You are on page 1of 8

Logopedics Phoniatrics Vocology

ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/ilog20

Speech intelligibility of Parkinson’s disease


patients evaluated by different groups of
healthcare professionals and naïve listeners

Joana Carvalho , Rita Cardoso , Isabel Guimarães & Joaquim J. Ferreira

To cite this article: Joana Carvalho , Rita Cardoso , Isabel Guimarães & Joaquim J. Ferreira
(2020): Speech intelligibility of Parkinson’s disease patients evaluated by different groups
of healthcare professionals and naïve listeners, Logopedics Phoniatrics Vocology, DOI:
10.1080/14015439.2020.1785546

To link to this article: https://doi.org/10.1080/14015439.2020.1785546

Published online: 07 Jul 2020.

Submit your article to this journal

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=ilog20
LOGOPEDICS PHONIATRICS VOCOLOGY
https://doi.org/10.1080/14015439.2020.1785546

ORIGINAL ARTICLE

Speech intelligibility of Parkinson’s disease patients evaluated by different


groups of healthcare professionals and naïve listeners
Joana Carvalhoa,b, Rita Cardosoa,b, Isabel Guimar~aesc and Joaquim J. Ferreiraa,b
a
Clinical Pharmacological Unit, Instituto de Medicina Molecular, Faculty of Medicine, University of Lisbon, Lisbon, Portugal; bCampus
Neurologico Senior, Torres Vedras, Portugal; cDepartment of Speech Therapy, Escola Superior de Saude de Alcoit~ao, Estoril, Portugal

ABSTRACT ARTICLE HISTORY


Introduction: Speech intelligibility, how well a listener comprehends the speaker’s message, is related Received 1 April 2020
to the listener’ expertise and type of the message conveyed. There is no evidence about speech intelli- Revised 8 June 2020
gibility in different groups of healthcare professionals and naïve listeners. Accepted 15 June 2020
Objectives: This study is the first to understand if there were differences in the speech intelligibility of
KEYWORDS
Parkinson’s Disease (PD) patients by different experienced and naïve listeners, according to the speech Dysarthria; intelligibility;
stimuli and dysarthria severity. audio-perceptive analysis;
Materials and methods: Randomly digitised audio-files (50 words and 50 sentences) of 10 PD Parkinson’s disease
patients, one without dysarthria and 9 with different dysarthria severities (3 each: mild, moderate and
severe dysarthria), were collected from a database of 60 PD patients’ audio-files. A jury panel was
formed by five different listeners groups including 10 speech and language therapists, 10 neurologists,
10 PD relatives, 12 PD patients, and 10 people from the general population. The jury panel transcribed
single words and sentences from the audio recordings, the percentage correctly understood was cal-
culated and the results were compared between the groups.
Results: Multiple comparisons showed significant speech intelligibility differences between healthcare
professionals and naïve listerners in words (highest effect size, g2 ¼ 0.7) and sentences (the highest
effect size: g2 ¼ 0.6). Pairwise comparisons revealed that those significant differences were specifically
in words with moderate and severe dysarthria and sentences with all severity levels of dysarthria.
Conclusion: The groups of healthcare professionals who work with dysarthria are more likely to
understand the PD patients’ speech than the groups of naïve listeners.

Introduction judging PD speech intelligibility, in a reliable manner [8] while


others found no significant difference in speech intelligibility
Speech intelligibility, how well a speaker’s acoustic signal
between experienced and in experienced listeners for dysarthric
can be accurately recovered by a listener [1,2] is commonly
speech [4]. Kreiman et al. [5] state that clinical training and
used as a functional index of the communicative perform-
experience might influence speech perception. In fact, experi-
ance on Parkinson’s disease (PD) patients with hypokinetic
dysarthria. Ninety percent of Parkinson’s disease (PD) enced listeners (78%) had higher speech intelligibility than
patients present reduced speech intelligibility that is naïve listeners (50%) in adults with dysarthria secondary to
expected to decline with disease progression [2,3]. Clearly, cerebral palsy [9]. Platt et al. [9] showed that naïve listeners
low levels of intelligibility may limit substantially the PD without formal training do not have specific internal standards
patients’ independence affecting their participation at the for judging pathological qualities.
society level and thus their quality of life [2]. While previous work has compared speech intelligibility
Speech intelligibility is typically measured by having lis- by experienced listeners versus naïve listeners [5,8] to the
teners orthographically transcribe words or sentences that authors best of knowledge, no previous investigation has
the speaker had read and a score is calculated dividing the specifically addressed the differences between groups of pro-
number of words or sentences understood and the total fessional experienced listeners. Moreover, studies mention-
number of words or sentences spoken [4,5]. ing naïve listeners are mainly with the general population
Prior research suggests that speech intelligibility may be [10–12] and few with PD patients [13–15] and PD rela-
affected by many variables like: tives [12].
(i) Listener’s experience and training [2,6,7]. There is con- (ii) Listener’s familiarity. Results of studies examining
flicting evidence regarding the role of listener experience (naïve familiarity with the content of the spoken message have
versus expert) in the assessment of dysarthric speech [4]. Some been equivocal likely because of methodological differences.
researchers demonstrated that naïve listeners are capable of Tjaden and Liss [14] showed that SLT undergraduate

