Skip to main content
NCBI home page
Journal of Speech, Language, and Hearing Research : JSLHR logoLink to Journal of Speech, Language, and Hearing Research : JSLHR
. 2025 Jun 4;68(7):3090–3106. doi: 10.1044/2025_JSLHR-24-00802

The Use of Conversational Repairs in Interactions Involving People With Parkinson's Disease: A Large Corpus Comparison Study

Katerina A Tetzloff a,✉, Tyson S Barrett a, Julie M Liss b, Stephanie A Borrie a
PMCID: PMC12263192  PMID: 40466626

Abstract

Purpose:

Interlocutors engage in acts of conversational repair to resolve trouble sources, or communication breakdowns. This is necessary for successful communication, allowing interlocutors to establish and maintain common ground. Here, we investigated the use of conversational repairs in the conversations of people with Parkinson's disease (PD) and concomitant dysarthria.

Method:

Conversational repairs were coded in a large corpus of 114 conversations involving a person with PD and a neurotypical (NT) partner (NT–PD dyads) and a comparison corpus of 80 conversations involving two NT partners (NT–NT dyads). Conversations varied across two contextual dimensions: conversational goal (informational vs. relational) and partner familiarity (familiar vs. unfamiliar).

Results:

Over the course of 10-min conversations, NT–PD and NT–NT dyads produced a similar number of conversational repairs; however, NT–PD dyads exhibited higher usage of repairs per conversational turn. In addition, dyads involving participants with PD with moderate dysarthria showed different repair patterns relative to dyads involving participants with PD with mild or mild–moderate dysarthria as well as NT–NT dyads—the key being that they used fewer self-initiated and more other-initiated repairs. Across all dyads, informational conversations had more repairs than relational conversations, and dyads involving participants with PD with moderate dysarthria relied more heavily on other-initiated repairs when conversing with a familiar partner relative to an unfamiliar partner.

Conclusions:

The use of repairs in the conversations of people with PD differs from that in NT–NT conversations. The key is that the conversations of people with PD have a greater number of repairs per conversational turn and the distribution of repairs differs when people with PD have more pronounced intelligibility impairments. Additionally, the study provides empirical support for longstanding claims regarding the use of repairs in NT–NT dyads as well as claims from case study conversations involving people with dysarthria.

Parkinson's disease (PD) is the fastest growing neurological disease in the world and is expected to double in incidence, reaching 12 million people by 2040 (Dorsey et al., 2018; Yang et al., 2020). People with PD experience restricted communicative participation (McAuliffe et al., 2017; Miller et al., 2006), which means that they have difficulty participating in life situations where information, ideas, feelings, and more are discussed (Yorkston et al., 2007). In other words, they have difficulty communicating with others. This restricted communicative participation can have significant consequences for one's well-being, including social isolation, loss of employment, loss of relationships, and difficulty accessing services. As such, improving communicative participation has been identified as a critical need by clinicians and their clients with PD (Yorkston et al., 2007). One of the key communication challenges for people with PD is the motor speech disorder of dysarthria. Prior research reports that as many as 90% of people with PD will develop dysarthria and reduced speech intelligibility as the disease progresses (Moya-Galé & Levy, 2019). While dysarthria has been identified as the predominant communication issue for people with PD, co-occurring deficits in high-level complex language function may be present in some individuals and contribute to broader communication challenges (Altmann & Troche, 2011).

Because PD does not necessarily shorten one's life span, there is a large window of opportunity to promote communicative participation, allowing for the maintenance of professional and personal relationships, and continued engagement in everyday life roles. Traditional behavioral treatment focuses on modifying the way in which a client with PD speaks to improve intelligibility (i.e., integrity of the degraded signal). The expectation is that increasing intelligibility of the person with PD will generalize to improved communicative participation (Baylor & Darling-White, 2020; Collis & Bloch, 2012; Torrence et al., 2016). However, empirical examination of this assumption has found that intelligibility improvements have only weak to moderate relationships with improvements in participation outcomes (Borrie et al., 2022; Sixt Börjesson et al., 2021; Spencer et al., 2020). That is, intelligibility impairment of the person with PD does not sufficiently explain the communicative participation barriers experienced by this growing population (Borrie et al., 2022). In other words, successful participation in conversation with others is not solely based on the intelligibility of one conversational partner.

