Lexical Tone

Mada [mda], a Plateau language spoken in central Nigeria, has a large number of seemingly inexplicable tone alternations that occur between the singular and plural forms of many nouns. These alternations find straightforward and reasonable explanations if one assumes the existence of underlying representations that include floating tone prefixes and certain nouns with toneless stems. Another fascinating aspect of Mada addressed in this paper is that despite any other evidence of consonant-tone interaction in the language, there is a direct correlation between the proposed floating tone singular prefixes and the surface realizations of the singular diminutive prefix: diminutive nouns with floating (L) prefixes take only və-, and those with floating (H) prefixes take only fə- .

This dissertation provides a comprehensive description of the tonal and intona- tional phonology of Lhasa Tibetan (LT) in the Autosegmental-Metrical framework. It is based on recorded data elicited from members of the Tibetan-Canadian community in Ottawa and Toronto.
The first two chapters of the dissertation contain background information about LT, a summary of previous research on LT tones and intonation, and an overview of the theoretical framework and conceptual tools used in the rest of the dissertation.
The third chapter deals with word tonology. I establish that the prosodic structure of LT brings evidence for four main constituents at or below the word level: a) the mora encodes vowel length contrasts, b) the syllable is the tone-bearing unit (TBU), c) the prosodic word, which is maximally binary, delimits the application of most tonal processes, and d) the prosodic word group, which matches grammatical words, is the domain of downstep. This prosodic structure provides evidence against the universality of the Prosodic Hierarchy (Selkirk, 2002; Nespor and Vogel, 2007) in that it has no phonological phrase, but has two word-level constituents. I then argue that LT has three lexical tones (H, LH, and L) – L being limited to some suffixes – and propose that these lexical tones are subject to tone rules applying within the prosodic word and the prosodic word group. These tone rules are similar to those proposed by Duanmu (1992), but have been improved to accurately predict the tone patterns of long polysyllabic words. Based on phonetic evidence, I also come to the conclusion that LT no longer has stress, and that the stress pattern found in other Bodic varieties has been reinterpreted as a part of the tonal system.
The fourth chapter analyses phrasal prosody. I argue that LT forms intona- tional phrases around clauses and marks them with  nal lengthening, pitch reset, and a limited set of boundary tones (H% and L%). Although communicative functions and information structure are mostly realized by means of  final particles and morphosyntactic devices in LT, I show that boundary tones, focal tones, and deac- centing interact with word tones to form complex melodic patterns.
In the  fifth chapter, I present a phonologically-based F0 synthesis model to verify the adequacy of the proposed Autosegmental-Metrical model of LT. This F0 synthesis model consists of three main components: a) the tonal targets defined in previous chapters, b) an F0 interpolation component based on the PENTA model (Xu, 2004), and c) an evaluation component allowing a comparison of the F0 contours of real LT utterances with resynthesized F0 contours of the same utterances. The F0 synthesis model is able to generate F0 contours that approximate the F0 contours of real LT recordings, suggesting that the proposed phonological model adequately captures the overall tonal and intonational phonology of LT.

This paper provides a systematic account of the emergence of contour tone, based on a ‘syllable-tone-register’ model and a large body of new firsthand acoustic tonal data. The emergence of tone is a process of pitch upgrading from an accompanying feature of phonation types to a distinctive feature. It involves three linguistic prerequisites, which bring about three processes: (L1) various phonation types, clear voice, breathy voice, falsetto, etc., which intrinsically induce pitch differences. (L2) Mono-syllabic (with great sonorous power) morpheme structure that enhance the pitch differences. (L3) Phonologicalization, which conceptualizes the enhanced pitch differences to a phonological category, tone. Furthermore, this paper proposes two ecological preconditions behind the linguistic prerequisites: (N1) Humid climates, which tend to induce a breathy voice. (N2) Small body-size, which tends to induce falsetto and creaky voice. A flowchart is designed to show how these prerequisites lead to different types of the world’s languages and finally output contour tone.

This thesis bridges accounts of tonogenesis to representations of contour tones in the dominant framework, autosegmental phonology. Accounts of tonogenesis reference phonetic features and structures that are unable to be represented in autosegmental phonology. As these features are required for the phonologization of contour tones, it is argued that they must also receive some representation in the synchrony. This is done under the Evolutionary Phonology framework of sound change. An attempt to reconcile the disparity between perceived phonetic features and synchronic structure is made by discussing the implications of the sequencing of targets within autosegmental contours. This innovation would better reflect the tonal systems of the world’s languages as well as other asymmetries in segmental phonology.

