Changes in speech prosodic characteristics in people with Parkinson’s disease after a singing intervention

(1)

Katja Marianne Hyppönen

Changes in speech prosodic characteristics in people with Parkinson’s disease after a singing intervention

University of Eastern Finland

Philosophical Faculty

Master’s thesis

March 2020

(2)

ITÄ-SUOMEN YLIOPISTO – UNIVERSITY OF EASTERN FINLAND

Tiedekunta – Faculty

Filosofinen tiedekunta Osasto – School Humanistinen osasto Tekijät – Author

Katja Marianne Hyppönen Työn nimi – Title

Puheen prosodiapiirteiden muutoksia Parkinsonin tautia sairastavilla lauluintervention jälkeen

Pääaine – Main subject

Yleinen kielitiede

Työn laji – Level Päivämäärä – Date

9.3.2020 Sivumäärä – Number of pages

61 Pro gradu -tutkielma X

Sivuainetutkielma Kandidaatin tutkielma Aineopintojen tutkielma Tiivistelmä – Abstract

Parkinsonin tauti on etenevä neurologinen sairaus, jonka pääoireita ovat liikkeiden hidastuminen, vapina ja lihasjäykkyys. Siihen liittyy usein myös ääni- ja puheoireita, joita kutsutaan hypokineettiseksi dysartriaksi. Sen tyypillisiin oireisiin kuuluu äänen voimakkuuden pieneneminen, monotonisuus sekä epäselvä artikulaatio. Tässä pro gradu -tutkielmassa on tarkoitus selvittää, voiko ryhmämuotoisista laulullisista menetelmistä olla hyötyä

Parkinsonin tautiin liittyvien äänioireiden kuntoutuksessa.

Tämä tutkielma on osa Tampereen yliopiston monitieteistä Kuuluva ääni -hanketta, joka pyrkii keräämään lisää tietoa Parkinsonin taudin ääni- ja kommunikaatiokuntoutuksesta. Hankkeeseen kuuluvaan laululliseen

kuntoutukseen osallistui 15 Parkinsonin tautia sairastavaa henkilöä. Interventio koostui tapaamisista (8 x 90 min), jotka sisälsivät laulullisia harjoituksia ja puheterapeuttisten menetelmien piirteitä (esim. Lee Silverman Voice Treatment).

Osallistujilta tallennettiin luentanäytteet ennen ja jälkeen intervention normaalissa ja emotionaalisessa luennassa.

Tallenteista analysoitiin perusäänentaajuus ja sen keskihajonta (f0, f0SD), harmonisuus, intensiteetti, äänen energian alle 1kHz oleva osuus, puhe- ja artikulaationopeus sekä nPVI. Kutakin muuttujaa tarkasteltiin ennen ja jälkeen intervention sekä normaalin ja emotionaalisen luennan välisen eron suhteen. Intervention jälkeen havaittiin tilastollisesti merkitseviä muutoksia miesosallistujilla äänen perustaajuudessa f0 emotionaalisessa luennassa sekä äänienergian alle 1kHz jäävässä osuudessa prosodisessa kapasiteetissa. Lisäksi koko ryhmällä havaittiin tilastollisesti merkitsevä muutos äänen energian alle 1kHz jäävässä osuudessa normaalissa ja emotionaalisessa luennassa sekä prosodisessa kapasiteetissa.

Tutkimustulosten perusteella laululliset menetelmät vaikuttavat hyödyllisiltä Parkinsonin tautia sairastavien äänikuntoutuksessa. Käytetty tutkimusasetelma tarjoaa tietoa vain intervention jälkeisistä muutoksista. Tästä syystä tarvitaan lisätutkimuksia, jotka sisältävät kontrolliryhmän, jotta saataisiin lisätietoa menetelmän

vaikuttavuudesta. Samoin tarvitaan tietoa terveiden verrokkien suoriutumisesta vastaavista tehtävistä viitearvojen saamiseksi.

Avainsanat – Keywords

Parkinsonin tauti, hypokineettinen dysartria, äänihäiriö, ryhmäkuntoutus, ääniterapia, laulu, prosodia

(3)

ITÄ-SUOMEN YLIOPISTO – UNIVERSITY OF EASTERN FINLAND

Tiedekunta – Faculty Philosophical Faculty

Osasto – School School of Humanities Tekijät – Author

Katja Marianne Hyppönen Työn nimi – Title

Changes in speech prosodic characteristics in people with Parkinson’s disease after a singing intervention Pääaine – Main

subject

General Linguistics

Työn laji – Level Päivämäärä – Date 9.3.2020

Sivumäärä – Number of pages

61 Pro gradu -tutkielma X

Sivuainetutkielma Kandidaatin tutkielma Aineopintojen tutkielma Tiivistelmä – Abstract

Parkinson’s disease is a progressive neurological disorder. Its main symptoms include tremor, muscular rigidity and bradykinesia but often voice and speech symptoms known as hypokinetic dysarthria are also present. Its typical symptoms include voice monotony, low voice intensity and imprecise articulation. Aim of this master’s thesis is to find out if it is possible to relieve the speech and voice symptoms caused by PD with the help of a singing intervention.

This master’s thesis is part of a multidisciplinary project “Kuuluva ääni” organized by Tampere University which aims to obtain information about voice and communication rehabilitation in PD. 15 individuals with idiopathic PD were selected to the singing intervention. The intervention consisted of sessions (8 x 90 minutes) with features of singing rehearsals and speech and voice rehabilitation methods like Lee Silverman Voice Treatment.

Speech samples were collected from reading tasks from each participant before and after the intervention in normal and emotional conditions. Fundamental frequency f0 and its variation f0SD, mean harmonicity, intensity, percentage of voice energy below 1kHz, total speech rate, articulation rate and nPVI were measured from each sample. Each variable was analysed at baseline and post-intervention. In addition, the difference between the normal and emotional conditions was measured from each variable. After the intervention statistically significant changes were observed in fundamental frequency f0 in emotional condition and in percentage of vocal energy below 1kHz in prosodic capacity in the subgroup of males. A statistically significant change was also observed in percentage of vocal energy below 1kHz in normal and emotional conditions and in prosodic capacity in whole group.

Based on the current findings singing interventions seem to be potential in speech and voice rehabilitation in PD.

Because this study was a single group repeated measures study, we can examine only the changes occurring after the intervention. A study with a control group is needed to confirm the effectivity of singing interventions. More information is also needed about performance of healthy controls in similar tasks in order to collect reference values.