According to synergistic models of successful participation in conversation, partners must functionally coordinate their interaction behaviors to ground the conversation, creating shared mutual beliefs, knowledge, and understanding (i.e., common ground), including resolving sources of misunderstanding (McCabe, 2007). One method for assuring common ground is the use of conversational repairs. Conversational repairs refer to interactions in which a conversational partner notes a miscommunication (also known as a “trouble source”) and then resolves it (Albert & De Ruiter, 2018; Sacks et al., 1974; Schegloff et al., 1977). When such communication breakdowns occur (e.g., misunderstanding or mishearing), repairs are necessary so that conversational partners can proceed with the shared understanding that common ground has been achieved. Given the joint nature of conversation, trouble sources are a mutual problem for both partners. As such, all conversational repairs involve two roles and two activities. The partner that produces the trouble source is referred to as “self,” and the recipient-partner is referred to as “other”; the two activities are initiating the repair and completing the repair. Thus, there is self-initiated self-repair (SISR), self-initiated other repaired (SIOR), other-initiated self-repaired (OISR), and other-initiated other-repaired (OIOR). Examples of each type of repair can be seen in Table 1, which were selected from our own corpus used in the current study.

Table 1.

Examples of each of the four types of conversational repair.

Repair Example
SISR P1: Mine has three of those things. No! Four of those things.
SIOR P1: and there are three yellow (4.1 s) I don't know what they're called.
P2: like seashells
OISR P1: Does yours have a green garbage can?
P2: What color?
P1: Green
OIOR P1: The sign is green, the same color as the house.
P2: You mean the café.
Open in a new tab

Note. SISR = self-initiated self-repair; SIOR = self-initiated other-repair; OISR = other-initiated self-repair; OIOR = other-initiated other-repair; P1 = self-partner 1; P2 = other-partner 2.

Past work on conversational repairs have been based heavily in qualitative studies. A seminal claim from this earlier work is that SISRs are tremendously more frequent than other types of conversational repairs, at least in the conversations of neurotypical partners (NT–NT dyads). In order to complete a SISR, the self-partner must first notice the trouble source via self-monitoring (Clark, 1994; Levelt, 1983) and then interrupt their flow of speech to create a new utterance that resolves the trouble source and any potential consequences for their recipient-partner. While robust empirical evidence is currently lacking, intuitively, the presence of SISRs makes sense because they are the only type of repair that can happen within the same conversational turn, making them the most efficient (Sacks et al., 1974; Schegloff et al., 1977).

It has also been claimed that OIOR should be the least frequent type of conversational repair in NT–NT dyads (Clift, 2016; Kitzinger, 2012; Schegloff, 2007; Schegloff et al., 1977). The paucity of OIORs may be explained by a study by Kendrick (2015), which showed that the timing of OIOR is 400–600 ms slower than SISRs, which typically take between 100 and 300 ms to complete. In that extended window of time, there is the opportunity for the creator of the trouble source (i.e., self-partner) to self-correct and resolve the miscommunication faster than if that waited for their recipient-partner to identify the ambiguity. It has been further suggested that other-initiated repairs (both OISR and OIOR) may be more cognitively demanding than SISRs, which may also contribute to why they are seemingly dispreferred (Mertens & de Ruiter, 2021).

In a few quantitative studies, the frequency of conversational repairs has been shown to differ based on the goals of the conversation. For example, conversations with greater informational needs (e.g., solving a task or a puzzle) have been shown to yield more repairs overall than those that are relational or rapport-building (Colman & Healy, 2011; Dideriksen et al., 2023; Dingemanse et al., 2015). It has been suggested that maintaining highly precise common ground is more important in informational versus relational conversations, and thus more repairs are required to assure this priority. On the other hand, rapport-building conversations can deal with more ambiguity and the desire to not interrupt the flow of conversation may be a higher priority in these conversations. Partner familiarity (i.e., whether one previously knew their partner or not) has not been shown to affect the use of conversational repairs for NT–NT dyads, with familiar conversational partners engaging in a comparable number of repairs as unfamiliar partners (Colman & Healy, 2011).

Research into the use of conversational repairs in the conversations of people with PD and concomitant dysarthria is much less robust. However, people with dysarthria from various neurological etiologies frequently experience communication breakdowns in their conversations, and existing case studies have suggested that this impacts the use of conversational repairs. For example, Bloch and Wilkinson (2011) examined the use of repairs in two individuals with dysarthria secondary to multiple sclerosis and motor neuron disease and found that other-initiated repairs were more frequent among these individuals relative to their NT counterparts; in other words, the person with dysarthria would say something, and the NT partner would have to ask them to repeat what they said because they did not understand the first time. Furthermore, it has been argued that SISRs, which are the most common type of repair in NT partners, were less effective for people with dysarthria because the articulatory and respiratory effort required to complete a self-repair may not be worth the exertion (Bloch & Barnes, 2020; Bloch & Wilkinson, 2004, 2009). Rather, other-initiated repairs, which function to highlight some difficulty that one participant is having in understanding the other's previous turn (Schegloff, 2007), have been found to be more prevalent in conversations of people with dysarthria (Bloch, 2006; Bloch & Wilkinson, 2004, 2011). Collectively, these small case studies suggest that the presence of dysarthria results in troubles with understanding, leading the NT partners to ask their partner to repeat themselves or clarify details so that common ground can be reestablished. Additionally, people with dysarthria may also be less likely to self-initiate repairs because they have been shown to struggle initiating speech, particularly in PD (Miller et al., 2006, 2011; Sapir, 2014). A longer pause preceding an intended self-repair provides a window of opportunity for an other-initiated repair to occur instead. Existing literature in this area, however, is all based on limited data of a few conversations among few participants.