In previous work examining heritage language phonology, heritage speakers have often patterned differently from native speakers and late-onset second language (L2) learners with respect to overall accent and segmentals. The current study extended this line of inquiry to suprasegmentals, comparing the properties of lexical tones produced by heritage, native, and L2 speakers of Mandarin living in the U.S. We hypothesized that heritage speakers would approximate native norms for Mandarin tones more closely than L2 speakers, yet diverge from these norms in one or more ways. We further hypothesized that, due to their unique linguistic experience, heritage speakers would sound the most ambiguous in terms of demographic background. Acoustic data showed that heritage speakers approximated native-like production more closely than L2 speakers with respect to the pitch contour of Tone 3, durational shortening in connected speech, and rates of Tone 3 reduction in non-phrase-final contexts, while showing the highest levels of tonal variability among all groups. Perceptual data indicated that heritage speakers' tones differed from native and L2 speakers' in terms of both intelligibility and perceived goodness. Consistent with the variability results, heritage speakers were the most difficult group to classify demographically. Taken together, these findings suggest that, with respect to tone, early heritage language experience can, but does not necessarily, result in a phonological advantage over L2 learners. Further, they add support to the view that heritage speakers are language users distinct from both native and L2 speakers.

We report data on post-focus realization of lexical tones in Standard Chinese, comparing them to their counterparts in the pre-focus and on-focus conditions reported in Chen and Gussenhoven (2008). While post-focus lexical tones are often realized with a compressed F 0 range as observed in prior studies, in some tonal contexts, their F 0 range is expanded. In addition, some post-focus tones also show a long-lasting effect of the preceding tone, but with a pattern different from the usual tonal coarticulation effect. What consistently differentiates the post-focus from the on-focus condition is the degree of distinctiveness in the lexical tonal contours, in particular when the preceding focused tone is High. These data suggest that F 0 range compression is not the only and primary characteristic of postfocus tonal realization. Rather, the observed post-focus effects may be viewed as the multifaceted manifestations of weak implementation of post-focus tonal targets, as they are associated with prosodically non-prominent constituents.

Lexical tones – pitches differentiating between word meanings in tonal languages – are particularly difficult for atonal language speakers to learn. To test the hypotheses of embodied cognition and spoken word recognition, we examined whether – and how – gesture could facilitate English speakers’ discrimination between Mandarin words differing in lexical
tone. Words were learned with gestures conveying tone pitch contours (pitch gestures), gestures conveying word meanings
(semantic gestures) or no gestures. The results demonstrated that pitch gestures enhanced English speakers’ discrimination
between the meanings of Mandarin words differing in tone, whereas semantic gestures hindered their identification of tones in learned words. These findings indicate that the visuospatial features of pitch gestures strengthen the relationship between English speakers’ representations of Mandarin lexical tones and word meanings, supporting the predictions of spoken word
recognition and embodied cognition.

Phonological field work is largely about establishing contrast in comparable environments. The notion of phonological contrast, however, can be confusing, particularly in its application to tone analysis. Does it mean phonemic contrast in the structuralist sense, or does it mean underlying contrast in the generative sense? Many linguists, in publications otherwise written from a generative perspective, support underlying tonal contrasts with minimal pairs and other data that are based on structuralist criteria. This paper critiques how tonal contrast is often supported in the literature and demonstrates that many supposed minimal pairs are invalid from a generative perspective. It further demonstrates that because many morphemes in tone languages consist solely of floating tones, the potential for these cannot be ignored when establishing comparable phonological environments.

In Mandarin Chinese, word meaning is partially determined by lexical tone . Previous studies suggest that lexical tone is processed as linguistic information and not as pure tonal information . The current study explored the online processing of lexical tones. Event-related potentials were obtained from 25 Mandarin speakers while they listened to normal and anomalous sentences containing one of three types of semantic anomalies created by manipulating the tone, the syllable, or both tone and syllable (double-anomaly) of sentence-final words. We hypothesized N400 effects elicited by all three types of anomalies and the largest by the double-anomaly. As expected, all three elicited N400 effects starting approximately 150 ms poststimulus and continuing until 1000 ms in some areas. Surprisingly, onset of the double-anomaly effect was approximately 50 ms later than the rest. Delayed detection of errors in this condition may be responsible for the apparent delay. Slight differences between syllable and tone conditions may be due to the relative timing of these acoustic cues.