Avainsanat – Keywords

Parkinson’s disease, hypokinetic dysarthria, voice disorder, group rehabilitation, speech and voice therapy, singing, prosody

(4)

TABLE OF CONTENTS:

1. Introduction 1

1.1 Typical voice-related symptoms of Parkinson’s disease 2

1.1.1 Monotony of pitch 3

1.1.1.1 Reduced fundamental frequency f0 and its variability 3

1.1.1.2 Reduced mean harmonicity 3

1.1.2 Reduced speech intensity 4

1.1.2.1 Speech intensity 4

1.1.2.2 Energy below 1 kHz 4

1.1.3 Changes in speech rate and rhythm 5

1.1.3.1 Total speech rate and articulation rate 5

1.1.3.2 Normalized pairwise variability index nPVI 5

1.2 Speech therapy and singing interventions in Parkinson’s disease 6

1.2.1 Common ground between language and music 6

1.2.2 Lee Silverman Voice Treatment 7

1.2.3 Singing interventions in Parkinson’s disease 7

2. Aim of the study and research questions 9

3. Method 10

3.1. Participants 10

3.2. Intervention 11

3.3. Data collection and outcome measures 12

3.4. Ethical considerations 13

3.5. Statistical analysis 13

4. Results 14

4.1. Monotony of pitch 14

4.1.1. Fundamental frequency f0 and its variation f0SD 14

4.1.2. Reduced mean harmonicity 16

4.2. Intensity features of speech 19

4.2.1. Speech intensity 19

4.2.2. Percentage of vocal energy below 1 kHz 21

4.3. Speech rate and rhythm 24

4.3.1. Total speech rate 24

4.3.2. Articulation rate 26

4.3.3. Normalized pairwise variability index nPVI 29

(5)

4.4. Conclusions 32

5. Discussion 33

5.1. Statistically significant findings 33

5.1.1. Fundamental frequency f0 33

5.1.2. Energy below 1 kHz 34

5.2. Statistically non-significant findings 35

5.2.1. Standard deviation of fundamental frequency f0SD 35

5.2.2. Reduced mean harmonicity 35

5.2.3. Speech intensity 36

5.2.4. Total speech rate, articulation rate 37

5.2.5. Normalized pairwise variability index nPVI 37

5.3. Methodological considerations 38

5.3.1. Participants and intervention 38

5.3.2. Data collection and outcome measures 39

5.3.3. Clinical implications and future research 39

6. References 41

Appendix 1 – Results of fundamental frequency f0 and its variation f0SD 51

Appendix 2- Results of mean harmonicity 52

Appendix 3 – Results of speech intensity 53

Appendix 4 – Results of percentage of vocal energy below 1kHz 54 Appendix 5 - Results of total speech rate and articulation rate 55 Appendix 6 – Results of normalized pairwise variability index nPVI 56

(6)

1 1. Introduction

Parkinson’s disease (PD) is a chronic, progressive neurodegenerative disease (Tarnanen, Pekkonen, & Atula, 2019; Hornykiewicz, 1998). In PD, dopaminergic neurons in substantia nigra gradually break down or die.

The reason for this is unknown. The loss of dopaminergic neurons leads to lack of neurotransmitter dopamine and damage of neural pathways controlling volitional movements. PD may also affect the basal ganglia and fronto-striatal networks, which participate in the preparation of internally generated

movements (Cunnington, Iansek, Bradshaw & Phillips, 1995; Morris, 2000). These pathologies are related to various motor and non-motor symptoms which often reduce the quality of life (Chaudhuri, Healy &

Schapira, 2006; Dickson, 2012). Typical motor symptoms in PD include stiffness, postural instability and tremor. Typical non-motor symptoms include depression, cognitive decline and mood and sleep disorders.

Impairment of speech and voice is also one of the typical symptoms (Hartelius & Svensson, 1994). This deficit is known as hypokinetic dysarthria and it affects approximately 70% of people with PD.

Speech and language therapy is often administered in order to ameliorate the voice and speech deficits caused by PD. As PD is a progressive neurological disease, it is often difficult to maintain the gains of speech and language therapy. Lee Silverman Voice Treatment is an example of voice and speech rehabilitation method used in PD with long term positive effects. It has been found effective up to 2 years in improving speech intensity (Ramig et al., 2001). Speech and voice rehabilitation may either target the impairments, e.g. dysprosody or quiet voice or it may aim to seek alternative or compensatory ways to make understandability as high as possible (Miller, 2017). In addition, in speech and voice rehabilitation it is important to educate the person with PD and their close relatives about ideal communicative environment, possible changes in communication due to PD and how to anticipate and manage them. To facilitate this, a person who has received a PD diagnosis, should be referred to a speech and voice therapeutist as soon as possible. However, due to scarcity of resources there may be a delay in it and a person with PD may be referred to a speech and language therapeutist only after significant deficits in speech and voice are present (Paronen & Vuomajoki, 2019).

Typical treatments of PD include dopaminergic medication and deep brain stimulation (DBS) (Miller, 2017). Although dopaminergic treatments have positive effects in limb movements, there are studies, where a positive influence in individual voice-speech parameters has been found but the effects have not transferred to improved communication or speech intelligibility (Brabenec, Mekyska, Galaz & Rektorova, 2017; Elfmarkova et al., 2016; Ho et al., 2008; Rusz et al., 2013b; Rusz, Tykalova & Klempir, 2016).

Concerning DBS, it has been observed that while individual speech motor or acoustic variables may

(7)

2

improve, no positive effect in communication or speech intelligibility has been observed (Wertheimer et al., 2014). Due to these findings, other modes of treatment and rehabilitation should be examined further to find ways to improve speech and language deficits and communication in people with PD.

1.1 Typical voice-related symptoms in Parkinson’s disease

Typical voice-related symptoms in PD are called hypokinetic dysarthria (Skodda, Rinsche & Schlegel, 2009;

Skodda, Visser & Schlegel, 2011). Its characteristics include hypohonia, bradykinesia and akinesia.

Prominent features of hypophonia include monotone voice, hoarse voice quality and reduced vocal

loudness. As a result of this, a person with PD may not be able to reach full voice intensity and their speech intelligibility may decrease (Miller, 2017). Bradykinesia in turn refers to deceleration of articulatory

movements while akinesia is manifested as delays in movement initiation. These may be manifested as imprecise articulations and bursts of rapid speech (Darley, Aronson & Brown, 1975; Duffy, 2004).

Dysprosody refers to deficits in speech signal to mediate intonational and affective information (Ma, Schneider, Hoffmann & Storch, 2015). It is a typical symptom of hypokinetic dysarthria and one of the core disruptions observed in speech in people with PD (Miller, 2017; Samra et al., 1969; Skodda et al., 2009). The above-mentioned symptoms of hypokinetic dysarthria and how they are manifested may have direct and indirect effects on speech prosody (De Bodt, Hernández-Dı ́az Huici & Heyning, 2002;

Klopfenstein, 2009). For instance, hypophonia may be manifested as flattened intonations in utterances, decreased contrast between unstressed and stressed syllables and lowered voice intensity (Ma et al., 2015). As a result of this, speech intelligibility may decrease as a person with PD has challenges in expressing their wishes in a comprehensible way while their interlocutor may have challenges in understanding and hearing what is being said as the voice intensity may be low and a lack of contrast between stressed and unstressed syllables may be present. This example shows clearly how PD may impair and deteriorate communication (Miller, 2017).

PD may have a general negative influence on communication already in the early stage of disease while the changes in speech and voice are still mild and speech intelligibility is rather preserved (Miller, Noble, Jones, Allcock & Burn, 2008; Schalling, Johansson, & Hartelius, 2018). This may be due to cognitive decline or a mood disorder. PD may also impair the pragmatic part of communication as people with PD often have difficulties in understanding metaphors or irony or recognizing true intentions of a speaker. In addition, individuals may lack facial expressions while speaking or expressing emotions as muscle rigidity affects muscles involved in facial expressions (LaPointe, Murdoch, & Stierwalt, 2010). Lack of facial gestures may cause further challenges to communications as the interlocutors may think the person with PD is uninterested because there are no facial gestures visible (Miller, 2017).

