Dictionary: <Dictionary> OUP UNCORRECTED PROOF – <STAGE>, 21/11/20, SPi chapter 2 Articulatory Measures of Prosody Taehong Cho and Doris Mücke 2.1 Introduction Over the past few decades, theories of prosody have been developed that have broadened the topic in such a way that the term ‘prosody’ does not merely pertain to low-level realiza- tion of suprasegmental features such as f0, duration, and amplitude, but also concerns high- level prosodic structure (e.g. Beckman 1996; Shattuck-Hufnagel and Turk 1996; Keating 2006; Fletcher 2010; Cho 2016). Prosodic structure is assumed to have multiple functions, such as a delimitative function (e.g. a prosodic boundary marking), a culminative function (e.g. a prominence marking), and functions deriving from the distribution of tones at both lexical and post-lexical levels. It involves dynamic changes of articulation in the laryngeal and supralaryngeal system, oſten accompanied by prosodic strengthening—that is, hyperartic- ulation of phonetic segments to enhance paradigmatic contrasts by a more distinct articula- tion, and sonority expansion to enhance syntagmatic contrasts by increasing periodic energy radiated from the mouth (see Cho 2016 for a review). Under a broad definition of prosody, therefore, prosody research in speech production concerns the interplay between phonetics and prosodic structure (e.g. Mücke et al. 2014, 2017). It embraces issues related to how abstract prosodic structure influences the phonetic implementation by the laryngeal and supralaryngeal systems, and how higher-level prosodic structure may in turn be recov- erable from or manifest in the variation in the phonetic realization. For example, a marking of a tonal event in the phonetic substance involves dynamic changes not only in the laryn- geal system (regulating the vocal fold vibration to produce f0 contours) but also in the supralaryngeal system (regulating movements of articulators to produce consonants and vowels in the textual string). With the help of articulatory measuring techniques, the way these two systems are coordinated in the spatio-temporal dimension is directly observable, allowing various inferences about the role of the prosodic structure in this coordination to be made. is chapter introduces a number of modern articulatory measuring techniques, along with examples across languages indicating how each technique may be used or has been 0004898684.INDD 15 11/21/2020 2:36:32 PM
Transcript
Dictionary: <Dictionary>
OUP UNCORRECTED PROOF – <STAGE>, 21/11/20, SPi
chapter 2
Articulatory Measur es of Prosody
Taehong Cho and Doris Mücke
2.1 Introduction
Over the past few decades, theories of prosody have been developed that have broadened the topic in such a way that the term ‘prosody’ does not merely pertain to low-level realiza-tion of suprasegmental features such as f0, duration, and amplitude, but also concerns high-level prosodic structure (e.g. Beckman 1996; Shattuck-Hufnagel and Turk 1996; Keating 2006; Fletcher 2010; Cho 2016). Prosodic structure is assumed to have multiple functions, such as a delimitative function (e.g. a prosodic boundary marking), a culminative function (e.g. a prominence marking), and functions deriving from the distribution of tones at both lexical and post-lexical levels. It involves dynamic changes of articulation in the laryngeal and supralaryngeal system, often accompanied by prosodic strengthening—that is, hyperartic-ulation of phonetic segments to enhance paradigmatic contrasts by a more distinct articula-tion, and sonority expansion to enhance syntagmatic contrasts by increasing periodic energy radiated from the mouth (see Cho 2016 for a review). Under a broad definition of prosody, therefore, prosody research in speech production concerns the interplay between phonetics and prosodic structure (e.g. Mücke et al. 2014, 2017). It embraces issues related to how abstract prosodic structure influences the phonetic implementation by the laryngeal and supralaryngeal systems, and how higher-level prosodic structure may in turn be recov-erable from or manifest in the variation in the phonetic realization. For example, a marking of a tonal event in the phonetic substance involves dynamic changes not only in the laryn-geal system (regulating the vocal fold vibration to produce f0 contours) but also in the supralaryngeal system (regulating movements of articulators to produce consonants and vowels in the textual string). With the help of articulatory measuring techniques, the way these two systems are coordinated in the spatio-temporal dimension is directly observable, allowing various inferences about the role of the prosodic structure in this coordination to be made.
This chapter introduces a number of modern articulatory measuring techniques, along with examples across languages indicating how each technique may be used or has been
0004898684.INDD 15 11/21/2020 2:36:32 PM
HYU
텍스트 상자
Cho, Taehong & Doris Mücke (2020). Articulatory Measures of Prosody. In C. Gussenhoven and A. Chen (Eds.), The Oxford Handbook of Language Prosody (pp. 16-38). Oxford University Press.
used on various aspects of prosody in the phonetics–prosody interplay. These include (i) laryngoscopy and electroglottography (EGG) to study laryngeal events associated with vocal fold vibration; (ii) systems such as the magnetometer (electromagnetic articulogra-phy, EMA), electropalatography (EPG), and ultrasound systems for exploring supralaryn-geal articulatory events; and (iii) aerodynamic measurement systems for recording oral/subglottal pressure and oral/nasal flow, and a device, called the RIP (Respitrace inductive plethysmograph) for recording respiratory activities.
2.2 Experimental techniques
2.2.1 Laryngoscopy
Laryngoscopy allows a direct observation of the larynx. A fibreoptic nasal laryngoscopy system (Ladefoged 2003; Hirose 2010) contains a flexible tube with a bundle of optical fibres which may be inserted through the nose, while the lens at the end of the fibreoptic bundle is usually positioned near the tip of the epiglottis above the vocal folds. Before the insertion of the scope, surface anaesthesia may be applied to the nasal mucosa and to the epipharyn-geal wall. The procedure is relatively invasive, requiring the presence of a physician during the experiment. A recent system for laryngoscopy provides high-speed motion pictures of the vibrating vocal folds with useful information about the laryngeal state and the glottal condition during phonation (e.g. Esling and Harris 2005; Edmondson and Esling 2006). The recording of laryngeal images by a laryngoscope is often made simultaneously with a recording of the electroglottographic and acoustic signals (see Hirose 2010: fig. 4.3). Because of its invasiveness and operating constraints, however, the use of laryngoscopy in phonetics research has been quite limited.
