JSLHR
Research Article
Articulatory Closure Proficiency in Patients With Parkinson’s Disease Following Deep Brain Stimulation of the Subthalamic Nucleus and Caudal Zona Incerta Fredrik Karlsson,a Katarina Olofsson,a Patric Blomstedt,a Jan Linder,a Erik Nordh,a and Jan van Doorna
Purpose: The present study aimed at comparing the effects of deep brain stimulation (DBS) treatment of the subthalamic nucleus (STN) and the caudal zona incerta (cZi) on the proficiency in achieving oral closure and release during plosive production of people with Parkinson’s disease. Method: Nineteen patients participated preoperatively and 12 months after DBS surgery. Nine patients had implantations in the STN, 7 bilaterally and 2 unilaterally (left). Ten had bilateral implantations in the cZi. Postoperative examinations were made off and on stimulation. All patients received simultaneous L-dopa treatment in all conditions. For a series of plosives extracted from a reading passage, absolute and relative measures of duration of frication and amplitude of plosive release were compared between conditions within each treatment group.
Results: Relative duration of frication increased in voiceless plosives in the on-stimulation condition in cZi patients. Similar trends were observed across the data set. Duration of prerelease frication and the release peak prominence increased in voiceless plosives on stimulation for both groups. Conclusion: The increased release prominence suggests that patients achieved a stronger closure gesture because of DBS but that the increased energy available resulted in increased frication.
T
Cmejla, Ruzickova, & Ruzicka, 2011). In particular, the production of plosive consonants is highly dependent on achieving full closure between articulators. Full closure allows a buildup of intra-oral pressure that is required for the plosive to be perceived as having been realized in an appropriate manner. Plosives are therefore particularly affected by the reduction in articulatory range in people with PD, which may cause incomplete closure between articulators (Ackermann & Ziegler, 1991). The most frequently discussed acoustic consequence of articulatory undershoot during closure is a strong inappropriate aperiodic component in place of the silent portion during occlusion (Ackermann & Ziegler, 1991; Chenausky et al., 2011; Dromey & Bjarnason, 2011; Forrest, 1989; Hartinger, Tripoliti, Hardcastle, & Limousin, 2011; Kent & Kim, 2003; Logemann & Fisher, 1981). Thus, the incomplete articulatory closure in plosives produced by people with PD leads to an increase in portions with aperiodic noise.
he hypokinetic dysarthria associated with Parkinson’s disease (PD) reduces proficiency in reaching articulatory targets. Specifically, imprecision in consonant articulation has been argued to be a hallmark of dysarthria (Kent & Kim, 2003). Numerous reports have noted that a reduction of articulatory extent affects many consonant types, especially those demanding close proximity between articulators (Ackermann & Ziegler, 1991; Chenausky, MacAuslan, & Goldhor, 2011; Dromey & Bjarnason, 2011; Karlsson et al., 2011; Kent, Weismer, Kent, Vorperian, & Duffy, 1999; McAuliffe, Ward, & Murdoch, 2006b; Rusz,
a
Umeå University, Sweden
Correspondence to Fredrik Karlsson: [email protected] Editor: Jody Kreiman Associate Editor: Julie Liss Received January 10, 2013 Revision received August 20, 2013 Accepted November 20, 2013 DOI: 10.1044/2014_JSLHR-S-13-0010
1178
Key Words: dysarthria, speech motor control, neurologic disorders
Disclosure: The authors have declared that no competing interests existed at the time of publication.
Journal of Speech, Language, and Hearing Research • Vol. 57 • 1178–1190 • August 2014 • A American Speech-Language-Hearing Association
The plosives are therefore less perceptually distinct from comparable fricatives, a process referred to as spirantization. A unified acoustic quantification does not exist for spirantization in plosives. Early assessments were based on visual information from the waveform display of the acoustic signal, and the plosive was marked as including spirantization based on the shape of the waveform (Weismer, 1984). Methods attempting to quantify the degree of spirantization in a plosive include measures of the intensity during closure (IDC) (Ackermann & Ziegler, 1991) and the intensity of closure relative to that of the following vowel, or to the mean intensity of the whole syllable (Chenausky et al., 2011; Dromey & Bjarnason, 2011). Results from these measures have been inconclusive, with patients with PD often showing more frequent (Weismer, 1984) and stronger (Ackermann & Ziegler, 1991) spirantization compared with controls, but with a substantial portion of patients falling well within the limits observed in controls. This suggests that PD may not always impair proficiency in achieving full articulatory closure during plosive production. It may, however, also be viewed as an indication that incomplete articulatory closure may not be appropriately captured by a single acoustic dimension or quantification. Alternative acoustic consequences of incomplete articulatory closure may be proposed, based on models of the dynamics of stop/plosive production (Stevens, 2000). A direct consequence of incomplete articulatory closure is a reduction of oral pressure build-up in the occlusion phase of the plosive. If full closure is not achieved, the energy provided by the respiratory system will not cause a sufficiently abrupt and forceful articulatory release, so that the acoustic transient will not be as acoustically prominent. Thus, in addition to the possible presence of spirantization, there may also be a reduced release transient amplitude in plosives produced with hypokinetic articulation, resulting in even less perceptual distinction from comparable fricatives. This acoustic property of hypokinetic speech movements has received much less attention than the spirantization component, but it has been noted in some previous reports (Auzou et al., 2000; Duez, 2007; Karlsson et al., 2011; Ozsancak, Auzou, Jan, & Hannequin, 2001; Wang, Kent, Duffy, Thomas, & Weismer, 2004). Slowed, more varied (McAuliffe, Ward, & Murdoch, 2006b), or less forceful (McAuliffe, Ward, & Murdoch, 2006a) articulatory release movements observed (through electropalatograpic recordings) for some consonant types with apical closure may, if extended to the production of plosives, further create more diffuse releases by increasing the time in near contact between articulators. Taken together, the acoustic consequences of hypokinetic articulation is very likely to cause plosives to be acoustically similar to fricatives, in accordance with earlier descriptions in the literature (Ackermann, Hertrich, Daum, Scharf, & Spieker, 1997; Ackermann & Ziegler, 1991; Auzou et al., 2000; Chenausky et al., 2011; Forrest, 1989; Hartinger et al., 2011; Logemann & Fisher, 1981; Roy, Leeper, Blomgren, & Cameron, 2001; Wang et al., 2004). Investigation of increased overall acoustic distinctiveness between plosives and fricatives is therefore seen as an intuitive and productive
methodology for studying articulatory closure proficiency as well as for quantifying treatment effects. Results from Ackermann and Ziegler (1991) indicate that plosives in unstressed syllables are more reduced in patients with PD compared to controls. Thus, the prosodic context in which the plosive is produced should be controlled for in investigations of proficiency in achieving articulatory closure. The cardinal symptoms of PD may be substantially reduced by deep brain stimulation (DBS), where electrodes are neurosurgically implanted in the subthalamic nucleus (STN) of the brain (see, e.g., Krack et al., 2003; Tripoliti et al., 2011). STN-DBS may, however, have negative longterm effects on perceived speech quality, especially if implantation is bilateral (see, e.g., Krack et al., 2003; Romito et al., 2009; Tripoliti et al., 2011). In accordance with the proposed lateralization of speech-related brain structures (Belin et al., 1998; Zatorre, Evans, & Meyer, 1992), temporal and segmental aspects of speech have been shown to be influenced mainly by unilateral (left) or bilateral STN stimulation (Schulz et al., 2012). The caudal zona incerta (cZi) has been suggested as an alternative target for treatment of motor symptoms of PD, with an initial report suggesting greater improvements in motor effects compared with STN-DBS (Plaha, Ben-Shlomo, Patel, & Gill, 2006). Detailed articulatory effects of cZi-DBS compared with STN-DBS have been the focus of a previous study from our group (Karlsson et al., 2011), comparing articulatory rate and precision off and on stimulation for the STN and cZi targets. Percent measurable voice onset time (VOT) (Auzou et al., 2000; Ozsancak et al., 2001) in diadochokinetic speech tasks was used to estimate articulatory precision, and the results showed that the patients treated with cZi-DBS exhibited a decreased proficiency in reaching the articulatory targets in plosives when stimulation was on. The findings also indicated that articulation rate may be improved by STN-DBS in simple speech motor tasks (Karlsson et al., 2011). Patients treated with cZi-DBS, however, showed a reduced articulation rate, as well as a reduction in quality of articulation (as measured by the percent measurable VOT metric) when stimulation was on compared with off. The patients treated with cZi-DBS additionally showed a decrease in articulatory proficiency, particularly at increased articulation rate, which was not found for those treated with STN-DBS. This lends further support to the conclusion that articulatory proficiency in patients treated with cZi-DBS may be more adversely affected by the stimulation than those treated with STN-DBS. However, although the percent measurable VOT metric of articulatory proficiency is a useful indicator of the overall success in achieving the articulatory target, it does not provide any information about which articulatory aspect has failed in the production of the plosive. Indeed, several underlying factors that may cause failed VOT measurements have been identified in the literature, including spirantization of the occlusion (Chenausky et al., 2011; Dromey & Bjarnason, 2011; Kent & Kim, 2003) and absence of a release burst (Ozsancak et al., 2001).
Karlsson et al.: Articulatory Closure Proficiency in Patients With PD
1179
The aim of the present study was, therefore, to provide a detailed, comprehensive investigation of acoustic indicators of patients’ proficiency in achieving full articulatory closure during plosive production. The acoustic quantities investigated were those that distinguish plosives from fricatives: the relative duration of aperiodic noise and the distinctiveness of the plosive release relative to the acoustic energy of the surrounding signal. Treatment effects of STN-DBS and cZi-DBS were compared in terms of their ability to support clear plosive productions by the patients.
Method Subjects Nineteen patients (15 males and four females) with idiopathic PD participated in this prospective, nonrandomized study. Nine patients had implantations in the STN (seven bilateral and two left-side unilateral), and 10 had implantations bilaterally in the cZi. An overview of the patients is presented in Table 1. Patients had been selected on clinical grounds for DBS surgery based on assessment of overall motor function; no consideration was taken with
Table 1. Summary of the 17 bilaterally and two unilaterally operated patients.