(8)

3 1.1.1 Monotony of pitch

1.1.1.1 Reduced fundamental frequency f0 and its variability

Reduced variability of fundamental frequency f0 is a typical finding in PD patients compared to healthy controls (Metter & Hanson, 1986; Goberman, 2005; Gamboa et al., 1997). This feature is also known as monopitch and it is manifested as speech with flattened intonation (Anand & Stepp, 2015; Miller, 2017).

This in turn may decrease speech intelligibility and impair the ability produce stress and emotional intonation in speech (Rusz, Cmejla, Ruzickova, & Ruzicka, 2011). There may be gender-specific differences in f0 variability as decrease in f0 variability may be more intense in females with PD (Skodda et al., 2009;

Skodda et al., 2011). Dopaminergic medications may increase fundamental frequency f0 variability in people with PD (Bowen, Hands, Pradhan & Stepp, 2013). However, there are also findings where dopaminergic medication has been found ineffective in treating voice deficits in PD (Miller, 2017).

Monotony of pitch in PD is a typical symptom of dysprosody in PD (Yorkston, Miller & Strand, 2004).

However, its underlying pathophysiology is still unclear (Skodda et al, 2011). Earlier research has found support for hypothesis suggesting that L-dopa-responsive laryngeal muscle hypokinesia may be the main mechanism behind it (Azevedo, Cardoso, Reis & Azevedo, 2003; Holmes, Oates, Phyland & Hughes, 2000;

Skodda et al., 2009). However, there are also findings which do not support this hypothesis (Goberman, Coelho & Robb, 2002). Variation in severity of dysprosody in people with PD may offer a partial explanation for diversity in results.

1.1.1.2 Reduced voice mean harmonicity

Mean harmonicity refers to signal-to-noise-ratio in human voice and it can be considered as an index of voice quality (Ferrand, 2002). The noise in the signal may result from turbulence created by inadequate closure or aperiodic vibration of vocal folds (de Krom, 1993). Voice harmonicity or harmonicity-to-noise ratio thus reflects the ratio between harmonic or periodic vibration and noise or aperiodic vibration in voice.

Previous findings concerning mean harmonicity suggest that mean harmonicity is stable during young age and adulthood but decreases during old age (Ferrand, 2002). Parkinson’s disease and aging have been observed to decrease voice harmonicity (Gamboa et al., 1997; Lortie, Rivard, Thibeault & Tremblay, 2017). This may be due to loss of vocal fold adduction where the vocal cords do not adduct sufficiently and air escapes through a gap between them (Dehqan, Scherer, Dashti, Ansari-Moghaddam & Fanaie, 2013;

Laukkanen & Leino, 1999). In case the vocal cords do not adduct sufficiently, the voice may sound breathy or hoarse. These may worsen speech intelligibility and cause difficulties to others to understand what is

(9)

4

being said. However, there are also findings without significant differences in harmonicity-to-noise ratio between people with PD and healthy controls (Kacha, Mertens, Grenez, Skodda & Schoentgen, 2017).

1.1.2 Reduced speech intensity 1.1.2.1 Speech intensity

Reduced speech intensity, also known as monoloudness, stands for disturbances in voice intensity variation (Anand & Stepp, 2015). It is typical for people with PD to have a reduced intensity range compared to healthy controls and it is also the most common speech and voice symptom in PD. (Ludlow & Bassich, 1984;

Schalling, Johansson & Hartelius, 2018). It may be manifested as quietness of voice and as challenges to reach and hold full speech intensity (Miller, 2017). In addition, it may cause challenges to express stress in syllables and words. As a result of it, the interlocutors of a person with PD may experience problems in hearing or comprehending what is being said. Based on this it can be said that reduced variability in speech intensity causes challenges in communication.

Speech intensity reduction in PD is a result of reduced range and increased rigidity of muscles in larynx, pharynx and thorax (Miller, 2017; Yorkston et al., 2004). This may impair phonation, prosody and respiration as the respiratory driving pressure and the vocal closure are reduced. However, it must be borne in mind that the voice mechanism itself is not impaired in people with PD but the self-monitoring.

This deficit in self-monitoring may result in decreased awareness of one’s own voice intensity and cause a person with PD to experience their speech intensity higher than what it really is (Theodoros, 2011).

Possible explanations for the deficit include abnormal sensory filtering associated with basal ganglia dysfunction and the influence of a motor-sensory inhibitory mechanism on auditory cortical activity during vocalisation. However, when given an external cue, a person with PD is able to increase their speech intensity for a short time.

1.1.2.2 Energy below 1 kHz

Energy below 1 kHz indicates the percentage of frequencies in the voice frequency spectre below 1 kHz. In a typical voice the majority of energy is located in low frequency regions and less energy is located in high frequency regions while the distribution is somewhat even (Laukkanen & Leino, 1999; Toivanen &

Seppänen, 2005). Changes in voice energy distribution may reflect changes in voice quality and affective state of the speaker. In a hypofunctional voice as in hypokinetic dysarthria it is typical that the majority of voice energy is located in the low frequency regions. When a hypofunctional voice is produced, the vocal cords may not close sufficiently and an air gap remains through which air may escape. Due to this the voice

(10)

5

may sound breathy and the voice intensity may be low. These features of voice may cause challenges in communication as a breathy voice may be considered to express sensitivity, femininity and softness while the speaker may not wish to express these qualities. In addition, low voice intensity may cause difficulties to interlocutors to hear what is being said.

1.1.3 Changes in speech rate and rhythm 1.1.3.1 Total speech rate and articulation rate

Typical rhythmical features of speech of a person with PD include disfluencies, inappropriate silences and short rushes of speech (White, Liss & Dellwo, 2010). In hypokinetic dysarthria, speech may be slow while the articulation rate remains relatively normal or subnormal. Speech rate abnormalities are expected in PD as the disease affects the basal ganglia which are assumed to control aspects related to time and space at the level of motor cortex (Brown & Marsden, 1998; Goberman & McMillan, 2005; Skodda & Schlegel, 2008). Furthermore, hypokinesia and increased rigidity of muscles in the speech production system may cause alterations in speech rate in people with PD.

Findings concerning speech rate and pausing in PD are inconsistent as there are findings with decreased and accelerated articulatory velocities (Ludlow, Connor & Bassich, 1987; Skodda et al., 2009).

The inconsistent findings may be due to several reasons, e.g. the individual differences in degree of

dysarthria and abnormalities between patients (Ho, Iansek, Marigliani, Bradshaw & Gates, 1999; Logemann, Fisher, Boshes & Blonsky, 1978; De Letter et al., 2010; Metter & Hanson, 1986). The stage of disease, cognitive function, overall impairment and medication may also influence the speech and articulation rates.

In addition, the small study sizes and methodologies used may also play a role in inconsistent findings.