Current Study

The contributions of this study are two-fold. The primary goal is to generate rigorous and robust conclusions about the use of conversational repairs in various conversation types involving a person with hypokinetic dysarthria secondary to PD and an NT communication partner (NT–PD dyads). To do this, we use a large clinical corpus of 114 NT–PD conversations, and a comparison corpus of the same types of NT–NT conversations from 80 age-matched dyads (NT–NT dyads). However, the use of a large corpus of NT–NT conversations makes it possible to address a secondary goal, to empirically investigate long-standing theoretical claims of the use of conversational repairs in NT populations. Using these two conversational corpora, the current study addressed the following research questions: (a) What is the distribution of conversational repair types in the conversations of NT–PD dyads and do these patterns differ from what is observed in the conversations of NT–NT dyads? (b) Are the patterns of conversational repair use in NT–PD dyads influenced by dysarthria severity? (c) Is the use of conversational repairs in NT–NT and NT–PD dyads differentially impacted by the context, specifically the conversational goal (informational vs. relational) and partner familiarity (familiar vs. unfamiliar)?

Given that SISRs have been assumed to be so ubiquitous in NT–NT conversations (Schegloff, 2007), we hypothesized that SISR would be the most common type of conversational repair in both NT–PD and NT–NT conversations; however, we also predicted that other-initiated repairs (OISRs and OIORs) would be more frequent in NT–PD relative to NT–NT conversations due to the presence of dysarthria and its impact on accurate speech perception (Liss et al., 1998), as well as other possible co-occurring challenges (e.g., cognition). We also hypothesized that the NT–PD conversations would contain more conversational repairs than NT–NT dyads overall, particularly for other-initiated repairs. We further predicted that as dysarthria severity increased, the number of SISRs would decrease and the number of OISR would increase, following what has been documented in previous small case studies (Bloch, 2006; Bloch & Wilkinson, 2004, 2011). Finally, regarding conversational context, we predicted that informational conversations would yield more conversational repairs in both types of dyads because of the need to have accurate common ground in order to successfully complete the task, and that conversations between familiar partners, particularly in NT–PD dyads, would yield fewer repairs than in NT–PD conversations with unfamiliar partners because familiar partners may be more accustomed to understanding the degraded speech of the person with PD.

Method

Data Collection

This study was approved by the institutional review board at Utah State University (Protocol #12515), and all participants gave informed consent. As part of a larger corpus study on speech coordination strategies and conversational success involving people with PD, we recorded 194 conversations from 74 unique dyads. Fifty-four dyads (114 conversations) included one partner with hypokinetic dysarthria secondary to PD and one NT partner, while 20 additional dyads (80 conversations) involved two NT partners. The study had a two by two within-subjects factorial design. Conversations differed by conversational goal (informational vs. relational) and partner familiarity (familiar vs. unfamiliar). This yielded four types of conversations: an informational (i.e., task-based) conversation with a familiar partner, an informational conversation with an unfamiliar partner, a relational (i.e., rapport-building) conversation with a familiar partner, and a relational conversation with an unfamiliar partner. Familiar partners were either spouses or close friends, while unfamiliar partners were strangers. Familiarity was considered an important factor for these conversations because past research has shown that there are salient differences between the conversations that people have with familiar partners versus acquaintances versus strangers (Planalp & Benson, 1992; Speer et al., 2024; Templeton et al., 2023). For all conversational pairs, partners were reported that their familiar partner was either a family friend or family member.

The informational conversations were elicited using the Diapix task (Baker & Hazan, 2011; Van Engen et al., 2010). This is a collaborative spot-the-difference task in which conversational partners must work together to try and identify 10 differences between a set of similar pictures. Each partner was given a picture and told not to show the other partner. They were then told that their goal was to identify 10 differences between the pictures in a 10-min period. If participants found all 10 differences with time to spare, they were given another set of pictures to work on. Different Diapix pictures were used for familiar versus unfamiliar conversations, so that the same participants could meaningfully engage in the task multiple times. The Diapix task was chosen for the informational conversations because it simulates problem-solving, where different individuals have different information and have to collaborate to share knowledge and information to achieve a common goal.