Patient
Sex
Age at surgery (years)
Implantation
Disease duration (years)
Pre-op
12 months post-op
Medication
LED (mg)
Medication L-Dopa, Entacapone, Pramipexole L-Dopa, Entacapone, Pramipexole L-Dopa, Pramipexole L-Dopa L-Dopa, Entacapone, Pramipexole L-Dopa, Entacapone, Pramipexole L-Dopa, Entacapone, Pramipexole
STN 1
M
52.5
Bilateral
11
L-Dopa, Entacapone
1,048
STN 2
M
68.3
Bilateral
9
1,463
STN 3 STN 4 STN 5
M M M
65.8 58.4 64.2
Unilateral (left) Bilateral Bilateral
7 7 5
STN 6
F
51.0
Bilateral
8
STN 7
F
53.3
Bilateral
8
STN 8 STN 9
M M
70.3 72.5
Bilateral Unilateral (left)
4 6
L-Dopa, Entacapone, Pramipexole L-Dopa, Pramipexole L-Dopa, Pramipexole L-Dopa, Entacapone, Pramipexole L-Dopa, Entacapone, Pramipexole Apomorphine hydrochloride,a Entacapone, Pramipexole L-Dopa, Entacapone L-Dopa, Entacapone, Pramipexole
7M/2F
61.8 ± 8.2
7.2 ± 2.1
cZi 1 cZi 2
M F
52.9 66.6
7 bilateral, 2 unilateral (left) Bilateral Bilateral
cZi 3 cZi 4 cZi 5
M M M
70.8 71.4 62.5
Bilateral Bilateral Bilateral
2 7 5
cZi 6
M
49.0
Bilateral
4
cZi 7
M
50.4
Bilateral
6
cZi 8
F
62.5
Bilateral
5
cZi 9
M
51.5
Bilateral
4
cZi 10
M
66.2
Bilateral
10
STN total
cZi total
8M/2F
60.4 ± 8.7
10 Bilateral
10 10
6.3 ± 2.7
900 702 1,996 1,628 2,241
2,046 975
L-Dopa, Ropinirole L-Dopa, Pramipexole
1,444 ± 565 L-Dopa, Pramipexole L-Dopa, Entacapone, Pramipexole, Rasagiline L-Dopa L-Dopa, Pramipexole L-Dopa, Entacapone, Pramipexole L-Dopa, Entacapone, Pramipexole L-Dopa, Entacapone, Pramipexole L-Dopa, Entacapone, Pramipexole L-Dopa, Pramipexole, Selegiline
750 410
848 924
L-Dopa, Pramipexole
1,298 1,098
0 b
864 ± 352
907 1,025 900 700 2,778 1,699 1,341
1,530 1,175 1,239 ± 763
L-Dopa, Entacapone, Pramipexole, Rasagiline L-Dopa L-Dopa, Pramipexole L-Dopa, Entacapone, Pramipexole L-Dopa, Entacapone, Pramipexole L-Dopa, Entacapone, Pramipexole L-Dopa, Pramipexole
600 450 1,397
LED (mg)
Trihexyphenidyl 2 mg 1 × 3
0 387 600 600 1,397 1,248 2,245 950 924
927 ± 654b
Note. Patients are listed in consecutive order, with mean and standard deviations for age and disease duration for the group. Ordinary preoperative and postoperative medications and calculated daily L-dopa equivalent doses (LED) ( Tomlinson et al., 2010) are further indicated for STN patients and for cZi Patients 1–9, along with corresponding group means and standard deviations. Patients STN 7 and cZi 10 received treatment for which LED estimates have not been made available ( Tomlinson et al., 2010). For STN 7, an approximation of pre-op LED is given. For Patient cZi 10, the actual dosage is indicated instead. STN = subthalamic nucleus; M = male; F = female; cZi = caudal zona incerta. a
Apomorphine hydrochloride is not included in Tomlinson et al. (2010). The calculations here are based on a 96-mg daily dosage, which is estimated to 960 mg daily LED. bPatient cZi 10 is excluded from these descriptive statistics.
1180 Journal of Speech, Language, and Hearing Research • Vol. 57 • 1178–1190 • August 2014
regard to speech status. A perioperative CT fused with the preoperative MRI confirmed correct location of the electrodes in all patients. The general motor function was assessed according to the Motor subscale (Part III) of the Unified Parkinson’s Disease Rating Scale (UPDRS III). All subjects gave their written informed consent after receiving information on the details of the study according to the Helsinki declaration. The study was approved by the Regional Ethical Review Board in Umeå (Dnr: 08-093M: 2008-08-18). All subjects were native speakers of Swedish.
Study Design All patients were evaluated before surgery (pre-op) and 12 months after surgery (post-op). Pre-op baseline assessments were performed with PD medication 1.5 times the patient’s normal morning dose in order to ensure that the patient was in the “on” state when evaluated (Deuschl et al., 2002). Post-op evaluations were performed under two conditions: off stimulation (Stim OFF) and on stimulation (Stim ON). Both recordings were made on the same day, each within the optimal time in the patients’ medication cycle, and 60 min after DBS had been switched on or off.
Speech Material The speech material on which this study was conducted consisted of readings of an 89-word standard Swedish passage. Due to a change in the protocol used in the longitudinal data collection program, the early recordings contained only a single reading of the passage at the beginning of the recording session. In all other recordings, the passage was read once at the beginning of the recording session and once more at the end of the session. One patient in the STN group (STN 1) received the earlier instructions in all recordings. Five patients in the STN group (STN 2 through STN 6) received the earlier instructions in the pre-op recording but not in post-op recordings. All patients in the cZi group received the recent instructions in all conditions. Recordings were made in a sound-treated booth, using a calibrated headmounted microphone (Sennheiser MKE 2 P-C), with a 15-cm mouth-to-microphone distance. The speech samples were recorded using a digital audio flash recorder (Marantz PMD 660) or a digital audio tape recorder (Panasonic SV 3800) at a 44.1- or 48-kHz sampling rate. A calibration tone (80 dB, 1 kHz) was used at the beginning of each recording in order to afford normalized comparisons of acoustic amplitude across recordings. A total of 107 readings of the standard text underwent a phonetic markup procedure in which all attempted productions of voiceless plosives in stressed words were identified. From the full set of productions, voiceless plosives produced in an utterance initial position were excluded in order to avoid the effect of durational measurements being unduly skewed by the difficulty in separating the silent initial portion of initial plosives from the silence preceding speech. Further, plosives produced before or after a fricative consonant were excluded
in order to avoid confounding results by the spreading of fricated portions from neighboring fricatives. An overview of the plosives finally extracted for subsequent acoustical analysis is presented in Table 2. The overall numbers of plosives for each group are different because the number of subjects is different (9 versus 10), and because some subjects in the STN group did not have a second reading for earlier recording sessions. An overview of the markup procedure and the component parts of the plosive is provided in Figures 1 and 2. In both figures, the markup of [k] produced within the word pojke (“a boy”) is shown, with Figure 1 illustrating a production with good articulatory proficiency, and Figure 2 a production with adverse effects on articulation.
Acoustic Analysis The extracted plosives and plosive components were analyzed acoustically in terms of duration and intensity. The relative duration with frication was computed based on the total duration of frication relative to the total duration of the plosive. The plosive release was identified as the transient with the strongest amplitude in the portion of the signal approximate to where an audible release was perceived. The amplitude of the identified strongest transient (release transient) was subsequently extracted and normalized relative to the calibration tone using the Praat software package (Boersma & Weenink, 2012). Based on the extracted acoustic properties, three parameters of plosive production were derived. First, the relative duration of frication was calculated as the fraction of the plosive duration. In cases where no frication was observed in the plosive, a relative duration of 0% was used. Second, the duration of frication immediately preceding the release transient (the prerelease frication) was measured. In cases where no release transient had been identified, the plosive was excluded from this particular analysis. An illustration of the prerelease frication measure is provided in Figure 2.
Table 2. Overview of the manually transcribed voiceless plosives included in the study, grouped by place of articulation. Group STN cZi Total
Condition
Bilabial
Dental
Velar
Total
Pre-op Post-op, Stim OFF Post-op, Stim ON Pre-op Post-op, Stim OFF Post-op, Stim ON
89 126 136 156 157 148 812
128 181 175 205 206 206 1,099
62 87 90 97 103 100 539
279 394 401 458 466 452 2,450
Note. The total number of each plosive is provided for each condition: preoperative (pre-op) and 12 months postoperative (post-op) made with and without stimulation (Stim ON, StimOFF) for each treatment group.
Karlsson et al.: Articulatory Closure Proficiency in Patients With PD
1181
Figure 1. Example markup of a [k] segment uttered inside a production of pojke (“boy”). The period involving aperiodic noise is marked with a “frication” label, and the portion of the signal in which the release transient is manifested is marked on a separate tier and highlighted in the signal. Patient STN 2 made the production 12 months post-op with stimulation off and on medication.
Third, the prominence of the release transient was calculated as the peak amplitude relative to the amplitude of the surrounding acoustic context. The prominence of the peak was used, rather than a direct measure of peak amplitude, due to the demand that the release peak amplitude not
include the amplitude of other sources, such as simultaneous aperiodic noise or voicing. Thus, it should provide a measure of the transient’s ability to separate perceptually the produced plosive from a production of a homorganic fricative.
Figure 2. Example markup of a [k] segment uttered inside a production of pojke (“boy”). The period involving aperiodic noise is marked with “frication” labels, and the portion of the signal in which the release transient is perceived is marked on a separate tier and highlighted in the signal. The duration of prerelease frication is further illustrated by a horizontal arrow. Patient STN 2 made the production 12 months post-op with stimulation on and on medication.
1182 Journal of Speech, Language, and Hearing Research • Vol. 57 • 1178–1190 • August 2014
In order to calculate the prominence (relative peak amplitude) of the transient, adjacent regions marked as having frication were divided into 10 sections. In cases where no audible release was present, the plosive was excluded from this particular analysis. The amplitudes of these 10 sections were subsequently calculated and normalized relative to the calibration tone in order to provide a repeated estimation of the acoustic energy near the plosive release. A linear model was subsequently fitted to these amplitude measurements. The release peak prominence was then computed as the deviation of the release peak amplitude (normalized by the amplitude of the calibration tone) from the level of the fitted model (i.e., the absolute value of the residual of the model corresponding to the timing of the release peak). The concept of peak prominence has been used previously in other contexts (see, e.g., Hillenbrand, Cleveland, & Erickson, 1994).
Statistical Analysis Measurements of relative duration of frication, duration of prerelease frication, and release prominence were analyzed statistically using linear mixed effects models (Arnau, Bono, & Vallejo, 2009) for each acoustic parameter. Effects of time of recording (pre-op or 12 months post-op) and stimulation condition (Stim OFF or Stim ON) were investigated for each acoustic parameter within each patient. Post hoc testing of significant main effects and interactions were conducted by multiple comparisons of means, using Tukey contrasts within each treatment group.
Reliability The reliability of the acoustic analysis was estimated by repeating the markup and calculations on a randomly selected subset (2.7%) of the data. The two resulting data sets
Figure 3. Mean relative duration of frication in produced voiceless plosives in each condition (pre-op on medication, 12 months post-op off and on stimulation, both on medication) for each group of patients. The number of plosives from which means were calculated is indicated for each stimulation condition. The confidence interval of each mean estimate is indicated by error bars.
Karlsson et al.: Articulatory Closure Proficiency in Patients With PD
1183
(the original and the set resulting from the repeated markup) were examined for test–retest consistency using Cronbach’s alpha for each parameter investigated. The results showed an “Excellent” internal consistency for relative duration of frication (62 items in the original analysis, 63 in the repeated analysis, a = 0.92), an “Acceptable” internal consistency for prerelease frication (32 items in the original analysis, 34 in the repeated analysis, a = 0.78) and “Good” for release transient prominence (53 items in the original analysis, 51 in the repeated analysis, a = 0.78).