1.1.3.2 Normalized pairwise variability index nPVI

Normalised pairwise variability index (nPVI) is an equation which measures how much durational contrast exists between successive events in an utterance (Daniele & Patel, 2013). It was originally developed to measure rhythmic differences in “syllabic-timed” and “stressed-timed” languages. Stress-timed languages have patterns with equal durations between stressed syllables, e.g. Germanic languages like German and English, whereas in syllable-timed languages syllables have equal durations, e.g. Romance languages like Spanish and French. (Abercrombie, 1967; Ling, Grabe, & Nolan, 2000). This classification has been later disputed. It has been argued that syllable-timing or stress-timing are not absolute but gradient features (Dauer, 1983). In addition, it has been later argued that neither a continuous or dichotomous classification

(11)

6

system can describe properties of languages which exhibit both syllable-timed and stress-timed properties (Nespor, 1990).

nPVI has also been applied to measure rhythmic variability in speech and music (Patel, Iversen &

Rosenberg, 2006). As studies examining music and language have shown music to reflect prosody of a composer’s mother tongue, it can be assumed that language-inherent rhythmic units like nPVI may assist in perception of speech rhythm (Späth et al., 2016). In addition, nPVI has been used as an acoustic measure in assessing speech prosody and intelligibility and to describe distinguishing rhythm patterns between

different types of dysarthric speech (Kim, Kent & Weismer, 2011; Liss et al., 2007; Liss et al., 2009). In earlier studies it has been observed that healthy speakers have higher nPVI scores than people with PD (Kim & Choi, 2017).

Many earlier studies concerning nPVI have measured duration of consonants or vowels (Nolan &

Asu, 2009). This can be explained e.g. by controversies in syllable division in certain languages. However, as syllables have a central role in phonological structure which in turn is related to rhythm of a language, it is worthwhile to study nPVI based on syllable durations (Liss et al., 2009).

1.2 Speech therapy and singing interventions in Parkinson’s disease 1.2.1 Common ground between language and music

Language and music seem to be universal across human cultures and unique to our species (Patel, 2008).

They seem to share many features, e.g. the ability to convey meaningful content, hierarchical structural organisation and encode learned categories with the help of motor and perceptual systems (Jackendoff &

Lerdahl, 2006). This has led scientists to speculate if these abilities are related and to which extent (Koelsch, 2011). One area where a shared mechanism has been proposed is that of pitch perception (Beach, 1991;

Price, Ostendorf, Schattuck-Hufnagel & Fong, 1991). Pitch is an essential part of spoken language and music. In language it is utilized to disambiguate syntactic structures and to convey semantic and pragmatic meaning while in music the melodic structure is conveyed via the relative pitch changes.

Research in cognitive neuroscience and psychology has found support for a shared mechanism between music and language concerning pitch processing. It has been observed that linguistic pitch patterns are encoded with higher fidelity in the auditory brainstem in musicians than in non-musicians (Wong, Skoe, Russo, Dees & Kraus, 2007). However, there are also opposing findings which suggest that pitch processing in music and language may be supported by distinct mechanisms. In addition, there are findings concerning brain injuries which impair language skills while the musical skills remain intact and vice versa (Peretz, Belleville & Fontaine, 1997).

(12)

7 1.2.2 Lee Silverman Voice Treatment

Lee Silverman Voice treatment (LSVT LOUD) is a speech rehabilitation program for PD. Its main aim is to improve voice intensity (Sapir, Spielman, Ramig, Story & Fox, 2007; Ramig, Sapir, Fox & Countryman, 2001).

Another prime target in LSVT LOUD is to assist people with PD to regain their internal cuing abilities. This is because people with PD have been observed to be able to increase their speech intensity when given an external cue. The program also aims to generalize these settings in order to transfer the skills learned in therapy to everyday life environments and to enable maintenance of possible gains acquired during therapy (Fox, Ebersbach, Ramig & Sapir, 2012).

LSVT LOUD consists of individual therapy sessions which take place four days per week for four consecutive weeks (Beck, 2019; Fox et al., 2012; Sapir, Ramig, & Fox, 2011). During each session the

individual is prompted to practise glides up and down on their pitch range, sustained vowels and frequently used functional phrases with a loud voice. The length of the utterances is increased gradually. The

participants are also encouraged to practise exercises daily outside therapy settings in order to gain more benefit from the program.

LSVT LOUD has been shown to be effective up to 2 years in improving hypophonia (Ramig et al., 2001). Short-term effects of the program have also been shown in an fMRI study where the speech intensity of participants increased and an increased involvement of cortical areas in right hemisphere was observed (Narayana et al., 2010). Other immediate changes observed after an LSVT LOUD intervention include improved speech intelligibility, stronger closure of vocal folds and improvement of communication (Ramig, & Dromey, 1996; Smith, Ramig, Dromey, Perez & Samandari, 1995). Improvements in facial expression, speech rate and consonant articulation have also been observed (Fox et al., 2006).

While LSVT LOUD has been found effective in improving speech deficits caused by PD, the program has certain weaknesses and limitations. The challenge of maintaining the improvements gained with the intervention can be considered as the most significant weakness, especially because PD is a progressive disease (Beck, 2019). Because of this the researchers have begun to examine other possible approaches with similar elements in order to find potential rehabilitation methods for speech deficits in PD.

1.2.3 Singing interventions in Parkinson’s disease

Earlier research provides evidence that singing may be a potential tool in speech and voice rehabilitation of people with PD (Barnish, Atkinson, Barran & Barnish, 2016). Singing may be potential in PD rehabilitation in

(13)

8

promoting and intensifying certain aspects of voice and its production (Haneishi, 2001). It may for instance support louder voice production and help individuals to train their respiratory capacities. Singing may also improve intonation, speech rate and timing because singing includes greater pitch range and more tempo variation than regular speech (Schlaug, Marchina & Norton, 2008). Because singing involves a slower articulation rate compared to regular speech, it may also improve coordination of articulation, phonation and respiration.

After the limitations of LSVT programs were recognized, researchers have begun to examine singing as a potential alternative in speech rehabilitation for people with PD (Beck, 2019). Singing is considered potential for this task as it aims proper abdominal breathing and sustained phonation and includes the full pitch range of an individual. However, studies concerning singing as a speech rehabilitation method in PD are still limited. The results so far reveal that singing interventions have been effective in improving speech loudness and carer-rated intelligibility (Haneishi, 2001), fundamental frequency f0 and its variation (Tanner, 2016), improved quality of voice prosody and fatigue ratings in reading tasks (di Benedetto, 2009) and speech intelligibility (Higgins, 2018). However, there are also findings where singing intervention has not been found effective in improving features of speech among people with PD. For example, di Benedetto (2009), Elefant (2012) and Yinger (2012) report no improvement for fundamental frequency f0 or its variation. Also, Elefant (2012), Shih (2012) and Tanner (2016) report no improvement in speech intensity after a singing intervention. The above-mentioned interventions have certain limitations for instance concerning sample sizes, participant randomization and control groups. In addition, there are differences in intensity between the interventions and whether they were carried out individually or in a group.

(14)

9 2. Aim of the study and research questions

This study is part of a multidisciplinary research project Kuuluva ääni (“Carrying voice”) organized by Tampere University. The project aims to obtain useful information about rehabilitation of communication and voice deficits in PD. The project started in 2018 and will continue until 2021. The research proposal has been approved by the Ethics Committee of Tampere region (decision 25/2018).