In prosody research, a laryngoscope may be used to explore the laryngeal mechanisms for controlling f0 in connection with stress, tones, and phonation types. Lindblom (2009) discussed a fibrescopic study in Lindqvist-Gauffin (1972a, 1972b) that examined laryngeal behaviour during glottal stops and f0 changes for Swedish word accents. The fibrescopic data were interpreted as indicating that there may be three dimensions involved in control-ling f0 and phonation types: glottal adduction-abduction, laryngealization (which involves the aryepiglottic folds), and activity of the vocalis muscle. With reference to more recent fibrescopic data (Edmondson and Esling 2006; Moisik and Esling 2007; Moisik 2008), Lindblom (2009) suggested that the glottal stop, creaky voice, and f0 lowering may involve the same kind of laryngealization to different degrees. Basing their argument on cross- linguistic fibrescopic data, Edmondson and Esling (2006) suggested that there are indeed different ‘valve’ mechanisms for controlling articulatory gestures that are responsible for cross-linguistic differences in tone, vocal register, and stress, and that languages may differ in choosing specific valve mechanisms.
A fibreoptic laryngoscopic study that explores the interplay between phonetics and prosodic structure is found in Jun et al. (1998), who directly observed the changing glot-tal area in the case of disyllabic Korean words with different consonant types, with the aim of understanding the laryngeal states associated with vowel devoicing. A change in the glottal area was in fact shown to be conditioned by prosodic position (accentual
phrase-initial vs. accentual phrase-medial), interpreted as gradient devoicing. Further fibreoptic research might explore the interplay between phonetics and prosodic struc-ture, in particular to study the connection between prosodic strengthening and laryngeal articulatory strengthening.
2.2.2 Electroglottography
The electroglottograph (EGG), also referred to as laryngograph, is a non-invasive device that allows for monitoring vocal fold vibration and the glottal condition during phonation (for more information see Laver 1980; Rothenberg and Mashie 1988; Rothenberg 1992; Baken and Orlikoff 2000; d’Alessandro 2006; Hirose 2010; Mooshammer 2010). It estimates the contact area between the vocal folds during phonation by measuring changes in the transverse electrical impedance of the current between two electrodes across the larynx placed on the skin over both sides of the thyroid cartilage (Figure 2.1). Given that a glottis filled with air does not conduct electricity, the electrical impedance across the larynx is roughly negatively correlated with the contact area. EGG therefore not only provides an accurate estimation of f0 but also measures parameters related to the glottal condition, such as open quotient (OQ, the percentage of the open glottis interval relative to the duration of the full abduction–adduction cycle), contact or closed quotient (CQ, the percentage of the closed glottis interval relative to the duration of the full cycle), and skewness quotient (SQ, the ratio between the closing and opening durations). EGG signals are often obtained simultaneously with acoustic and airflow signals, so that the glottal condition can be holistically estimated. While readers are referred to d’Alessandro (2006) for a review of how voice source parameters (including those derived from EGG signals) may be used in prosody
0.40.30.20.1
0–0.1–0.2–0.3
0 2 4 6 8Time (ms)
Normal voice
10 12 14 16 18Incr
easin
g vo
cal f
old
cont
act =
=> 0.20.15
0.10.0.5
0–0.05
–01–0.15
0 10 20 30 40Time (ms)
Creaky voice
50 60 70 80Incr
easin
g vo
cal f
old
cont
act =
=>
0.40.30.20.1
0–0.1–0.2–0.3–0.4
0 2 4 6 8Time (ms) Time (ms)
Breathy voice
10 12 14 16 18 20Incr
easin
g vo
cal f
old
cont
act =
=> 0.30.20.1
0–0.1–0.2–0.3–0.4
0 2 4 6 8
Loud voice
10 12 14 16 18Incr
easin
g vo
cal f
old
cont
act =
=>
Figure 2.1 Waveforms corresponding to vocal fold vibrations in electroglottography (examples by Phil Hoole at IPS Munich) for different voice qualities. High values indicate increasing vocal fold contact. Photo taken at IfL Phonetics Lab, Cologne.
analysis, in what follows we will discuss a few cases in which EGG is used in exploring the interplay between phonetics and prosody.
An EGG may be used to explore variation in voice quality strengthening as a function of prosodic structure. For example, Garellek (2014) explored the effects of pitch accent (phrase-level stress) and boundary strength on the voice quality of vowels in vowel-initial words in English and Spanish. Word-initial vowels under pitch accent were found to have an increase in EGG contact (as reflected in the CQ) in both English and Spanish, showing laryngealized voice quality. Garellek’s study of the phonetics–prosody interplay built on the assumption that fine-grained phonetic detail, this time in the articulatory dimension of glottis, is modulated differently by different sources of prosodic strengthening (prominence vs. bound-ary). Interestingly, however, both languages showed a decrease in EGG contact at the begin-ning of a larger prosodic domain (e.g. intonational phrase-initial vs. word-initial). This runs counter to the general assumption that domain-initial segments are produced with a more forceful articulation (e.g. Fourgeron 1999, 2001; Cho et al. 2014a) and that phrase-initial vowels are more frequently glottalized than phrase-medial ones (e.g. Dilley et al. 1996; Di Napoli 2015), which would result in an increase in EGG contact. Moreover, contrary to Garellek’s observation, Lancia et al. (2016) reported that vowel-initial words in German showed more EGG contact only when the initial syllable was unstressed, which indicates that more research is needed to understand this discrepancy from cross-linguistic perspec-tives. EGG was also used for investigating glottalization at phrase boundaries in Tuscan and Roman Italian, pointing to the fact that these glottal modifications are used as prosodic markers in a gradient fashion (Di Napoli 2015).
Another EGG study that relates the glottal condition to prominence is Mooshammer (2010). It examined various parameters obtained from EGG signals in order to explore how word-level stress and sentence-level accent may be related to vocal effort in German. The author showed that a vowel produced with a global vocal effort (i.e. with increased loud-ness) was similar to a vowel with lexical stress at least in terms of two parameters, OQ and glottal pulse shape (obtained by applying a version of principal component analysis), inde-pendent of higher-level accent (due to focus). A focused vowel, on the other hand, was produced with a decrease in SQ compared to an unfocused vowel, showing a more sym-metrical vocal pulse shape. To the extent these results hold, lexical stress and accent in German may be marked by different glottal conditions. However, given that an accented vowel is in general produced with an increase in loudness, as has been found across lan-guages (including German, e.g. Niebuhr 2010), further research is required to explore the exact relationship between vocal effort and accent, both of which apparently increase loudness.