Results The relative duration of frication in voiceless plosives is displayed in Figure 3. Statistical testing confirmed a significant main effect of stimulation condition ( pre-op, Stim OFF, and Stim ON), F(2, 2417) = 7.1, p < .001; and stimulation target (STN or cZi), F(1, 17) = 6.1, p < .05. Further, a significant Stimulation Condition × Stimulation Target interaction effect was found, F(2, 2287) = 3.6, p < .05. Post hoc testing confirmed a significantly higher relative duration of frication post-op in the Stim ON condition compared
with Stim OFF ( p = .01) and pre-op ( p < .05) for voiceless plosives produced by patients in the cZi group. Consistent with the findings of the statistical testing, Figure 4 shows increased relative duration of frication for patients in the cZi group. In the STN group, the effects are more mixed, with four patients showing a decrease, two showing a moderate effect, and three showing an overall increase in relative duration of frication. A separate analysis of the duration of frication manifested before the articulatory release transient (prerelease frication) is presented in Figure 5. Statistical testing showed a significant main effect of stimulation condition, F(2, 2427) = 14.1, p < .001; but not for stimulation target, F(1, 17) = 3.2, p = .09. Post hoc analysis confirmed a significant increase in prerelease frication in plosives produced with Stim ON compared with Stim OFF ( p < .01) and pre-op ( p < .01) for the cZi group, but for no other investigated contrasts. As seen in Figure 6, the effect observed for the cZi group is consistent with the pattern of nine out of 10 patients. Patient cZi 7 showed an opposite effect. The within-group variability in the STN group in terms of effect of DBS is substantially larger than within the cZi group.
Figure 4. Mean relative duration of frication during plosive production off stimulation (Stim OFF) and on stimulation (Stim ON) for each patient within each treatment group (STN or cZi). The plosives were produced the same day (1.5 hr apart) 12 months post-op, with patients on medication.
1184 Journal of Speech, Language, and Hearing Research • Vol. 57 • 1178–1190 • August 2014
Figure 5. Mean duration of prerelease frication in voiceless plosives in each condition (pre-op on medication, 12 months post-op off and on stimulation, both on medication) for each group of patients. The number of plosives from which means were calculated is indicated for each stimulation condition. The confidence interval of each mean estimate is indicated by error bars.
The results concerning the release prominence in the produced plosives are presented in Figure 7. For this aspect of plosive production, a significant main effect was found for stimulation condition only, F(2, 2048) = 6.7, p < .01. Post hoc analysis confirmed a significantly higher release transient prominence Stim ON compared with Stim OFF (p < .01) and pre-op (p < .01) recordings for plosives produced by patients in the cZi treatment group. The increase in release prominence observed for STN patients Stim ON compared with Stim OFF and pre-op was not shown to be significant (p = .07; p = .11). Figure 8 shows that, similar to the case of relative duration of frication and prerelease frication, the effect of DBS in terms of release prominence is more coherent in the cZi group compared with the STN group. An overview of the patient-specific effects of either cZi-DBS or STN-DBS treatment (Figures 4, 6, and 8) is provided in Table 3, including postoperative Stim OFF levels as well as the effects of DBS observed for each of the three investigated parameters of plosive production. Further, the effects observed in the parameters are summarized in Table 3 into a description of the overall outcome, where the major effects of DBS are encoded. In the STN group, one patient showed an overall positive effect, demonstrating on average
a more prominent release peak and less average frication. Five patients agreed well with a pattern of an increase in the distinctiveness of the release peak but showed negative effects in terms of increased presence of aperiodic noise. One patient showed an increase in release peak distinctiveness, with no strong effect on the presence of frication. For the remaining two, less distinct release peaks and less frication during plosives were observed. In the cZi treatment group, one patient showed negative effects across all investigated parameters (i.e., a less prominent release peak and more pronounced presence of frication). Six patients showed increased distinctiveness of the release peak together with an increased presence of aperiodic noise in the plosive. Two patients showed a decreased prominence of the release peak, and one showed an increase in the release peak prominence (all without showing a strong effect on the presence of frication).
Discussion Increased spirantization and a reduced plosive release are frequently observed features of the productions of plosives by people with PD. The effect of STN-DBS or
Karlsson et al.: Articulatory Closure Proficiency in Patients With PD
1185
Figure 6. Mean duration of prerelease frication off stimulation (Stim OFF) and on stimulation (Stim ON) for each patient within each treatment group (STN or cZi). The plosives were produced the same day (1.5 hr apart) 12 months post-op, with patients on medication.
cZi-DBS on these acoustic effects of a hypokinetic articulation has, however, not been reported previously. The results presented here show a beneficial effect of cZi-DBS on patients’ ability to manifest a plosive release transient that is distinct from the surrounding acoustic energy. After the effect of energy of other origins than that of the release of articulatory closure was factored out, voiceless plosives were shown to be produced with a stronger release transient on stimulation compared with off stimulation. This effect was observed both on the group level as well as in the individual results of seven of 10 patients. Release transient prominence also increased in patients treated with STN-DBS, but the larger within-group variation resulted in the effect not being significant at the group level. In patients who received STN-DBS, six of nine patients showed an increase in release transient distinctiveness. The data presented here therefore provide evidence that both cZi-DBS and STN-DBS result in a more distinctive release transient that spectrally distinguishes plosives from their corresponding fricatives. The finding is more consistent for patients receiving cZi-DBS. With a more prominent release transient being present in plosives during DBS, the likelihood of plosives being
erroneously perceived as fricatives could be expected to be greatly reduced for patients under both treatments. However, the beneficial effect of DBS did not extend to all aspects of plosive production. The results showed an overall strong presence of frication in plosives on stimulation for both STN-DBS and cZi-DBS treatments. In the STN group, the presence of frication was also strongly present in the Stim OFF and pre-op conditions. The stimulation effects for individual STN-DBS patients also showed large variation, with four subjects improving, four worsening, and one showing no clear effect. For patients receiving cZi-DBS, however, the effect of stimulation is much clearer. Patients showed a strong increase in adverse frication of plosives Stim ON compared with Stim OFF. This effect was observed in seven out of 10 patients (the remaining three showed no clear effect). Similarly, there was more frication before the release transient with Stim ON compared with Stim OFF. Thus, when stimulation was turned on, plosives were produced with a higher degree of frication and longer periods of fricative energy. The observed production pattern is, therefore, not consistent with an effective articulatory movement for plosives that have strong supraglottal pressure in typical speech. Further, reduced perceptual distinctiveness of the plosive
1186 Journal of Speech, Language, and Hearing Research • Vol. 57 • 1178–1190 • August 2014
Figure 7. Mean release prominence in voiceless plosives in each condition (pre-op on medication, 12 months post-op off and on stimulation, both on medication) for each group of patients. The number of plosives from which means were calculated is indicated for each stimulation condition. The confidence interval of each mean estimate is indicated by error bars.
relative to fricatives could be expected due to this adverse effect of DBS treatment. Taken in combination, the two findings indicate that, although the patients may have been able to achieve an acoustically distinct release transient, indicative of a more successful articulatory closure, they showed proportionally more acoustic energy in articulatory states prior to and following full articulatory closure. A stronger presence of this latter adverse effect was observed for the cZi-DBS group, but a similar effect was also present in the STN-DBS group. Thus, the results are consistent with a conclusion that patients were able to generate an increased total articulatory energy during plosive production on stimulation compared with off stimulation and were able to manifest a more forceful release where required. However, an increased proportion of the energy was spent during incomplete articulatory closure, reducing the articulatory quality of the plosive. This effect was seen both for the STN-DBS and cZi-DBS groups, but the cZi-DBS patients appeared to be more affected by the adverse aspects of increased frication than were the STNDBS patients.
The results presented here extend the results of the one previously published report in which effects of STN-DBS or cZi-DBS on articulatory proficiency have been compared (Karlsson et al., 2011). The cZi-DBS patients were observed to be more adversely affected by stimulation in terms of performance in speech-related motor task (oral diadochokinesis) compared with STN-DBS patients (Karlsson et al., 2011), especially when the quality of the produced plosives is considered. The results from the present investigation show that what is affected, in general, is the proficiency in suppressing aperiodic noise in the portion of the plosive where full closure between articulators would have been expected in clear speech. As a consequence, patients treated with DBS should be expected to be perceived as imprecise in their speech during stimulation. Patients with cZi-DBS were more uniformly affected by this reduction in articulatory proficiency, whereas those with STN-DBS showed a large degree of within-group variation. The large degree of individual variation in the STN group agrees well with previous reports focusing on the spirantization of plosives in stimulated patients (Dromey &
Karlsson et al.: Articulatory Closure Proficiency in Patients With PD
1187
Figure 8. Mean plosive release prominence off stimulation (Stim OFF) and on stimulation (Stim ON) for each patient within each treatment group (STN or cZi). The plosives were produced the same day (1.5 hr apart) 12 months post-op, with patients on medication.
Bjarnason, 2011). As an increase in frication generally may result in skewing of the spectrum toward higher frequencies, the results presented here could further help explain the increased long-term average spectrum (LTAS) averages due to STN-DBS (Tripoliti et al., 2011). The increased release peak amplitude and increased presence of frication agree well with both the improved articulatory muscle force and the improved peak force precision observed previously in patients treated with STN-DBS (Pinto, Pollák, Benabid, Gentil, & Fraix, 2003). However, the release prominence data presented here show a stronger presence of individual variation than the evaluation of articulator force (Pinto et al., 2003). Further, whereas articulatory muscle force estimated the muscular ability of the articulator, the release prominence quantity investigated here was able to estimate the acoustic consequence of a potentially more forceful closure action between articulators, resulting in a perceptually clearer plosive rather than one perceived as a near-fricative. One could note that the effects observed for patients with DBS frequently agree well with the nature of speech articulation effects observed due to other causes, such as intoxication (Pisoni & Martin, 1989). Thus, although it is unlikely that this reduction in
quality alone would lead to a large reduction of speech intelligibility, the effects observed here may certainly affect patients adversely in their everyday communication.
Conclusion This article has provided evidence of a more prominent acoustic release, indicative of a more successful articulatory closure, in plosives produced by patients with Parkinson’s disease who have undergone either STN-DBS or cZi-DBS treatment. Adverse effects in terms of an increased presence of frication during plosive production were, however, also observed on stimulation compared with off stimulation. It is proposed that more energy is made available during the occlusion phase on stimulation but that effective use of this energy is prevented by slowed or inappropriate articulatory closure and release action. As a result of these acoustic effects, patients may be perceived as having a reduced speech articulation. Patients in both the STN-DBS and the cZi-DBS groups showed similar effects of their DBS treatment on plosive production, but the adverse effects appeared to be more homogeneously present in the cZi-DBS group.