This current study aims to examine prosodic features of speech in Finnish adults with PD and possible changes in them following a group singing intervention. Although prosody in Finnish language has been studied, research concerning the effects of PD in prosody in Finnish language is sparse. The research questions focus on monopitch, monoloudness and speech rate and rhythm, as they are all typical features of dysprosody in hypokinetric dysarthria and areas of speech affected by PD. (Anand & Stepp, 2015; Ludlow

& Bassich, 1984; Skodda & Schlegel, 2008). The research questions are as follows:

1. Does the group singing intervention change the participants’ pitch, pitch variability and

harmonicity-to-noise-ratio as measured with fundamental frequency f0, fundamental frequency variation (f0SD) and mean harmonicity?

2. Does the group singing intervention change the participants’ voice production as measured with speech intensity and percentage of vocal energy below 1 kHz?

3. Does the group singing intervention change the participants’ speech rate and rhythmicity as measured with total speech rate, articulation rate and normalized pairwise variability index nPVI?

These variables are measured in baseline and post-intervention in normal and emotional conditions. The difference between normal and emotional condition refers to the remaining prosodic ability of the individual. This difference is subsequently called as prosodic capacity. One aim of this study is to see if singing serves as a cue in allowing a person with PD to utilize their prosodic abilities they have but have been unable to use due to PD.

(15)

10 3. Method

3.1 Participants

Fifteen individuals with idiopathic PD were selected to the group singing intervention. Their background information is shown in table 1 for females and in table 2 for males. The inclusion criterion was a diagnosis of idiopathic Parkinson’s disease. The exclusion criteria were Deep Brain Stimulation (DBS) treatment, dementia and communication disorders preceding the PD diagnosis (for instance dysphonia, stuttering, aphasia, dysarthria and apraxia of speech). Of the 15 participants, 8 were male and 7 were female. The average age was 72 years, ranging from 54 to 83 years. Time from diagnosis was on average 8 years, and it ranged from 2 to 19 years.

The severity of the participants’ symptoms and the stage of disease progression were assessed with Hoehn & Yahr scale, where stage 0 indicates no signs of disease and stage 5 indicates a very high level of disability. The participants’ stages ranged from 1 to 4, average stage being 1.9. The level of possible

cognitive impairment was measured with Mini Mental Status Examination (MMSE). The MMSE scores range from 0 to 30. Scores equal to or greater than 24 points indicate a normal cognition, whereas scores below 24 points can indicate cognitive impairment. All participants had an MMSE score higher than 24 (with the exception of one participant who only had a score of 21 but was nevertheless included in the study, as his performance did not diverge from that of the others).

Geriatric depression scale (GDS) was used to examine depressive symptoms among the participants.

Score 0-10 refers to normal condition whereas score 11-20 refers to mild depression and scores above that to moderate or severe depression. Of 15 participants, 3 can be considered mildly depressed, as their scores ranged from 13 to 19.

A short-scale version of Voice Handicap Index (VHI) was used to assess the participants’ self-rated psychosocial impact of voice impairment. The scores range from 0 to 36, with higher values indicating a more severe voice-related handicap. The scores ranged from 1 to 20, average being 14. Nine participants had received speech therapy earlier. All participants reported that medication did not have any effect on their speech. Eight of the participants had previous singing experience.

(16)

11 Table 1 – Female participants

ID Age

Disease duration (years)

Hoehn

&

Yahr

MMSE GDS VHI

Earlier speech therapy due to

PD

Singing experience

Participation to intervention

Engagement to home exercises

K1 75 12 1 29 7 11 no yes 8/8 29/42

K2 67 16 1.5 28 5 1 no no 8/8 10/42

K4 65 2 1 27 13 17 no yes 8/8 29/42

K7 63 7 1.5 27 6 12 yes N/A 4/8 10/42

K9 71 6 2 27 8 2 no N/A 4/8 10/42

KK1 78 9 3 27 19 12 yes yes 8/8 25/42

KK3 54 7 1.5 29 2 16 no no 6/8 38/42

Mean (SD)

67.6 (8.0)

8.4 (4.5)

1.6 (0.7)

27.7 (1.0)

8.6 (5.7)

10.1

(6.3) - - - -

Table 2 – Male participants

ID Age

Disease duration (years)

Hoehn

&

Yahr

MMSE GDS VHI

Earlier speech therapy due to

PD

Singing experience

Participation to intervention

Engagement to home exercises

K3 74 10 2,5 26 16 18 no yes 8 35

K5 69 3 3 29 0 11 yes no 8 32

K8 78 7 1,5 29 8 19 yes yes 8 27

K11 76 5 2 27 4 18 yes no 4 6

K12 73 11 1 26 3 17 yes yes 3 5

KK2 83 6 1 28 5 17 yes no 6 38

KK4 76 19 2,5 21 5 18 yes yes 7 17

KK5 76 3 4 29 9 20 yes yes 8 21

Mean (SD)

75.6 (4.0)

8.0 (5.3)

2.2 (1.0)

26.9 (2.7)

6.3 (4.8)

17.3

(2.7) - - - -

3.2 Intervention

The intervention consisted of 90-minute sessions which took place once per week for 8 subsequent weeks excluding one week break due to a national holiday. The sessions took place in a classroom at Tampere

(17)

12

University. The group sessions were carried out by a BA student in music pedagogy with 20 years of experience in choir leading and theatre.

The structure of each session was the following:

1. Greetings, sharing experiences concerning home exercises and filling in a questionnaire about vocal status and emotional state

2. Warm-up exercises, posture check 3. Vocal warm-up exercises

4. Singing exercises

5. Closing conversation and filling in a questionnaire about vocal status and emotional state.

The group sessions shared features with an evidence-based speech therapy method LSVT LOUD.

The main aims of LSVT LOUD include increase of vocal loudness and improvement of internal cueing abilities (Sapir et al., 2011; Fox et al., 2012). The current intervention was designed to target similar goals.

During the sessions the participants were advised to apply different emotions and big movements to voice production and to use loud voice to make exercises more intense. They were also given an audio CD and written instructions to facilitate practising at home. Engagement to home exercises was monitored by the leader by interviewing each participant at the start of each session and by checking how much they had practised at home between the group sessions.

3.3 Data collection and outcome measures

The current study is a single group repeated measures study in which possible changes in voice occurring after a group singing intervention are examined based on acoustic measures. The participants read the story

“The North Wind and the Sun” which was recorded with AKG C544L head-mounted condenser microphone and digitized at 44,100 samples per second by using Focusrite iTrack Solo audio interface. The 4 cm mouth- to-microphone distance at a 45-degree angle was controlled by using a ruler from the corner of the participants’ mouths. The recordings took place in quiet office rooms. Each participant was asked to read the story first with normal voice, then in empathized, emotional voice, as playing a role in a radioplay. Both recordings were carried out both before and after the intervention so there were four samples for each participant.

A caption from the middle of the text was cut off from each sample with Audacity 2.3.1 Software for further analysis (Mazzoni, 2019). Only these captions were analyzed in this study. The selected captions were then annotated with Praat (Boersma & Weenink, 2013) in word, syllable and phone levels. The annotated

(18)

13

captions were then analyzed with a script which measures prosodic features of speech (Tavi, 2019). The measured variables were analyzed from each sample. This was done to see if there were any significant changes in measured variables and in prosodic capacities in each variable before and after the intervention.