Congenital amusia is a neurogenetic disorder that affects the processing of musical pitch in speakers of non-tonal languages like English and French. We assessed whether this musical disorder exists among speakers of Mandarin Chinese who use pitch to alter the meaning of words. Using the Montreal Battery of Evaluation of Amusia, we tested 117 healthy young Mandarin speakers with no self-declared musical problems and 22 individuals who reported musical difficulties and scored two standard deviations below the mean obtained by the Mandarin speakers without amusia. These 22 amusic individuals showed a similar pattern of musical impairment as did amusic speakers of non-tonal languages, by exhibiting a more pronounced deficit in melody than in rhythm processing. Furthermore, nearly half the tested amusics had impairments in the discrimination and identification of Mandarin lexical tones. Six showed marked impairments, displaying what could be called lexical tone agnosia, but had normal tone production. Our results show that speakers of tone languages such as Mandarin may experience musical pitch disorder despite early exposure to speech-relevant pitch contrasts. The observed association between the musical disorder and lexical tone difficulty indicates that the pitch disorder as defining congenital amusia is not specific to music or culture but is rather general in nature.

The aim of this study is to investigate how word accents from stress-accent languages (American English, German, Czech), pitch-accent languages (Lithuanian, Swedish, Slovene, Croatian Čakavian) and tone languages (Mandarin Chinese, Thai) are perceived by Croatian listeners with a four pitch-accent system. Perception tests were carried out on a selected group of listeners (N=31) who had more than 94% of correct answers in the pre-testing of recognising the four Croatian accents. After listening to the foreign word twice, Croatian subjects made the choice to which of the four Croatian accents (long falling, short falling, long rising, short rising) the foreign word prosody was most similar.
Results show that isolated words with long vowels from English, German and Czech are associated with the Croatian long accents, and words with short vowels mostly with the Croatian short accents. Words with long vowels from all stress-accent languages Croatian listeners match to the rising (short and long) pitch accents (Am. English 69,9%, German 72%, Czech 68,3%). The main reason for the perception of an ascending tone contour could be the tone feature of focus marking. The tone feature of neutral focus marking in stress-accent languages can be variable (flat, falling or rising) while in Croatian as a neutral it has to be realised as falling.
Swedish accents (accent 1 and accent 2) are not grouped: both accents are associated with the Croatian accent patterns most different from each other (39% with the short rising, and more than 25% with the long rising). Lithuanian circumflex in around 60% of words is perceived as rising (46,8% as long rising, 12,9% as short rising), while Lithuanian acute in more than 70% of tokens is perceived as similar to the Croatian falling accents. With respect to the duration feature, Slovene acute matches both the Croatian long accents (47,8%) and the short accent (52,2%), Slovene circumflex is matched more to the short accents (83,3%). With respect to the tone feature, Slovene acute is perceived as similar to the rising tones (85,5%), and circumflex to the falling tones (59,7%). Čakavian acute is mostly associated with the Štokavian long rising accent (86,3%). Chinese tone 1 (high tone level) is perceived as Croatian rising tones (71,5%, mostly as short rising), tone 2 (high rising) is also perceived as rising (83%, mostly as long rising), tone 4 (high falling) is perceived as falling (76%, mostly as short falling) and tone 3 (so-called low-dipping, which changes from modal to creaky phonation type) in 57% is perceived as similar and matched to the Croatian long rising, and in 21% to the short falling. Thai long vowels are perceived as Croatian long, and short vowels as Croatian short accents (>90%). Thai high tone in 90% of words and Thai rising tone in 87% of words are perceived as Croatian long rising, Thai falling tone, mid tone and low tone are in more than 80% of words perceived as Croatian falling pitch-accents.
These results of perception similarity can help us to compare results of acoustic analysis of acute-accent in different pitch-accent languages. Furthermore, these results can be useful for application in foreign language teaching. For Croatian students who learn foreign languages with lexical pitch-accents or tones it will be easier to know which accent can mach the specific Croatian accent, and which is not similar and needs special effort in learning.

Disyllabic lexical tone sandhi for the Wu (Chinese) variety of Longquan 龙泉 is described for one male speaker. The tone sandhi described is typologically more similar to that of Northern Min varieties than Wu varieties. It is argued that the retention of creaky tonation on the second syllable is a marker of stress-a phonological phenomenon hitherto not yet described for Chinese.