The relational conversations were elicited using an adapted version of the Relationship Closeness Induction Task (RCIT; Sedikides et al., 1999). Like the Diapix task, the RCIT was 10 min long. The RCIT uses a validated list of questions and conversational partners are instructed to discuss these questions in as natural of a way as possible. The questions include more general life questions (e.g., “What are your hobbies?”) and more thought-evoking questions (e.g., “If you could have one wish granted, what would it be?”); the initial introductory questions (e.g., “What is your name?”) were removed so that all questions were still applicable for discussion between familiar partners. Participants were told to spend roughly 4 min discussing the more general questions, and the remaining 6 min on the more though-provoking questions. The RCIT task simulates how we get to know someone new in real life, making it a good simulation of a relational-based conversation. The order of conversations (familiar informational, familiar relational, unfamiliar informational, unfamiliar relational) was counterbalanced across participants.

A trained research assistant directed and set-up all conversations. For all conversations, conversational partners were seated facing each other and fitted with a wireless CVL Lavalier microphone synced with either a Shure BLX188 Dual Lavalier System or a Shure SM35 Performance Headset Condenser microphone connected to a Zoom H4N Portable Digital Recorder; the microphone was situated approximately 1 in. from the speaker's mouth. Separate audio channels were used to record each conversational partner, and standard settings (48 kHz; 16-bit sampling rate) were utilized for audio recordings. Partners were seated approximately 6 ft apart from each other, to ensure that the microphones only picked up on that specific talker's voice and not that of their conversational partner.

Participants

There was a total of 92 NT participants, and their mean age as 70 years; 63 (68%) were female. None of the NT participants had a history of speech, language, or cognitive impairment. There was also a total of 35 participants with PD. All participants, both NT and PD, were native speakers of American English and had normal or corrected-to-normal hearing. All participants with PD had a medical diagnosis of PD from a neurologist and a clinical diagnosis of hypokinetic dysarthria of either mild (n = 14), mild–moderate (n = 14), or moderate (n = 7) severity, determined via perceptual analysis by a licensed speech-language pathologist not associated with the current study; this severity rating was based on clinical impressions from a read speech sample from each speaker with PD. The mean age of the participants with PD was 71.5 years, and 13 (37%) were female. All but one participant with PD reported taking medication for their PD. Five participants with PD had a history of deep brain stimulation for their PD, which occurred between one and 3 years prior to participation in this study. The mean number of years since PD diagnosis was 7 years (SD = 5.67). Twenty-six of the 35 participants with PD completed a self-rating of their motor impairment; 10 self-rated as mild, 14 as moderate, and two as severe. Importantly, all participants with PD completed the Communicative Participation Item Bank (CPIB; Baylor et al., 2009), which is a patient report outcome measure, quantifying the impact of one's communication disorders on their ability to participate in everyday interactions. This assessment is scored from 0 (extreme difficulty with communication) to 30 (no difficulty with communication). The self-ratings for the participants with PD in the present study was on average 19.6 (SD = 5.60), indicating that these individuals face substantial challenges in their daily interactions.

Twenty-two of the participants with PD each participated in four conversations, two with the same familiar and unfamiliar partner, for both conversation types (see Figure 1); 12 participants with PD had an informational and relational conversation with a familiar partner only, and one participant with PD had both conversations with an unfamiliar partner only.1 A total of 20 NT participants (some the same as those in NT–PD conversations) also had informational and relational conversations with a familiar and an unfamiliar NT partner (NT–NT conversations). Of the NT–PD unique dyads, 45 were male–female, seven were female–female, and two were male–male; of the NT–NT dyads, 35 were male–female, four were female–female, and one was male–male.

Figure 1.

A flow diagram of study design involving a participant interacting with unfamiliar and familiar partners. It is followed by four 10-minute conversations. Conversation 1. Interaction between participant and familiar partner in informational task. Conversation 2. Interaction between participant and familiar partner in relationship closeness induction task. Conversation 3. Interaction between participant and unfamiliar partner in informational task. Conversation 4. Interaction between participant and unfamiliar partner in relationship closeness induction task.

Open in a new tab

Diagram of study design, showing the process for participants to engage in four conversations with two partners. RCIT = Relationship Closeness Induction Task.

The average speech rate data for participants in syllables per second are shown in Table 2. Furthermore, cognitive data from the National Institutes of Health Cognition Toolbox (Weintraub et al., 2013) were collected from all participants in the NT–PD dyads, which can be seen in Table 3. These data highlight that all cognitive variables were comparable between NT and PD participants with the exception of processing speed, which was significantly lower for PD participants. These data were collected on the same date as the conversations were recorded and were administered by a trained research assistant.

Table 2.

Speech rate data in syllables per second.

Talkers Mean SPS (SE) – informational conversations Mean SPS (SE) – relational conversations
NT with NT partners 3.97 (0.02) 4.10 (0.02)
NT with PD partners 4.05 (0.02) 4.03 (0.02)
PD with mild dysarthria 3.96 (0.03) 4.23 (0.03)
PD with mild-to-moderate dysarthria 4.02 (0.03) 4.37 (0.04)
PD with moderate dysarthria 3.84 (0.04) 4.09 (0.04)
Open in a new tab

Note. SPS = syllables per second; SE = standard error; NT = neurotypical; PD = Parkinson's disease.