2.3 Aerodynamic and respiratory movement measures
Aerodynamic devices most widely used for phonetic research use oral and nasal masks (often called Rothenberg masks, following Rothenberg 1973) through which the amount of oral/nasal flow can be obtained in a fairly non-invasive way. Intraoral pressure may be
simultaneously obtained by inserting a small pressure tube between the lips inside the oral mask; this records the pressure of the air in the mouth (e.g. Ladefoged 2003; Demolin 2011). One aerodynamic measure that is more directly related to prosody may be subglottal pres-sure, as sufficient subglottal pressure is required for initiating and maintaining vocal fold vibration (van den Berg 1958) and an increase in subglottal pressure is likely to result in an increase in loudness (sound pressure level) and f0 (e.g. Ladefoged and McKinney 1963; Lieberman 1966; Ladefoged 1967). It is not, however, easy to measure subglottal pressure directly; this involves either a tracheal puncture (i.e. inserting a pressure transducer needle in the trachea; see Ladefoged 2003: fig. 3) or inserting a rubber catheter with a small bal-loon through the nose and down into the oesophagus at the back of the trachea (e.g. Ladefoged and McKinney 1963). Non-invasive methods to estimate subglottal pressure have been developed by using intraoral pressure and volume flow (Rothenberg 1973; Smitheran and Hixon 1981; Löfqvist et al. 1982), but these have limited applicability in prosody research, because certain conditions (e.g. a CVCV context) must be met to obtain reliable data.
Aerodynamic properties of speech sounds can be compared with respiratory activities such as lung volume, which may be obtained with a so-called RIP, or Respitrace inductive plethysmograph (e.g. Gelfer et al. 1987; Hixon and Hoit 2005; Fuchs et al. 2013, 2015). In this technique, subjects wear two elastic bands (approximately 10 cm wide vertically), one around the thoracic cavity (the rib cage) and one around the abdominal cavity (Figure 2.2). The bands expand and recoil as the volume of the thoracic and abdominal cavities changes during exhalation and inhalation, such that the electrical resistance of small wires attached to the bands (especially the upper band) is used to estimate the change in the lung volume
2Er malt Tania, aber nicht Sonja. (He paints Tanja, but not Sonja.)
1
0
Aco
ustic
s
–1
–2
1
0.8 Inhalation Exhalation
0.6
�or
ax in
V
0.4
0.2
0.8
0.6
0.4
Abd
omen
in V
0.2
0
3.5 4 4.5 5 5.5 6 6.5 7
3.5 4 4.5 5 5.5 6 6.5 7
3.5 4 4.5 5 5.5Time in s
6 6.5 7
Figure 2.2 Volume of the thoracic and abdominal cavities in a Respitrace inductive plethysmo-graph during sentence production, inhalation and exhalation phase.
(Photo by Susanne Fuchs at Leibniz Zentrum, ZAS, Berlin.)
during speech production. Gelfer et al. (1987) used the subglottal pressure measurements (Ps) obtained directly from the trachea to examine the nature of global f0 declination (e.g. Pierrehumbert 1979; Cooper and Sorenson 1981). Based on comparison of Ps with f0 and estimated lung volume (obtained with a RIP), Gelfer et al. suggested that Ps is a controlled variable in sentence production, and f0 declination comes about as a consequence of con-trolling Ps. Most recently, however, Fuchs et al. (2015) assessed respiratory contributions to f0 declination in German by using the same RIP technique, and suggested that f0 declination may not stem entirely from physiological constraints on the respiratory system but may additionally be modulated by speech planning as well as by communicative constraints as suggested in Fuchs et al. (2013). This finding is in line with Arvaniti and Ladd (2009) for Greek.
Some researchers have measured airflow and intraoral pressure as an index of respiratory force, because they are closely correlated with subglottal pressure. For instance, oral flow (usually observed during a vowel or a continuant consonant) and oral pressure (usually observed during a consonant) are often interpreted as being correlated with the degree of prominence (e.g. Ladefoged 1967, 2003). Exploring boundary-related strengthening effects on the production of three-way contrastive stops in Korean (lenis, fortis, aspirated; e.g. Cho et al. 2002), for example, Cho and Jun (2000) observed systematic variation of airflow measured just after the release of the stop as a function of boundary strength. However, the detailed pattern was better understood as supporting the three-way obstruent contrast. This again implies that variation in the respiratory force as a function of prosodic structure is further modulated in a language-specific way, in this case by the segmental phonology of the language.
In a similar vein, nasal flow has been investigated by researchers in an effort to understand how the amount of nasal flow produced with nasal sounds may be regulated by prosodic structure (e.g. Jun 1996; Fougeron and Keating 1997; Gordon 1997; Fougeron 2001). From an articulatory point of view, Fougeron (2001) hypothesized that the articulatory force associ-ated with prosodic strengthening may have the effect of elevating the velum, resulting in a reduction of nasal flow. Results from French (Fougeron 2001), Estonian (Gordon 1997), and English (Fougeron and Keating 1997) indeed show that nasal flow tends to be reduced in domain-initial position, in line with Fougeron’s articulatory strengthening-based account. (See Cho et al. 2017 for a suggestion that reduced nasality for the nasal consonant may be interpreted in terms of paradigmatic vs. syntagmatic enhancement due to prominence and domain-initial strengthening, respectively.) These studies again indicate that an examination of nasal flow would provide useful data on how low-level segmental realization is condi-tioned by higher-order prosodic structural factors.
2.4 Point-tracking techniques for articulatory movements
Point-tracking techniques allow for the measuring of positions and movements of articula-tors over time by attaching small pellets (or sensors) to flesh points of individual articula-tors. The point-tracking systems that have been used in the phonetic research include the
magnetometer, the X-ray microbeam, and the Optotrak (an optoelectronic system). The point-tracking systems track movements of individual articulators, including the upper and lower lips and the jaw as well as the tongue and velum, over time during speech production, although the optoelectronic system is limited to an external use (i.e. for tracking the lips and the jaw; see Stone 2010 for a brief summary of each of these systems).
Among the point-tracking systems, the electromagnetic articulograph (EMA), also gen-erally referred to as magnetometer, has been steadily developed and more widely used in recent years than the other two techniques because the former is less costly and more accessible, and provides a more rapid tracking rate compared to the X-ray microbeam system. The EMA system uses alternating electromagnetic fields that are generated from multiple transmitters placed around the subject’s head. The early two-dimensional magnetometer system (e.g. Perkell et al. 1992) used three transmitters, but Carstens’ most recent system (AG 501; see Hoole 2014) uses nine transmitters that provide multi-dimensional articulatory data. In this technique, a number of receiver coils (sensors, as small as 2 × 3 mm) are glued on the articulators (Figure 2.3), usually along the midsagittal plane, but capturing more dimensions is also possible in the three-dimensional systems. Note that the NDI Wave System (Berry 2011) uses sensors containing multiple coils (e.g. Tilsen 2017; Shaw and Kawahara 2018). The basic principle is that the strength of the electromagnetic field in a receiver (sensor) is inversely related to its distance from each of the transmitters around the head at different frequencies. Based on this principle, the system calculates the sensor’s volt-ages at different frequencies and obtains the distances of each sensor from the transmitters, allowing the positions of the sensors to be estimated in the two-dimensional XY or the three-dimensional XYZ coordinate plane plus two angular coordinates (see Zhang et al. 1999 for details of the technical aspects of two-dimensional EMA; Hoole and Zierdt 2010 and Hoole 2014 for EMA systems that use three or more dimensions; and Stone 2010 for a more general discussion on EMA). Given its high temporal and spatial resolution (at a sample rate up to 1,250 Hz in a recent EMA system), an EMA is particularly useful in investigating dynamic aspects of overlapping vocal tract actions that are coactive over time (for quantita-tive analysis of EMA data see Danner et al. 2018; Tomaschek et al. 2018; Wieling 2018). In prosody research, an EMA is particularly useful in the exploration of the phonetics–prosody
Figure 2.3 Lip aperture in electromagnetic articulography. High values indicate that lips are open during vowel production. Trajectories are longer, faster, and more displaced in target words in con-trastive focus (lighter grey lines) compared to out of focus.