CONTACT Joana Carvalho joana.margarida.cr@gmail.com Clinical Pharmacological Unit, Instituto de Medicina Molecular, Faculty of Medicine, University of
Lisbon; Campus Neurologico Senior, Torres Vedras, Lisbon, Portugal
ß 2020 Informa UK Limited, trading as Taylor & Francis Group
2 J. CARVALHO ET AL.

students previously familiarized with dysarthric speech Listeners –jury panel


obtained significantly better speech intelligibility scores than
SLT undergraduate students without familiarity with dys- Jury panel
arthric speech. In contrast, Walshe et al. [13–15] found non- A jury panel formed by healthcare professionals (10 speech
significant differences in speech intelligibility across three and language therapists and 10 neurologists) and naïve lis-
groups of listeners familiarized with the speech stimuli used teners (12 PD patients, 10 PD relatives, and 10 people from
in the research. Even so, the adult speakers with dysarthria the general population) was organized. The elements of the
presented the lowest listeners’ intelligibility ratings. jury panel had to fulfill the following criteria: (i) All listen-
(iii) Type of speech stimuli (the speech content or the ers had to report normal hearing abilities and to be blind to
degree of dysarthria). Speech intelligibility is driven by mul- any speaker or speech stimuli list variables; (ii) Health care
tiple sources including voice (e.g. pitch, timbre, and timing) professionals had to report working experience with adults’
and linguistic (e.g. lexicality, phonological and syntactic) dysarthria; (iii) The PD relatives had to report daily contact
cues. Semantic predictability dramatically affects a listener’s with PD patient; (iv) The general population (GP) had to
ability to understand speech, with sentences more predict- report no daily contact with PD patients, family or acquain-
able than single words resulting in higher intelligibility tances with disease; (v) The PD patients had to be without
scores [16]. This evidence is confirmed both for normal and diagnosed cognitive deficit that prevented collaboration in
dysarthric speech [6,17–19]. When considering the disor- the study and without having undergone to deep brain
dered speech research shows that the severity grade corre- stimulation.
lates with intelligibility rating in that higher intelligibility
scores correspond to lower grade dysarthria [20]. Also,
some authors stress the benefit of knowing the syntax over Materials
the semantic cue provided, particularly for mild dysarthria Ten digitized audio files randomly selected from 10 male
[10]. Different listeners’ judgments of normal and nearly PD patients reading single words and sentences were drawn
normal voices are comparable possibly due to the similar from a database of 60 PD patients’ audio-files without dys-
exposure to what constitutes a normal voice [5,14]. arthria and with mild to severe dysarthria. Only men were
However, for the intermediate grade dysphonia even experi- recruited since there are differences between sexes in terms
enced raters are subject to great variability [10]. of speech characteristics which could bias the results. The
(iv) Listeners biases (e.g. fatigue, hearing sensitivity, PD patients were native Portuguese speakers, who had a
attention lapses) are considered random effects that can mean age of 72.2 years ± 7.9 SD and 8.8 years of education
cause a drift in perceptual judgments [10]. Wild et al. [21] ± 5.8 SD. All presented a medical diagnosis of idiopathic
unequivocally demonstrated that clear speech can be proc- PD according to the UK Parkinson’s Disease Brain Bank
essed even with low attentional level but degraded speech is Criteria [22]. Time from diagnosis ranged from 4 to 19
processed depending on the listener’s state of attentiveness. years and based on the Hoehn and Yahra disease stage
In this study the primary aim was on the listeners and between 1 and 5. The PD disability and impairment was
their capacity to understand PD patients’ speech. The main discrete for the PD patient without hypokinetic dysarthria
hypothesis was that professional experienced listeners (MDS-UPDRS overall score equal to 113) and slight to
(Neurologists and Speech and Language Therapists (SLTs)) moderate for the nine PD patients with hypokinetic dysarth-
would have different hit percentages in speech intelligibility ria (MDS-UPDRS overall score between 148 and 225). All
than naïve listeners (PD patients, PD relatives, and General patients were taking medication and were assessed in the
Population, GP). Within the naïve speakers those with PD “ON” condition.
speech familiarity are also expected to have different hit per- The dysarthria diagnosis and level of severity was per-
centages than those without PD speech familiarity. If intelligi- formed by an SLT expert in the assessment and treatment
bility varies within different groups of listeners it is of acquired sensorimotor speech disorders, using the
important to take all of them into account when intelligibility Frenchay Dysarthria Assessment – 2nd edition (FDA-2)
is being assessed. Another aim is to understand the dysarthria European Portuguese version protocol [23] and a Clinical
severity effect in speech intelligibility. Given research suggest- Global Impression (CGI) scale. Data had been recorded in a
ing that speech dimensions deteriorate in the course of the quiet room with professional speech recording equipment.
PD disease it was hypothesized that severe dysarthria would Two other SLTs experienced in PD perform an audio-per-
yield worst speech intelligibility than mild and moderate dys- ceptive analysis of the recording files also using CGI. The
arthria severities. Understanding the speech intelligibility in patients included in this study were those were 100% agree-
the different levels of dysarthria may provide specific thera- ment between the raters was achieved.
peutic clues to improve the PD patients’ communication. The digitised audio-files selected from each patient
include five single words and five sentences chosen from a
set of 12 words and sentences read by each patient. Some of
Materials and methods
the files presented a spurious noise that was impossible to
A cross-sectional study was approved by an ethical commit- filter. To harmonise the quantity of stimuli between PD
tee and the study was conducted in accordance with the patients and levels of dysarthria severity levels it was only
Declaration of Helsinki. used 10 from each. It is worth mentioning that the
LOGOPEDICS PHONIATRICS VOCOLOGY 3