Table 3.

Cognitive variables from Parkinson's disease (PD) and neurotypical (NT) participants of NT–PD dyads.

NIH cognitive test PD M (SE) NT M (SE) p value
Working memory (list sorting) 98.17 102.30 .18
Inhibitory control of attention (flanker test) 88.45 92.95 .11
Cognitive flexibility (dimensional change card sort test) 103.55 109.39 .11
Processing speed (pattern comparison) 82.03 100.82 < .001
Vocabulary knowledge (picture vocabulary test) 107.62 107.02 .84
Open in a new tab

Note. NIH = National Institutes of Health; SE = standard error.

Conversational Analysis

Author K.A.T. listened to the audio files and coded each instance of a conversational repair by denoting which participant caused the trouble source, which participant initiated the repair, and which participant completed the repair. Trouble sources were identified when either participant indicated difficulty with producing or understanding an utterance, and repairs were identified when either participant addressed the trouble source and subsequently completed the repair; the same participant did not necessarily initiate and complete the repair. The repairs were then coded as SISR, SIOR, OISR, or OIOR depending on who caused the trouble source, initiated the repair, and completed the repair either within the same turn or in a subsequent (and not necessarily immediately) occurring turn following the trouble source. A research assistant was trained to code the conversational repairs and independently did so for 20% (n = 39) of the conversations. A free marginal multirater kappa coefficient was calculated to determine interrater reliability and yielded a value of .83, indicating very strong interrater reliability (Landis & Koch, 1977).

Statistical Analysis

To assess the distribution of conversational repairs in our NT–NT and NT–PD conversations, we used linear mixed-effects regression modeling to flexibly estimate differences in the dependent variable while accounting for the repeated measures within conversations and participants. The number of repairs was the dependent measure, with the type of repair and dyad type (NT–NT or NT–PD) as an interaction term; random intercepts were included by the unique pairings of conversational participants.2 We also assessed whether the number of repairs differed by dyad severity (NT–NT, NT–mild, NT–mild-to-moderate, NT–moderate) and whether that varied by repair type (i.e., we specified an interaction between dyad type and repair type). As before, a random intercept was included, this time by conversation.3 Both of these models were also run with repair rate (i.e., number of conversational repairs per 10 conversational turns) as the dependent measure as well.

The final research question targeted how conversational context, both goal and partner familiarity, affected the distribution of conversational repairs in both the NT–NT and the NT–PD dyads (by severity). As with previous research questions, we used linear mixed-effects regression modeling. As before, the number of repairs was the dependent variable. We then used a “top down” approach: first testing the four-way interactions of the independent variables: type of repair, dyad severity, partner familiarity (familiar vs. unfamiliar), and conversational goal (informational vs. relational), then increasing parsimony with three-way interactions, and so on. This top-down approach was used until an interaction was significant or only main effects remained.4

For all models, we used the likelihood ratio test to evaluate the interactions within each model (as compared to a more parsimonious model), and Tukey's honestly significant difference (HSD) post hoc tests were used to conduct pairwise comparisons. All analyses were conducted in R (R Core Team, 2020) with the packages “lme4” (Bates et al., 2015), “lmerTest” (Kuzentsova et al., 2017), and “emmeans” (Lenth, 2024).

Results

Descriptive Measures

The number of turns and the duration of each turn were analyzed for each NT–NT and NT–PD conversation and are shown in Table 4. Overall, there were more turns in informational versus relational conversations, and turn duration was longer in relational versus informational conversations. NT and PD participants took a comparable number of and duration of turns, apart from PD with moderate dysarthria taking fewer and shorter turns.

Table 4.

Mean number of turns and turn duration by dysarthria severity in NT–NT and NT–PD conversations.

Variable Mean number of turns (SE) – informational conversations Mean number of turns (SE) – relational conversations Mean duration of turns in seconds (SE) – informational conversations Mean duration of turns in seconds (SE) – relational conversations
NT with NT partners 122.17 (7.90) 100.04 (6.18) 1.87 (0.02) 2.24 (0.02)
NT with PD partners 137.96 (4.27) 109.00 (4.08) 1.84 (0.03) 2.37 (0.11)
PD with mild dysarthria 135.42 (4.07) 111.68 (6.36) 1.89 (0.03) 2.11 (0.03)
PD with mild-to-moderate dysarthria 132.60 (3.84) 94.50 (6.54) 1.86 (0.06) 2.18 (0.07)
PD with moderate dysarthria 105.67 (9.96) 83.61 (7.64) 1.65 (0.04) 2.13 (0.06)
Open in a new tab

Note. NT = neurotypical; PD = Parkinson's disease; SE = standard error.