(Photo by Fabian Stürtz at IfL Phonetics Lab, Cologne.)
interplay in the case of tone-segment alignment and articulatory signatures of prosody structure. A few examples are provided below. It is worth noting, however, that before the EMA system was widely used, prosodically conditioned dynamic aspects of articulation had been investigated by researchers using the X-ray microbeam (e.g. Browman and Goldstein 1995; de Jong 1995; Erickson 1995) and an optoelectronic device (e.g. Edwards et al. 1991; Beckman and Edwards 1994).
An EMA may be used to investigate the timing relation of tonal gestures with supralaryn-geal articulatory gestures (D’Imperio et al. 2007b). Tone gestures are defined as dynamic movements in f0 space that can be coordinated with articulatory actions of the oral tract within the task dynamics framework (Gao 2009; Mücke et al. 2014). For example, Katsika et al. (2014) showed that boundary tones in Greek are lawfully timed with the vocalic gesture of the pre-boundary (final) syllable, showing an anti-phase (sequential) coupling relationship between the tone and the vocalic gesture in interaction with stress distribution over the phrase-final word (see also Katsika 2016 for related data in Greek). As for the tone-gesture alignment associated with pitch accent, Mücke et al. (2012) reported that Catalan employs an in-phase coupling relation (i.e. roughly simultaneous initiation of gestures) between the tone and the vocalic gesture with a nuclear pitch accent LH. By contrast, the tone–gesture align-ment in a language with a delayed nuclear LH rise, such as German, is more complex (e.g. L and H may compete to be in phase with the vocalic gesture, with in-phase L inducing a delayed peak). Moreover, quite a few studies have shown that tone–segment alignment may be captured better with articulatory gestural landmarks rather than with acoustic ones, in line with a gestural account of tone–segment alignment (e.g. Mücke et al. 2009, 2012; Niemann et al. 2014; see also Gao 2009 for lexical tones in Mandarin). More broadly, an EMA is a device that provides useful information about the nature of tone–segment coordination, allowing for various assumptions of the segmental anchoring hypothesis (see chapter 6).
EMA has also been extensively used to investigate supralaryngeal articulatory character-istics of prosodic strengthening in connection with prosodic structure. Results of EMA studies have shown that articulation is systematically modified by prominence, largely in such a way as to enhance paradigmatic contrast (e.g. Harrington et al. 2000; Cho 2005, 2006a; see de Jong 1995 for similar results obtained with the X-ray microbeam system). On a related point, Cho (2005) showed that strengthening of [i] by means of adjustments of the tongue position manifested different kinematic signatures of the dual function (boundary vs. prominence marking) of prosodic structure (see Tabain 2003 and Tabain and Perrier 2005 for relevant EMA data in French; Mücke and Grice 2014 in German; and Cho et al. 2016 in Korean). A series of EMA studies has also examined the nature of supralaryngeal articula-tion at prosodic junctures, especially in connection with phrase-final (pre-boundary) lengthening (Edwards et al. 1991; Byrd 2000; Byrd et al. 2000, 2006; Byrd and Saltzman 2003; Krivokapić 2007; Byrd and Riggs 2008; Krivokapić and Byrd 2012; Cho et al. 2014b; Katsika 2016). These kinematic data have often been interpreted in terms of gesture types as associated with different aspects of the prosodic structure, like ‘π-gestures’ for slowing down the local tempo at boundaries and ‘μ-gestures’ for temporal and spatial variations under stress and accent. Krivokapić et al. (2017) have extended the analysis of vocal tract actions to manual gestures, combining an EMA with motion capture, and have shown that manual gestures are tightly coordinated with pitch accented syllables and boundaries. Both vocal tract actions and manual gestures (pointing gestures) undergo lengthening under prominence. Scarborough et al. (2009) is an example of the combined use of an EMA and
an optoelectronic system to examine the relationship between articulatory and visual (facial) cues in signalling lexical and phrasal stress in English. Together, these studies have provided insights into the theory of the phonetics–prosody interplay in general, and the dynamic approaches have improved our understanding of the human communicative sound system (e.g. Mücke et al. 2014; see also recent EMA studies on other aspects of speech dynamics, such as Hermes et al. 2017; Pastätter and Pouplier 2017).
2.4.1 Ultrasound
Ultrasound imaging (also referred to as (ultra)sonography) is a non-invasive technique used in phonetic research to produce dynamic images of the sagittal tongue shape, which allows for investigating vocal tract characteristics, tongue shape, and tongue motion over time (for reviews see Gick 2002; Stone 2010). It uses the reflective properties of ultra-high-frequency sound waves (which humans cannot hear) to create images of the inside of the body (Figure 2.4). The high-frequency sound wave penetrates through the soft tissues and fluids, but it bounces back off surfaces or tissues of a different density as well as air. The variation in reflected echoes is processed by computer software and displayed as a video image. It has some limitations, however (see Stone 2010). For example, due to its relatively low sampling rate (generally lower than 90 Hz), it does not allow for the investigation of sophisticated dynamic characteristics of tongue movements (unlike EMA, which provides a sample rate up to 1,250 Hz). In addition, it is generally unable to capture the shape of the tongue tip and the area beyond tissue or air that reflects ultrasound. However, because ultrasound imaging allows researchers to examine real-time detailed lingual postures not easily captured by methods such as EPG and EMA (including the tongue groove and the tongue root; see Lulich et al. 2018), and because some systems are portable and inexpensive (Gick 2002), it has increasingly been used in various phonetic studies (Stone 2010; Carignan 2017; Ahn 2018; Strycharczuk and Sebregts 2018; Tabain and Beare 2018), and also in studies involving young children (Noiray et al. 2013).
