Breathy voice during nasality: A cross-linguistic study

Journal of Phonetics 59 (2016) ]]]–]]]

Contents lists available at ScienceDirect

Journal of Phonetics journal homepage: www.elsevier.com/locate/phonetics

Research Article

Breathy voice during nasality: A cross-linguistic study Marc Garellek a,n, Amanda Ritchart a, Jianjing Kuang b a b

Department of Linguistics, UC San Diego, 9500 Gilman Drive 0108, La Jolla, CA92093-0108, USA Department of Linguistics, University of Pennsylvania, 619 Williams Hall, 255 S 36th Street, Philadelphia, PA19104-6305, USA

A R T I C L E

I N F O

Article history: Received 9 December 2015 Received in revised form 2 September 2016 Accepted 14 September 2016

Keywords: Nasality Voice quality Phonation Enhancement Sound change

A B S T R A C T

In some languages, there is a diachronic correspondence between nasal and breathy sounds, whose origin is often attributed to the acoustic similarities between nasal and breathy vowels. In this study, we test whether nasal consonants and vowels are also produced with breathier voice quality than their oral counterparts in three Yi (Loloish) languages: Bo, Luchun Hani, and Southern Yi. We analyzed oral vs. nasal vowels and consonants using electroglottographic and acoustic measures of phonation. Results indicate that nasal consonants are often breathier than laterals, as are vowels following nasals when compared to vowels following oral consonants. These findings support the assumption that at least some of these nasal-breathy sound changes involve a stage in which the two articulations co-occur. We claim that the production of breathy voice quality during nasals can arise through listener misperception or phonetic enhancement. These findings also contribute to the understanding of nasality as an abstract feature that involves multiple articulations. & 2016 Elsevier Ltd. All rights reserved.

1. Introduction There are several documented cases of sound change involving interactions between nasal and breathy sounds. In one direction, nasalization can be derived diachronically from breathy sounds – notably [h] and other voiceless fricatives (Blevins & Garrett, 1993; Colarusso, 1988, pp. 44–45). For instance, Hindi [sãp] ‘snake,’ which has a nasal vowel following the voiceless fricative, comes from Sanskrit [sarpa], which has an oral vowel (Ohala & Busà, 1995). Similarly in Nyole, [ŋ] is thought to derive from np via nh (Schadeberg, 1989). In the other direction, breathy sounds can be derived from nasal ones (Blevins, 2004, Section 6.1.1; Blevins & Garrett, 1993; Ladefoged et al., 1976). For example, intervocalic nn - [h] in some varieties of Basque (Igartua, 2011). These processes are sometimes referred to as ‘spontaneous/aspirate nasalization’ and ‘nasal aspiration’, respectively. Together, the bidirectional relationship between nasal and glottal sounds is often called ‘rhinoglottophilia’ (Blevins & Garrett, 1993; Matisoff, 1975; Rogers, 2011). Researchers generally attribute such changes to perceptual misparsing due to the acoustic similarity between nasality and breathiness. However, there could also be an articulatory basis for such changes. In this paper, we investigate this possibility by focusing on whether nasal sounds are produced with breathy voice quality. There is strong support from patterns of sound change that listener misperception could be at least partly responsible for interactions between nasal and breathy sounds (Blevins & Garrett, 1993; Matisoff, 1975; Ohala, 1975, 1981; Ohala & Busà, 1995). For example, Ohala and Busà (1995) hypothesize that present-day English ‘goose’ derives from ‘gans-’ because listeners misattributed the acoustic characteristics of vowel nasality as belonging to the [s] rather than the [n]. The reasons for this are (1) that a glottal spreading gesture is often part of the phonetic implementation of voiceless fricatives (Löfqvist & McGarr, 1987), which may contribute to the percept of breathiness, and (2) that nasal vowels share some common acoustic correlates with breathy vowels (relative to modal ones). Nasal vowels are differentiated from oral ones by a wide variety of acoustic measures (Beddor, 1993; Carignan, 2014; Carignan, Shosted, Fu, Liang, & Sutton, 2015; Delattre & Monnot, 1968; Delvaux, Demolin, Harmegnies, & Soquet, 2008; Pruthi & Espy-Wilson, 2007; Shosted, Carignan, & Rong, 2012; Styler, 2015). Crucially, these include a weaker first formant relative to the nasal pole (usually the first or second harmonic), which is caused by velopharyngeal coupling (Chen, 1997; Maeda, 1993; Stevens, 2000; Simpson, 2012; Zellou & Tamminga, 2014).

n

Corresponding author. E-mail address: [email protected] (M. Garellek).

0095-4470/$ - see front matter & 2016 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.wocn.2016.09.001

2

M. Garellek et al. / Journal of Phonetics 59 (2016) ]]]–]]]

Fig. 1. Sample audio spectra of vowels from a female speaker of Southern Yi. The harmonic most affected by the first formant is indicated by the ‘F1’ label. Lax vowels have a weaker F1 relative to the first harmonic than tense vowels. Within a given phonation type, nasalized vowels also have weaker F1 than non-nasalized vowels.

Breathy voice is likewise distinguished from modal voice by a weaker first formant and louder first harmonic. Unlike for nasality, the increased spectral tilt for breathy vowels (relative to non-breathy ones) is not caused by coupling of different vocal tract cavities, but instead by the larger open quotient of vocal fold vibrations – that is, a larger proportion of the vibratory cycle during which the glottis is open – as well as by the presence of a posterior glottal gap (Bickley, 1982; Garellek, 2014; Garellek & Keating, 2011; Gordon & Ladefoged, 2001; Hanson, Stevens, Kuo, Chen, & Slifka, 2001; Klatt & Klatt, 1990; Kreiman et al., 2012; Zhang, 2016). Thus, nasal and breathy vowels are acoustically similar, despite the fact that these similarities arise from different articulations. The spectral similarities are illustrated in Fig. 1 using sample audio spectra from Southern Yi, one of the languages investigated in this study. Southern Yi has nasalized vowels that are either tense – more constricted – or lax – more breathy. In these examples, the amplitude of the harmonic closest to F1 (either the second or third harmonic) is lower relative to that of the first harmonic in both nasalized vowels relative to oral ones (regardless of phonation type) and in lax vowels relative to tense ones (regardless of whether the vowel is also nasalized).

1.1. Phonetic enhancement Listener misperception by itself can account for changes that involved the addition of a novel gesture (as in the change from Sanskrit [sarpa] to Hindi [sãp] ‘snake’, Ohala & Busà, 1995): the initial vowel in Sanskrit was likely breathy from the glottal spreading during the [s], and listeners misattributed the acoustic characteristics of breathiness as resulting from a nasalized vowel; these listeners would subsequently produce the word with a lowered velum, as well as with breathiness from the preceding [s]. On the other hand, changes that involve the replacement of one articulatory gesture for another likely require an additional intermediate stage. Take, for instance, the change in some varieties of Basque, in which intervocalic nn - [h] (Igartua, 2011). In this case, it is highly unlikely that listeners misperceived [n] as [h], because the two segments differ along many acoustic dimensions: compared with intervocalic [h], intervocalic [n] is voiced and has both coronal and nasal cues leading into and out of the consonant. Therefore, under a listener-misperception account, such a change should require an intermediate stage in which the nasal is also breathy, before nasality is lost: nn (listeners misperceive nasals as being breathy nasals) - [h] (listeners misperceive breathy nasals as breathy but non-nasal). The last stage could also be due to gestural reduction, which is a common process in sound change (Garrett & Johnson, 2013).