1188 Journal of Speech, Language, and Hearing Research • Vol. 57 • 1178–1190 • August 2014
Table 3. Summary of patient-specific outcomes of deep brain stimulation (DBS). Release prominence Patient
Stim OFF
Stim ON
DBS outcome
Relative duration of frication Stim OFF
STN 1
High
Decrease
Worsening
Mid
STN 2
Low
Increase
Improvement Low
STN 3
High
STN 4
Mid
Moderate Worsening decrease Decrease Worsening
STN 5
Low
STN 6
High High
DBS outcome
Stim ON
Prerelease frication Stim OFF
Stim ON
Decrease
Improvement Mid
Decrease
Increase
Worsening
Increase
Low
Moderate Improvement Mid decrease Decrease Improvement High
Comparable
Improvement Low
Comparable No effect
Low
Mid
Moderate increase Increase
Improvement Mid
Increase
Worsening
Mid
Moderate increase Moderate increase Increase
STN 7
Mid
Increase
Improvement Low
Increase
Worsening
Low
Increase
STN 8
High
Increase
Improvement Low
Increase
Worsening
Mid
Increase
STN 9
Mid
Increase
Improvement High
Decrease
Improvement High
Decrease
cZi 1
High
Increase
Improvement Mid
Increase
Worsening
Mid
Increase
cZi 2
Mid
Improvement Low
Increase
Worsening
Mid
Comparable
cZi 3
Low
Moderate increase Increase
Improvement Low
Worsening
Low
cZi 4
High
Decrease
Worsening
Moderate increase Increase
Worsening
High
cZi 5
Mid
Increase
Improvement Mid
Increase
Worsening
High
Moderate increase Moderate increase Increase
cZi 6
Mid
Increase
Worsening
Mid
Comparable
cZi 7
Mid
Comparable No effect
High
Decrease
cZi 8
Mid
Moderate Improvement Low increase Moderate Worsening Mid decrease Increase Improvement Low
Increase
Low
Increase
cZi 9 cZi 10
High High
Increase Decrease
Comparable No effect Comparable No effect
Mid Mid
Increase Comparable
Mid
Improvement High Worsening High
Worsening
DBS outcome
Overall DBS outcome
Improvement Less distinct release peak, less frication Worsening Improved release peak, more frication No effect Less distinct release peak, less frication Worsening Improved release peak, more frication Worsening Improved release peak Worsening Improved release peak, more frication Worsening Improved release peak, more frication Worsening Improved release peak, more frication Improvement Overall positive outcome Worsening Improved release peak, more frication No effect Improved release peak, more frication Worsening Improved release peak, more frication Worsening Overall negative outcome Worsening Improved release peak, more frication No effect Improved release peak, more frication Improvement Less distinct release peak Worsening Improved release peak, more frication Worsening Improved release peak No effect Less distinct release peak
Note. Patient-specific outcomes of DBS are given in terms of plosive release prominence, prerelease frication duration, and relative duration of frication during the realization of the plosive. For each parameter investigated, an overview of the patient’s starting level at Stim OFF (12 months post-op, on medication) and change direction in Stim ON recordings (made 1.5 hr later, on medication) compared with the Stim OFF baseline is provided. The development is categorized (“Improvement,” “Worsening,” or “No change”) for each investigated parameter. A broad characterization of the effect of DBS is further provided for each patient.
Acknowledgments We would like to thank deep brain stimulation nurse specialist Anna Fredricks for assistance in providing patient information and research engineer Anders Asplund for preparation of the speech recordings. We wish to acknowledge the support of grants from the Swedish Research Council (Grant No. 2009-946) and Magnus Bergvall’s Foundation in Sweden. A visiting professorship at the Rehabilitation Research Chair, King Saud University, Riyadh, Saudi Arabia, to Erik Nordh is gratefully acknowledged.
References Ackermann, H., Hertrich, I., Daum, I., Scharf, G., & Spieker, S. (1997). Kinematic analysis of articulatory movements in
central motor disorders. Movement Disorders, 12, 1019–1027. doi:10.1002/mds.870120628 Ackermann, H., & Ziegler, W. (1991). Articulatory deficits in Parkinsonian dysarthria: An acoustic analysis. Journal of Neurology, Neurosurgery & Psychiatry, 54, 1093–1098. doi:10.1136/jnnp.54.12.1093 Arnau, J., Bono, R., & Vallejo, G. (2009). Analyzing small samples of repeated measures data with the mixed-model adjusted F test. Communications in Statistics—Simulation and Computation, 38, 1083–1103. doi:10.1080/03610910902785746 Auzou, P., Ozsancak, C., Morris, R. J., Jan, M., Eustache, F., & Hannequin, D. (2000). Voice onset time in aphasia, apraxia of speech and dysarthria: A review. Clinical Linguistics & Phonetics, 14, 131–150. doi:10.1080/026992000298878 Belin, P., Zilbovicius, M., Crozier, S., Thivard, L., Fontaine, A., Masure, M. C., & Samson, Y. (1998). Lateralization of speech
Karlsson et al.: Articulatory Closure Proficiency in Patients With PD
1189
and auditory temporal processing. Journal of Cognitive Neuroscience, 10, 536–540. Boersma, P., & Weenink, D. (2012). Praat (Version 5.3.04) [Computer software]. Retrieved from http://www.praat.org/ Chenausky, K., MacAuslan, J., & Goldhor, R. (2011). Acoustic analysis of PD speech. Parkinson’s Disease, 2011, 1–13. doi:10.4061/2011/435232 Deuschl, G., Fogel, W., Hahne, M., Kupsch, A., Müller, D., Oechsner, M., . . . Volkmann, J. (2002). Deep-brain stimulation for Parkinson’s disease. Journal of Neurology, 249, iii36–iii39. doi:10.1007/s00415-002-1308-x Dromey, C., & Bjarnason, S. (2011). A preliminary report on disordered speech with deep brain stimulation in individuals with Parkinson’s disease. Parkinson’s Disease, 2011, 1–11. doi:10.4061/2011/796205 Duez, D. (2007, August). Acoustic analysis of occlusive weakening in Parkinsonian French speech. Paper presented at the 16th International Congress of Phonetic Sciences, Saarbrücken, Germany. Forrest, K. (1989). Kinematic, acoustic, and perceptual analyses of connected speech produced by Parkinsonian and normal geriatric adults. The Journal of the Acoustical Society of America, 85, 2608. doi:10.1121/1.397755 Hartinger, M., Tripoliti, E., Hardcastle, W. J., & Limousin, P. (2011). Effects of medication and subthalamic nucleus deep brain stimulation on tongue movements in speakers with Parkinson’s disease using electropalatography: A pilot study. Clinical Linguistics & Phonetics, 25, 210–230. doi:10.3109/ 02699206.2010.521877 Hillenbrand, J. M., Cleveland, R. A., & Erickson, R. L. (1994). Acoustic correlates of breathy vocal quality. Journal of Speech and Hearing Research, 37, 769. Karlsson, F., Unger, E., Wahlgren, S., Blomstedt, P., Linder, J., Nordh, E., . . . van Doorn, J. (2011). Deep brain stimulation of caudal zona incerta and subthalamic nucleus in patients with Parkinson’s disease: Effects on diadochokinetic rate. Parkinson’s Disease, 2011, 1–10. doi:10.4061/2011/605607 Kent, R. D., & Kim, Y.-J. (2003). Toward an acoustic typology of motor speech disorders. Clinical Linguistics & Phonetics, 17, 427–445. doi:10.1080/0269920031000086248 Kent, R. D., Weismer, G., Kent, J. F., Vorperian, H. K., & Duffy, J. R. (1999). Acoustic studies of dysarthric speech: Methods, progress, and potential. Journal of Communication Disorders, 32, 141–186. doi:10.1016/S0021-9924(99)00004-0 Krack, P., Batir, A., Van Blercom, N., Chabardes, S., Fraix, V., Ardouin, C., . . . Pollák, P. (2003). Five-year follow-up of bilateral stimulation of the subthalamic nucleus in advanced Parkinson’s disease. New England Journal of Medicine, 349, 1925–1934. doi:10.1056/NEJMoa035275 Logemann, J. A., & Fisher, H. B. (1981). Vocal tract control in Parkinson’s disease. Journal of Speech and Hearing Disorders, 46, 348. McAuliffe, M. J., Ward, E. C., & Murdoch, B. E. (2006a). Speech production in Parkinson’s disease: I. An electropalatographic investigation of tongue-palate contact patterns. Clinical Linguistics & Phonetics, 20, 1–18. McAuliffe, M. J., Ward, E. C., & Murdoch, B. E. (2006b). Speech production in Parkinson’s disease: II. Acoustic and electropalatographic investigation of sentence, word and segment durations. Clinical Linguistics & Phonetics, 20, 19–33. doi:10.1080/ 0269-9200400001069
Ozsancak, C., Auzou, P., Jan, M., & Hannequin, D. (2001). Measurement of voice onset time in dysarthric patients: Methodological considerations. Folia Phoniatrica et Logopaedica, 53, 48–57. doi:10.1159/000052653 Pinto, S., Pollák, P., Benabid, A.-L., Gentil, M., & Fraix, V. (2003). Bilateral subthalamic stimulation effects on oral force control in Parkinson’s disease. Journal of Neurology, 250, 179–187. doi:10.1007/s00415-003-0966-7 Pisoni, D. B., & Martin, C. S. (1989). Effects of alcohol on the acoustic-phonetic properties of speech: Perceptual and acoustic analyses. Alcoholism: Clinical and Experimental Research, 13, 577–587. doi:10.1111/j.1530-0277.1989.tb00381.x Plaha, P., Ben-Shlomo, Y., Patel, N. K., & Gill, S. S. (2006). Stimulation of the caudal zona incerta is superior to stimulation of the subthalamic nucleus in improving contralateral parkinsonism. Brain, 129(Pt. 7), 1732–1747. doi:10.1093/brain/awl127 Romito, L. M., Contarino, M. F., Vanacore, N., Bentivoglio, A. R., Scerrati, M., & Albanese, A. (2009). Replacement of dopaminergic medication with subthalamic nucleus stimulation in Parkinson’s disease: Long-term observation. Movement Disorders, 24, 557–563. doi:10.1002/mds.22390 Roy, N., Leeper, H. A., Blomgren, M., & Cameron, R. M. (2001). A description of phonetic, acoustic, and physiological changes associated with improved intelligibility in a speaker with spastic dysarthria. American Journal of Speech-Language Pathology, 10, 274. doi:10.1044/1058-0360(2001/025) Rusz, J., Cmejla, R., Ruzickova, H., & Ruzicka, E. (2011). Quantitative acoustic measurements for characterization of speech and voice disorders in early untreated Parkinson’s disease. The Journal of the Acoustical Society of America, 129, 350–367. doi:10.1121/1.3514381 Schulz, G. M., Hosey, L. A., Bradberry, T. J., Stager, S. V., Lee, L.-C., Pawha, R., . . . Braun, A. R. (2012). Selective left, right and bilateral stimulation of subthalamic nuclei in Parkinson’s disease: Differential effects on motor, speech and language function. Journal of Parkinson’s Disease, 2, 29–40. Stevens, K. N. (2000). Acoustic phonetics. Cambridge, MA: MIT Press. Tomlinson, C. L., Stowe, R., Patel, S., Rick, C., Gray, R., & Clarke, C. E. (2010). Systematic review of levodopa dose equivalency reporting in Parkinson’s disease. Movement Disorders, 25, 2649–2653. doi:10.1002/mds.23429 Tripoliti, E., Zrinzo, L. U., Martinez-Torres, I., Frost, E., Pinto, S., Foltynie, T., . . . Limousin, P. (2011). Effects of subthalamic stimulation on speech of consecutive patients with Parkinson disease. Neurology, 76, 80–86. doi:10.1212/WNL. 0b013e318203e7d0 Wang, Y.-T., Kent, R. D., Duffy, J. R., Thomas, J. E., & Weismer, G. (2004). Alternating motion rate as an index of speech motor disorder in traumatic brain injury. Clinical Linguistics & Phonetics, 18, 57–84. Weismer, G. (1984). Articulatory characteristics of Parkinsonian dysarthria: Segmental and phrase-level timing, spirantization, and glottal-supraglottal coordination. In M. McNeil, J. Rosenbeck, & A. Aronson (Eds.), The dysarthrias: Physiology, acoustics, perception, management ( pp. 101–130). San Diego, CA: CollegeHill Press. Zatorre, R. J., Evans, A., & Meyer, E. (1992, May 8). Lateralization of phonetic and pitch discrimination in speech processing. Science, 256, 846–849. doi:10.1126/science.256.5058.846
1190 Journal of Speech, Language, and Hearing Research • Vol. 57 • 1178–1190 • August 2014
Copyright of Journal of Speech, Language & Hearing Research is the property of American Speech-Language-Hearing Association and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.