application of FDA-2 EP version intelligibility task implies significant values. The same methodology was applied to
the performance of 12 words and sentences randomly sentences correctly understood.
chosen from a set of 109 words and 60 sentences [23]. The inter-rater reliability was calculated by means of per-
These tasks include a group of samples with strong correla- cent agreement. Results demonstrate an almost perfect
tions in terms of distribution of frequency of phonemes, agreement between the jury panel in each group in terms of
word length, and type of syllabic structure involved. A total intelligibility of single words (all groups with 0.9 percent
stimulus of 50 different words (5 without dysarthria and 15 agreement) and sentences (GP and PD patients with 0.8
for each dysarthria severity level) and 50 different sentences percent agreement and healthcare professionals and PD rela-
(5 without dysarthria and 15 for each dysarthria severity tives with 0.9).
level) were gathered at the end. All descriptive and inferential results were obtained using
SPSSV Software, version 20.0 (SPSS, Inc., Chicago, IL) and
R

p-values less than .05 were considered significant.


Procedures
For this study, the peak amplitude of the selected audio Results
stimuli was normalized with a software with the initial
default setting is 1 dB. Reliability of the dysarthria severity Jury panel
was reanalyzed by two other SLTs, also with expertise in Fifty-two listeners aged 25–76 years participated in the
PD, by means of an audio-perceptive analysis of the record- study. The healthcare professionals are significantly younger
ing files. (p ¼ .001) than the naïve listeners. Within the healthcare
The jury panel was informed about the purpose and pro- professionals and the naïve listeners groups the mean age
ceedings of this study by reading an information sheet and was not significantly different (p > .05) (Table 1). The
if the participant wished to continue an informed consent majority of the listeners were females except in the PD
form was signed. The audio-files – single words and senten- patients’ listeners who were mainly men (Table 1). The
ces – were presented separately, once only, using stereo healthcare professionals (all with higher education) have a
headphones connected to a laptop at 50% volume level. similar level of education as half of the PD patients and a
Between each word there was a four second interval and higher level of education than the majority of the other
between each sentence a seven second interval. The panel naïve jury panel (PD relatives and GP). Globally, GP is the
members were instructed to write down what they heard. naïve listeners with the lower level of education. The SLTs
All listeners were blind to the background information of reported a mean of working experience of 17.9 years ±3.9
the speakers and to the other listeners ratings. with adults with dysarthria other than PD patients. The
Neurologists have a mean of working experience of
Statistical analysis 16.4 years ±3.9 of working with movement disorders, par-
ticularly with PD.
The transcribed single words and sentences were considered
“correct” if there was a perfect match between all words and
phonemes spoken and written by the jury, or “incorrect” if Professional experienced versus naï ve listeners
there was not. The primary outcome was the total words Descriptive statistics revealed differences in the hit percent-
correctly understood and it was calculated and presented as age and mean number of words correctly understood
a hit percentage, mean values, and standard deviation 95% between groups. In ascending order PD patients < GP < PD
confidence interval (CI) per group, and the hit percentage relatives < SLTs < Neurologists (Table 2).
per level of dysarthria. Statistical analysis indicate that there are significant over-
Normal distribution was investigated with the Shapiro all differences between the medians of the total words cor-
Wilk test and homogeneity of the variances with Levene’s rectly annotated by the five groups v2 (4, N ¼ 52) ¼ 19.0,
test. Since the homogeneity of the variance was not verified, p ¼ .001with a mean rank of 18.7 for the GP, 33.2 for SLTs,
a non-parametric test was chosen to compare results 41.7 for Neurologists, 22.0 for PD relatives and 18.7 for PD
between the groups – a Kruskall–Wallis test (adjusted for patients. The effect size was not high, g2¼ v2/N  1,
tied ranks) – and follow-up tests were conducted to evaluate g2¼0.4. Because the overall test is significant, pairwise com-
pairwise differences among the five groups and the effect parisons among the five groups were completed and showed
size for pairwise comparisons was presented with statistically significant differences only between: (i) the healthcare

Table 1. Demographic data for the jury panel.