M. Garellek et al. / Journal of Phonetics 59 (2016) ]]]–]]]

3

Ultimately then, in a change like nn - [h], we also expect an intermediate stage in which the nasal consonant is produced with breathy voice. This stage could derive not only from listener misperception, but also from phonetic enhancement. It is often claimed that distinct articulations are coproduced in the realization of an abstract linguistic feature, in part because their acoustic similarities are mutually enhancing and can be integrated perceptually (Diehl & Kluender, 1989; Kingston & Diehl, 1994; Kingston, Diehl, Kirk, & Castleman, 2008; Keyser & Stevens, 2006; Stevens & Keyser, 1989, 2010). Indeed, though the production of vowel nasality via breathy voice has been claimed to be an enhancement strategy (for the acoustic reasons described earlier; Stevens & Keyser, 1989; Keyser & Stevens, 2006), to our knowledge no one has claimed that voice quality adjustments are used to enhance nasal consonants. In fact, Stevens and Keyser (1989, pp. 92–93) state that breathiness is expected to weaken the acoustic attributes of sonorant consonants (including nasals), because vocal fold vibration is less ‘reliable’ during breathy voice than modal voice, and because the increased airflow is likely to generate frication, compromising the feature [ +sonorant]. However, we argue that, like nasal vowels, nasal consonants could also be enhanced by breathy voice. In terms of their internal perceptual cues, nasal consonants are characterized as having mostly low-frequency energy (Stevens, 2000). Since breathy voice is also characterized by its low-frequency energy, breathy nasal consonants might sound ‘more nasal’ than non-breathy nasal consonants. Moreover, breathy voice during nasal consonants could help enhance their external, coarticulatory, cues. Consonant nasality is known to spread onto adjacent ‘contextually nasalized’ vowels (Beddor, 2009; Beddor & Krakow, 1999; Beddor, McGowan, Boland, Coetzee, & Brasher, 2013; Cohn, 1993; Solé, 1995), and this spreading of nasality serves as a cue to the preceding nasal (Beddor, 1993; Beddor & Krakow, 1999; Beddor et al., 2013; Malécot, 1960). Assuming that breathy voice is linked temporally to the primary nasal feature or gesture (e.g., lowered velum), any sound with nasal airflow – including contextually nasalized vowels – will also be accompanied by breathy voice. The association of breathy voice with nasal consonants could therefore be motivated as an enhancement of nasality, assuming both nasality and breathy voice spread onto adjacent vowels. However, contextual nasalization varies within and across languages in complex ways that cannot always be attributed to coproduction of the lowered velum gesture (Scarborough & Zellou, 2013; Scarborough, Zellou, Mirzayan, & Rood, 2015; Solé, 1995). So if nasal airflow is absent during vowels adjacent to nasal consonants, we could still expect some enhancement of nasality in these environments; that is, even if nasal consonants are not themselves breathy, adjacent vowels might still be breathy in order to enhance the coarticulatory information during vowels. Thus, the co-occurrence of breathy voice on a vowel adjacent to a nasal consonant might help cue the listener to the presence of the nasal, regardless of whether the preceding nasal is breathy or whether the primary nasality gesture cooccurs with the vowel. Either scenario – if breathy voice is associated with the nasal consonant or just with contextually nasalized vowels – can help account for the intermediate stage in diachronic nasal aspiration. An enhancement account is also compatible with listener misperception, because the articulatory covariation between nasal and glottal gestures could be driven by, or lead to, listener misperception. On the one hand, listeners may sometimes misperceive nasals as being both nasal and breathy, introducing articulatory variation that results in the use of breathy nasals as a means of phonetic enhancement; on the other hand, enhancement can initially arise for reasons unrelated to listener misperception (e.g., in order to make speech more recoverable to the listener; Keyser & Stevens, 2006; Stevens & Keyser, 2010), resulting in nasals being produced with breathiness. Thus, nasality and breathiness might interact diachronically because listeners misattribute the acoustic signature of one gesture with the other (Ohala, 1975, 1981), and/or because breathiness is sometimes part of the phonetic implementation of nasality as an enhancement strategy. Crucially, both scenarios require a stage in which speakers coproduce nasality and breathy voice. In this study then, we test whether nasal sounds are in fact produced with breathy voice. 1.2. Articulatory representation of nasality If nasal sounds are indeed breathier than oral sounds, this would have ramifications for how nasals are represented articulatorily. Nasal manner of articulation is typically described as being produced with a lowered velum, and oral sounds as being produced with a raised velum (International Phonetic Association, 1999). However, it has become increasingly clear that nasality is phonetically multidimensional. Focusing on the oral–nasal contrast in vowels, many researchers have shown that nasal vowels also involve specific lingual and pharyngeal modulations in addition to velum lowering. These non-velar articulatory adjustments during a nasal sound are thought to enhance its acoustic attributes, and possibly also the oral–nasal contrast (Carignan et al., 2015; da Matta Machado, 1993; Demolin, Delvaux, Methens, & Soquet, 2003; Shosted, 2015; Shosted et al., 2012). If nasal consonants and contextually nasalized vowels are indeed breathy, then this study will provide further evidence of the articulatory multidimensionality of nasality, and that this multidimensionality can be explained at least in part by strategies of phonetic enhancement. 1.3. The current study This study investigates whether nasality is produced with breathy voice, and if so, in which specific contexts of nasality. Specifically, we hypothesize that (a) nasal consonants are breathier than their non-nasal counterparts, and (b) ‘nasalized’ oral vowels (vowels adjacent to a nasal consonant) are breathier than vowels adjacent to a non-nasal consonant. In this study, we use laterals as the non-nasal counterpart: we avoided voiced obstruents because their more sizable constriction and necessary aerodynamic constraints are expected to lead to stronger source–filter interactions (Davidson, 2016; Ní Chasaide & Gobl, 1993; Westbury, 1983). These hypotheses will be tested for three Yi (Loloish) languages (Bo, Luchun Hani, and Southern Yi), with publicly available electroglottographic (EGG) and audio recordings. The acoustic measures derived from the audio signal are expected to be influenced

M. Garellek et al. / Journal of Phonetics 59 (2016) ]]]–]]]

4

by both voice quality and nasality, whereas the EGG reflects laryngeal articulations only. The statistical analyses conducted in the study are designed as to distinguish nasality with and without breathy voice, allowing us to determine the extent to which acoustic correlates of nasality might be due to laryngeal configuration. We chose these languages for practical as well as theoretical reasons. Practically, these languages had sufficient EGG data already available, as we describe in Section 2.1; in terms of theoretical considerations, these languages also pose interesting questions for listener misperception and nasal enhancement via breathiness. The Yi languages contrast two phonation categories, tense vs. lax, where lax vowels are breathier than tense ones (Kuang & Keating, 2014; Kuang, 2011b). However, these languages are currently undergoing a sound change whereby listeners do not always use changes in voice quality as cues to the contrast (Kuang & Cui, 2016). Thus, if lax phonation is no longer a cue to the contrast, listeners might be more likely to misattribute its acoustic effects to nasality. If, on the other hand, breathy voice is used by speakers to enhance nasality, we expect that breathy nasals should be more likely to occur with contrastively lax sounds (rather than tense ones), because the ‘additional’ breathiness from the nasality would not further weaken the lax vs. tense contrasts. One could also argue that the presence of a phonation contrast in these languages might inhibit voice quality enhancement of nasality altogether. However, we consider this unlikely, because distinct linguistic contrasts often share voice quality correlates: for example, both stressed and glottalized vowels show decreased spectral tilt in Ixpantepec Nieves Mixtec (Carroll, 2015); in Gujarati, Hmong, and Mazatec, both aspirated stops and contrastive phonation can lead to breathiness on vowels (Esposito & Khan, 2012; Garellek, 2012; Garellek & Keating, 2011).