3.4 Ethical considerations

An ethical review was carried out for the project “Kuuluva ääni”. A data management plan was also written which contains information about project content, types of data collected and their storage, ethical considerations and copyrights. The collected data is stored in material banks and research archives where only the researchers have access to. During the active phase of research (2018-2025) the data is not available for open distribution. After the active phase the material will be relocated to a societal database which will be responsible for its storage, use, privacy protection, archive copies, information security, communication, terms of use and distribution.

Every participant signed a written consent form and was informed about the study’s purpose and course before participating. They were also informed about the possibility to withdraw from participation at any stage. The participants have been treated with anonymity so that they cannot be identified from any research reports. This principle of privacy protection has been observed in all phases of research.

3.5 Statistical analysis

All statistical analysis was carried out with IBM SPSS software version 25.0 (IBM, 2017). The differences between baseline and post-intervention and changes in prosodic capacity in different variables were analysed with related-Samples Wilcoxon Signed Rank Test. These analyses were carried out for the whole group (N=15) and for subgroups of female (N=7) and male (N=8) participants. Statistical significance was indicated by p ≤ 0.05.

In addition to related-Samples Wilcoxon Signed Rank Test, a linear mixed statistical model with t- tests using Satterthwaite-method was used to analyse results of fundamental frequency f0. This analysis was carried out by using phone specific raw data. As its results were clearly in line with results provided by related-Samples Wilcoxon Signed Rank Tests, rest of the analyses were carried out using only the latter tests.

(19)

14 4. Results

4.1 Monotony of pitch

4.1.1 Fundamental frequency f0 and its variation f0SD

Analysis of fundamental frequency f0 (N=15) showed no statistically significant change from baseline to post-intervention either in normal (Z=67 000, p=0.691) or emotional (Z=74 000, p=0.427) condition nor change in the post-intervention prosodic capacity of fundamental frequency f0 (Z=55 000, p=0.776) in the whole group. Changes in fundamental frequency f0 from baseline to post-intervention in whole group are shown in table 3. See Appendix 1 for results.

Table 3 – Changes in fundamental frequency f0 and its variation f0SD, whole group

Fundamental frequency f0 [Hz] Variation in fundamental frequency f0SD ID

Difference in normal reading

Difference in emotional

reading

Difference in capacity

reading

Difference in capacity

K1 1.49 -7.00 8.49 1.39 -7.02 8.41

K2 -0.14 2.40 -2.54 -6.35 -1.05 -5.30

K3 0.22 -10.10 10.32 0.94 -4.35 5.29

K4 -7.29 -22.44 15.15 2.06 -20.27 22.33

K5 5.03 -5.41 10.44 -0.48 0.88 -1.36

K7 16.71 32.29 -15.58 14.83 24.08 -9.25

K8 -3.85 6.78 -10.63 -0.91 6.2 -7.11

K9 -2.24 7.62 -9.86 -3.50 1.76 -5.26

K11 -20.85 -8.91 -11.94 -6.55 -5.78 -0.77

K12 -13.13 -14.88 1.75 5.54 -2.03 7.57

KK1 15.73 12.99 2.74 -1.66 4.41 -6.07

KK2 -13.78 -11.55 -2.23 0.57 2.48 -1.91

KK3 -0.85 16.87 -17.72 -0.76 -0.70 -0.06

KK4 -9.65 -30.78 21.13 -2.20 -6.04 3.84

KK5 24.86 -8.20 33.06 -14.26 -0.04 -14.22

Mean (SD)

-0.52 (12.34)

-2.69 (16.10)

2.172 (14.44)

-0.76 (6.31)

-0.50 (9.25)

-0.26 (8.89)

Analysis of fundamental frequency f0 in the subgroup of females (N=7) did not show statistically significant change from baseline to post-intervention either in normal (Z=12 000, p=0.735) or emotional condition (Z=8 000, p= 0.310) or change in the post-intervention the prosodic capacity of fundamental

(20)

15

frequency f0 (Z=18 000, p=0.499). Changes in fundamental frequency f0 from baseline to post-intervention in in the subgroup of females are shown in table 4. See Appendix 1 for results.

Table 4 – Changes in fundamental frequency f0 and its variation f0SD, females

Fundamental frequency f0 [Hz] Variation in fundamental frequency f0SD ID

reading

Difference in capacity

reading

K1 1.49 -7.0 8.49 1.39 -7.02 8.41

K2 -0.14 2.4 -2.54 -6.35 -1.05 -5.3

K4 -7.29 -22.44 15.15 2.06 -20.27 22.33

K7 16.71 32.29 -15.58 14.83 24.08 -9.5

K9 -2.24 7.62 -9.86 -3.5 1.76 -5.26

KK1 15.73 12.99 2.74 -1.66 4.41 -6.07

KK3 -0.85 16.87 -17.72 -0.76 -0.7 -0.06

Mean (SD)

3.34 (9.21)

6.10 (17.57)

-2.76 (12.36)

0.86 (6.79)

0.17 (13.31)

0.69 (11.14)

Analysis of fundamental frequency f0 from baseline to post-intervention showed a statistically significant change in emotional condition (Z=34 000, p=0.025) in the subgroup of males (N=8). No

statistically significant change was observed from baseline to post-intervention in normal condition (Z=24 000, p=0.401) nor concerning the post-intervention prosodic capacity of fundamental frequency f0 (Z=13 000, p=0.484). Changes in fundamental frequency f0 from baseline to post-intervention in the subgroup of males are shown in table 5. See Appendix 1 for results.

Table 5 – Changes in fundamental frequency f0 and its variation f0SD, males

Fundamental frequency f0 [Hz] Variation of fundamental frequency f0SD ID

reading

Difference in normal reading

reading

K3 0.22 -10.1 10.32 0.94 -4.35 5.29

K5 5.03 -5.41 10.44 -0.48 0.88 -1.36

K8 -3.85 6.78 -10.63 -0.91 6.2 -7.11

K11 -20.85 -8.91 -11.94 -6.55 -5.78 -0.77

K12 -13.13 -14.88 1.75 5.54 -2.03 7.57

KK2 -13.78 -11.55 -2.23 0.57 2.48 -1.91

(21)

16

KK4 -9.65 -30.78 21.13 -2.2 -6.04 3.84

KK5 24.86 -8.2 33.06 -14.26 -0.04 -14.22

Mean (SD)

-3.89 (14.28)

-10.38 (10.45)

6.49 (15.50)

-2.17 (5.93)

-1.09 (4.29)

-1.08 (7.07)

Analysis of fundamental frequency standard deviation f0SD (N=15) showed no statistically significant change from baseline to post-intervention either in normal (Z=73 000, p=0.460) or emotional condition (Z=69 000, p=0.609) No statistically significant change was observed in the post-intervention prosodic capacity of standard deviation of fundamental frequency f0SD (Z=70 000, p=0.570) in the whole group.

Changes in standard deviation of fundamental frequency (f0SD) from baseline to post-intervention in the whole group are shown in table 3. See Appendix 1 for results.