Education, N (%)
N Age mean ± SD (range) Gender (%) F (female) j M (male) Elementary, <4 years Secondary, 5–11 years Higher, >12 years
PD patients 12 65.6 ± 7.2 (55–76) F ¼ 33.3% j M ¼ 66.7% 2 (16.7%) 4 (33.3%) 6 (50%)
PD Relatives 10 65.0 ± 6.0 (56–75) F ¼ 100% j M ¼ 0% 2 (20%) 4 (40%) 4 (40%)
GP 10 62.6 ± 6.2 (57–76) F ¼ 40% j M ¼ 60% 5 (50%) 5 (50%) 0
SLTs 10 40.7 ± 12.5 (25–57) F ¼ 90% j M ¼ 10% 0 0 10 (100%)
Neurologists 10 43.9 ± 7.4 (34–56) F ¼ 40% j M ¼ 60% 0 0 10 (100%)
4 J. CARVALHO ET AL.

Table 2. Listeners’ hit percentage (%). Table 3. Speech intelligibility (%) according to dysarthria severity.
Words Sentences Dysarthria
Listeners % M ± SD 95% CI % M ± SD 95% CI Normal speech Mild Moderate Severe
PD patients 56.2 28.1 ± 5.7 24.5–31.7 61.2 30.6 ± 4.8 27.6–33.6 PD patients Words 93.3 57.8 71.1 27.2
PD relatives 60.6 30.3 ± 4.5 27.1–33.5 66.4 33.2 ± 4.0 30.3–36.1 Sentences 81.7 71.1 68.3 37.2
GP 60.2 30.1 ± 2.3 28.5–31.7 67.8 33.9 ± 4.0 31.1–36.8 PD relatives Words 98 68.7 75.3 26.7
SLTs 66.8 33.4 ± 2.4 31.7–35.1 73.0 36.5 ± 2.7 34.6–38.4 Sentences 74 73.3 78 45.3
Neurologists 70.4 35.2 ± 1.9 33.8–36.6 75.0 37.5 ± 2.0 36.1–38.9 GP Words 98 64 72.7 30.7
M ± SD: mean ± standard deviation; CI: confidence interval. Sentences 82 76 79.3 43.3
SLT Words 100 77.3 83.3 28.7
Sentences 84 84 82.7 48.7
professionals (SLTs and Neurologists) and the GP (v2 (1, Neurologists Words 98 86 87.3 28.0
Sentences 92 82 81.3 56
N ¼ 20) ¼ 6.6, p ¼ .01 g2¼0.3 and v2 (1, N ¼ 20) ¼ 12.9, p
¼ .001g2¼0.7 respectively); (ii) Neurologists and PD rela-
tives (v2 (1, N ¼ 20) ¼ 8.7, p ¼ .001 g2¼0.5) and PD
patients (v2 (1, N ¼ 22) ¼ 9.0, p ¼ .001g2¼0.4). than sentences for the control speech stimuli, without dys-
A specific analysis within the naïve jury groups matched arthria; (ii) equal in words and sentences for the moderate
by educational level showed non-significant results between dysarthria; (iii) lower in words than sentences for mild and
the groups at the elementary level of education (v2 (3, severe dysarthria.
N ¼ 9) ¼ 10.350, p ¼ .410) and at the secondary level of Comparing the percentage magnitude between normal
education (v2 (3, N ¼ 12) ¼ 18.200, p ¼ .312) for words. speech intelligibility and: (i) severe dysarthria it reduced
Table 2 shows that speech intelligibility for sentences is 67.3% in words and 36.9% in sentences; (ii) moderate dys-
slightly different between groups. In ascending order: PD arthria it reduced 19.4% in words and 5.1% in sentences;
patients < PD relatives < GP < SLTs < Neurologists. Statistical (iii) mild dysarthria it reduced 30.4% in words and 5.6% in
analysis indicate that there are significant overall differences sentences (Table 5).
between the medians of the total sentences correctly anno-
tated by the five groups v2 (4, N ¼ 52) ¼ 18.0, p ¼ .001with a
mean rank of 26.0 for GP, 32.4 for SLTs, 39.9 for Discussion
Neurologists, 22.4 for PD relatives, and 14.3 for PD patients. Professional experienced versus naive listeners
The effect size was not high, g2¼ v2/N  1, g2 ¼ 0.4. Because
the overall test is significant, pairwise comparisons among The current findings show a significant difference between