Research Article
Articulatory Closure Proficiency in Patients With Parkinson’s Disease Following Deep Brain Stimulation of the Subthalamic Nucleus and Caudal Zona Incerta Fredrik Karlsson,a Katarina Olofsson,a Patric Blomstedt,a Jan Linder,a Erik Nordh,a and Jan van Doorna
Purpose: The present study aimed at comparing the effects of deep brain stimulation (DBS) treatment of the subthalamic nucleus (STN) and the caudal zona incerta (cZi) on the proficiency in achieving oral closure and release during plosive production of people with Parkinson’s disease. Method: Nineteen patients participated preoperatively and 12 months after DBS surgery. Nine patients had implantations in the STN, 7 bilaterally and 2 unilaterally (left). Ten had bilateral implantations in the cZi. Postoperative examinations were made off and on stimulation. All patients received simultaneous L-dopa treatment in all conditions. For a series of plosives extracted from a reading passage, absolute and relative measures of duration of frication and amplitude of plosive release were compared between conditions within each treatment group.
Results: Relative duration of frication increased in voiceless plosives in the on-stimulation condition in cZi patients. Similar trends were observed across the data set. Duration of prerelease frication and the release peak prominence increased in voiceless plosives on stimulation for both groups. Conclusion: The increased release prominence suggests that patients achieved a stronger closure gesture because of DBS but that the increased energy available resulted in increased frication.
T
Cmejla, Ruzickova, & Ruzicka, 2011). In particular, the production of plosive consonants is highly dependent on achieving full closure between articulators. Full closure allows a buildup of intra-oral pressure that is required for the plosive to be perceived as having been realized in an appropriate manner. Plosives are therefore particularly affected by the reduction in articulatory range in people with PD, which may cause incomplete closure between articulators (Ackermann & Ziegler, 1991). The most frequently discussed acoustic consequence of articulatory undershoot during closure is a strong inappropriate aperiodic component in place of the silent portion during occlusion (Ackermann & Ziegler, 1991; Chenausky et al., 2011; Dromey & Bjarnason, 2011; Forrest, 1989; Hartinger, Tripoliti, Hardcastle, & Limousin, 2011; Kent & Kim, 2003; Logemann & Fisher, 1981). Thus, the incomplete articulatory closure in plosives produced by people with PD leads to an increase in portions with aperiodic noise.
he hypokinetic dysarthria associated with Parkinson’s disease (PD) reduces proficiency in reaching articulatory targets. Specifically, imprecision in consonant articulation has been argued to be a hallmark of dysarthria (Kent & Kim, 2003). Numerous reports have noted that a reduction of articulatory extent affects many consonant types, especially those demanding close proximity between articulators (Ackermann & Ziegler, 1991; Chenausky, MacAuslan, & Goldhor, 2011; Dromey & Bjarnason, 2011; Karlsson et al., 2011; Kent, Weismer, Kent, Vorperian, & Duffy, 1999; McAuliffe, Ward, & Murdoch, 2006b; Rusz,
a
Umeå University, Sweden
Correspondence to Fredrik Karlsson: [email protected] Editor: Jody Kreiman Associate Editor: Julie Liss Received January 10, 2013 Revision received August 20, 2013 Accepted November 20, 2013 DOI: 10.1044/2014_JSLHR-S-13-0010
1178
Key Words: dysarthria, speech motor control, neurologic disorders
Disclosure: The authors have declared that no competing interests existed at the time of publication.
Journal of Speech, Language, and Hearing Research • Vol. 57 • 1178–1190 • August 2014 • A American Speech-Language-Hearing Association
The plosives are therefore less perceptually distinct from comparable fricatives, a process referred to as spirantization. A unified acoustic quantification does not exist for spirantization in plosives. Early assessments were based on visual information from the waveform display of the acoustic signal, and the plosive was marked as including spirantization based on the shape of the waveform (Weismer, 1984). Methods attempting to quantify the degree of spirantization in a plosive include measures of the intensity during closure (IDC) (Ackermann & Ziegler, 1991) and the intensity of closure relative to that of the following vowel, or to the mean intensity of the whole syllable (Chenausky et al., 2011; Dromey & Bjarnason, 2011). Results from these measures have been inconclusive, with patients with PD often showing more frequent (Weismer, 1984) and stronger (Ackermann & Ziegler, 1991) spirantization compared with controls, but with a substantial portion of patients falling well within the limits observed in controls. This suggests that PD may not always impair proficiency in achieving full articulatory closure during plosive production. It may, however, also be viewed as an indication that incomplete articulatory closure may not be appropriately captured by a single acoustic dimension or quantification. Alternative acoustic consequences of incomplete articulatory closure may be proposed, based on models of the dynamics of stop/plosive production (Stevens, 2000). A direct consequence of incomplete articulatory closure is a reduction of oral pressure build-up in the occlusion phase of the plosive. If full closure is not achieved, the energy provided by the respiratory system will not cause a sufficiently abrupt and forceful articulatory release, so that the acoustic transient will not be as acoustically prominent. Thus, in addition to the possible presence of spirantization, there may also be a reduced release transient amplitude in plosives produced with hypokinetic articulation, resulting in even less perceptual distinction from comparable fricatives. This acoustic property of hypokinetic speech movements has received much less attention than the spirantization component, but it has been noted in some previous reports (Auzou et al., 2000; Duez, 2007; Karlsson et al., 2011; Ozsancak, Auzou, Jan, & Hannequin, 2001; Wang, Kent, Duffy, Thomas, & Weismer, 2004). Slowed, more varied (McAuliffe, Ward, & Murdoch, 2006b), or less forceful (McAuliffe, Ward, & Murdoch, 2006a) articulatory release movements observed (through electropalatograpic recordings) for some consonant types with apical closure may, if extended to the production of plosives, further create more diffuse releases by increasing the time in near contact between articulators. Taken together, the acoustic consequences of hypokinetic articulation is very likely to cause plosives to be acoustically similar to fricatives, in accordance with earlier descriptions in the literature (Ackermann, Hertrich, Daum, Scharf, & Spieker, 1997; Ackermann & Ziegler, 1991; Auzou et al., 2000; Chenausky et al., 2011; Forrest, 1989; Hartinger et al., 2011; Logemann & Fisher, 1981; Roy, Leeper, Blomgren, & Cameron, 2001; Wang et al., 2004). Investigation of increased overall acoustic distinctiveness between plosives and fricatives is therefore seen as an intuitive and productive
methodology for studying articulatory closure proficiency as well as for quantifying treatment effects. Results from Ackermann and Ziegler (1991) indicate that plosives in unstressed syllables are more reduced in patients with PD compared to controls. Thus, the prosodic context in which the plosive is produced should be controlled for in investigations of proficiency in achieving articulatory closure. The cardinal symptoms of PD may be substantially reduced by deep brain stimulation (DBS), where electrodes are neurosurgically implanted in the subthalamic nucleus (STN) of the brain (see, e.g., Krack et al., 2003; Tripoliti et al., 2011). STN-DBS may, however, have negative longterm effects on perceived speech quality, especially if implantation is bilateral (see, e.g., Krack et al., 2003; Romito et al., 2009; Tripoliti et al., 2011). In accordance with the proposed lateralization of speech-related brain structures (Belin et al., 1998; Zatorre, Evans, & Meyer, 1992), temporal and segmental aspects of speech have been shown to be influenced mainly by unilateral (left) or bilateral STN stimulation (Schulz et al., 2012). The caudal zona incerta (cZi) has been suggested as an alternative target for treatment of motor symptoms of PD, with an initial report suggesting greater improvements in motor effects compared with STN-DBS (Plaha, Ben-Shlomo, Patel, & Gill, 2006). Detailed articulatory effects of cZi-DBS compared with STN-DBS have been the focus of a previous study from our group (Karlsson et al., 2011), comparing articulatory rate and precision off and on stimulation for the STN and cZi targets. Percent measurable voice onset time (VOT) (Auzou et al., 2000; Ozsancak et al., 2001) in diadochokinetic speech tasks was used to estimate articulatory precision, and the results showed that the patients treated with cZi-DBS exhibited a decreased proficiency in reaching the articulatory targets in plosives when stimulation was on. The findings also indicated that articulation rate may be improved by STN-DBS in simple speech motor tasks (Karlsson et al., 2011). Patients treated with cZi-DBS, however, showed a reduced articulation rate, as well as a reduction in quality of articulation (as measured by the percent measurable VOT metric) when stimulation was on compared with off. The patients treated with cZi-DBS additionally showed a decrease in articulatory proficiency, particularly at increased articulation rate, which was not found for those treated with STN-DBS. This lends further support to the conclusion that articulatory proficiency in patients treated with cZi-DBS may be more adversely affected by the stimulation than those treated with STN-DBS. However, although the percent measurable VOT metric of articulatory proficiency is a useful indicator of the overall success in achieving the articulatory target, it does not provide any information about which articulatory aspect has failed in the production of the plosive. Indeed, several underlying factors that may cause failed VOT measurements have been identified in the literature, including spirantization of the occlusion (Chenausky et al., 2011; Dromey & Bjarnason, 2011; Kent & Kim, 2003) and absence of a release burst (Ozsancak et al., 2001).
Karlsson et al.: Articulatory Closure Proficiency in Patients With PD
1179
The aim of the present study was, therefore, to provide a detailed, comprehensive investigation of acoustic indicators of patients’ proficiency in achieving full articulatory closure during plosive production. The acoustic quantities investigated were those that distinguish plosives from fricatives: the relative duration of aperiodic noise and the distinctiveness of the plosive release relative to the acoustic energy of the surrounding signal. Treatment effects of STN-DBS and cZi-DBS were compared in terms of their ability to support clear plosive productions by the patients.
Method Subjects Nineteen patients (15 males and four females) with idiopathic PD participated in this prospective, nonrandomized study. Nine patients had implantations in the STN (seven bilateral and two left-side unilateral), and 10 had implantations bilaterally in the cZi. An overview of the patients is presented in Table 1. Patients had been selected on clinical grounds for DBS surgery based on assessment of overall motor function; no consideration was taken with
Table 1. Summary of the 17 bilaterally and two unilaterally operated patients.