Analysis of fundamental frequency standard deviation f0SD in the subgroup of females (N=7) showed no statistically significant change from baseline to post-intervention in normal (Z=15 000, p=0.866) or emotional condition (Z=14 000, p=1.000). No statistically significant change was observed in the post- intervention prosodic capacity of standard deviation of fundamental frequency f0SD in the subgroup of females (Z=16 000, p=0.735). Changes in standard deviation of fundamental frequency (f0SD) from baseline to post-intervention in the subgroup of subgroup of women are shown in table 4. See Appendix 1 for results.

Analysis of fundamental frequency standard deviation f0SD in the subgroup of males (N=8) showed no statistically significant changes from baseline to post-intervention in normal (Z=24 000, p=0.401) or emotional condition (Z=22 000, p=0.575) or concerning the post-intervention prosodic capacity of standard deviation of fundamental frequency f0SD (Z=20 000, p=0.779). Changes in standard deviation of

fundamental frequency (f0SD) from baseline to post-intervention in the subgroup of males are shown in table 5. See Appendix 1 for results.

4.1.2 Mean harmonicity

Analysis of mean harmonicity (N=15) did not show statistically significant change from baseline to post- intervention in normal (Z=58 000, p=0.910) or emotional condition (Z=47 000, p=0.460) in the whole group.

In addition, no statistically significant change was observed in the post-intervention prosodic capacity of mean harmonicity (Z=60 000, p=1.000). Changes in mean harmonicity from baseline to post-intervention in the whole group are shown in table 6. See Appendix 2 for results.

(22)

17 Table 6 – Changes in mean harmonicity, whole group

Mean harmonicity [dB]

ID Difference in

normal reading

Difference in emotional reading

Difference in prosodic capacity

K1 -0.03 0.71 -0.74

K2 0.57 -2.84 3.41

K3 -1.20 -1.00 -0.20

K4 -1.31 -2.64 1.33

K5 -3.32 1.59 -4.91

K7 0.69 -0.13 0.82

K8 -1.94 -0.07 -1.87

K9 1.77 0.83 0.94

K11 -0.78 -0.31 -0.47

K12 -0.66 0.49 -1.15

KK1 -0.09 -0.19 0.10

KK2 0.82 -0.70 1.52

KK3 -0.88 0.67 -1.55

KK4 -0.34 -0.19 -0.15

KK5 0.91 -1.31 2.22

Mean (SD) -0.39 (1.28) -0.34 (1.23) -0.05 (1.97)

In the subgroup of females (N=7) (see table 7), no statistically significant change in mean

harmonicity was observed from baseline to post-intervention either in normal condition (Z=8 000, p=0.310) or emotional condition (Z=7 000, p=0.237). In addition, no statistically significant change was observed in the post-intervention prosodic capacity of mean harmonicity (Z=8 000, p=0,310). Changes in mean harmonicity from baseline to post-intervention in the subgroup of females are shown in table 7. See Appendix 2 for results.

(23)

18 Table 7 – Changes in mean harmonicity, females

normal reading

K1 -0.03 0.71 -0.74

K2 0.57 -2.84 3.41

K4 -1.31 -2.64 1.33

K7 0.69 -0.13 0.82

K9 1.77 0.83 0.94

KK1 -0.09 -0.19 0.1

KK3 -0.88 0.67 -1.55

Mean (SD) 0.10 (1.03) -0.51(1.56) 0.62 (1.59)

In the subgroup of males (N=8) (see table 8), no statistically significant change was observed in mean harmonicity from baseline to post-intervention either in normal condition (Z=26 000, p=0.263) or emotional condition (Z=18 000, p=1.000). In addition, no statistically significant change was observed in post-intervention prosodic capacity of mean harmonicity (Z=24 000, p=0.401). Changes in mean

harmonicity from baseline to post-intervention in the subgroup of males are shown in table 8. See Appendix 2 for results.

Table 8 – Changes in mean harmonicity, males

normal reading

K3 -1.20 -1.00 -0.20

K5 -3.32 1.59 -4.91

K8 -1.94 -0.07 -1.87

K11 -0.78 -0.31 -0.47

K12 -0.66 0.49 -1.15

(24)

19 4.2 Intensity features of speech

4.2.1 Speech intensity

Analysis of speech intensity (N=15) showed no statistically significant change neither from baseline to post- intervention in normal (Z=63 000, p=0.865) or emotional (Z=62 000, p=0.910) condition nor in the post- intervention prosodic capacity of speech intensity (Z=55 000, p=0.776) in the whole group. Changes in speech intensity from baseline to post-intervention in the whole group are shown in table 9. See Appendix 3 for results.

Table 9 – Changes in speech intensity, whole group

Speech Intensity [dB]

normal reading

K1 -6.81 -5.86 -0.95

K2 -8.00 -7.04 -0.96

K3 3.30 1.59 1.71

K4 6.55 5.58 0.97

K5 -1.18 -2.49 1.31

K7 -4.27 -7.26 2.99

K8 -5.05 -2.22 -2.83

K9 4.32 8.63 -4.31

K11 -4.28 -4.76 0.48

K12 4.26 4.10 0.16

KK1 -2.55 -1.00 -1.55

KK2 8.12 8.73 -0.61

KK2 0.82 -0.70 1.52

KK4 -0.34 -0.19 -0.15

KK5 0.91 -1.31 2.22

Mean (SD) -0.81 (1.39) -0.19 (0.91) -0.63 (2.18)

(25)

20

KK3 2.55 -5.86 8.41

KK4 -10.91 -17.08 6.17

KK5 9.24 6.99 2.25

Mean (SD) -0.31 (6.25) -1.20 (7.20) 0.88 (3.25)

In the subgroup of females (N=7), the analysis of speech intensity showed no statistically significant change neither from baseline to post-intervention in normal (Z=17 000, p=0.612) or emotional condition (Z=16 000, p=0.735) nor in the post-intervention prosodic capacity of speech intensity (Z=19 000, p=0.398).

Changes in speech intensity from baseline to post-intervention in the subgroup of females are shown in table 10. See Appendix 3 for results.

Table 10 – Changes in speech intensity, females

normal reading

K1 -6.81 -5.86 -0.95

K2 -8.00 -7.04 -0.96

K4 6.55 5.58 0.97

K7 -4.27 -7.26 2.99

K9 4.32 8.63 -4.31

KK1 -2.55 -1.00 -1.55

KK3 2.55 -5.86 8.41

Mean (SD) -1.17 (5.68) -1.83 (6.51) 0.66 (4.09)

(26)

21

In the subgroup of males (N=8), the analysis of speech intensity showed no statistically significant change from baseline to post-intervention in normal (Z=18 000, p=1.000) or emotional condition (Z=18 000, p=1.000) or in the post-intervention prosodic capacity of speech intensity (Z=10 000, p=0.263). Changes in speech intensity from baseline to post-intervention in the subgroup of males are shown in table 11. See Appendix 3 for results.