the five groups were completed and significant differences professional experienced listeners (Neurologists and SLTs) and
were found between: (i) SLTs and PD patients (v2 (1, naïve listeners in speech intelligibility. These results lend sup-
N ¼ 22) ¼ 7.8, p ¼ .01 g2¼0.4); and (ii) Neurologists and port to the published literature that refers professional experi-
PD patients (v2(1, N ¼ 22) ¼ 13.318, p ¼ .001 g2¼0.6) and enced listeners effects in the intelligibility of speech [9,11].
PD relatives (v2 (1, N ¼ 20) ¼ 7.612, p ¼ .01 g2¼0.6). Although not significant, within the professionals experienced
Within the naïve jury groups matched by educational listeners, Neurologists had slightly higher intelligibility hit per-
level showed non-significant results for the sentences centages than SLTs. Considering that both have professional
between the groups at the elementary level of education (v2 experience in adults’ dysarthria a possible explanation for this
(3, N ¼ 9) ¼ 18.0, p ¼ .324) and at the secondary level of difference is the Neurologists familiarization with PD patients
education (v2 (3, N ¼ 12) ¼ 13.180, p ¼ .518). for more than five years against no experience with PD
Overall healthcare professionals (SLTs and Neurologists) patients for the SLTs. Moreover, the Neurologists, with expert-
had the highest hit percentage for all speech stimuli while ise in PD, in this study, seem to be highly sensitive to PD
PD patients had the lowest (Table 3). speech (words and sentences) and to corroborate this is the
The statistical analysis between the medians of the total significant intelligibility difference they make from the PD
words according to dysarthria severity correctly annotated patients and PD relatives listeners. Conversely, SLTs without
by the five groups showed significant differences only for experience in PD only distinguish themselves from GP (in the
the words with moderate (v2 (4, N ¼ 52) ¼ 10.1, p ¼ .039) words stimuli) and PD patients (in the sentences stimuli).
and severe (v2 (4, N ¼ 52) ¼ 17.1, p ¼ .002) dysarthria and These innovative results seem to indicate the need for expertise
for the sentences with mild (v2 (4, N ¼ 52) ¼ 11.733, and experience not only in adults’ dysarthria but also in the
p ¼ .019), moderate (v2 (4, N ¼ 52) ¼ 19.443, p ¼ .001) and underlying condition, namely PD [9,11].
severe (v2 (4, N ¼ 52) ¼ 17.254, p ¼ .002) dysarthria. One striking finding was the homogeneity of the ratings
Table 4 reveals the dysarthria severity levels significant across the naive listeners rather than an expected difference
differences between health care professionals and the naïve between familiarized listeners as the PD patients and PD
listeners for moderate and severe dysarthria. relatives and nonfamiliarized listeners as the GP group.
These results stand in contrast with those reported by a
number of authors who argued that prior familiarization
Speech stimuli effect
with a specific speaker led to increased intelligibility scores
Table 5 summarizes all the hit percentage of intelligibility [13,14]. However, the PD patients and PD relatives groups,
for the speech stimuli showing that it is: (i) higher in words in this study, were not familiar with the patients’ recorded
LOGOPEDICS PHONIATRICS VOCOLOGY 5