Patient
Sex
Age at surgery (years)
Implantation
Disease duration (years)
Pre-op
12 months post-op
Medication
LED (mg)
Medication L-Dopa, Entacapone, Pramipexole L-Dopa, Entacapone, Pramipexole L-Dopa, Pramipexole L-Dopa L-Dopa, Entacapone, Pramipexole L-Dopa, Entacapone, Pramipexole L-Dopa, Entacapone, Pramipexole
STN 1
M
52.5
Bilateral
11
L-Dopa, Entacapone
1,048
STN 2
M
68.3
Bilateral
9
1,463
STN 3 STN 4 STN 5
M M M
65.8 58.4 64.2
Unilateral (left) Bilateral Bilateral
7 7 5
STN 6
F
51.0
Bilateral
8
STN 7
F
53.3
Bilateral
8
STN 8 STN 9
M M
70.3 72.5
Bilateral Unilateral (left)
4 6
L-Dopa, Entacapone, Pramipexole L-Dopa, Pramipexole L-Dopa, Pramipexole L-Dopa, Entacapone, Pramipexole L-Dopa, Entacapone, Pramipexole Apomorphine hydrochloride,a Entacapone, Pramipexole L-Dopa, Entacapone L-Dopa, Entacapone, Pramipexole
7M/2F
61.8 ± 8.2
7.2 ± 2.1
cZi 1 cZi 2
M F
52.9 66.6
7 bilateral, 2 unilateral (left) Bilateral Bilateral
cZi 3 cZi 4 cZi 5
M M M
70.8 71.4 62.5
Bilateral Bilateral Bilateral
2 7 5
cZi 6
M
49.0
Bilateral
4
cZi 7
M
50.4
Bilateral
6
cZi 8
F
62.5
Bilateral
5
cZi 9
M
51.5
Bilateral
4
cZi 10
M
66.2
Bilateral
10
STN total
cZi total
8M/2F
60.4 ± 8.7
10 Bilateral
10 10
6.3 ± 2.7
900 702 1,996 1,628 2,241
2,046 975
L-Dopa, Ropinirole L-Dopa, Pramipexole
1,444 ± 565 L-Dopa, Pramipexole L-Dopa, Entacapone, Pramipexole, Rasagiline L-Dopa L-Dopa, Pramipexole L-Dopa, Entacapone, Pramipexole L-Dopa, Entacapone, Pramipexole L-Dopa, Entacapone, Pramipexole L-Dopa, Entacapone, Pramipexole L-Dopa, Pramipexole, Selegiline
750 410
848 924
L-Dopa, Pramipexole
1,298 1,098
0 b
864 ± 352
907 1,025 900 700 2,778 1,699 1,341
1,530 1,175 1,239 ± 763
L-Dopa, Entacapone, Pramipexole, Rasagiline L-Dopa L-Dopa, Pramipexole L-Dopa, Entacapone, Pramipexole L-Dopa, Entacapone, Pramipexole L-Dopa, Entacapone, Pramipexole L-Dopa, Pramipexole
600 450 1,397
LED (mg)
Trihexyphenidyl 2 mg 1 × 3
0 387 600 600 1,397 1,248 2,245 950 924
927 ± 654b
Note. Patients are listed in consecutive order, with mean and standard deviations for age and disease duration for the group. Ordinary preoperative and postoperative medications and calculated daily L-dopa equivalent doses (LED) ( Tomlinson et al., 2010) are further indicated for STN patients and for cZi Patients 1–9, along with corresponding group means and standard deviations. Patients STN 7 and cZi 10 received treatment for which LED estimates have not been made available ( Tomlinson et al., 2010). For STN 7, an approximation of pre-op LED is given. For Patient cZi 10, the actual dosage is indicated instead. STN = subthalamic nucleus; M = male; F = female; cZi = caudal zona incerta. a
Apomorphine hydrochloride is not included in Tomlinson et al. (2010). The calculations here are based on a 96-mg daily dosage, which is estimated to 960 mg daily LED. bPatient cZi 10 is excluded from these descriptive statistics.
1180 Journal of Speech, Language, and Hearing Research • Vol. 57 • 1178–1190 • August 2014
regard to speech status. A perioperative CT fused with the preoperative MRI confirmed correct location of the electrodes in all patients. The general motor function was assessed according to the Motor subscale (Part III) of the Unified Parkinson’s Disease Rating Scale (UPDRS III). All subjects gave their written informed consent after receiving information on the details of the study according to the Helsinki declaration. The study was approved by the Regional Ethical Review Board in Umeå (Dnr: 08-093M: 2008-08-18). All subjects were native speakers of Swedish.
Study Design All patients were evaluated before surgery (pre-op) and 12 months after surgery (post-op). Pre-op baseline assessments were performed with PD medication 1.5 times the patient’s normal morning dose in order to ensure that the patient was in the “on” state when evaluated (Deuschl et al., 2002). Post-op evaluations were performed under two conditions: off stimulation (Stim OFF) and on stimulation (Stim ON). Both recordings were made on the same day, each within the optimal time in the patients’ medication cycle, and 60 min after DBS had been switched on or off.
Speech Material The speech material on which this study was conducted consisted of readings of an 89-word standard Swedish passage. Due to a change in the protocol used in the longitudinal data collection program, the early recordings contained only a single reading of the passage at the beginning of the recording session. In all other recordings, the passage was read once at the beginning of the recording session and once more at the end of the session. One patient in the STN group (STN 1) received the earlier instructions in all recordings. Five patients in the STN group (STN 2 through STN 6) received the earlier instructions in the pre-op recording but not in post-op recordings. All patients in the cZi group received the recent instructions in all conditions. Recordings were made in a sound-treated booth, using a calibrated headmounted microphone (Sennheiser MKE 2 P-C), with a 15-cm mouth-to-microphone distance. The speech samples were recorded using a digital audio flash recorder (Marantz PMD 660) or a digital audio tape recorder (Panasonic SV 3800) at a 44.1- or 48-kHz sampling rate. A calibration tone (80 dB, 1 kHz) was used at the beginning of each recording in order to afford normalized comparisons of acoustic amplitude across recordings. A total of 107 readings of the standard text underwent a phonetic markup procedure in which all attempted productions of voiceless plosives in stressed words were identified. From the full set of productions, voiceless plosives produced in an utterance initial position were excluded in order to avoid the effect of durational measurements being unduly skewed by the difficulty in separating the silent initial portion of initial plosives from the silence preceding speech. Further, plosives produced before or after a fricative consonant were excluded
in order to avoid confounding results by the spreading of fricated portions from neighboring fricatives. An overview of the plosives finally extracted for subsequent acoustical analysis is presented in Table 2. The overall numbers of plosives for each group are different because the number of subjects is different (9 versus 10), and because some subjects in the STN group did not have a second reading for earlier recording sessions. An overview of the markup procedure and the component parts of the plosive is provided in Figures 1 and 2. In both figures, the markup of [k] produced within the word pojke (“a boy”) is shown, with Figure 1 illustrating a production with good articulatory proficiency, and Figure 2 a production with adverse effects on articulation.
Acoustic Analysis The extracted plosives and plosive components were analyzed acoustically in terms of duration and intensity. The relative duration with frication was computed based on the total duration of frication relative to the total duration of the plosive. The plosive release was identified as the transient with the strongest amplitude in the portion of the signal approximate to where an audible release was perceived. The amplitude of the identified strongest transient (release transient) was subsequently extracted and normalized relative to the calibration tone using the Praat software package (Boersma & Weenink, 2012). Based on the extracted acoustic properties, three parameters of plosive production were derived. First, the relative duration of frication was calculated as the fraction of the plosive duration. In cases where no frication was observed in the plosive, a relative duration of 0% was used. Second, the duration of frication immediately preceding the release transient (the prerelease frication) was measured. In cases where no release transient had been identified, the plosive was excluded from this particular analysis. An illustration of the prerelease frication measure is provided in Figure 2.
Table 2. Overview of the manually transcribed voiceless plosives included in the study, grouped by place of articulation. Group STN cZi Total
Condition
Bilabial
Dental
Velar
Total
Pre-op Post-op, Stim OFF Post-op, Stim ON Pre-op Post-op, Stim OFF Post-op, Stim ON
89 126 136 156 157 148 812
128 181 175 205 206 206 1,099
62 87 90 97 103 100 539
279 394 401 458 466 452 2,450
Note. The total number of each plosive is provided for each condition: preoperative (pre-op) and 12 months postoperative (post-op) made with and without stimulation (Stim ON, StimOFF) for each treatment group.
Karlsson et al.: Articulatory Closure Proficiency in Patients With PD
1181
Figure 1. Example markup of a [k] segment uttered inside a production of pojke (“boy”). The period involving aperiodic noise is marked with a “frication” label, and the portion of the signal in which the release transient is manifested is marked on a separate tier and highlighted in the signal. Patient STN 2 made the production 12 months post-op with stimulation off and on medication.
Third, the prominence of the release transient was calculated as the peak amplitude relative to the amplitude of the surrounding acoustic context. The prominence of the peak was used, rather than a direct measure of peak amplitude, due to the demand that the release peak amplitude not
include the amplitude of other sources, such as simultaneous aperiodic noise or voicing. Thus, it should provide a measure of the transient’s ability to separate perceptually the produced plosive from a production of a homorganic fricative.
Figure 2. Example markup of a [k] segment uttered inside a production of pojke (“boy”). The period involving aperiodic noise is marked with “frication” labels, and the portion of the signal in which the release transient is perceived is marked on a separate tier and highlighted in the signal. The duration of prerelease frication is further illustrated by a horizontal arrow. Patient STN 2 made the production 12 months post-op with stimulation on and on medication.
1182 Journal of Speech, Language, and Hearing Research • Vol. 57 • 1178–1190 • August 2014
In order to calculate the prominence (relative peak amplitude) of the transient, adjacent regions marked as having frication were divided into 10 sections. In cases where no audible release was present, the plosive was excluded from this particular analysis. The amplitudes of these 10 sections were subsequently calculated and normalized relative to the calibration tone in order to provide a repeated estimation of the acoustic energy near the plosive release. A linear model was subsequently fitted to these amplitude measurements. The release peak prominence was then computed as the deviation of the release peak amplitude (normalized by the amplitude of the calibration tone) from the level of the fitted model (i.e., the absolute value of the residual of the model corresponding to the timing of the release peak). The concept of peak prominence has been used previously in other contexts (see, e.g., Hillenbrand, Cleveland, & Erickson, 1994).
Statistical Analysis Measurements of relative duration of frication, duration of prerelease frication, and release prominence were analyzed statistically using linear mixed effects models (Arnau, Bono, & Vallejo, 2009) for each acoustic parameter. Effects of time of recording (pre-op or 12 months post-op) and stimulation condition (Stim OFF or Stim ON) were investigated for each acoustic parameter within each patient. Post hoc testing of significant main effects and interactions were conducted by multiple comparisons of means, using Tukey contrasts within each treatment group.
Reliability The reliability of the acoustic analysis was estimated by repeating the markup and calculations on a randomly selected subset (2.7%) of the data. The two resulting data sets
Figure 3. Mean relative duration of frication in produced voiceless plosives in each condition (pre-op on medication, 12 months post-op off and on stimulation, both on medication) for each group of patients. The number of plosives from which means were calculated is indicated for each stimulation condition. The confidence interval of each mean estimate is indicated by error bars.
Karlsson et al.: Articulatory Closure Proficiency in Patients With PD
1183
(the original and the set resulting from the repeated markup) were examined for test–retest consistency using Cronbach’s alpha for each parameter investigated. The results showed an “Excellent” internal consistency for relative duration of frication (62 items in the original analysis, 63 in the repeated analysis, a = 0.92), an “Acceptable” internal consistency for prerelease frication (32 items in the original analysis, 34 in the repeated analysis, a = 0.78) and “Good” for release transient prominence (53 items in the original analysis, 51 in the repeated analysis, a = 0.78).