The connection between pitch, phonation and plosive voicing has been observed in many of the languages of the world, including those of the Himalayas. The paper examines different analyses of largely phonetically similar data from various Tibetan and Tamangic languages, comparing and evaluating the substantially different analyses offered in different descriptive traditions, arguing that principles of economy and descriptive adequacy allow us to choose between alternative analyses for many languages that have been described with different types of contrasts involving plosive manner and pitch height.

This study examined the role of acquisition order and crosslinguistic similarity in influencing transfer at the initial stage of perceptually acquiring a tonal third language (L3). Perception of tones in Yoruba and Thai was tested in adult sequential bilinguals representing three different first (L1) and second language (L2) backgrounds: L1 Mandarin-L2 English (MEBs), L1 English-L2 Mandarin (EMBs), and L1 English-L2 intonational/non-tonal (EIBs). MEBs outperformed EMBs and EIBs in discriminating L3 tonal contrasts in both languages, while EMBs showed a small advantage over EIBs on Yoruba. All groups showed better overall discrimination in Thai than Yoruba, but group differences were more robust in Yoruba. MEBs' and EMBs' poor discrimination of certain L3 contrasts was further reflected in the L3 tones being perceived as similar to the same Mandarin tone; however, EIBs, with no knowledge of Mandarin, showed many of the same similarity judgments. These findings thus suggest that L1 tonal experience has a particularly facilitative effect in L3 tone perception, but there is also a facilitative effect of L2 tonal experience. Further, crosslinguistic perceptual similarity between L1/L2 and L3 tones, as well as acoustic similarity between different L3 tones, play a significant role at this early stage of L3 tone acquisition.

Contemporary word prosody in Croatian dialects possesses three different rising tones: "long rising" and "short rising", which appears diachronically later, and "acute" which is older. Acute appears mostly in Čakavian dialects, but also in some Kajkavian and in Oldštokavian dialects, and it has been called "neoacute", "new acute", "čakavian acute", "Posavina acute" and "twisted". In this paper we analyse phonetic realizations of acute tone in recordings of Oldčakavian dialects (from Pitve and Pučišća), in Kakjavian dialects (from Đurđevac and Molve) and in Oldštokavian dialects (from Bački Monoštor, Stari Mikanovci and Gundinci). From speech recordings we extracted words with realizations of acute and they were auditorily verified by a dialectological expert. For further acoustic analysis we selected words with the least influence of sentence intonation on the lexical tone. In the vowel with acute tone we analysed frequency direction, the frequency range and the duration. If present we analysed the tone level of the postaccented syllable. The analysis shows variability in acute realization within the speakers, and within the dialect. In Čakavian and Oldštokavian dialects the postaccented syllable can be low or high, whereas in Kajkavian dialects it is mostly high. Tone contour in the stressed vowel varies: it can be realized as flat, falling, rising and falling-rising tone contour.

In this paper I claim that while the phonological category is responsible for the segmental effects, the exact form of the vowel is an historical accident, contemporarily maximizing the tonal distinctions, but not requiring synchronic derivation.

The present study examines how lexical tone and focus contribute to the formation and alignment of f0contours in speech. This was done through an investigation of f0contour formation in short Mandarin sentences. These sentences all consisted of five syllables with varying tones on the middle three syllables. The sentences were produced by eight Mandarin speakers with four different focus patterns: focus on the first, second, or last word, or with no narrow focus. The f0patterns of these sentences were examined through point-by-point f0tracing, graphical comparison of averaged f0contours, f0-contour-syllable alignment analysis, and analysis of maximum, minimum f0, and slope of f0contours. The results indicate that (a) while the lexical tone of a syllable is the most important determining factor for the local f0contour of the syllable, focus extensively modulates the global shape of the f0curve, which in turn affects the height and even the shape of local contours; (b) the tones of adjacent syllables also extensively influence both the shape and height of the f0contour of a syllable, with the preceding tone exerting more influence than the following tone; (c) despite extensive variations in shape and height, the f0contour of a tone remains closely aligned with the associated syllable; and (d) both focus and tonal interaction may generate substantial f0decline over the course of an utterance. These findings seem to be able to reduce the unpredictability in the formation and alignment of f0contours in speech.