Table 4. Pairwise comparisons according to dysarthria and type of listener.


Speech stimuli Listeners Mann–Whitney p Value
Words with moderate dysarthria SLT versus PD U ¼ 29.500 .041
Neurologists versus PD U ¼ 17.000 .004
Words with severe dysarthria SLT versus PD U ¼ 20.500 .025
SLT versus GP U ¼ 22.000 .033
Neurologists versus PD relatives U ¼ 9.000 .002
Neurologists versus GP U ¼ 8.500 .002
Sentences with mild dysarthria SLT versus PD U ¼ 23.000 .013
Neurologists versus PD U ¼ 15.000 .003
Sentences with moderate dysarthria SLT versus PD U ¼ 11.000 .001
Neurologists versus PD U ¼ 10.000 .001
Sentences with severe dysarthria SLT versus PD U ¼ 26.000 .020
Neurologists versus PD U ¼ 11.500 .001
Neurologists versus GP U ¼ 13.500 .005
Neurologists versus PD relatives U ¼ 19.000 .017

Table 5. Percentage of words and sentences correctly transcribed per level than words for all dysarthria severities. Evidence highlights
of dysarthria. that a sentence, with the presence of context provided by the
Words Sentences semantics improves the listeners’ accuracy than a single word
Dysarthria Dysarthria [24,25]. Nevertheless, for the control speech stimuli, without
Without Mild Moderate Severe Without Mild Moderate Severe dysarthria, sentences seemed to be more difficult to understand
97.7% 67.3% 78.3% 30.4% 82.7% 77.1% 77.6% 45.8%
than words and it seems reasonable to speculate if: (i) the talk-
ers’ speech rate (e.g. accelerated speech) that reduces articula-
speech. Unfortunately the body of research with naïve lis- tory accuracy and consequently the speech intelligibility; (ii)
teners like the ones in this study was not found in the lit- the listeners’ sensitivity to normal age-related changes in
erature therefore comparisons are not possible. speech, perhaps especially changes related to voice or prosody
Several hypotheses can be considered to explain these as stated in a study from Sussman and Tjaden [26]; (iii) the
results. First, it may be possible that PD relatives, in this normal speech is under represented (one) sample against three
study, had a reduced exposure to the speech dimensions pre- for each dysarthria severity.
sent in the stimuli considering that PD speech dysarthria is Listeners recognized better the normal speech (between
highly variable among speakers. Second, it is possible that PD 83 and 98%) than the disordered dysarthric speech (between
relatives are used to contextual intelligibility (intelligibility 30 and 78%) which is in line with the results from other
associated with visual cues, e.g. physical context, facial expres- study [12]. The severe dysarthria speech stimuli are the
sion, body language and gestures) and not only to isolated most difficult to understand with a declination in intelligi-
acoustic speech. In fact, research shows that visual informa- bility, when compared to the normal speech, higher than
tion gives the listener additional cues to decode speech [6]. 67% in words and more than 37% in sentences. This result
Third, although no statistically significant difference exists seems obvious considering the expected speech decline with
within the naïve listeners groups it should be noted that, in disease progression and therefore an understanding higher
this study, PD patients’ speech intelligibility hit percentage is negative impact [2,3]. Another unexpected result is that
the lowest. The published literature states that people with words with mild dysarthria seem more difficult to under-
dysarthria may perceive their own speech differently to those stand than those with moderate dysarthria (10% more diffi-
without dysarthria [16] and suggests that PD patients lack cult) and sentences are similarly understood (5% worst for
insight into their speech difficulties [17]. However, the aim of moderate dysarthria). It is possible that the results obtained
this study was about PD patients’ perceptions of others’ dys- are related to (i) the reference levels of mild and moderate
arthric speech without visual aspects of communication [1]. dysarthria used to categorize the speech stimuli. Dysarthria
Finally, it can be speculated if this results from random severity categorization depends on the presence of segmental
effects such as listening fatigue and attention lapses have and supra segmental degradation patterns that impact the
interfered with speech intelligibility [10]. integrity of the speech signal. This categorization was per-
Speech intelligibility is a key component in human com- formed by SLTs with expertise and experience in PD in the
munication and the results of this study point out to need presence of the PD patients. So, it is possible that SLT esti-
for specific training and experience of healthcare professio- mates may be biased by the knowledge of the speakers’
nals. Also important is the naive listeners’ education to background and/or familiarity with the FDA-2 intelligibility
endow them with knowledge, strategies, and tools so that tasks content and/or by the presence of the speaker at the
the PD patient can be more easily understood. assessment data (audiovisual intelligibility)[13–15]; (ii) the
transcription system used, in this study, may not be sensi-
tive to measure small intelligibility differences [13–15].
Speech stimuli effect
Considering the role and importance of intelligibility as
One expected finding was the type of speech stimuli, words an outcome in SLT field, more research is need to deter-
versus sentences, effect in intelligibility. In this study, globally, mine the degree of impairment and how to monitor the
higher values were obtained in the intelligibility of sentences changes of speech intelligibility.
6 J. CARVALHO ET AL.