Primary Analyses

Our first research question examined the number of and distribution of each type of conversational repair in the NT–NT and NT–PD conversations. The average number of each type of repair across all NT–NT and NT–PD conversations can be seen in Table 5. As illustrated in the left panel of Figure 2, for NT–NT conversations, SISR was the most frequent type of repair (for pairwise comparisons with OISR, OIOR, and SIOR, all ßs > 5.94, ps < .001). In these NT–NT conversations, the other three types of repairs occurred at relatively the same frequency.

Table 5.

Mean number of each type of repair in NT–NT and NT–PD conversations.

Type of repair NT–NT M (SE) NT–PD M (SE)
SISR 6.78 (0.37) 6.99 (0.32)
SIOR 0.20 (0.06) 0.21 (0.05)
OISR 0.84 (0.15) 1.58 (0.20)
OIOR 0.62 (0.12) 0.42 (0.07)
Open in a new tab

Note. NT = neurotypical; SE = standard error; PD = Parkinson's disease; SISR = self-initiated self-repair; SIOR = self-initiated other-repair; OISR = other-initiated self-repair; OIOR = other-initiated other-repair.

Figure 2.

Two combinations of a bar chart and a scatter plot for NT NT and NT PD. In both graphs, the horizontal axis represents the four repair types and the vertical axis represents the number of repairs per 10 minutes conversation ranges from negative 1 to 19. In graph 1, the bars have the number of repairs per 10 minutes conversation data as follows. SISR: 7. SIOR: 0.3. OISR: 1. OIOR: 0.9. The data points at SISR scattered between 0 and 16, SIOR between negative 1 and 3, OISR between negative 1 and 6, and OIOR between negative 1 and 6. In graph 2, the bars have the number of repairs per 10 minutes conversation data as follows. SISR: 7.2. SIOR: 0.3. OISR: 2. OIOR: 0.7. The data points at SISR scattered between 0 and 19, SIOR between negative 1 and 3, OISR between negative 1 and 13, and OIOR between negative 1 and 5.

Open in a new tab

Conversational repairs in NT–NT and NT–PD conversations. NT = neurotypical; PD = Parkinson's disease; SISR = self-initiated self-repair; SIOR = self-initiated other-repair; OISR = other-initiated self-repair; OIOR = other-initiated other-repair.

A similar distribution of repair types was observed for the NT–PD conversations, shown in the right panel of Figure 2. For the NT–PD conversations, SISR was again the most frequent type of repair (for pairwise comparisons with OISR, OIOR, SIOR all ßs > 5.4 and all ps < .001), and the second most frequent type of repair was OISR, which was significantly more common than OIOR (ß = 1.16, p < .001), and SIOR (ß = 1.38, p < .001). SIOR and OIOR were the least frequent repair types, and their low frequency was comparable in these NT–PD conversations. The likelihood ratio test of our linear model revealed that the overall number of repairs did not significantly differ between NT–NT and NT–PD dyads, F(1, 93) = 1.61, p = .21, and there was no significant interaction between dyad type and type of repair, F(3, 659) = 1.91, p = .13; see Figure 2.

Given that the number of turns differed between groups, with participants with PD and concomitant moderate dysarthria taking fewer conversational turns (see Table 4 above), we then looked at the distribution of conversational repairs based on repair rate, or number of repairs per 10 conversational turns. When looking at the repair, differences between NT and PD participants are revealed. The average repair rate for each type of repair across NT–NT and NT–PD conversations can be seen in Table 6. Figure 3 also shows the repair rates for each type of conversational repair in NT–NT and NT–PD conversations. When looking at repair rate, it is evident that NT–PD conversations have more conversational repairs per 10 turns than NT–NT conversations, as evidenced by the likelihood ratio test, F(3, 659) = 2.72, p = .04; pairwise comparisons revealed that this was particularly the case for OISRs, which happened at a more frequent rate in NT–PD conversations than in NT–NT conversations.

Table 6.

Mean repair rate (number of repairs per 10 conversational turns) in NT–NT and NT–PD conversations.

Type of repair NT–NT M (SE) NT–PD M (SE)
SISR 0.27 (0.01) 0.30 (0.01)
SIOR 0.007 (0.002) 0.009 (0.002)
OISR 0.03 (0.005) 0.07 (0.008)
OIOR 0.02 (0.005) 0.02 (0.003)
Open in a new tab

Note. NT = neurotypical; SE = standard error; PD = Parkinson's disease; SISR = self-initiated self-repair; SIOR = self-initiated other-repair; OISR = other-initiated self-repair; OIOR = other-initiated other-repair.

Figure 3.