Results The relative duration of frication in voiceless plosives is displayed in Figure 3. Statistical testing confirmed a significant main effect of stimulation condition ( pre-op, Stim OFF, and Stim ON), F(2, 2417) = 7.1, p < .001; and stimulation target (STN or cZi), F(1, 17) = 6.1, p < .05. Further, a significant Stimulation Condition × Stimulation Target interaction effect was found, F(2, 2287) = 3.6, p < .05. Post hoc testing confirmed a significantly higher relative duration of frication post-op in the Stim ON condition compared
with Stim OFF ( p = .01) and pre-op ( p < .05) for voiceless plosives produced by patients in the cZi group. Consistent with the findings of the statistical testing, Figure 4 shows increased relative duration of frication for patients in the cZi group. In the STN group, the effects are more mixed, with four patients showing a decrease, two showing a moderate effect, and three showing an overall increase in relative duration of frication. A separate analysis of the duration of frication manifested before the articulatory release transient (prerelease frication) is presented in Figure 5. Statistical testing showed a significant main effect of stimulation condition, F(2, 2427) = 14.1, p < .001; but not for stimulation target, F(1, 17) = 3.2, p = .09. Post hoc analysis confirmed a significant increase in prerelease frication in plosives produced with Stim ON compared with Stim OFF ( p < .01) and pre-op ( p < .01) for the cZi group, but for no other investigated contrasts. As seen in Figure 6, the effect observed for the cZi group is consistent with the pattern of nine out of 10 patients. Patient cZi 7 showed an opposite effect. The within-group variability in the STN group in terms of effect of DBS is substantially larger than within the cZi group.
Figure 4. Mean relative duration of frication during plosive production off stimulation (Stim OFF) and on stimulation (Stim ON) for each patient within each treatment group (STN or cZi). The plosives were produced the same day (1.5 hr apart) 12 months post-op, with patients on medication.
1184 Journal of Speech, Language, and Hearing Research • Vol. 57 • 1178–1190 • August 2014
Figure 5. Mean duration of prerelease frication in voiceless plosives in each condition (pre-op on medication, 12 months post-op off and on stimulation, both on medication) for each group of patients. The number of plosives from which means were calculated is indicated for each stimulation condition. The confidence interval of each mean estimate is indicated by error bars.
The results concerning the release prominence in the produced plosives are presented in Figure 7. For this aspect of plosive production, a significant main effect was found for stimulation condition only, F(2, 2048) = 6.7, p < .01. Post hoc analysis confirmed a significantly higher release transient prominence Stim ON compared with Stim OFF (p < .01) and pre-op (p < .01) recordings for plosives produced by patients in the cZi treatment group. The increase in release prominence observed for STN patients Stim ON compared with Stim OFF and pre-op was not shown to be significant (p = .07; p = .11). Figure 8 shows that, similar to the case of relative duration of frication and prerelease frication, the effect of DBS in terms of release prominence is more coherent in the cZi group compared with the STN group. An overview of the patient-specific effects of either cZi-DBS or STN-DBS treatment (Figures 4, 6, and 8) is provided in Table 3, including postoperative Stim OFF levels as well as the effects of DBS observed for each of the three investigated parameters of plosive production. Further, the effects observed in the parameters are summarized in Table 3 into a description of the overall outcome, where the major effects of DBS are encoded. In the STN group, one patient showed an overall positive effect, demonstrating on average
a more prominent release peak and less average frication. Five patients agreed well with a pattern of an increase in the distinctiveness of the release peak but showed negative effects in terms of increased presence of aperiodic noise. One patient showed an increase in release peak distinctiveness, with no strong effect on the presence of frication. For the remaining two, less distinct release peaks and less frication during plosives were observed. In the cZi treatment group, one patient showed negative effects across all investigated parameters (i.e., a less prominent release peak and more pronounced presence of frication). Six patients showed increased distinctiveness of the release peak together with an increased presence of aperiodic noise in the plosive. Two patients showed a decreased prominence of the release peak, and one showed an increase in the release peak prominence (all without showing a strong effect on the presence of frication).
Discussion Increased spirantization and a reduced plosive release are frequently observed features of the productions of plosives by people with PD. The effect of STN-DBS or
Karlsson et al.: Articulatory Closure Proficiency in Patients With PD
1185
Figure 6. Mean duration of prerelease frication off stimulation (Stim OFF) and on stimulation (Stim ON) for each patient within each treatment group (STN or cZi). The plosives were produced the same day (1.5 hr apart) 12 months post-op, with patients on medication.
cZi-DBS on these acoustic effects of a hypokinetic articulation has, however, not been reported previously. The results presented here show a beneficial effect of cZi-DBS on patients’ ability to manifest a plosive release transient that is distinct from the surrounding acoustic energy. After the effect of energy of other origins than that of the release of articulatory closure was factored out, voiceless plosives were shown to be produced with a stronger release transient on stimulation compared with off stimulation. This effect was observed both on the group level as well as in the individual results of seven of 10 patients. Release transient prominence also increased in patients treated with STN-DBS, but the larger within-group variation resulted in the effect not being significant at the group level. In patients who received STN-DBS, six of nine patients showed an increase in release transient distinctiveness. The data presented here therefore provide evidence that both cZi-DBS and STN-DBS result in a more distinctive release transient that spectrally distinguishes plosives from their corresponding fricatives. The finding is more consistent for patients receiving cZi-DBS. With a more prominent release transient being present in plosives during DBS, the likelihood of plosives being
erroneously perceived as fricatives could be expected to be greatly reduced for patients under both treatments. However, the beneficial effect of DBS did not extend to all aspects of plosive production. The results showed an overall strong presence of frication in plosives on stimulation for both STN-DBS and cZi-DBS treatments. In the STN group, the presence of frication was also strongly present in the Stim OFF and pre-op conditions. The stimulation effects for individual STN-DBS patients also showed large variation, with four subjects improving, four worsening, and one showing no clear effect. For patients receiving cZi-DBS, however, the effect of stimulation is much clearer. Patients showed a strong increase in adverse frication of plosives Stim ON compared with Stim OFF. This effect was observed in seven out of 10 patients (the remaining three showed no clear effect). Similarly, there was more frication before the release transient with Stim ON compared with Stim OFF. Thus, when stimulation was turned on, plosives were produced with a higher degree of frication and longer periods of fricative energy. The observed production pattern is, therefore, not consistent with an effective articulatory movement for plosives that have strong supraglottal pressure in typical speech. Further, reduced perceptual distinctiveness of the plosive
1186 Journal of Speech, Language, and Hearing Research • Vol. 57 • 1178–1190 • August 2014
Figure 7. Mean release prominence in voiceless plosives in each condition (pre-op on medication, 12 months post-op off and on stimulation, both on medication) for each group of patients. The number of plosives from which means were calculated is indicated for each stimulation condition. The confidence interval of each mean estimate is indicated by error bars.
relative to fricatives could be expected due to this adverse effect of DBS treatment. Taken in combination, the two findings indicate that, although the patients may have been able to achieve an acoustically distinct release transient, indicative of a more successful articulatory closure, they showed proportionally more acoustic energy in articulatory states prior to and following full articulatory closure. A stronger presence of this latter adverse effect was observed for the cZi-DBS group, but a similar effect was also present in the STN-DBS group. Thus, the results are consistent with a conclusion that patients were able to generate an increased total articulatory energy during plosive production on stimulation compared with off stimulation and were able to manifest a more forceful release where required. However, an increased proportion of the energy was spent during incomplete articulatory closure, reducing the articulatory quality of the plosive. This effect was seen both for the STN-DBS and cZi-DBS groups, but the cZi-DBS patients appeared to be more affected by the adverse aspects of increased frication than were the STNDBS patients.
The results presented here extend the results of the one previously published report in which effects of STN-DBS or cZi-DBS on articulatory proficiency have been compared (Karlsson et al., 2011). The cZi-DBS patients were observed to be more adversely affected by stimulation in terms of performance in speech-related motor task (oral diadochokinesis) compared with STN-DBS patients (Karlsson et al., 2011), especially when the quality of the produced plosives is considered. The results from the present investigation show that what is affected, in general, is the proficiency in suppressing aperiodic noise in the portion of the plosive where full closure between articulators would have been expected in clear speech. As a consequence, patients treated with DBS should be expected to be perceived as imprecise in their speech during stimulation. Patients with cZi-DBS were more uniformly affected by this reduction in articulatory proficiency, whereas those with STN-DBS showed a large degree of within-group variation. The large degree of individual variation in the STN group agrees well with previous reports focusing on the spirantization of plosives in stimulated patients (Dromey &
Karlsson et al.: Articulatory Closure Proficiency in Patients With PD
1187
Figure 8. Mean plosive release prominence off stimulation (Stim OFF) and on stimulation (Stim ON) for each patient within each treatment group (STN or cZi). The plosives were produced the same day (1.5 hr apart) 12 months post-op, with patients on medication.
Bjarnason, 2011). As an increase in frication generally may result in skewing of the spectrum toward higher frequencies, the results presented here could further help explain the increased long-term average spectrum (LTAS) averages due to STN-DBS (Tripoliti et al., 2011). The increased release peak amplitude and increased presence of frication agree well with both the improved articulatory muscle force and the improved peak force precision observed previously in patients treated with STN-DBS (Pinto, Pollák, Benabid, Gentil, & Fraix, 2003). However, the release prominence data presented here show a stronger presence of individual variation than the evaluation of articulator force (Pinto et al., 2003). Further, whereas articulatory muscle force estimated the muscular ability of the articulator, the release prominence quantity investigated here was able to estimate the acoustic consequence of a potentially more forceful closure action between articulators, resulting in a perceptually clearer plosive rather than one perceived as a near-fricative. One could note that the effects observed for patients with DBS frequently agree well with the nature of speech articulation effects observed due to other causes, such as intoxication (Pisoni & Martin, 1989). Thus, although it is unlikely that this reduction in
quality alone would lead to a large reduction of speech intelligibility, the effects observed here may certainly affect patients adversely in their everyday communication.
Conclusion This article has provided evidence of a more prominent acoustic release, indicative of a more successful articulatory closure, in plosives produced by patients with Parkinson’s disease who have undergone either STN-DBS or cZi-DBS treatment. Adverse effects in terms of an increased presence of frication during plosive production were, however, also observed on stimulation compared with off stimulation. It is proposed that more energy is made available during the occlusion phase on stimulation but that effective use of this energy is prevented by slowed or inappropriate articulatory closure and release action. As a result of these acoustic effects, patients may be perceived as having a reduced speech articulation. Patients in both the STN-DBS and the cZi-DBS groups showed similar effects of their DBS treatment on plosive production, but the adverse effects appeared to be more homogeneously present in the cZi-DBS group.
