether variability in speaking rate in L1 or L2 differs as a function of task (e.g., read vs. spontaneous speech) and complexity of the task (e.g., more or less complicated reading passages). The results of this study will inform our understanding of the myriad complex factors that influence non-native speech.

This study investigated the production of the four Mandarin tones by a group of school-aged Spanish learners of Chinese (n=12) and a group of native Chinese children (n=4) with a mean age of 9.5 years. The participants were recorded in a quiet room at the school premises while performing an imitation task in which they produced 32 monosyllabic words embedded in a carrier phrase. Time-normalized pitch contours were extracted at 20 consecutive points, converted to logarithmic Z-scores to normalize F0 variation across talkers and submitted to growth curve analysis to compare the surface F0 contours of the four tones. A significant difference in the F0 shapes produced by the two groups was found for all four tones, but a significant difference in F0 height was found only for Tones 2 and 3. The findings suggested that native-like production of pitch contour may be more challenging than pitch height due to their relatively more complex f0-related laryngeal muscle activities and lesser attention to the former than the latter F0 dimension among non-native tone listeners.

This chapter introduces the three degrees of price discrimination as well as a discussion of the attendant arbitrage possibilities. The focus of the chapter is on second-degree price discrimination, which is illustrated with versioning and bundling. There is also an extensive discussion of two-part tariffs with application to the pricing of a system consisting of a consumable and a durable.