2. Method 2.1. The EGG corpus Electroglottography (EGG) indirectly measures changes in the vocal fold contact area during voicing via a low-amplitude, high frequency current that is passed between two electrodes placed on the neck. The opening and closing of the vocal folds produce changes in the impedance: more contact between the folds leads to less impedance and thus greater relative EGG contact (Baken & Orlikoff, 2000). EGG is non-invasive, does not interfere with natural speech, and can measure changes in voice quality at the voice source, which is crucial for our study: because nasal and breathy vowels share acoustic attributes, measuring acoustic correlates specific to laryngeal voice quality during nasal vowels is difficult (Simpson, 2012). EGG thus enables us to measure voice quality patterns before the source is filtered by the rest of the vocal tract. The language data in this study (which also appear in Kuang, 2011b, 2013a, 2013b; Kuang & Keating, 2014) come from public materials from the “Production and Perception of Linguistic Voice Quality” project at UCLA.1 Simultaneous EGG and audio signals were recorded in a quiet room directly to a computer via its sound card. Recordings were made in stereo using Audacity. The audio signal came from a Shure SM10A head-mounted microphone and was the first channel of the recordings; the EGG signal, recorded using a Glottal Enterprises EG-2 electroglottograph, was the second channel. EGG recordings were collected following commonly used practices (Baken & Orlikoff, 2000; Rothenberg & Mashie, 1988): electrode placement was centered superior-inferiorly on the neck so as to optimize signal strength (as indicated on the machine), the collar was worn tight enough around the neck to prevent the electrodes from moving, and the gain was adjusted to prevent clipping. The data for the three Yi languages were collected by the third author and Jiangping Kong in 2009 in Yunnan province, China. For Bo, four female and six male speakers were recorded, and the analysis in this paper includes 966 tokens. For Southern Yi, six female and six male speakers were recorded, and the analysis includes 541 tokens. For Luchun Hani, four female and five male speakers were recorded, and the analysis includes 1512 tokens. Bo, Southern Yi, and Luchun Hani are all Yi (Loloish) languages with a tense-lax phonation contrast on syllables, which are always CV in structure. Vowels may contrast in nasality, but due to the rarity of nasal vowels in the corpus we do not analyze them in this study. The Yi languages all have three lexical tones (low, mid, and high). For Southern Yi in particular, the high tone is not attested in this particular subset of the corpus. Tone and phonation are orthogonal contrasts in these languages. For each of the languages, we extracted five classes of segments: (1) oral vowels, which were preceded by an oral consonant; (2) contextually nasalized vowels, which were preceded by a nasal; (3) oral sonorants (in this corpus, always [l]), and (4) nasal sonorants [m, n, ŋ]. The oral and nasal sonorants were always word-initial. Transcriptions of sample target words are shown in Table 1. 2.2. Labeling and measures The target words were already labeled in the corpus. Two coders extracted the target words and adjusted the segmentation when appropriate. Vowels preceded by a voiceless onset were segmented from the onset to offset of voicing. Vowels preceded by a sonorant were segmented from the abrupt increase in energy (corresponding to the transition between sonorant and vowel) until the voicing offset. Sonorants were also segmented from the start of voicing until the vowel onset. The EGG waveforms of the segmented sounds were analyzed using EggWorks. The subsequent analysis will focus on two articulatory measures of breathy voice: contact quotient and peak increase in contact. These measures were chosen because they 1

Available at http://www.phonetics.ucla.edu/voiceproject/voice.html

M. Garellek et al. / Journal of Phonetics 59 (2016) ]]]–]]]

5

Table 1 Transcription of sample target words in Southern Yi illustrating the crucial phonation dimensions (tense and lax) and nasality dimensions (oral sonorant, oral vowel, nasal sonorant, nasalized vowel) in the three languages. Oral sonorant + oral vowel

Nasal sonorant + nasalized vowel

Tense

‘tiger’

‘how much’

Lax

‘tea leaf’

‘get sick’

are known to distinguish tense from lax voice in these languages (Keating, Esposito, Garellek, Khan, & Kuang, 2011; Kuang, 2011b; Kuang & Keating, 2014). Contact quotient (CQ) is a ratio of the portion of time during a glottal cycle during which the vocal folds are in contact, as measured indirectly via changes in EGG current impedance. Constricted voice qualities usually show a greater CQ value (which reflects a more closed glottis), whereas breathy voice usually shows a smaller CQ value. Previous studies have shown that CQ varies with contrastive voice quality changes in a variety of languages, including the languages in this study (DiCanio, 2009; Esposito, 2010; Khan, 2012; Kuang & Keating, 2014). However, the measure has certain shortcomings. Notably, it depends on accurate estimates of the glottal closing and opening moments from the EGG waveform, but the latter especially can be hard to estimate. Thus, in the present study, CQ was measured using the hybrid method (following Howard, 1995), which defines the point of vocal fold closure as the peak in the derivative of the EGG signal, and uses a 25% peak-to-peak amplitude threshold for detecting the point of vocal fold opening (following Orlikoff, 1991). This hybrid method for measuring contact was used because thresholds at 20% and 25% are most strongly correlated with vocal fold contact measured via direct imaging of the glottis (Herbst & Ternström, 2006), and because this particular version of EGG contact quotient was found to be the most robust in characterizing differences in voice quality in the Yi languages (Kuang, 2011b). Peak increase in contact (PIC) is the amplitude of the positive peak in the derivative EGG signal, corresponding to the highest rate of increase of vocal fold contact (Michaud, 2004). Michaud (2004) used the measure (which he called Derivative-EGG closure peak amplitude) to study focus prosody, but later studies showed that contrastive phonation categories differ in PIC, with breathy voice having the highest values (Esposito, 2012; Keating et al., 2011; Kuang, 2011b). This is thought to be due to the unconstricted nature of breathy voice; the vocal folds are slackened and spread, allowing for faster (though shorter) contacting of the vocal folds compared with more modal or constricted voice qualities. It is important to note that the EGG signal measures relative, and not absolute, contact. Lower values of CQ and higher values of PIC are correlated with increased breathiness, but are not direct measures of the phonation's articulation or perceptual attributes. Nonetheless, EGG provides the least invasive way to study phonation in read speech at the voice source before it is filtered by the vocal tract, and therefore is ideal for the purposes of the present study. The audio waveforms were also analyzed acoustically using VoiceSauce (Shue, Keating, Vicenik, & Yu, 2011), in order to extract F0, as well as A1-P0, an acoustic measure of nasality (Chen, 1997; Scarborough & Zellou, 2013; Styler, 2015). We used this measure to determine whether vowels that follow a nasal consonant are indeed acoustically nasalized. A1-P0 corresponds to the difference in amplitude between the harmonic closest to the first oral formant (A1) and the first nasal pole (P0, the harmonic nearest 250-300 Hz). Nasal vowels have a more prominent nasal pole, so we expect lower values of A1-P0 if the vowel is nasal. VoiceSauce does not automatically compute A1-P0 calculations, but we used a similar method as Styler (2015). Briefly, A1 (the harmonic closest to the first formant) was generated by VoiceSauce, which uses the STRAIGHT algorithm (Kawahara, de Cheveigné, & Patterson, 1998) to calculate the harmonic frequencies and amplitudes and Snack (Sjölander, 2004) to calculate formant frequencies and bandwidths. P0 (the first nasal pole) corresponds to H1 or H2, depending on which harmonic is louder. Following Styler (2015), the amplitude of P0 was corrected for vowel formants, which in VoiceSauce is done using the correction by Iseli, Shue, and Alwan (2007). Given that P0 usually falls between 250 and 300 Hz, and that high vowels have an F1 in that range, A1-P0 is usually not used for high vowels. Instead, A1-P1 (the second nasal pole, around 1000 Hz) is normally used (Chen, 1997). VoiceSauce does not automatically calculate harmonics near P1, and recent work suggests that this measure is not a good discriminant of vowel nasality in high vowels (Styler, 2015). Thus, our acoustic analysis of vowel nasality using A1-P0 focuses only on non-high vowels. Lastly, we also used VoiceSauce to compare the spectral tilt of lateral and nasal consonants. As with A1-P0 for vowels, we chose an acoustic measure that should vary as a function of both nasality and breathiness. Acoustically, nasals differ from laterals primarily in terms of their mid-frequency spectral configurations (see Fig. 2). Nasals have a low frequency first pole around 250–300 Hz, with relatively weak higher frequencies. In contrast, laterals have their first pole around 360 Hz, and a cluster of three poles between 1500 and 4000 Hz (Stevens, 2000). Therefore, nasality differences between laterals and nasals should be reflected especially in terms of spectral tilt in the mid-frequency (1000–3000 Hz) region of the spectrum, with laterals having lower spectral tilt than nasals. We therefore measured H4-2kHz, the difference in amplitude between the fourth harmonic and the harmonic closest to 2000 Hz in the audio spectrum. This measure should distinguish nasals from laterals, but is also known to correlate with perceptible changes in voice quality, with breathier voice qualities showing higher values of the measure (Garellek, Keating, Esposito, & Kreiman, 2013; Garellek, Samlan, Gerratt, & Kreiman, 2016).