Lehnert-LeHouillier et al. (2010) used an ultrasound imaging system to investigate prosodic strengthening as shown in the tongue shape for mid vowels in domain-initial position with different prosodic boundary strengths in English. They found a cumulatively increasing magnitude of tongue lowering as the boundary strength increased for vowels in
Figure 2.4 Tongue shapes in ultrasound.(Photo by Aude Noiray at the Laboratory for Oral Language Acquisition.)
0004898684.INDD 23 11/21/2020 2:36:36 PM
Dictionary: <Dictionary>
OUP UNCORRECTED PROOF – <STAGE>, 21/11/20, SPi
24 TAEHONG CHO AND DORIS Mücke
vowel-initial (VC) syllables, but not in consonant-initial (CVC) syllables. Based on these results, the authors suggested that boundary strengthening is localized in the initial seg-ment, whether consonantal or vocalic. In a related ultrasound study that examined lingual shapes for initial vowels in French, Georgeton et al. (2016) showed that prosodic strength-ening of domain-initial vowels is driven by the interaction between language factors (such as the phonetic distinctiveness in the perceptual vowel space) and physiological constraints imposed on the different tongue shapes across speakers.
In an effort to explore the articulatory nature of the insertion of a schwa-like element into a consonantal sequence, Davidson and Stone (2003) used an ultrasound imaging technique to investigate how phonotactically illegal consonantal sequences may be repaired. Their tongue motion data in the production of /zC/ sequences in pseudo-Polish words suggested that native speakers of English employed different repair mechanisms for the illegal sequences, but in a gradient fashion in line with a gestural account—that is, as a result of the interpolation between the flanking consonants without showing a positional target in the articulatory space (Browman and Goldstein 1992a). It remains to be seen how such a gradi-ent repair may be further conditioned by prosodic structure.
An ultrasound imaging system may be used in conjunction with laryngoscopy. Moisik et al. (2014) employed simultaneous laryngoscopy and laryngeal ultrasound (SLLUS) to examine Mandarin tone production. In SLLUS (see also Esling and Moisik 2012), laryngos-copy is used to obtain real-time video images of the glottal condition, which provide infor-mation about laryngeal state, while laryngeal ultrasound is simultaneously used to record changes in larynx height. Results showed no positive correlation between larynx height and f0 in the production of Mandarin tones except for low f0 tone targets that were found to be accompanied by larynx raising due to laryngeal constriction (as low tone often induces creakiness). This study implies that larynx height may be controlled to help facilitate f0 change, especially under circumstances in which f0 targets may not be fully accomplished (e.g. due to vocal fold inertia). Despite the invasiveness of laryngoscopy, the innovative technique was judged to be particularly useful in exploring the relation between f0 regula-tion and phonation type and their relevance to understanding the production of tones and tonal register targets.
2.4.2 Electropalatography
Electropalatography (EPG) is a technique that allows for monitoring linguo-palatal contact (i.e. contact between the tongue and the hard palate) and its dynamic change over time dur-ing articulation. The subject wears a custom-fabricated artificial palate, usually made of a thin acrylic, held in place by wrapping around the upper teeth (Figure 2.5). Several dozen electrodes are placed on the artificial palate, and the electrodes that are contacted by the tongue during articulation send signals to an external processing unit, indexing details of tongue activity during articulation. Unlike the EMA, which is usually used to track articulatory movements in the midsagittal plane, an EPG records the tongue contact any-where on the entire palate (i.e. the target of the tongue movement). EPG is generally limited to investigating the production of consonants and high vowels, which involve tongue
0004898684.INDD 24 11/21/2020 2:36:36 PM
Dictionary: <Dictionary>
OUP UNCORRECTED PROOF – <STAGE>, 21/11/20, SPi
ARTICULATORY MEASURES OF PROSODY 25
contact between the lateral margins of the tongue and the sides of the palate near the upper molars (Byrd et al. 1995; Gibbon and Nicolaidis 1999; Stone 2010; see Ünal-Logacev et al. 2018 for the use of EPG with an aerodynamic device).
In prosodic research, EPG has been used to measure prosodic strengthening as evi-denced by the production of lingual consonants across languages, in English (Fougeron and Keating 1997; Cho and Keating 2009), French (Fougeron 2001), Korean (Cho and Keating 2001), Taiwanese (Hsu and Jun 1998), and German (Bombien et al. 2010) (see Keating et al. 2003 for a comparison of four languages). Results of these studies have shown that consonants have a greater degree of linguo-palatal contact when they occur in initial position of a higher prosodic domain and when they occur in stressed and accented syllables. These studies have provided converging evidence that low-level articulatory realization of consonants is fine tuned by prosodic structure, and as a consequence prosodic structure itself is expressed at least to some extent by systematic low-level articulatory variation of consonantal strengthening, roughly proportional to prosodic strengthening as stemming from the prosodic structure.
EPG has also been used in exploring effects of syllable position and prosodic boundaries on articulation of consonantal clusters (Byrd 1996; Bombien et al. 2010). Byrd (1996), for example, showed that a consonant in English is spatially reduced (with lesser linguo-palatal contact) in the coda as compared to the same consonant in the onset, and that an onset cluster (e.g. /sk/) overlaps less than a coda or a heterosyllabic cluster of the same segmental make-up. In relation to this study, Bombien et al. (2010) examined articulation of initial consonant clusters in German and reported that at least for some initial clusters (e.g. /kl/ and /kn/), boundary strength induced spatial and/or temporal expansion of the initial con-sonant, whereas stress caused temporal expansion of the second consonant, and both resulted in less overlap. These two studies together imply that coordination of consonantal gestures in clusters is affected by elements of prosodic structure, such as syllable structure, stress, and prosodic boundaries.
/t/ /s/
/k/ /ç/
Figure 2.5 Contact profiles in electropalatography for different stops and fricatives. Black squares indicate the contact of the tongue surface with the palate (upper rows = alveolar articulation, lower row = velar articulation).
(Photo taken at IfL Phonetics Lab, Cologne.)
0004898684.INDD 25 11/21/2020 2:36:37 PM
Dictionary: <Dictionary>
OUP UNCORRECTED PROOF – <STAGE>, 21/11/20, SPi
26 TAEHONG CHO AND DORIS Mücke
2.5 Summary of articulatory measurement techniques
Table 2.1 summarizes advantages and disadvantages of several articulatory measurement techniques that we have introduced in this chapter.