Two combinations of a bar chart and a scatter plot for NT NT and NT PD. In both graphs, the horizontal axis represents the four repair types and the vertical axis represents the number of repairs per 10 minute conversational turns ranges from 0 to 1. In graph 1, the bars have the number of repairs per 10 minute conversational turns data as follows. SISR: 0.27. SIOR: 0.00. OISR: 0.05. OIOR: 0.04. The data points at SISR scattered between 0.00 and 0.65, SIOR between 0.00 and 0.12, OISR between 0.00 and 0.25, and OIOR between 0.00 and 0.25. In graph 2, the bars have the number of repairs per 10 minute conversational turns data as follows. SISR: 0.30. SIOR: 0.02. OISR: 0.10. OIOR: 0.03. The data points at SISR scattered between 0.00 and 0.90, SIOR between 0.00 and 0.25, OISR between 0.00 and 0.50, and OIOR between 0.00 and 0.25.

Open in a new tab

Conversational repair rates in NT–NT and NT–PD conversations. NT = neurotypical; PD = Parkinson's disease; SISR = self-initiated self-repair; SIOR = self-initiated other-repair; OISR = other-initiated self-repair; OIOR = other-initiated other-repair.

Our second research question examined how dysarthria severity impacted the number of and distribution of conversational repairs in the NT–PD conversations. This analysis revealed differences by dyad severity (NT–NT, NT–mild, NT–mild-to-moderate, NT–moderate), which are illustrated in Table 7 and Figure 4. A likelihood ratio test of the linear model showed a significant interaction between type of repair and dyad severity, F(9, 653) = 5.40, p < .001. Tukey's HSD post hoc tests demonstrated that NT–NT dyads had a similar number of SISRs as NT–mild and NT–mild-to-moderate dyads (−0.87 < ßs < −0.49, ps > .51). However, NT–moderate dyads had significantly fewer SISRs than all other dyad types (ßs < −1.51, ps < .06). NT–moderate dyads showed significantly more OISRs than NT–NT (ß = 1.90, p = .002) and NT–mild dyads (ß = 1.89, p = .008); NT–mild-to-moderate and NT–moderate dyads were not different with respect to OISRs (ß = 1.09, p = .68). SIORs and OIORs were comparable across all dyad types.

Table 7.

Mean number of each type of repair by dyad type.

Type of repair Dyad – M (SE)
NT–NT NT–mild NT–mild-to-moderate NT–moderate
SISR 6.78 (0.37) 7.66 (0.50) 7.28 (0.55) 5.11 (0.49)
SIOR 0.20 (0.06) 0.28 (0.09) 0.18 (0.06) 0.12 (0.08)
OISR 0.84 (0.15) 0.90 (0.21) 1.68 (0.29) 2.78 (0.63)
OIOR 0.62 (0.12) 0.48 (0.12) 0.50 (0.10) 0.19 (0.10)
Open in a new tab

Note. SE = standard error; NT = neurotypical; SISR = self-initiated self-repair; SIOR = self-initiated other-repair; OISR = other-initiated self-repair; OIOR = other-initiated other-repair.

Figure 4.

A dot plot with error bars for four different repair types versus the mean number of repairs per 10-minutes conversation ranges from 0 to 9. The four repair types are SISR, SIOR, OISR, and OIOR. The dyad conditions: NT NT, NT mild, NT mild to moderate, and NT moderate are represented by different shades. The data are as follows. SISR. NT NT: 6.8. NT mild: 7.7. NT mild to moderate: 7.3. NT moderate: 5.3. SIOR. NT NT: 0.2. NT mild: 0.3. NT mild to moderate: 0.2. NT moderate: 0.1. OISR. NT NT: 0.9. NT mild: 0.95. NT mild to moderate: 1.8. NT moderate: 2.8. OIOR. NT NT: 0.6. NT mild: 0.5. NT mild to moderate: 0.5. NT moderate: 0.2.

Open in a new tab

Mean number of repairs per conversation by dysarthria severity. NT = neurotypical; SISR = self-initiated self-repair; SIOR = self-initiated other-repair; OISR = other-initiated self-repair; OIOR = other-initiated other-repair.

Similar data but looking at repair rate rather than total number of repairs can be seen in Table 8 and Figure 5. When looking at repair rate rather than number of repairs per conversation, dyad type again showed a significant interaction with type of repair, based on a likelihood test, F(9, 653) = 6.09, p < .0001. Pairwise interactions indicated that NT–mild dyads had significantly more repairs than NT–NT dyads (ß = 0.06, p = .02) and NT–moderate dyads (ß = 0.09, p = .0003). Additionally, NT–moderate dyads had a significantly greater repair rate than NT–NTs in terms of OISRs (ß = 0.21, p < .0001), as well as NT–mild dyads (ß = 0.009, p = .001).

Table 8.

Mean repair rate by dyad type.