M. Garellek et al. / Journal of Phonetics 59 (2016) ]]]–]]]

6

Fig. 2. Sample audio spectra of sonorant consonants from a female speaker of Southern Yi. Laterals have lower mid-frequency spectral tilt, as shown by the red dashed line illustrating the approximate slope of H4-2kHz. (For interpretation of the references to color in this figure caption, the reader is referred to the web version of this paper.)

2.3. Statistical analyses Linear mixed-effect models with Satterthwaite approximation (using the lmerTest package in R) were fit for each language. For each language, there were four models for the four dependent variables: mean H4-2kHz (over the entire target segment, for all sonorants), mean A1-P0 (over the entire target segment, only for non-high vowels), mean CQ (over the entire target segment, for all vowels), and mean PIC (over the course of the entire target segment, for all vowels).2 The dependent variables were averaged over the target segment to smooth over abrupt changes in voice measures that are often found at segment boundaries. Each model had the following fixed effects: mean F0 (standardized), Phonation of target segment (dummy coded), and Nasal Category (e.g., oral consonant/vowel, nasal consonant/vowel, also dummy coded). The reference value for Phonation and Nasal category was changed to assess particular comparisons. F0 was included in the models because it is known to be a good predictor of CQ (Kuang, 2013a). For the acoustic measures H4-2kHz and A1-P0, which are expected to be influenced by both voice quality and nasality, we also included mean CQ and PIC as fixed effects. Thus if there is a significant effect of nasality on these measures, it will hold regardless of the influence of the voice source. For all models, we also included the interaction of Phonation and Nasal Categories, in case any difference in voice quality between nasal categories depends on the particular phonation type. For example, we might expect breathiness to only cooccur with contrastively lax vowels, because breathiness would enhance both the nasal and phonation contrasts. In Southern Yi, an additional fixed effect of Aspiration was included to control for vowels that were preceded by aspirated stops (which can influence vowel breathiness). Tone was not included in the models, because it never emerged as a significant predictor of the measures independent of F0. The same holds for effects of vowel quality. Random intercepts by Speaker and Word were included in the model. Random slopes by speaker maximally included F0, CQ and PIC (for the acoustic measures), Phonation, and Nasal Category (with their interaction); random slopes by word included only acoustic measures (F0, CQ and PIC) because neither Phonation nor Nasal categories could vary for a given word. If the model did not converge, we simplified the random effects structure, but always kept both F0 (the most consistent predictor of the acoustic and EGG measures) and Nasal category (the crucial predictor). Results are considered significant when p <0:05. The model structure and outputs can be found as supplementary materials.3

3. Results Below we report results for H4-2kHz, A1-P0, and CQ and PIC within each language, with a summary of results at the end. To improve readability, only the crucial comparisons (nasal vs. oral vowels, nasal vs. oral consonants) are described in the text below. A summary of the findings can be found in Table 5. Further statistical information and figures can be found as supplementary materials (see footnote 3).

3.1. Bo A summary of the statistical comparisons for Bo is shown in Table 2. As expected, nasal consonants in Bo had higher midfrequency spectral tilt than lateral ones, regardless of the phonation category (tense consonants: β ¼0.59, SE ¼0.24, p<0:05; lax consonants: β ¼1.47, SE ¼0.25, p<0:0001). Therefore, even when the effect of voice quality is controlled for (using CQ and PIC as control factors), nasals have higher mid-frequency spectral tilt than laterals. In terms of articulatory adjustments, results show that 2 3

For CQ and PIC, results are qualitatively the same when non-low vowels are excluded from the models. http://idiom.ucsd.edu/ mgarellek/files/Garellek_etal_nasals_supplementary_materials.html

M. Garellek et al. / Journal of Phonetics 59 (2016) ]]]–]]]

7

Table 2 Summary of statistical comparisons for Bo. Asterisks indicate significance of at least p< 0.05. Nasal type

Measure

Phonation

β

SE

p

Consonant

H4-2kHz

Tense Lax Tense Lax Tense Lax

0.59 1.47 0.04 0.02 0.27 0.33

0.24 0.25 0.11 0.14 0.19 0.23

0.03n < 0.001n 0.73 0.91 0.17 0.15

Tense Lax Tense Lax Tense Lax

0.72 0.10 0.27 0.08 0.06 0.04

0.26 0.24 0.09 0.09 0.17 0.17

0.01n 0.69 < 0.001n 0.36 0.71 0.80

CQ PIC

Vowel

A1-P0 CQ PIC

Contact quotient (standardized) ( Breathier)

Vowels

Sonorants

*

2

1 Oral V Progressively Nasalized V Oral C

0

Nasal C

-1

-2

Lax

Tense

Lax

Tense

Fig. 3. Mean CQ (standardized) for Bo. Progressively nasalized tense vowels are significantly lower than oral tense vowels (as indicated by the asterisk), consistent with breathier phonation.

nasals vs. laterals do not differ in either CQ or PIC. Thus, nasal consonants have higher spectral tilt than lateral ones, but this acoustic effect is not driven by breathier articulation as measured via EGG. The A1-P0 analysis for Bo vowels reveals that tense vowels following nasals have lower A1-P0 than corresponding vowels following non-nasal consonants (tense vowels: β ¼ 0.72, SE ¼0.26, p<0:05). Thus tense vowels in Bo show acoustic correlates of nasality when following a nasal consonant. However, lax vowels show no acoustic correlates of nasality after nasal consonants. To determine if the coarticulatory nasality during tense nasalized vowels is accompanied by articulatory adjustments in voice quality, we next analyze CQ and PIC in the same contexts. Results show that Bo tense vowels have lower CQ after nasals than non-nasal sounds (tense vowels: β ¼ 0.27, SE ¼0.09, p<0:01), but no effect is found for Bo lax vowels (see Fig. 3). The PIC analysis also revealed no significant differences as a function of nasality. In sum, in addition to being acoustically nasalized, tense vowels after nasal consonants also show articulatory correlates of breathier phonation via lower CQ.