Table 2.1 Advantages and disadvantages of articulatory measuring techniques
Devices Advantages Disadvantages
Laryngoscopyprovides high-speed motion pictures of larynx activities
Direct observation of vocal fold vibrations; drawing inferences about different glottal states and activity of the vocalis muscle
Relatively invasive; not ideal for obtaining a large quantity of data from a large population
EGGmonitors glottal states using electrical impedance
Non-invasive; relatively easy to handle for monitoring vocal fold vibration and different glottal states during phonation with a larger subject pool
Not as accurate as laryngoscopy; indirect observation of larynx activity estimated by a change in electrical impedance across the larynx
RIPestimates lung volume change during speech by measuring expansion and recoil of elastic bands wrapped around the thoracic and abdominal cavities
Non-invasive device; useful for testing the relationship between the respiratory process and global prosodic planning in conjunction with acoustic data
Difficult to capture fine detail at the segmental level; useful for testing prosodic effects associated with large prosodic constituents, such as an intonational phrase
EMAtracks positions and movements of sensors that are attached on several articulators (e.g. tongue, lips, jaw, velum), within an electromagnetic field
Data obtained at high sampling rates; useful for examining kinematics for both consonants and vowels; simultaneous observation of multiple articulators; high temporal and spatial resolution; applicable to manual and facial movements in prosody; three-dimensional devices available; used for recording a larger population with a recent device
Limited to point tracking; usually used to track movements of points along the midsagittal plane; difficult to capture the surface structure (e.g. the complete shape of the tongue); can only be used to observe the anterior vocal tract; quite invasive with sensors; possibly impedes natural articulation
Ultrasoundprovides imaging of tongue position and movement
Data obtained at relatively high sampling rates (though not as high as a recent EMA system); generally non-invasive; used for measuring the real-time lingual postures during vowel and consonant production; a portable device is available
Difficult to image the tongue tip (usually about 1 cm), some parts of the vocal tract (e.g. the palatal and pharyngeal wall), and the bony articulator (e.g. the jaw); because tongue contours are tracked relative to probe position, it is difficult to align them to a consistent reference, such as hard palate structure
0004898684.INDD 26 11/21/2020 2:36:37 PM
Dictionary: <Dictionary>
OUP UNCORRECTED PROOF – <STAGE>, 21/11/20, SPi
ARTICULATORY MEASURES OF PROSODY 27
2.6 Conclusion
Prosody research has benefited from various experimental techniques over the past decades and has as a result extended in its scope, embracing investigations of various phonetic events that occur in both laryngeal and supralaryngeal speech events in the interplay between segmental phonetics and prosodic structure. Careful examination of fine-grained phonetic detail that arises in the phonetics–prosody interplay has no doubt advanced our understanding of the regulatory mechanisms and principles that underlie the dynamics of articulation and prosody. Prosodic features and segmental features are not implemented independently. The coordination between the two systems is modulated in part by univer-sally driven physiological and biomechanical constraints imposed on them, but also in part by linguistic and communicative factors that must be internalized in the phonological and phonetic grammars in language-specific ways. Future prosody research with the help of articulatory measuring techniques of the type introduced in this chapter will undoubtedly continue to uncover the physiological and cognitive underpinnings of the articulation of prosody, illuminating the universal versus language-specific nature of prosody that under-lies human speech.
Devices Advantages Disadvantages
EPGmeasures linguo-palatal contact patterns using individual artificial palates with incorporated touch-sensitive electrodes
Useful for examining linguo-palatal contact patterns along the parts of the palate over time, especially for coronal consonants; provides information about the width, length, and curvature of constriction made along the palatal region
Custom-made artificial palates needed for individual subjects; often impedes natural speech; restricted to complete tongue– palate contacts; no information about articulation beyond the linguo-palatal contact (e.g. non-high vowels, labials, velars)
Optoelectronic device (Optotrak)provides three- dimensional motion data (using optical measurement tech-niques) with active markers placed on surfaces
Non-invasive; data with high temporal and spatial resolution; useful for capturing movements of some articulators (jaw, chin, lips) as well as all other visible gestures (e.g. face, head, hand)
Limited to point tracking; limited to external use (i.e. to ‘visible’ movements); impossible to capture articulation inside the vocal tract; the system requires line of sight to track the markers
Real-time MRIprovides real-time imaging data of complex structures and articulatory move-ments of the vocal tract using strong magnetic fields
Non-invasive with high spatial resolution; useful for rich anatomical data analysis as well as for analysis of articulatory movements of the entire vocal tract (including the pharyngeal area)
Relatively poor time resolution; subjects are recorded in supine posture, which may influence speech; concurrent audio from production is very difficult to acquire because of scanner noise
0004898684.INDD 27 11/21/2020 2:36:37 PM
Dictionary: <Dictionary>
OUP UNCORRECTED PROOF – <STAGE>, 21/11/20, SPi
28 TAEHONG CHO AND DORIS Mücke
Acknowledgements
This work was supported in part by the Global Research Network programme through the Ministry of Education of the Republic of Korea and the National Research Foundation of Korea (NRF-2016S1A2A2912410), and in part by the German Research Foundation as an aspect of SFB 1252 ‘Prominence in Language’ in the project A04 ‘Dynamic Modelling of Prosodic Prominence’ at the University of Cologne.
0004898684.INDD 28 11/21/2020 2:36:37 PM
1
References
Ahn, S. (2018). The role of tongue position in laryngeal contrasts: An ultrasound study of English and
Brazilian Portuguese. Journal of Phonetics 71: 451-467.
Arvaniti, A. and D.R. Ladd (2009). Greek wh-questions and the phonology of intonation. Phonology
26: 43-74.
Baken, P. J. and R. F. Orlikoff (2000). Clinical Measurement of Speech and Voice. San Diego: Singular
Publishing Group.
Beckman, M. E. (1996). The parsing of prosody. Language and Cognitive Processes 11: 17–67.
Berry, J. (2011). Accuracy of the NDI wave speech research system. Journal of Speech, Language, and
Hearing Research 54(5): 1295-1301.
Bombien, L., C. Mooshammer, P. Hoole and B. Kühnert (2010). Prosodic and segmental effects on
EPG contact patterns of word-initial German clusters. Journal of Phonetics 38: 388–403.
Browman, C. and L. Goldstein (1992). “Targetless” schwa: an articulatory analysis. D.R. Ladd and G.
Docherty (eds), Papers in Laboratory Phonology II. CUP. 26-67.
Browman, C. and L. Goldstein (1995). Gestural syllable position effects in American English. In F.
Bell-Berti and L. J. Raphael (Eds., Producing Speech: Contemporary Issues. For Katherine Safford
Harris. AIP Press: Woodbury, NY.
Byrd, D. (1996). Influences on articulatory timing in consonant sequences. Journal of Phonetics 24:
209-244.
Byrd, D. (2000). Articulatory vowel lengthening and coordination at phrasal junctures. Phonetica 57:
3–16.