The tones of Tamang (Sino-Tibetan family) involve both F0 and voice quality characteristics: two of the four tones (tones 3 and 4) were reported to be breathy in studies from the 1970s. For the present research (thirty years later), audio and electroglottographic data were collected from 5 speakers of the Risiangku dialect in their 30s or 40s. Voice quality is estimated by computing the glottal open quotient. The present results bear on 788 syllables (from a corpus of 6,500). They show that in the speech of three speakers (M2, M3, M5), tones 3 and 4 have a higher open quotient (which provides an indirect cue to the degree of breathiness) than tones 1 and 2, with tone 3 more clearly so than tone 4, especially for speaker M2. The difference in open quotient between the four tones for the other two speakers is negligible or inconsistent. The Tamang data are compared with similar data from Naxi, which possesses level tones, and from Vietnamese, which possesses pitch-plus-voice-quality t...

The tones of Tamang (Sino-Tibetan family) involve both F0 and voice quality characteristics: two of the four tones (tones 3 and 4) were reported to be breathy in studies from the 1970s. For the present research (thirty years later), audio and electroglottographic data were collected from 5 speakers of the Risiangku dialect in their 30s or 40s. Voice quality is estimated by computing the glottal open quotient. The present results bear on 788 syllables (from a corpus of 6,500). They show that in the speech of three speakers (M2, M3 ...

This paper is a preliminary report on an ongoing research on Màda, a Platoid language spoken predominantly in Nasarawa State and few adjoining states; Kaduna and Plateau states, Nigeria. This paper discusses the phonological feature of tone and its relevance in the phonology of Màda. It examines the structure of tone bearing unit with a view to revealing the possible sequence of tone placement, types and patterns as well as the functions of tone in Màda. The motivation for this study stems from the fact that tone is contrastive and plays a significant role in determining the pitch-related suprasegmental pattern in the language. The expectation of this paper, however, is to describe tone and tone groups in Màda, hence no appeal is made to a specific theoretical model as this attempt is only a descriptive study. Data for this study was sourced through monolingual, direct elicitation and participant observation methods. The findings show that Màda attests three level tones with the high tone unmarked and two-tone glides-high-low/^/ and low-high /˅/ respectively. It further shows that tone in Màda is morphophonemic and highly productive such that pitch variation is capable of initiating contrast in the meaning of lexical items (morphotonemics). Syntactically, this study posits that, tonal contrast has a high functional load in the grammar of Màda. For instance, tone changes a-272-simple declarative sentence to interrogative sentence type, makes a distinction between singular and plural, and changes infinitives to verbal nouns. Therefore, this paper, among other things, document and decipher Màda tone system. We recommend that minority languages be studied and documented to forestall language endangerment.

Does (the presence or complexity of) tone inversely cor- relate with dryness of climate? The authors (Everett et al.) suggest that the absence of ambient humidity in the air negatively correlates with the presence of (com- plex?) lexical tone, partly because of the effect that dry air has to increase the difficulty in achieving precise articulatory targets.