3.2. Luchun Hani A summary of the statistical comparisons for Luchun Hani is shown in Table 3. As expected, nasal consonants in Luchun Hani had higher mid-frequency spectral tilt than lateral ones, regardless of the phonation category (tense consonants: β ¼0.79, SE¼ 0.19, p<0:001; lax consonants: β ¼ 0.61, SE¼ 0.20, p<0:01). Therefore, even when the effect of voice quality is controlled for (using CQ and PIC as control factors), Luchun Hani nasals have higher mid-frequency spectral tilt than laterals. In terms of articulatory adjustments, results show that nasal consonants have lower CQ than lateral ones when lax (β ¼ 0.61, SE ¼0.16, p <0:001). We found no significant effects for consonants in terms of PIC. In sum, Luchun Hani nasal consonants generally have higher spectral tilt than laterals (regardless of their phonation type), but this acoustic effect is accompanied by articulatory correlates of breathiness (lower CQ) for lax nasals only, as shown in Fig. 4. The A1-P0 analysis for Luchun Hani reveals that vowels following nasals have lower values of the measure than vowels following non-nasal consonants, regardless of the phonation category (tense vowels: β ¼  0.49, SE¼ 0.16, p< 0:01; lax vowels: β ¼  0.57, SE ¼0.15, p<0:01). Thus, vowels following nasal consonants in the language show acoustic correlates of nasalization. To determine if the coarticulatory nasality during nasalized vowels is accompanied by articulatory adjustments in voice quality, we next analyze CQ and PIC in the same contexts. Results show that lax vowels have lower CQ after nasals than after non-nasals (lax vowels: β ¼ 0.44, SE¼ 0.15, p<0:01), but no effect is found for tense vowels (see Fig. 4). The PIC analysis also revealed no significant

M. Garellek et al. / Journal of Phonetics 59 (2016) ]]]–]]]

8

Table 3 Summary of statistical comparisons for Luchun Hani. Asterisks indicate significance of at least p < 0.05. Nasal type

Measure

Phonation

β

SE

p

Consonant

H4-2kHz

Tense Lax Tense Lax Tense Lax

0.79 0.61 0.25 0.61 0.22 0.03

0.19 0.20 0.16 0.16 0.15 0.15

< 0.001n < 0.01n 0.14 < 0.001n 0.17 0.84

Tense Lax Tense Lax Tense Lax

0.49 0.57 0.07 0.44 0.01 0.04

0.16 0.15 0.16 0.15 0.10 0.10

< 0.01n < 0.01n 0.68 < 0.01n 0.90 0.73

CQ PIC

Vowel

A1-P0 CQ PIC

Contact quotient (standardized) ( Breathier)

Vowels

2

Sonorants

*

*

1 Oral V Progressively Nasalized V Oral C

0

Nasal C

-1

-2

Lax

Tense

Lax

Tense

Fig. 4. Mean CQ (standardized) for Luchun Hani. Progressively nasalized lax vowels are significantly lower than oral lax vowels, and lax nasal consonants are significantly lower than lax laterals (as indicated by asterisks), consistent with breathier phonation.

Table 4 Summary of statistical comparisons for Southern Yi. Asterisks indicate significance of at least p < 0.05. Nasal type

Measure

Phonation

β

SE

p

Consonant

H4-2kHz

Tense Lax Tense Lax Tense Lax

0.99 0.92 0.05 0.23 0.58 0.47

0.28 0.27 0.18 0.18 0.20 0.20

< 0.01n < 0.01n 0.79 0.22 < 0.01n < 0.05n

Tense Lax Tense Lax Tense Lax

0.59 0.52 0.36 0.02 0.38 0.02

0.19 0.21 0.21 0.16 0.21 0.22

< 0.01n 0.05n 0.09 0.88 0.09 0.92

CQ PIC

Vowel

A1-P0 CQ PIC

differences as a function of nasality. Therefore, in addition to being acoustically nasalized, Luchun Hani lax vowels after nasal consonants also show articulatory correlates of breathier phonation. Tense vowels on the other hand show acoustic correlates of nasality without any articulatory correlates of breathiness. 3.3. Southern Yi A summary of the statistical comparisons for Southern Yi is shown in Table 4. Nasal consonants in Southern Yi had higher midfrequency spectral tilt than laterals, regardless of the phonation category (tense consonants: β ¼0.99, SE ¼0.28, p<0:01; lax consonants: β ¼0.92, SE¼ 0.27, p<0:01). Therefore, even when the effect of voice quality is controlled for (using CQ and PIC as control factors), Southern Yi nasal consonants have higher mid-frequency spectral tilt than laterals. In terms of articulatory adjustments, results show that nasal consonants have higher PIC than laterals, regardless of phonation type (tense consonants:

M. Garellek et al. / Journal of Phonetics 59 (2016) ]]]–]]]

Vowels

*

Peak increase in contact (standardized) (Breathier )

2

9

Sonorants

* Nasal Category

0

Oral V Progressively Nasalized V Oral C Nasal C

-2

Lax

Tense

Lax

Tense

Fig. 5. Mean PIC (standardized) for Southern Yi. Nasal consonants are significantly higher than oral ones (as indicated by the asterisks), consistent with breathier phonation.

Table 5 Summary of results for acoustic measures (A1-P0 for vowels, H4-2K for consonants) and EGG measures (CQ and PIC). Parentheses mark which phonation type and/or EGG measure showed significant differences. ✓ indicates a significant result at p< 0:05; indicates no significant differences for any phonation type or measure.

Bo Luchun Hani Southern Yi

Acoustic EGG Acoustic EGG Acoustic EGG

Nasalized ½  V more breathy/nasal than oral [V]

Nasal consonant more breathy/nasal than lateral

✓ (Tense) ✓ (Tense, CQ) ✓ ✓ (Lax, CQ) ✓

✓ ✓ ✓ (Lax, CQ) ✓ ✓ (PIC)

β ¼0.58, SE ¼0.20, p<0:01; lax consonants: β ¼0.47, SE ¼0.20, p<0:05). We found no significant effects for consonants in terms of CQ. In sum, Southern Yi nasal consonants generally have higher spectral tilt than non-nasal laterals (regardless of their phonation type), and this acoustic effect is also accompanied by articulatory correlates of breathiness (higher PIC, as seen in Fig. 5). The A1-P0 analysis for Southern Yi reveals that vowels following nasals have lower A1-P0 than vowels following consonants, regardless of the phonation category (tense vowels: β ¼  0.59, SE ¼0.19, p<0:01; lax vowels: β ¼ 0.53, SE ¼0.21, p<0:05). Thus, vowels following nasal consonants in this language show acoustic correlates of nasalization. In terms of articulatory adjustments, results show that tense and lax vowels following nasal vs. non-nasal sounds do not differ in either CQ or PIC. Therefore, Southern Yi vowels show acoustic correlates of nasalization when following a nasal consonant, but show no articulatory correlates of breathiness. 3.4. Summary of results A summary of the statistical comparisons can be found in Table 5. Although the acoustic measures (A1-P0 for vowels, H4-2kHz for consonants) show consistent effects of nasality showing increased spectral tilt, not all nasal sounds are accompanied by changes in phonation. However, when the voice quality of nasal sounds differs articulatorily from their oral counterparts, either CQ or PIC shows values that are correlated with increased breathiness for the nasal sounds compared with the oral sounds.

4. Discussion and conclusion The results of this study provide articulatory evidence for correlates of breathy voice during nasal consonants and contextually nasalized vowels. However, in any given language studied here, not all nasal sounds are necessarily breathier than their oral counterparts. The implications of these findings for phonetic enhancement, nasal representation, and sound change are discussed below. 4.1. Breathy voice during nasality To our knowledge, there is no articulatory reason that can explain why nasals would be produced with breathy voice; nasality can be produced minimally by velum lowering, whereas breathy voice is produced in the larynx. We also believe that there is no phonological explanation for this co-occurrence; nasality and phonation contrasts are orthogonal in these languages. Thus, we claim that the likeliest explanations of the coproduction of nasals with breathy voice relate to listener misperception and phonetic enhancement. Nasal and breathy vowels share common acoustic properties, such as increased first formant bandwidth and