Table 11 – Changes in speech intensity, males

ID

Difference in normal reading

K3 3.3 1.59 1.71

K5 -1.18 -2.49 1.31

K8 -5.05 -2.22 -2.83

K11 -4.28 -4.76 0.48

K12 4.26 4.10 0.16

KK2 8.12 8.73 -0.61

KK4 -10.91 -17.08 6.17

KK5 9.24 6.99 2.25

Mean (SD) 0.44 (7.00) -0.64 (8.16) 1.08 (2.59)

4.2.2 Percentage of vocal energy below 1 kHz

Analysis of percentage of vocal energy below 1 kHz showed a statistically significant change from baseline to post-intervention both in normal (Z=22 000, p=0.031) and emotional conditions (Z=16 000, p=0.012) in the whole group (n=15). In addition, a statistically significant change in the post-intervention prosodic capacity of percentage of vocal energy below 1 kHz was observed (Z=99 000, p=0.027). Changes in percentage of vocal energy below 1 kHz from baseline to post-intervention in whole group are shown in table 12. See Appendix 4 for results.

(27)

22

Table 12 – Changes in percentage of energy below 1 kHz, whole group Energy below 1 kHz [%]

normal reading

K1 0.001 0 0.001

K2 0.002 0.002 0

K3 0 0.001 -0.001

K4 0.015 0.017 -0.002

K5 -0.009 0.002 -0.011

K7 -0.003 0.025 -0.028

K8 0.005 0.010 -0.005

K9 0.018 0.017 0.001

K11 0.003 0.007 -0.004

K12 0.001 0.008 -0.007

KK1 0.039 0.033 0.006

KK2 0.003 0.001 0.002

KK3 0.001 0.010 -0.009

KK4 0.005 0.027 -0.022

KK5 0.005 0.007 -0.002

Mean (SD) 0.006 (0.011) 0.011 (0.010) -0.005 (0.009)

For the subgroup of females (N=7), analysis of percentage of vocal energy below 1 kHz showed no statistically significant change from baseline to post-intervention either in normal (Z=3 000, p=0.063) or emotional condition (Z=4 000, p=0.091). In addition, there was no statistically significant change in the post-intervention prosodic capacity of percentage of vocal energy below 1 kHz (Z=20 000, p=0.310).

Changes in percentage of vocal energy below 1 kHz from baseline to post-intervention in the subgroup of females are shown in table 13. See Appendix 4 for results.

(28)

23

Table 13 – Changes in percentage of energy below 1 kHz, females Energy below 1 kHz [%]

normal reading

K1 0.001 0 0.001

K2 0.002 0.002 0

K4 0.015 0.017 -0.002

K7 -0.003 0.025 -0.028

K9 0.018 0.017 0.001

KK1 0.039 0.033 0.006

KK3 0.001 0.010 -0.009

Mean (SD) 0.010 (0.015) 0.015 (0.012) -0.004 (0.011)

For the subgroup of males (N=8), the analysis of percentage of vocal energy below 1 kHz showed no statistically significant change from baseline to post-intervention either in normal (Z=12 000, p=0.401) or emotional condition (Z=7 000, p=0.123). A statistically significant change was observed in the post-

intervention prosodic capacity of percentage of vocal energy below 1 kHz (Z=33 000, p=0.036). Changes in percentage of vocal energy below 1 kHz from baseline to post-intervention in the subgroup of males are shown in table 14. See Appendix 4 for results.

Table 14 – Changes in percentage of energy below 1 kHz, males Energy below 1 kHz [%]

normal reading

K3 0 0.001 -0.001

K5 -0.009 0.002 -0.011

K8 0.005 0.010 -0.005

(29)

24

K11 0.003 0.007 -0.004

K12 0.001 0.008 -0.007

KK2 0.003 0.001 0.002

KK4 0.005 0.027 -0.022

KK5 0.005 0.007 -0.002

Mean (SD) 0.002 (0.005) 0.008 (0.008) -0.006 (0.007)

4.3 Speech rate and rhythm 4.3.1 Total speech rate

Analysis of total speech rate (N=15) showed no statistically significant change from baseline to post- intervention either in normal (Z=58 500, p=0.932) or in emotional (Z=68 500, p=0.629) condition in the whole group. No statistically significant change was observed in the post-intervention prosodic capacity in total speech rate (Z=47 000, p=0.460). Changes in total speech rate from baseline to post-intervention in the whole group are shown in table 15. See Appendix 5 for results.

Table 15 – Changes in total speech rate, whole group

Total Speech Rate [syllables/second]

ID Difference in normal

reading Difference in emotional reading Difference in prosodic capacity

K1 0.43 0.53 -0.1

K2 0.17 0.16 0.01

K3 -0.56 0.23 -0.79

K4 -0.25 0.33 -0.58

K5 0.51 -0.44 0.95

K7 -0.03 -0.09 0.06

K8 0.1 -0.03 0.13

K9 0.43 0.01 0.42

(30)

25

K11 0.66 0.59 0.07

K12 0.18 -0.57 0.75

KK1 -0.17 -0.16 -0.01

KK2 -0.49 -0.24 -0.25

KK3 0.1 0.01 0.09

KK4 -0.52 -1.11 0.59

KK5 -0.38 -0.31 -0.07

Mean (SD) 0.012 (0.40) -0.07267 (0.44) 0.084667 (0.46)

In the subgroup of females (N=7), no statistically significant change in total speech rate was found from baseline to post-intervention either in normal (Z=9 500, p=0.445) or emotional condition (Z=7 500, p=0.270). In addition, no statistically significant change was observed in the post-intervention prosodic capacity in total speech rate (Z=13 500, p=0.933). Changes in total speech rate from baseline to post- intervention in the subgroup of females are shown in table 16. See Appendix 5 for results.

Table 16 – Changes in total speech rate, females

ID Difference in normal reading

Difference in

emotional reading Difference in prosodic capacity

K1 0.43 0.53 -0.1

K2 0.17 0.16 0.01

K4 -0.25 0.33 -0.58

K7 -0.03 -0.09 0.06

K9 0.43 0.01 0.42

KK1 -0.17 -0.16 -0.01

KK3 0.1 0.01 0.09

Mean (SD) 0.10 (0.27) 0.11 (0.24) -0.02 (0.30)

(31)

26

In the subgroup of males (N=8), no statistically significant change in total speech rate was observed from baseline to post-intervention either in normal (Z=20 000, p=0.779) or emotional condition (Z=27 000, p=0.208). In addition, no statistically significant change was observed in the post-intervention prosodic capacity in total speech rate (Z=12 500, p=0.441). Changes in total speech rate from baseline to post- intervention in the subgroup of males are shown in table 17. See Appendix 5 for results.

Table 17 – Changes in total speech rate, males

ID Difference in normal reading

K3 -0.56 0.23 -0.79

K5 0.51 -0.44 0.95

K8 0.1 -0.03 0.13

K11 0.66 0.59 0.07

K12 0.18 -0.57 0.75

KK2 -0.49 -0.24 -0.25

KK4 -0.52 -1.11 0.59

KK5 -0.38 -0.31 -0.07

Mean (SD) -0.06 (0.49) -0.24 (0.52) 0.17 (0.57)

4.3.2 Articulation rate

Analysis of articulation rate (N=15) showed no statistically significant change from baseline to post- intervention either in normal (Z=51 500, p=0.629) or emotional (Z=65 000, p=0.776) condition. No

statistically significant change was observed in the post-intervention prosodic capacity in articulation rate (Z=58 000, p=0.910). Changes in articulation rate from baseline to post-intervention in the whole group are shown in table 18. See Appendix 5 for results.