Limitations Notes on contributors


Although the present results are innovative for the know- Joana Carvalho Degree in Speech Therapy from the School of Health
ledge of PD patients’ speech by different groups of listeners of the University of Aveiro. Master in neuroscience from the Faculty
of Medicine of the University of Lisbon. Post-graduated in oropharyn-
and show some promising trends there are certain limita- geal dysphagia by the Epap Institute (I.EPAP). Working as a speech
tions that must be mentioned. First, the cross-sectional therapist at Campus Neurologico Senior since 2014. Researcher at the
nature of this study and the small-scale sample of listeners Instituto de Medicina Molecular, dedicated to the study of speech and
limits the generalizability of the results. Nevertheless, this swallowing disorders in movement disorders. Involvement in financed
study provided innovative data by using five different project in the field of the article: FraLusoPark-Dysarthria in Parkinson
Disease project coordinated by Serge Pinto (Laboratoire Parole et
groups of listeners. Second, listeners’ hearing accuracy was Langage, Universite Aix-Marselle/CNRS) and Joaquim Ferreira
based on self-report, therefore comparable hearing accuracy (Institute of Molecular Medicine, Faculty of Medicine of Lisbon)
among and between groups cannot be guaranteed. Third, (http://dx.doi.org/10.1136/bmjopen-2016-012885).
the significantly lower level of education in the GP group
Rita Cardoso Degree in Speech Therapy from the School of Health of
compared to the other naïve listeners groups might have the University of Aveiro. Master in neuroscience from the Faculty of
been a major limitation to the ability to transcript words Medicine of the University of Lisbon. Post-graduated in oropharyngeal
and sentences correctly. In Portugal, 52.8% of individuals dysphagia by the Epap Institute (I.EPAP). Coordinator of the speech
aged 65 years or older have only a primary school level of therapist therapy service at the Senior Neurological Campus in Torres
Vedras, where she works exclusively with neurological pathology.
education [27]. Admittedly, less educated individuals have Lecturer of Speech Therapy at the Alcoit~ao Health School of Sciences.
lower frequency of reading and writing habits. However, Member of the Portuguese Society for Movement Diseases (SPDMOV)
this conjecture is beyond the scope of this study. Fourth, and the Portuguese Society for Speech Therapy (SPTF). Researcher at
the type of error of the sentence (e.g. phonetical or lexical) the Instituto de Medicina Molecular, dedicated to the study of speech
was not taken into account and if it had changed the mean- and swallowing disorders in movement disorders. Involvement in
financed project in the field of the article: FraLusoPark-Dysarthria in
ing of the sentence. Therefore, it is not possible to inference Parkinson Disease.
if the type of error influenced the intelligibility rating of all
or a particularly group of listeners although it was not the Isabel Guimar~ aes Bachelor of Speech Therapy; Post-graduated in
Lecturing and Administration from Alcoit~ao Health School of Sciences;
purpose of this study. Moreover, the aim was to use the real PhD in Experimental Phonetics - University College London, currently
PD speech signal. Fourth, the stimuli audio-data were Professor at the Alcoit~ao Health School of Sciences, Coordinator of the
selected from PD patients in ON state. Master in Speech Therapy (specialization in Orofacial Motricity and
Swallowing); Senior researcher at the Unit of Clinical Pharmacology
and Therapeutics - Institute of Molecular Medicine - University of
Lisbon; Founder and former editor-in-chief of the Portuguese Journal
Suggestions for future research of Speech Therapy; Currently scientific Vice-President of the
Portuguese Society of Speech Therapists; Involvement in financed pro-
Future research should increase the strength of these results
ject in the field of the article: FraLusoPark-Dysarthria in
and seek improvement directed towards: (i) enhancing formal Parkinson Disease.
examination through a hearing screening prior to presenting
Joaquim J. Ferreira Neurologist, having completed his medical training
speech stimuli. Prevalence of age-related hearing loss, in the
and subsequent PhD in Neurology at the Faculty of Medicine of the
United States, is approximately one in three people age 65–74 University of Lisbon, currently Professor of Neurology and Clinical
and nearly one in two people age 75[28]; (ii) distinguish Pharmacology and Director of the Laboratory of Clinical
between intelligibility (acoustic signal only) and comprehensi- Pharmacology and Therapeutics at the Faculty of Medicine of the
bility (visual and acoustic information) data from PD rela- University of Lisbon, director of a Research Group at the Institute of
Molecular Medicine and Clinical Director of the CNS-Campus
tives and GP listeners can be a useful index to better predict
Neurol
ogico Senior, co-director of the Movement Diseases Panel of the
how PD patients are perceived by naïve listeners [17]; (iii) European Academy of Neurology and former President of the
identifying speech intelligibility analysis with ON and OF European Section of the International Society for Parkinson’s Disease
state PD patients, and also (iv) determining speech intelligi- and other Diseases of the Movement.
bility in dysarthria secondary to Multisystemic Atrophy,
Supranuclear Palsy and cerebellar ataxia. ORCID
Joaquim J. Ferreira http://orcid.org/0000-0003-3950-5113

Acknowledgments
The authors gratefully acknowledge all participants for their voluntary References
collaboration in this study. [1] Middag C. Automatische analyse van pathologische spraak
[Automatic analysis of pathological speech]. Universiteit Gent;
2013.
Disclosure statement [2] Hustad KC. The relationship between listener comprehension
and intelligibility scores for speakers with dysarthria. J Speech
This research did not receive any specific grant from funding agencies Lang Hear Res. 2008;51:562–573.
in the public, commercial, or not-for-profit sectors. None of the [3] Duffy JR. Motor speech disorders: substrates, differential diag-
authors have any conflicts of interest or financial disclosures to report. nosis and management. 3rd ed. St. Louis (MO): Elsevier; 2013.
LOGOPEDICS PHONIATRICS VOCOLOGY 7