10

M. Garellek et al. / Journal of Phonetics 59 (2016) ]]]–]]]

a stronger first harmonic (Chen, 1997; Simpson, 2012; Styler, 2015). Nasal consonants, like breathy sounds, are characterized by mostly low-frequency energy (Stevens, 2000). Thus, listeners may misperceive nasals as also having breathy voice, and/or speakers may produce nasals with breathy voice to enhance the acoustic properties of nasality. Our findings thus support the assumption of enhancement theories that articulatory gestures are coproduced in part because their acoustic similarities are mutually enhancing (Diehl & Kluender, 1989; Kingston & Diehl, 1994; Keyser & Stevens, 2006; Stevens & Keyser, 1989, 2010; Stevens, Keyser, & Kawasaki, 1986). On the other hand, one particular model of phonetic enhancement – Enhancement Theory – predicts that breathiness would not be used to enhance sonorant consonants, specifically because the higher airflow might decrease sonority by generating frication (Stevens & Keyser, 1989, Section 5.6). However, it is unclear whether all types of breathiness increase airflow to the point of generating frication; impressionistically, the lax vowels in these languages are not very breathy-sounding. Indeed, it is perhaps for this reason that breathiness can also be used to enhance nasality, as we discuss in more detail below. More importantly, Stevens and Keyser (1989) do not mention how breathiness could nonetheless enhance the nasal feature associated with nasal consonants. We find articulatory evidence for breathy voice during nasal consonants in two of the three languages (for Luchun Hani, only lax nasals showed this effect). We believe that there are two possible explanations for this result. First, if the breathy voice during nasal onsets is coarticulated onto the following vowel, then the acoustic correlates of coarticulatory vowel nasalization might be enhanced. This is a possible explanation for Luchun Hani, where both lax nasal consonants and lax nasalized vowels show lower CQ. However, in Southern Yi, nasal consonants are breathier (have higher PIC) despite the fact that contextually nasalized vowels show no voice quality differences compared with non-nasalized vowels. It is possible in this case that the breathy voice during nasal consonants enhances internal cues to consonantal nasality, independent of the nasal's effect on the following vowel. As mentioned earlier, nasal sonorant energy is concentrated in the lower frequencies, just like breathy voice (Klatt & Klatt, 1990; Stevens, 2000), so it is conceivable that nasal consonants are perceived as more nasal if they are also breathy. Further experimental work is needed to test specifically whether nasal consonants are identified more effectively when they are coproduced with breathy voice, much like breathy vowels are known to be perceived as nasal-sounding (Arai, 2006; Hawkins & Stevens, 1985; Klatt & Klatt, 1990; Lintz & Sherman, 1961). Nonetheless, our findings suggest that breathy voice could serve as an enhancement of nasality, regardless of the type of segment (vowel vs. consonant) bearing the feature. The results of this study also show interesting interactions between nasality and phonation type in vowels. The three Yi languages contrast tense and lax phonation, and two of the three languages show articulatory evidence for coproduction of nasalized vowels with breathy phonation. It is therefore unsurprising that voice quality changes associated with nasality interact to some extent with contrastive phonation in these languages. Following an enhancement account, we hypothesized that if breathy nasals were found, they might be more likely to occur with contrastively lax vowels (rather than tense ones), because the breathiness derived from nasality would not weaken the phonation contrasts. Indeed, in Luchun Hani the lax vowels and consonants are breathier if nasal. On the other hand, in Bo it is the tense vowels – not the lax ones – that show breathier voice quality when nasalized. In fact, lax vowels in Bo show no acoustic evidence for vowel nasalization, which suggests that coarticulatory nasalization is limited in certain contexts, for reasons which are still unknown. Overall, it is possible that we find only limited evidence of nasals being breathier than oral sounds because speakers must make use of voice quality changes to signal other linguistic contrasts, and that speakers of Bo make use of breathier voice quality during tense nasals because the tense-lax contrast is being cued through other means such as f0 and vowel quality (Kuang, 2011a; Kuang & Cui, 2016). We also found that Bo tense vowels show EGG correlates of breathiness even when the preceding nasal (the source of the vowel's nasalization) does not. These findings suggest that the EGG correlates of breathiness may not be tied to a ‘lowered velum’ nasal gesture that overlaps with a following vowel; instead, we interpret these results as evidence that nasality as an abstract feature involves multiple articulations. When this feature overlaps with multiple segments (e.g., with a nasal consonant and the following vowel), it may be realized differently according to the segment. Researchers have shown that nasal vowels also involve specific lingual and pharyngeal modulations that may serve to enhance distinctions between oral and nasal vowels (Carignan et al., 2015; da Matta Machado, 1993; Demolin et al., 2003; Shosted, 2015; Shosted et al., 2012). Our study therefore builds on this work by showing that nasality may also involve specific laryngeal articulations (i.e. breathy voice), possibly because glottal spreading and velopharyngeal coupling yield similar acoustic effects and thus reinforce the contrast (Kingston & Diehl, 1994; Kingston et al., 2008). Of course, languages (and speakers of a particular language) may differ in their precise articulatory implementation of nasality. We do not wish to suggest that speakers of all languages will produce nasals with breathy voice; indeed, our study is the first to look for (and find) articulatory correlates to breathiness during nasal sounds, and not all statistical comparisons in these languages show correlates of breathiness during nasals. But we claim that if nasal sounds do indeed involve a specific laryngeal articulation distinct from oral ones, this laryngeal articulation should tend towards breathy voice, because it is confusable with the acoustic attributes of nasality, which it may also serve to enhance. 4.2. Implications for sound change As discussed earlier, nasal and breathy sounds sometimes interact diachronically in languages of the world. Although such diachronic interactions are somewhat rare, they must be accounted for. Researchers have suggested that they are due to the acoustic similarity between nasality and breathy voice (Blevins, 2004; Blevins & Garrett, 1993; Matisoff, 1975; Ohala & Busà, 1995). Our study is in line with this possible path of sound change, assuming that listener misperception results in a stage where speakers produce nasals with breathiness, and eventually the nasality disappears entirely, as in nn (misperception of nasal consonant as

M. Garellek et al. / Journal of Phonetics 59 (2016) ]]]–]]]

11

also being breathy) -ɦ (loss of nasality). The ‘promotion’ of breathiness to primary articulator can later arise through listener misperception (Ohala, 1981; Ohala & Busà, 1995) or gestural reduction (Garrett & Johnson, 2013). Such a diachronic change could account for cases where nasality is replaced by a breathy sound, as in the case of certain dialects of Basque (Igartua, 2011). However, the intermediate stage of coproduction of nasal consonants with breathy voice can also be explained in terms of phonetic enhancement. Speakers might enhance nasal consonants by producing them as breathy. This could in principle enhance the nasal consonant's internal cues (by amplifying the nasal consonants low-frequency energy), or its coarticulatory cues: the breathy voice adjacent to the consonant can cue the listener to the presence of a nasal gesture (Arai, 2006; Keyser & Stevens, 2006; Klatt & Klatt, 1990; Stevens & Keyser, 1989). As we discussed earlier, an enhancement explanation of the cooccurrence of nasality with breathiness is also compatible with listener misperception. For instance, the listener can misperceive nasals as being both nasal and breathy, and then the listener-turned-speaker produces the sound with both nasality and breathy voice. Assuming that these breathy nasals are more ‘nasal-sounding’, speakers in the community then use breathy nasals as a means of phonetic enhancement. Phonetic enhancement could also explain changes in the opposite direction, whereby a breathy sound becomes nasalized. As discussed earlier, ‘spontaneous nasalization’ adjacent to sounds produced with glottal spreading is attested in languages of the world (Blevins & Garrett, 1993; Matisoff, 1975; Ohala & Busà, 1995; Rogers, 2011). It too can be accounted for in terms of speakers’ coproduction of breathiness with nasality, as in nha - nh̃ ã (enhancement of breathiness) - hã. However, it should be noted that our study does not bear directly on the question of whether breathy sounds (like [h]) are more likely to be produced with nasalization, in particular with velopharyngeal coupling. The fact that these diachronic interactions between nasal and breathy sounds are not highly attested across languages might further support an enhancement account: producing nasal sounds as breathy is only one possible enhancement strategy of many, such as durational, lingual, or pharyngeal adjustments (Carignan et al., 2015; da Matta Machado, 1993; Demolin et al., 2003; Scarborough et al., 2015; Shosted, 2015; Shosted et al., 2012; Solé, 1995). If the enhancement strategy depends on factors like cue informativity (Kirby, 2013), then it is possible that breathiness does not frequently replace nasalization because voice quality differences are confusable with other contrasts in the language like stop aspiration, contrastive phonation, or tone. Ultimately, however, we are unable to determine whether listener misperception, phonetic enhancement, or both misperception and enhancement are responsible for the presence of breathier nasals in these three languages. We leave this question for future research.