On the stress rule, tone, extrametricality and accent shift in In Neoštokavian Stanimir Rakić, Beograd. 1. According to the traditional view the rising accents in Neoštokavian1 may take any position in the word except the last one, while falling accents redundantly fall on the first syllable. In view of accents, one can distinguish five main kinds of suffixes in NŠ: cyclic, receptive, extracyclic, extrametrical and dominant (s. Rakić (1991) which all influence the word position of accents in a particular way. Morphological analysis reveals that only a limited number of native words have more than two syllables (Rakic 2008), and that in monomorphemic words accents normally fall on of the last three syllables. Capitalizing on this observation we can try to formulate the lexical rule for the accent of monomorphemic nouns. (1) а. If the last syllable is heavy, the stress falls on the penult. b. If the penult is heavy, the stress falls on it. c. If neither a) nor b) are fulfilled, the stress falls on the antepenult, but in lexically determined cases the penult is stressed. For the notion of extrametricality we have to stipulate that the final heavy syllable in NS is counted as light, and that the final light syllable as extrametrical. We also admit that in NS extrametricality is partly lexically determined, especially in foreign borrowings, so that the additional provision in 1c) is necessary.The extrametricality rule accounts for the difference between 1a) and 1b). The trochaic system based on (1) accounts for a number of shortening rules in NS (s. Rakic 2010). Most of these shortening rules are morphologically conditioned, i.e. triggered by particular kind of affixes. That the rule (1) does not have many exceptions was revealed already in seventies by Magner and Matejka (1971) " s inability to find a suitable example of lexical contrast in the place of the rising accents for their perception test. As in other metrical systems, stressed syllables are assigned tonal melodies which may be of two kinds: H and HL. Word internal syllables are usually assigned the melody (H), while initial syllable may get either (H) or (HL). Which word will get which kind of melody on the initial syllable is lexically determined. The important difference is that only (H) melody spreads to the right producing the so-called rising accents (e.g. short-rising vo da " water " and long rising gla va " head "), while H from (HL) melody can spread to the left if some proclitics is added. (HL) represents the so-called falling accents, but after H spreading to the left it changes into a rising accent (e.g. short-falling bra t " brother " , but short-rising o d brata " from the brother " , s. the phonetic interpreetation of rising accents by Lehiste and Ivić 1986). The spreading of H tone to the right is a lexical rule, and its spreading to the left may be lexical or postlexical. These rules do not apply equally in all NS dialects. In Belgrade dialect, in many disyllabic words, the spreading of (H) melody to the right is optional, especially in short syllables so that often no strict distinction between short-rising and short-falling accents is drawn. In Belgrade dialect, and generally in ekavian NS dialects, the spreading to the left is restricted to prefixes, i.e. has lexical character. On the other hand, in the ijekavian dialects of the southwest of former Yugoslavia the spreading to the left regularly applies both to prefixes and to proclitics , although, it seems, nowhere without some exceptions. This means that lexically assigned H spreads to the right or to the left, although not completely consistently throughout different NS dialects. The spreading of falling accents to the left as in the examples u gra d " in the town " , pre ko zi da requires a different explanation. In order to explain this shift, we can assume that some words get tone and stress only postlexically. We can assume that at the postlexical level of phonological word, some words get " initial H " on the first syllable, and that at postlexical level stress is subsumed to tone. Whether the phonological word includes some particle or not depends on syntax, i.e. in ta j gra d " that town " , the accent falls on the noun, while in u gra d " in the town " , the accent falls on the preposition. The shape of the phonological word is determined by syntax, and we can assume that at postlexical level , the stress falls on the syllable marked with tone. This seems to be the situation in some ijekavian NS dialects described in Inkelas and Zec (1988), which they surprisingly ascribe to Belgrade dialect.

The present research addresses three issues in the phonetics of Hanoi Vietnamese: (i) What are the airflow characteristics of the obstruent-final rhymes? (ii) Rhymes with final nasal consonants /m/, /n/, /ŋ/ show no acoustic trace of the final nasal segment when they carry tone B2 (orthographic tone nặng); does the nasal airflow on these rhymes differ significantly from what is observed on obstruent-final rhymes (which, from a phonological point of view, do not have any nasal segment)? (iii) How often are nasal-final, glottally constricted rhymes followed by a nasal release, and is this release similar to the nasal release which has been reported for rhymes with final obstruents /p/, /t/, /k/? Nasal airflow measurements of 391 syllables read by one speaker (and supporting data from two other speakers) go to show that nasal airflow 2 Nasal release, nasal finals and tonal contrasts in Vietnamese does not distinguish successfully between nasal-final rhymes carrying tone B2 and obstruent-final rhymes, and that unvoiced nasal release is more frequent after obstruents /p/, /t/, /k/ than after B2-tone rhymes ending in nasals. By contrast, simultaneously recorded oral airflow brings out a clear difference between these two sets of rhymes, confirming that voice quality plays a key role in maintaining the distinction between these rhymes, which are similar from the point of view of oral articulation.

Background: Tone languages such as Thai and Mandarin Chinese use differences in fundamental frequency (F 0 , pitch) to distinguish lexical meaning. Previous behavioral studies have shown that native speakers of a non-tone language have difficulty discriminating among tone contrasts and are sensitive to different F 0 dimensions than speakers of a tone language. The aim of the present ERP study was to investigate the effect of language background and training on the non-attentive processing of lexical tones. EEG was recorded from 12 adult native speakers of Mandarin Chinese, 12 native speakers of American English, and 11 Thai speakers while they were watching a movie and were presented with multiple tokens of low-falling, mid-level and high-rising Thai lexical tones. High-rising or low-falling tokens were presented as deviants among mid-level standard tokens, and vice versa. EEG data and data from a behavioral discrimination task were collected before and after a two-day perceptual categorization training task.