1188 Journal of Speech, Language, and Hearing Research • Vol. 57 • 1178–1190 • August 2014
Table 3. Summary of patient-specific outcomes of deep brain stimulation (DBS). Release prominence Patient
Stim OFF
Stim ON
DBS outcome
Relative duration of frication Stim OFF
STN 1
High
Decrease
Worsening
Mid
STN 2
Low
Increase
Improvement Low
STN 3
High
STN 4
Mid
Moderate Worsening decrease Decrease Worsening
STN 5
Low
STN 6
High High
DBS outcome
Stim ON
Prerelease frication Stim OFF
Stim ON
Decrease
Improvement Mid
Decrease
Increase
Worsening
Increase
Low
Moderate Improvement Mid decrease Decrease Improvement High
Comparable
Improvement Low
Comparable No effect
Low
Mid
Moderate increase Increase
Improvement Mid
Increase
Worsening
Mid
Moderate increase Moderate increase Increase
STN 7
Mid
Increase
Improvement Low
Increase
Worsening
Low
Increase
STN 8
High
Increase
Improvement Low
Increase
Worsening
Mid
Increase
STN 9
Mid
Increase
Improvement High
Decrease
Improvement High
Decrease
cZi 1
High
Increase
Improvement Mid
Increase
Worsening
Mid
Increase
cZi 2
Mid
Improvement Low
Increase
Worsening
Mid
Comparable
cZi 3
Low
Moderate increase Increase
Improvement Low
Worsening
Low
cZi 4
High
Decrease
Worsening
Moderate increase Increase
Worsening
High
cZi 5
Mid
Increase
Improvement Mid
Increase
Worsening
High
Moderate increase Moderate increase Increase
cZi 6
Mid
Increase
Worsening
Mid
Comparable
cZi 7
Mid
Comparable No effect
High
Decrease
cZi 8
Mid
Moderate Improvement Low increase Moderate Worsening Mid decrease Increase Improvement Low
Increase
Low
Increase
cZi 9 cZi 10
High High
Increase Decrease
Comparable No effect Comparable No effect
Mid Mid
Increase Comparable
Mid
Improvement High Worsening High
Worsening
DBS outcome
Overall DBS outcome
Improvement Less distinct release peak, less frication Worsening Improved release peak, more frication No effect Less distinct release peak, less frication Worsening Improved release peak, more frication Worsening Improved release peak Worsening Improved release peak, more frication Worsening Improved release peak, more frication Worsening Improved release peak, more frication Improvement Overall positive outcome Worsening Improved release peak, more frication No effect Improved release peak, more frication Worsening Improved release peak, more frication Worsening Overall negative outcome Worsening Improved release peak, more frication No effect Improved release peak, more frication Improvement Less distinct release peak Worsening Improved release peak, more frication Worsening Improved release peak No effect Less distinct release peak
Note. Patient-specific outcomes of DBS are given in terms of plosive release prominence, prerelease frication duration, and relative duration of frication during the realization of the plosive. For each parameter investigated, an overview of the patient’s starting level at Stim OFF (12 months post-op, on medication) and change direction in Stim ON recordings (made 1.5 hr later, on medication) compared with the Stim OFF baseline is provided. The development is categorized (“Improvement,” “Worsening,” or “No change”) for each investigated parameter. A broad characterization of the effect of DBS is further provided for each patient.
Acknowledgments We would like to thank deep brain stimulation nurse specialist Anna Fredricks for assistance in providing patient information and research engineer Anders Asplund for preparation of the speech recordings. We wish to acknowledge the support of grants from the Swedish Research Council (Grant No. 2009-946) and Magnus Bergvall’s Foundation in Sweden. A visiting professorship at the Rehabilitation Research Chair, King Saud University, Riyadh, Saudi Arabia, to Erik Nordh is gratefully acknowledged.
References Ackermann, H., Hertrich, I., Daum, I., Scharf, G., & Spieker, S. (1997). Kinematic analysis of articulatory movements in
central motor disorders. Movement Disorders, 12, 1019–1027. doi:10.1002/mds.870120628 Ackermann, H., & Ziegler, W. (1991). Articulatory deficits in Parkinsonian dysarthria: An acoustic analysis. Journal of Neurology, Neurosurgery & Psychiatry, 54, 1093–1098. doi:10.1136/jnnp.54.12.1093 Arnau, J., Bono, R., & Vallejo, G. (2009). Analyzing small samples of repeated measures data with the mixed-model adjusted F test. Communications in Statistics—Simulation and Computation, 38, 1083–1103. doi:10.1080/03610910902785746 Auzou, P., Ozsancak, C., Morris, R. J., Jan, M., Eustache, F., & Hannequin, D. (2000). Voice onset time in aphasia, apraxia of speech and dysarthria: A review. Clinical Linguistics & Phonetics, 14, 131–150. doi:10.1080/026992000298878 Belin, P., Zilbovicius, M., Crozier, S., Thivard, L., Fontaine, A., Masure, M. C., & Samson, Y. (1998). Lateralization of speech
Karlsson et al.: Articulatory Closure Proficiency in Patients With PD
1189
and auditory temporal processing. Journal of Cognitive Neuroscience, 10, 536–540. Boersma, P., & Weenink, D. (2012). Praat (Version 5.3.04) [Computer software]. Retrieved from http://www.praat.org/ Chenausky, K., MacAuslan, J., & Goldhor, R. (2011). Acoustic analysis of PD speech. Parkinson’s Disease, 2011, 1–13. doi:10.4061/2011/435232 Deuschl, G., Fogel, W., Hahne, M., Kupsch, A., Müller, D., Oechsner, M., . . . Volkmann, J. (2002). Deep-brain stimulation for Parkinson’s disease. Journal of Neurology, 249, iii36–iii39. doi:10.1007/s00415-002-1308-x Dromey, C., & Bjarnason, S. (2011). A preliminary report on disordered speech with deep brain stimulation in individuals with Parkinson’s disease. Parkinson’s Disease, 2011, 1–11. doi:10.4061/2011/796205 Duez, D. (2007, August). Acoustic analysis of occlusive weakening in Parkinsonian French speech. Paper presented at the 16th International Congress of Phonetic Sciences, Saarbrücken, Germany. Forrest, K. (1989). Kinematic, acoustic, and perceptual analyses of connected speech produced by Parkinsonian and normal geriatric adults. The Journal of the Acoustical Society of America, 85, 2608. doi:10.1121/1.397755 Hartinger, M., Tripoliti, E., Hardcastle, W. J., & Limousin, P. (2011). Effects of medication and subthalamic nucleus deep brain stimulation on tongue movements in speakers with Parkinson’s disease using electropalatography: A pilot study. Clinical Linguistics & Phonetics, 25, 210–230. doi:10.3109/ 02699206.2010.521877 Hillenbrand, J. M., Cleveland, R. A., & Erickson, R. L. (1994). Acoustic correlates of breathy vocal quality. Journal of Speech and Hearing Research, 37, 769. Karlsson, F., Unger, E., Wahlgren, S., Blomstedt, P., Linder, J., Nordh, E., . . . van Doorn, J. (2011). Deep brain stimulation of caudal zona incerta and subthalamic nucleus in patients with Parkinson’s disease: Effects on diadochokinetic rate. Parkinson’s Disease, 2011, 1–10. doi:10.4061/2011/605607 Kent, R. D., & Kim, Y.-J. (2003). Toward an acoustic typology of motor speech disorders. Clinical Linguistics & Phonetics, 17, 427–445. doi:10.1080/0269920031000086248 Kent, R. D., Weismer, G., Kent, J. F., Vorperian, H. K., & Duffy, J. R. (1999). Acoustic studies of dysarthric speech: Methods, progress, and potential. Journal of Communication Disorders, 32, 141–186. doi:10.1016/S0021-9924(99)00004-0 Krack, P., Batir, A., Van Blercom, N., Chabardes, S., Fraix, V., Ardouin, C., . . . Pollák, P. (2003). Five-year follow-up of bilateral stimulation of the subthalamic nucleus in advanced Parkinson’s disease. New England Journal of Medicine, 349, 1925–1934. doi:10.1056/NEJMoa035275 Logemann, J. A., & Fisher, H. B. (1981). Vocal tract control in Parkinson’s disease. Journal of Speech and Hearing Disorders, 46, 348. McAuliffe, M. J., Ward, E. C., & Murdoch, B. E. (2006a). Speech production in Parkinson’s disease: I. An electropalatographic investigation of tongue-palate contact patterns. Clinical Linguistics & Phonetics, 20, 1–18. McAuliffe, M. J., Ward, E. C., & Murdoch, B. E. (2006b). Speech production in Parkinson’s disease: II. Acoustic and electropalatographic investigation of sentence, word and segment durations. Clinical Linguistics & Phonetics, 20, 19–33. doi:10.1080/ 0269-9200400001069
Ozsancak, C., Auzou, P., Jan, M., & Hannequin, D. (2001). Measurement of voice onset time in dysarthric patients: Methodological considerations. Folia Phoniatrica et Logopaedica, 53, 48–57. doi:10.1159/000052653 Pinto, S., Pollák, P., Benabid, A.-L., Gentil, M., & Fraix, V. (2003). Bilateral subthalamic stimulation effects on oral force control in Parkinson’s disease. Journal of Neurology, 250, 179–187. doi:10.1007/s00415-003-0966-7 Pisoni, D. B., & Martin, C. S. (1989). Effects of alcohol on the acoustic-phonetic properties of speech: Perceptual and acoustic analyses. Alcoholism: Clinical and Experimental Research, 13, 577–587. doi:10.1111/j.1530-0277.1989.tb00381.x Plaha, P., Ben-Shlomo, Y., Patel, N. K., & Gill, S. S. (2006). Stimulation of the caudal zona incerta is superior to stimulation of the subthalamic nucleus in improving contralateral parkinsonism. Brain, 129(Pt. 7), 1732–1747. doi:10.1093/brain/awl127 Romito, L. M., Contarino, M. F., Vanacore, N., Bentivoglio, A. R., Scerrati, M., & Albanese, A. (2009). Replacement of dopaminergic medication with subthalamic nucleus stimulation in Parkinson’s disease: Long-term observation. Movement Disorders, 24, 557–563. doi:10.1002/mds.22390 Roy, N., Leeper, H. A., Blomgren, M., & Cameron, R. M. (2001). A description of phonetic, acoustic, and physiological changes associated with improved intelligibility in a speaker with spastic dysarthria. American Journal of Speech-Language Pathology, 10, 274. doi:10.1044/1058-0360(2001/025) Rusz, J., Cmejla, R., Ruzickova, H., & Ruzicka, E. (2011). Quantitative acoustic measurements for characterization of speech and voice disorders in early untreated Parkinson’s disease. The Journal of the Acoustical Society of America, 129, 350–367. doi:10.1121/1.3514381 Schulz, G. M., Hosey, L. A., Bradberry, T. J., Stager, S. V., Lee, L.-C., Pawha, R., . . . Braun, A. R. (2012). Selective left, right and bilateral stimulation of subthalamic nuclei in Parkinson’s disease: Differential effects on motor, speech and language function. Journal of Parkinson’s Disease, 2, 29–40. Stevens, K. N. (2000). Acoustic phonetics. Cambridge, MA: MIT Press. Tomlinson, C. L., Stowe, R., Patel, S., Rick, C., Gray, R., & Clarke, C. E. (2010). Systematic review of levodopa dose equivalency reporting in Parkinson’s disease. Movement Disorders, 25, 2649–2653. doi:10.1002/mds.23429 Tripoliti, E., Zrinzo, L. U., Martinez-Torres, I., Frost, E., Pinto, S., Foltynie, T., . . . Limousin, P. (2011). Effects of subthalamic stimulation on speech of consecutive patients with Parkinson disease. Neurology, 76, 80–86. doi:10.1212/WNL. 0b013e318203e7d0 Wang, Y.-T., Kent, R. D., Duffy, J. R., Thomas, J. E., & Weismer, G. (2004). Alternating motion rate as an index of speech motor disorder in traumatic brain injury. Clinical Linguistics & Phonetics, 18, 57–84. Weismer, G. (1984). Articulatory characteristics of Parkinsonian dysarthria: Segmental and phrase-level timing, spirantization, and glottal-supraglottal coordination. In M. McNeil, J. Rosenbeck, & A. Aronson (Eds.), The dysarthrias: Physiology, acoustics, perception, management ( pp. 101–130). San Diego, CA: CollegeHill Press. Zatorre, R. J., Evans, A., & Meyer, E. (1992, May 8). Lateralization of phonetic and pitch discrimination in speech processing. Science, 256, 846–849. doi:10.1126/science.256.5058.846
1190 Journal of Speech, Language, and Hearing Research • Vol. 57 • 1178–1190 • August 2014
Copyright of Journal of Speech, Language & Hearing Research is the property of American Speech-Language-Hearing Association and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.