Acknowledgments We would like to thank Scott Ragland and Stefan Wooding for their help with data segmentation and labeling, as well as Jamie White, Adam Chong, and audiences at the 2016 Annual Meeting of the Linguistic Society of America and the Southern California April Meeting on Phonology for helpful feedback. We are also very grateful to Taehong Cho, Christopher Carignan, and two anonymous reviewers for their careful reviews and suggestions on earlier drafts of the paper. References Arai, T. (2006). Cue parsing between nasality and breathiness in speech perception. Acoustical Science and Technology, 25, 298–301. Baken, R. J., & Orlikoff, R. F. (2000). Clinical measurement of speech and voice. San Diego: Singular Publishing Group. Beddor, P. S. (1993). The perception of nasal vowels. In M. K. Huffman, & R. A. Krakow (Eds.), Nasals, nasalization, and the velum. Phonetics and phonology (Vol. 5, pp. 171–196). San Diego; Academic Press. Beddor, P. S. (2009). A coarticulatory path to sound change. Language, 85, 785–821. Beddor, P. S., & Krakow, R. A. (1999). Perception of coarticulatory nasalization by speakers of English and Thai: Evidence for partial compensation. Journal of the Acoustical Society of America, 106, 2868–2887. Beddor, P. S., McGowan, K. B., Boland, J. E., Coetzee, A. W., & Brasher, A. (2013). The time course of perception of coarticulation. Journal of the Acoustical Society of America, 133, 2350–2366. Bickley, C. (1982). Acoustic analysis and perception of breathy vowels. MIT speech communication working papers (Vol. 1, pp. 71–81). Blevins, J. (2004). Evolutionary phonology: The emergence of sound patterns. Cambridge: Cambridge University Press. Blevins, J., & Garrett, A. (1993). The evolution of Ponapeic nasal substitution. Oceanic Lingustics, 32, 199–236. Carignan, C. (2014). An acoustic and articulatory examination of the "oral" in "nasal": The oral articulations of French nasal vowels are not arbitrary. Journal of Phonetics, 46, 23–33. Carignan, C., Shosted, R. K., Fu, M., Liang, Z.-P., & Sutton, B. P. (2015). A real-time MRI investigation of the role of lingual and pharyngeal articulation in the production of the nasal vowel system of French. Journal of Phonetics, 50, 34–51. Carroll, L. S. (2015). Ixpantepec Nieves Mixtec word prosody (Ph.D. thesis). San Diego: University of California. Chen, M. Y. (1997). Acoustic correlates of English and French nasalized vowels. Journal of the Acoustical Society of America, 102, 2360–2370. Cohn, A. C. (1993). Nasalisation in English: Phonology or phonetics. Phonology, 10, 43–81. Colarusso, J. (1988). The Northwest Caucasian languages: A phonological survey. New York: Garland. da Matta Machado, M. (1993). Fenômenos de nasalização vocálica em português: Estudo cine-radiográfico. Cadernos de estudos lingüísticos, 25, 113–127. Davidson, L. (2016). Variability in the implementation of voicing in American English obstruents. Journal of Phonetics, 54, 35–50. Delattre, P., & Monnot, M. (1968). The role of duration in the identification of French nasal vowels. International Review of Applied Linguistics, 6, 267–288. Delvaux, V., Demolin, D., Harmegnies, B., & Soquet, A. (2008). The aerodynamics of nasalization in French. Journal of Phonetics, 36, 578–606. Demolin, D., Delvaux, V., Methens, T., & Soquet, A. (2003). Determination of velum opening for French nasal vowels by magnetic resonance imaging. Journal of Voice, 17, 454–467. DiCanio, C. T. (2009). The phonetics of register in Takhian Thong Chong. Journal of the International Phonetic Association, 39, 162–188. Diehl, R. L., & Kluender, K. R. (1989). On the objects of speech perception. Ecological Psychology, 1, 121–144. Esposito, C. M. (2010). Variation in contrastive phonation in Santa Ana Del Valle Zapotec. Journal of the International Phonetic Association, 40, 181–198. Esposito, C. M. (2012). An acoustic and electroglottographic study of White Hmong phonation. Journal of Phonetics, 40, 466–476. Esposito, C. M., & Khan, S. (2012). Contrastive breathiness across consonants and vowels: A comparative study of Gujarati and White Hmong. Journal of the International Phonetic Association, 42, 123–143. Garellek, M. (2012). The timing and sequencing of coarticulated non-modal phonation in English and White Hmong. Journal of Phonetics, 40, 152–161. Garellek, M. (2014). Voice quality strengthening and glottalization. Journal of Phonetics, 45, 106–113. Garellek, M., & Keating, P. (2011). The acoustic consequences of phonation and tone interactions in Jalapa Mazatec. Journal of the International Phonetic Association, 41, 185–205.