Byrd, D., A. Kaun, S. Narayanan and E. Saltzman (2000). Phrasal signatures in articulation. In M. Broe
and J. Pierrehumbert (eds.), Papers in laboratory phonolgy V: Acquisition and the Lexicon.
Cambridge, UK: Cambridge University Press, 70–88.
Byrd, D. and D. Riggs (2008). Locality interactions with prominence in determining the scope of phrasal
lengthening. Journal of the International Phonetic Association 38: 187-202.
Byrd, D. and E. Saltzman (2003). The elastic phrase: modeling the dynamics of boundary-adjacent
lengthening. Journal of Phonetics 31: 149–180.
Byrd, D., Flemming, E., Mueller, C.A. and Tan, C.C. (1995). Using regions and indices in EPG data
reduction. Journal of Speech and Hearing Research 38: 821-827.
Carignan, C. (2017). Covariation of nasalization, tongue height, and breathiness in the realization of F1
of Southern French nasal vowels. Journal of Phonetics 63: 87-105.
Cho, T. (2004). Prosodically-conditioned strengthening and vowel-to-vowel coarticulation in English.
Journal of Phonetics 32: 141–76.
Cho, T. (2005). Prosodic strengthening and featural enhancement: evidence from acoustic and
articulatory realizations of /a,i/ in English. Journal of the Acoustical Society of America 117(6):
3867–78.
Cho, T. (2006). Manifestation of prosodic structure in articulation: evidence from lip kinematics in
English. In L. M. Goldstein, D. H. Whalen and C. T. Best (eds), Laboratory phonology 8: Varieties
of phonological competence. Berlin: Mouton de Gruyter, 519–48.
Cho, T. (2008). Prosodic strengthening in transboundary V-to-V lingual movement in American
English. Phonetica 65: 45–61.
Cho, T. (2016). Prosodic boundary strengthening in the phonetics-prosody interface. Language and
Linguistics Compass 10(3): 120-141.
2
Cho, T. and S.-A. Jun (2000). Domain-initial strengthening as featural enhancement: Aerodynamic
evidence from Korean. Chicago Linguistics Society 36: 31–44.
Cho, T., S.-A. Jun, and P. Ladefoged (2002). Acoustic and aerodynamic correlates of Korean stops and
fricatives. Journal of Phonetics 30: 193–228.
Cho, T. and P. Keating (2001). Articulatory and acoustic studies of domain-initial strengthening in
Korean. Journal of Phonetics 29: 155–90.
Cho, T. and P. Keating (2009). Effects of initial position versus prominence in English. Journal of
Phonetics 37: 466–85.
Cho, T., D. Kim and S. Kim (2017). Prosodically-conditioned fine-tuning of coarticulatory vowel
nasalization in English. Journal of Phonetics, 64,71-89.
Cho, T., Y. Lee and S. Kim (2014a). Prosodic strengthening on the /s/-stop cluster and the phonetic
implementation of an allophonic rule in English. Journal of Phonetics 46: 128–46.
Cho, T., M. Son & S. Kim (2016). Articulatory reflexes of the three-way contrast in labial stops and
kinematic evidence for domain-initial strengthening in Korean. Journal of the International
Phonetic Association 26: 129-155.
Cho, T., Y. Yoon and S. Kim (2014b). Effects of prosodic boundary and syllable structure on the
temporal realization of CV gestures in Korean. Journal of Phonetics 44: 96-109.
Cooper, W. E., and Sorenson, J. M. (1981). Fundamental frequency in sentence production. New York:
Springer-Verlag.
Demolin, D. (2011). Aerodynamic techniques for phonetic fieldwork. Proceedings of the 17th
International Congress of Phonetic Sciences. Hong Kong, 84-87.
D' Imperio, M., Espesser, R., Loevenbruck, H., Menezes, C., Nguyen, N., and Welby, P. (2007). Are
tones aligned with articulatory events? Evidence from Italian and French. In J. Cole, and J. I. Hualde
(eds), Laboratory phonology 9 (phonology and phonetics). Berlin, Germany: Walter de Gruyter,
577–608.
Danner, S.G., A.V. Barbosa and L. Goldstein (2018).Quantitative analysis of multimodal speech data
Journal of Phonetics 71: 268-283.
Davidson, L. and M. Stone (2003). Epenthesis versus gestural mistiming in consonant cluster
production: an ultrasound study. Proc. of the West Coast Conference on Formal Linguistics 22,
165-178.
de Jong K. (1995). The supraglottal articulation of prominence in English: linguistic stress as localized
hyperarticulation. Journal of the Acoustical Society of America 97: 491–504.
Di Napoli, J. (2015). Glottalization at phrase boundaries in Tuscan and Roman Italian. In J. Romero
and M. Riera (eds), The Phonetics/Phonology Interface: Sounds, representations, methodologies.
Amsterdam: John Benjamins, 125–148.
Dilley, L., S. Shattuck-Hufnagel and M. Ostendorf (1996). Glottalization of word-initial vowels as a
function of prosodic structure. Journal of Phonetics 24: 423-444.
Edmondson, J. A. and J. H. Esling (2006). The valves of the throat and their functioning in tone, vocal
register and stress: laryngoscopic case studies, Phonology 23: 157–191.
Edwards, J.E., M.E. Beckman and J. Fletcher (1991). The articulatory kinematics of final lengthening.
Journal of the Acoustical Society of America 89: 369–82.
Esling J.H. and J.H. Harris (2005). States of the glottis: An articulatory phonetic model based on
laryngoscopic observations. In Wj. J. Hardcastle and J. Mackenzie Beck (eds), A Figure of Speech:
A Festschrift for John Laver, LEA: New Jersey, 345-383.
Esling, J.H. and S.R. Moisik. (2012). Laryngeal aperture in relation to larynx height change: An analysis
using simultaneous laryngoscopy and laryngeal ultrasound. In D. Gibbon, D. Hirst and N.
Campbell (eds), Rhythm, melody and harmony in speech: Studies in honour of Wiktor Jassem
3
[Speech and Language Technology, 14/15], Poznań: Polskie Towarzystwo Fonetyczne, 117-127.
Fletcher, J. 2010. The prosody of speech: timing and rhythm. In W.J. Hardcastle, J. Laver and F.E.
Gibbon (eds), The Handbook of Phonetic Sciences, second edition. Oxford: Blackwell, 523–602.
Fougeron, C. (1999). Prosodically conditioned articulatory variations: a review. UCLA Working Papers
in Phonetics 97: 1–74.
Fougeron, C. (2001). Articulatory properties of initial segments in several prosodic constituents in
French. Journal of Phonetics 29: 109–35.