[4] Bunton K, Kent RD, Duffy JR, et al. Listener agreement for [17] Yorkston KM, Strand EA, Kennedy MRT. Comprehensibility of
auditory – perceptual ratings of dysarthria. J Speech Lang Hear dysarthric speech: implications for assessment and treatment
Res. 2007;50:1481–1496. planning. Am J Speech Lang Pathol. 1996;5:55–66.
[5] Kreiman J, Gerratt BR, Precoda K. Listener experience and per- [18] Mcgarr NS. The effect of context on the intelligibility of hear-
ception of voice quality. J Speech Hear Res. 1990;33:103–115. ing and deaf children’s speech. Lang Speech. 1981;24:255–264.
[6] Hustad KC. A closer look at transcription intelligibility for [19] Yorkston KM, Beukelman DR. Communication efficiency of
speakers with dysarthria: evaluation of scoring paradigms and dysarthric speakers as measured by sentence intelligibility and
linguistic errors made by listeners. Am J Speech Lang Pathol. speaking rate. J Speech Hear Disord. 1981;46:296–301.
2006;15:268–277. [20] Weismer G, Martin RE. 3. Acoustic and perceptual approaches
[7] Oldfield RC. The assessment and analysis of handedness: the to the study of intelligibility. In: Kent RD, editor. Intelligibility
Edinburgh inventory. Neuropsychologia. 1971;9:97–113. in speech disorders: theory, measurement and management.
[8] Anand S, Stepp CE. Listener perception of monopitch, natural- Philadelphia (PA): John Benjamins; 1992. p. 67–118.
ness, and intelligibility for speakers with Parkinson’s disease. [21] Wild CJ, Yusuf A, Wilson DE, et al. Effortful listening: the
J Speech Lang Hear Res. 2015;58:1134–1144. processing of degraded speech depends critically on attention.
[9] Platt LJ, Andrews G, Young M, et al. Dysarthria of adult cere- J Neurosci. 2012;32:14010–14021.
bral palsy: I. Intelligibility and articulatory impairment. [22] Gibb WR, Lees AJ. The relevance of the Lewy body to the
J Speech Hear Res. 1980;23:28–40. pathogenesis of idiopathic Parkinson’s disease. J Neurol
[10] Eadie TL, Baylor CR. The effect of perceptual training on inex- Neurosurg Psychiatry. 1988;51:745–752.
perienced listeners’ judgments of dysphonic voice. J Voice. [23] Cardoso R, Guimar~aes I, Santos H, et al. Frenchay dysarthria
2006;20:527–544.
assessment (FDA-2) in Parkinson’s disease: cross-cultural adap-
[11] Barkmeier-Kraemer J. Intelligibility of dysarthric speakers:
tation and psychometric properties of the European Portuguese
audio-only and audio-visual presentations. University of Iowa;
version. J Neurol. 2017;264:21–31.
1988 [accessed 2017 Mar 7]. Available from: https://www.
[24] Walshe M, Miller N, Leahy M, et al. Intelligibility of dysarthric
researchgate.net/publication/35891421_Intelligibility_of_dysarth-
speech: perceptions of speakers and listeners. Int J Lang
ric_speakers_audio-only_and_audio-visual_presentations
[12] Miller N, Allcock L, Jones D, et al. Prevalence and pattern of Commun Disord. 2008;43:633–648.
perceived intelligibility changes in Parkinson’s disease. J Neurol [25] Yorkston KM, Beukelman DR. A clinician-judge technique for
Neurosurg Psychiatry. 2007;78:1188–1190. quantifying dysarthric speech based on single-word intelligibil-
[13] Hustad KC, Cahill MA. Effects of presentation mode and ity. J Commun Disord. 1980;13:15–31.
repeated familiarization on intelligibility of dysarthric speech. [26] Sussman JE, Tjaden K. Perceptual measures of speech from
Am J Speech Lang Pathol. 2003;12:198–208. individual with Parkinson’s disease and Multiple Sclerosis:
[14] Tjaden K, Liss JM. The influence of familiarity on judgments of intelligibility and beyond. J Speech Lang Hear Res. 2012;55:
treated speech. Am J Speech Lang Pathol. 1995;4:39–48. 1208–1220.
[15] Liss JM, Spitzer SM, Caviness JN, et al. The effects of familiar- [27] dos Santos FFM; PORDATA Base de Dados Portugal
ization on intelligibility and lexical segmentation in hypokinetic Contempor^aneo, n.d. [accessed 2020 Jun 8]. https://www.por-
and ataxic dysarthria. J Acoust Soc Am. 2014;112:3022–3030. data.pt/Portugal/Populaç~ao+residente+com+15+a+64+anosþ
[16] Garcia JM, Cannito MP. Top down influences on the intelligi- e+65+e+mais+anos+por+nıvel+de+escolaridade+completoþ
bility of a dysarthric speaker: addition of natural gestures and mais+elevado+(percentagem)-2266-179430
situational context. In: Robin D, Yorkston K, Beukelman D, [28] National Institute on Aging (NIA). Hearing loss: a common
editors. Disorders of motor speech: assessment, treatment, and problem for older adults. n.d. [accessed 2020 May 21]. https://
clinical characterization. Baltimore (MD): Paul H. Brookes; www.nia.nih.gov/health/hearing-loss-common-problem-older-
1996. p. 89–104. adults

You might also like