12

M. Garellek et al. / Journal of Phonetics 59 (2016) ]]]–]]]

Garellek, M., Keating, P., Esposito, C. M., & Kreiman, J. (2013). Voice quality and tone identification in White Hmong. Journal of the Acoustical Society of America, 133, 1078–1089. Garellek, M., Samlan, R., Gerratt, B. R., & Kreiman, J. (2016). Modeling the voice source in terms of spectral slopes. Journal of the Acoustical Society of America, 139, 1404–1410. Garrett, A., & Johnson, K. (2013). Phonetic bias in sound change. In A. C. L. Yu (Ed.), Origins of sound change: Approaches to phonologization (pp. 51–97). Oxford: Oxford University Press. Gordon, M., & Ladefoged, P. (2001). Phonation types: A cross-linguistic overview. Journal of Phonetics, 29, 383–406. Hanson, H. M., Stevens, K. N., Kuo, H.-K. J., Chen, M. Y., & Slifka, J. (2001). Towards models of phonation. Journal of Phonetics, 29, 451–480. Hawkins, S., & Stevens, K. N. (1985). Acoustic and perceptual correlates of the non-nasal/nasal distinction for vowels. Journal of the Acoustical Society of America, 77, 1560–1575. Herbst, C., & Ternström, S. (2006). A comparison of different methods to measure the EGG contact quotient. Logopedics Phoniatrics Vocology, 31, 126–138. Howard, D. M. (1995). Variation of electrolaryngographically derived closed quotient for trained and untrained adult female singers. Journal of Voice, 9, 163–172. Igartua, I. (2011). La aspiración de origen nasal en la evolución fonológica del euskera: un caso de ‘rhinoglottophilia’. Anuario del Seminario de Filología Vasca “Julio de Urquijo”, 42, 171–189. International Phonetic Association (1999). Handbook of the International Phonetic Association. Cambridge: Cambridge University Press. Iseli, M., Shue, Y.-L., & Alwan, A. (2007). Age, sex, and vowel dependencies of acoustic measures related to the voice source. Journal of the Acoustical Society of America, 121, 2283–2295. Kawahara, H., de Cheveigné, A., & Patterson, R. D. (1998). An instantaneous-frequency-based pitch extraction method for high-quality speech transformation: Revised TEMPO in the STRAIGHT-suite. In ICSLP-1998 (p. 0659). Keating, P., Esposito, C., Garellek, M., Khan, S., & Kuang, J. (2011). Phonation contrasts across languages. In Proceedings of the international congress of phonetic sciences (pp. 1046– 1049), Hong Kong. Keyser, S. J., & Stevens, K. N. (2006). Enhancement and overlap in the speech chain. Language, 82, 33–63. Khan, S. (2012). The phonetics of contrastive phonation in Gujarati. Journal of Phonetics, 40, 780–795. Kingston, J., & Diehl, R. L. (1994). Phonetic knowledge. Language, 70, 419–454. Kingston, J., Diehl, R. L., Kirk, C. J., & Castleman, W. A. (2008). On the internal perceptual structure of distinctive features: The [voice] contrast. Journal of Phonetics, 36, 28–54. Kirby, J. (2013). The role of probabilistic enhancement in phonologization. In A. C. L. Yu (Ed.), Origins of sound change: Approaches to phonologization (pp. 228–246). Oxford: Oxford University Press. Klatt, D. H., & Klatt, L. C. (1990). Analysis, synthesis, and perception of voice quality variations among female and male talkers. Journal of the Acoustical Society of America, 87, 820–857. Kreiman, J., Shue, Y.-L., Chen, G., Iseli, M., Gerratt, B. R., Neubauer, J., et al. (2012). Variability in the relationships among voice quality, harmonic amplitudes, open quotient, and glottal area waveform shape in sustained phonation. Journal of the Acoustical Society of America, 132, 2625–2632. Kuang, J. (2011a). Phonation contrast in two register contrast languages and its influence on vowel and tone. In Proceedings of the 17th international congress of phonetic sciences (pp. 1146–1149), Hong Kong. Kuang, J. (2011b). Production and perception of the phonation contrast in Yi (Master's thesis). UCLA. Kuang, J. (2013a). Phonation in tonal contrasts (Ph.D. thesis). UCLA. Kuang, J. (2013b). The tonal space of contrastive five level tones. Phonetica, 70, 1–23. Kuang, J., & Cui, A. (2016). Relative cue weighting in perception and production of a sound change in progress. Talk presented at LabPhon 2016, Cornell, NY. Kuang, J., & Keating, P. (2014). Vocal fold vibratory patterns in tense versus lax phonation contrasts. Journal of the Acoustical Society of America, 136, 2784–2797. Ladefoged, P., Williamson, K., Elugbe, B., & Uwalaka, A. A. (1976). The stops of Owerri Igbo. Studies in African Linguistics, 6(Supp.), 147–163. Lintz, L. B., & Sherman, D. (1961). Phonetic elements and perception of nasality. Journal of Speech and Hearing Research, 4, 381–396. Löfqvist, A., & McGarr, N. (1987). Laryngeal dynamics in voiceless consonant production. In T. Baer, C. Sasaki, & K. Harris (Eds.), Laryngeal function in phonation and respiration (pp. 391–402). Boston: College Hill. Maeda, S. (1993). Acoustics of vowel nasalization and articulatory shifts in French nasal vowels. In M. K. Huffman, & R. A. Krakow (Eds.), Nasals, nasalization, and the velum. Phonetics and phonology (Vol. 5, pp. 147–167). Academic Press. Malécot, A. (1960). Vowel nasality as a distinctive feature in American English. Language, 36, 222–229. Matisoff, J. A. (1975). Rhinoglottophilia: The mysterious connection between nasality and glottality. In C. A. Ferguson, L. M. Hyman, & J. J. Ohala (Eds.), Nasálfest: Papers from a symposium on nasals and nasalization (pp. 265–287). Stanford: Stanford University Language Universals Project. Michaud, A. (2004). A measurement from electroglottography: DECPA , and its application in prosody. In Proceedings of speech prosody 2004 (pp. 633–636), Nara, Japan. Ní Chasaide, A., & Gobl, C. (1993). Contextual variations of the vowel voice source as a function of adjacent consonants. Language and Speech, 36, 303–330. Ohala, J. J. (1975). Phonetic explanations for nasal sound patterns. In C. A. Ferguson, L. M. Hyman, & J. J. Ohala (Eds.), Nasálfest: Papers from a symposium on nasals and nasalization (pp. 289–316). Stanford: Stanford University Language Universals Project. Ohala, J. J. (1981). The listener as a source of sound change. In C.S. Masek, R. A. Hendrick, & M. F. Miller (Eds.), Papers from the parasession on language and behavior (pp. 178–203). Chicago: Chicago Linguistic Society. Ohala, J. J., & Busà, M. G. (1995). Nasal loss before voiceless fricatives: A perceptually-based sound change. Rivista di Linguistica, 7, 125–144. Orlikoff, R. F. (1991). Assessment of the dynamics of vocal fold contact from the electroglottogram. Journal of Speech and Hearing Research, 34, 1066–1072. Pruthi, T., & Espy-Wilson, C. Y. (2007). Acoustic parameters for the automatic detection of vowel nasalization. In Proceedings of interspeech 2007 (pp. 1925–1928). Antwerp, Belgium. , Rogers, C. (2011). Spontaneous nasalization in Wichi . Paper delivered at the CUNY workshop on endangered language phonology. Rothenberg, M., & Mashie, J. J. (1988). Monitoring vocal fold abduction through vocal fold contact area. Journal of Speech and Hearing Research, 31, 338–351. Scarborough, R., & Zellou, G. (2013). Clarity in communication: “Clear” speech authenticity and lexical neighborhood density effects in speech production and perception. Journal of the Acoustical Society of America, 134, 3793–3807. Scarborough, R., Zellou, G., Mirzayan, A., & Rood, D. S. (2015). Phonetic and phonological patterns of nasality in Lakota vowels. Journal of the International Phonetic Association, 45, 289–309. Schadeberg, T. C. (1989). The velar nasal in Nyole. Annales Aequatoria, 10, 169–179. Shosted, R. (2015). Nasal vowels are not [ + NASAL] oral vowels. In J. Smith, & T. Ihsane (Eds.), Romance linguistics 2012: Selected papers from the 42nd linguistic symposium on romance languages (LSRL) (pp. 63–76). Shosted, R., Carignan, C., & Rong, P. (2012). Managing the distinctiveness of phonemic nasal vowels: Articulatory evidence from Hindi. Journal of the Acoustical Society of America, 131, 455–465. Shue, Y.-L., Keating, P. A., Vicenik, C., & Yu, K. (2011). VoiceSauce: A program for voice analysis. In Proceedings of the international congress of phonetic sciences (pp. 1846–1849), Hong Kong. Simpson, A. (2012). The first and second harmonics should not be used to measure breathiness in male and female voices. Journal of Phonetics, 40, 477–490. Sjölander, K. (2004). Snack sound toolkit. Stockholm; KTH. Solé, M.-J. (1995). Spatio-temporal patterns of velopharyngeal action in phonetic and phonological nasalization. Language and Speech, 38, 1–23. Stevens, K. N. (2000). Acoustic phonetics. Cambridge: MIT Press. Stevens, K. N., & Keyser, S. J. (1989). Primary features and their enhancement in consonants. Language, 65, 81–106. Stevens, K. N., & Keyser, S. J. (2010). Quantal theory, enhancement and overlap. Journal of Phonetics, 38, 10–19. Stevens, K. N., Keyser, S. J., & Kawasaki, H. (1986). Toward a phonetic and phonological theory of redundant features. In J. S. Perkell, & D. H. Klatt (Eds.), Invariance and variability in speech processes (pp. 426–449). New York: Psychology Press. Styler, W. (2015). On the acoustical and perceptual features of vowel nasality (Ph.D. thesis). University of Colorado. Westbury, J. R. (1983). Enlargement of the supraglottal cavity and its relation to stop consonant voicing. Journal of the Acoustical Society of America, 73, 1322–1336. Zellou, G., & Tamminga, M. (2014). Nasal coarticulation changes over time in Philadelphia English. Journal of Phonetics, 47, 18–35. Zhang, Z. (2016). Cause-effect relationship between vocal fold physiology and voice production in a three-dimensional phonation model. Journal of the Acoustical Society of America, 139, 1493–1507.