Fouigeron C., and P. A. Keating (1996). Variations in velic and lingual articulation depending on
prosodic position: Results for 2 French speakers. UCLA Working Papers in Phonetics 92: 88-96.
Fouigeron C., and P. A. Keating (1997). Articulatory strengthening at edges of prosodic domains.
Journal of the Acoustical Society of America 106: 3728–40.
Fuchs, S., C. Petrone, J. Krivokapić and P. Hoole (2013). Acoustic and respiratory evidence for
utterance planning in German. Journal of Phonetics 41: 29-47.
Fuchs, S., C. Petrone, A. Rochet-Capellan, U.D. Reichel, and L.L. Koenig (2015). Assessing respiratory
contributions to f0 declination in German across varying speech tasks and respiratory demands.
Journal of Phonetics 52: 35-45.
Garellek, M. (2014). Voice quality strengthening and glottalization. Journal of Phonetics 45: 106-113
Gao, M. (2009). Gestural Coordination among Vowel, Consonant and Tone Gestures in Mandarin
Chinese. Chinese Journal of Phonetics 2: 43-50.
Gelfer, C.E., K.S. Harris and T. Baer (1987). Controlled variables in sentence intonation. In T. Baer, C.
Sasaki, and K.S. Harris (eds.), Laryngeal function in phonation and respiration. Little Brown,
Boston, MA, 422-435.
Georgeton, L., T. Kocjančič Antolík and C. Fougeron (2016). Effect of Domain Initial Strengthening
on Vowel Height and Backness Contrasts in French: Acoustic and Ultrasound Data. Journal of
Speech, Language, and Hearing Research 59: 1575-1586.
Gibbon, F. E., Nicolaidis, K. (1999). Palatography. In W. J. Hardcastle, N. Hewlet (eds.) Coarticulation:
theory, data and techniques. Cambridge: CUP, 229–245.
Gick, B. (2002). The use of ultrasound for linguistic phonetic fieldwork. Journal of the International
Phonetic Association 32: 113-121.
Goldstein, L., Chitoran, I., and Selkirk, E. (2007). Syllable structure as coupled oscillator modes:
Evidence from Georgian vs. Tashlhiyt Berber. Proceedings of the XVI International Congress of
Phonetic Sciences. Saarbrücken, Germany, 241–244.
Gordon, M. (1997). Phonetic correlates of stress and the prosodic hierarchy in Estonian, 1997, in J.
Ross, I. Lehiste (eds), Estonian Prosody: Papers from a Symposium. Tallinn: Institute of Estonian
Language,100-24.
Harrington, J., J. Fletcher, and M. Beckman (2000). Manner and place conflicts in the articulation of
accent in Australian English. In M. Broe & J. Pierrehumbert (eds), Papers in Laboratory
Phonology V: Acquisition and the Lexicon. Cambridge, UK: CUP, 40–55.
Hermes, A., D. Mücke, and A. Bastian (2017). The variability of syllable patterns in Tashlhiyt Berber
and Polish. Journal of Phonetics 64: 127-144.
Hirose, H. (2010). Investigating the physiology of laryngeal structures. . In William J. Hardcastle, John
Laver, and Fiona E. Gibbon (eds), The Handbook of Phonetic Sciences, 2nd Edition. Wiley-
Blackwell, 130-152.
Hixon, T. and Hoit, J. (2005). Evaluation and management of speech breathing disorders: Principles
and methods. Redington Brown.
4
Hoole, P.and H. Fang (2004). Tone-vowel interaction in standard Chinese. In Proc. Intl. symposium on
tonal aspects of languages. Beijing: ISCA. http://www.phonetik.uni-
muenchen.de/~hoole/pdf/hoole_tal.pdf.
Hoole, P. & Zierdt, A. (2010). Five-dimensional articulography. In B. Maassen, P. van Lieshout (eds.),
Speech Motor Control: New developments in basic and applied research. OUP, 331-349
Hsu, C.-S. and S.-A. Jun (1998). Prosodic strengthening in Taiwanese: Syntagmatic or paradigmatic?
UCLA Working Papers in Phonetics 96: 69–89.
Jun, S.-A, M. Beckman and H.-J. Lee (1998). Fiberscopic evidence for the influence on vowel devoicing
of the glottal configurations for Korean obstruents. UCLA Working Papers in Phonetics 96: 43-68
Katsika A., J. Krivokapić, C. Mooshammer C., M. Tiede and L. Goldstein (2014). The coordination of
boundary tones and their interaction with prominence. Journal of Phonetics 44: 62–82.
Katsika A., (2016). The role of prominence in determining the scope of boundary-related lengthening
in Greek. Journal of Phonetics 55:149-181.
Keating, P., T. Cho, C. Fougeron and C.-S. Hsu (2003). Domain-initial strengthening in four languages.
In J. Local, R. Ogden, and R. Temple (eds), Papers in laboratory phonology 6: Phonetic
interpretations, Cambridge, UK: Cambridge University Press, 145–63.
Keating, P. (2006). Phonetic Encoding of Prosodic Structure. In J. Harrington & M. Tabain (eds),
Speech production: Models, phonetic processes, and techniques, Macquarie Monographs in
Cognitive Science. Psychology Press, New York and Hove, 167-186.
Krivokapić, J., and D. Byrd (2012). Prosodic boundary strength: an articulatory and perceptual
study. Journal of Phonetics 40: 430–42.
Krivokapić, J., M. Tiede and M. Tyrone (2017). A kinematic study of prosodic structure in articulatory
and manual gestures: Results from a novel method of data collection. Laboratory Phonology 8(1):
3, 1–26, DOI: https://doi.org/10.5334/labphon.75.
Ladd, D.R. (2008). Intonational Phonology. Cambridge University Press.
Ladefoged, P. (1967). Three Area of Experimental Phonetics. Oxford: Oxford University Press.
Ladefoged, P. (2003). Phonetic Data Analysis: An Introduction to Fieldwork and Instrumental
Techniques. Wiley-Blackwell.
Ladefoged, P. And N.P. McKinney 1963. Loudness, Sound Pressure, and Subglottal Pressure in Speech.
Journal of the Acoustical Society of America 35: 454-460.
Lancia, L., D. Voigt and G. Krasovitskiy (2016). Characterization of laryngealization as irregular vocal
fold vibration and interaction with prosodic prominence. Journal of Phonetics 54: 80-97.
Laver, J. (1980) The Phonetic Description of Voice Quality. Cambridge University Press, Cambridge.
Lehnert-LeHouillier, H., McDonough, J., and McAleavey, S. (2010). Prosodic strengthening in
American English domain-initial vowels. Proceedings of Speech Prosody 2010.