Type of repair Dyad – M (SE)
NT–NT NT–mild NT–mild-to-moderate NT–moderate
SISR 0.27 (0.01) 0.33 (0.02) 0.31 (0.02) 0.24 (0.02)
SIOR 0.007 (0.002) 0.01 (0.005) 0.007 (0.002) 0.005 (0.004)
OISR 0.03 (0.005) 0.04 (0.008) 0.07 (0.01) 0.12 (0.03)
OIOR 0.02 (0.005) 0.02 (0.005) 0.02 (0.004) 0.009 (0.004)
Open in a new tab

Note. SE = standard error; NT = neurotypical; SISR = self-initiated self-repair; SIOR = self-initiated other-repair; OISR = other-initiated self-repair; OIOR = other-initiated other-repair.

Figure 5.

A dot plot with error bars for four different repair types versus the mean number of repairs per 10-minute conversational turns ranges from 0.0 to 0.35. The four repair types are SISR, SIOR, OISR, and OIOR. The dyad conditions: NT NT, NT mild, NT mild to moderate, and NT moderate are represented by different shades. The data are as follows. SISR. NT NT: 0.27. NT mild: 0.33. NT mild to moderate: 0.31. NT moderate: 0.28. SIOR. NT NT: 0.01. NT mild: 0.02. NT mild to moderate: 0.01. NT moderate: 0.01. OISR. NT NT: 0.03. NT mild: 0.04. NT mild to moderate: 0.06. NT moderate: 0.13. OIOR. NT NT: 0.03. NT mild: 0028. NT mild to moderate: 0.027. NT moderate: 0.01.

Open in a new tab

Mean repair rate by dysarthria severity. NT = neurotypical; SISR = self-initiated self-repair; SIOR = self-initiated other-repair; OISR = other-initiated self-repair; OIOR = other-initiated other-repair.

Our final research question aimed to see how conversational context (conversational goal and partner familiarity) affects the number and distribution of conversational repairs in both NT–NT and NT–PD conversations; for this question, we only examined repair rate and not total number of repairs per conversation. These results can be seen in Table 9 and Figure 6 (NT–NT conversations) and Figure 7 (NT–PD conversations). There was a significant effect of conversational context, F(1, 633) = 30.79, p < .001, where informational conversations yielded significantly more repairs than relational conversations overall across the NT–NT and NT–PD conversations. There was also a significant three-way interaction between repair type, partner familiarity, and dyad type, F(9, 633) = 1.93, p < .05, which is illustrated in Figure 8. Most notably, NT–moderate dyads exhibited significantly a lower rate of SISRs and a higher rate of OISRs in familiar versus unfamiliar conversations. Additionally, NT–NT and NT–mild dyads had a greater repair rate for SISRs in unfamiliar conversations.

Table 9.

Mean number repair rate for each type of repair by conversational context in NT–PD conversations.

Dyad Type of repair Conversational context – M (SE)
Informational
 Relational
Familiar Unfamiliar Familiar Unfamiliar
NT–NT SISR 0.24 (0.02) 0.34 (0.03) 0.21 (0.02) 0.31 (0.03)
SIOR 0.02 (0.007) 0.01 (0.004) 0.00 (0.00) 0.00 (0.00)
OISR 0.05 (0.01) 0.04 (0.01) 0.02 (0.007) 0.01 (0.007)
OIOR 0.05 (0.009) 0.04 (0.01) 0.003 (0.003) 0.006 (0.004)
NT–mild SISR 0.32 (0.03) 0.42 (0.05) 0.27 (0.06) 0.31 (0.05)
SIOR 0.02 (0.007) 0.02 (0.009) 0.02 (0.01) 0.00 (0.00)
OISR 0.08 (0.02) 0.02 (0.006) 0.03 (0.01) 0.01 (0.005)
OIOR 0.04 (0.007) 0.02 (0.01) 0.02 (0.01) 0.00 (0.00)
NT–mild-to-moderate SISR 0.31 (0.02) 0.32 (0.06) 0.32 (0.05) 0.27 (0.04)
SIOR 0.01 (0.006) 0.004 (0.005) 0.03 (0.003) 0.00 (0.00)
OISR 0.10 (0.02) 0.09 (0.04) 0.05 (0.02) 0.03 (0.01)
OIOR 0.03 (0.007) 0.03 (0.01) 0.01 (0.005) 0.007 (0.007)
NT–moderate SISR 0.24 (0.04) 0.29 (0.04) 0.20 (0.03) 0.24 (0.04)
SIOR 0.00 (0.00) 0.03 (0.02) 0.00 (0.00) 0.00 (0.00)
OISR 0.31 (0.07) 0.10 (0.03) 0.10 (0.03) 0.03 (0.01)
OIOR 0.02 (0.01) 0.02 (0.01) 0.00 (0.00) 0.00 (0.00)
Open in a new tab

Note. SE = standard error; SISR = self-initiated self-repair; SIOR = self-initiated other-repair; OISR = other-initiated self-repair; OIOR = other-initiated other-repair.

Figure 6.

Four combinations of a bar chart and a scatter plot arranged 2 rows and 2 columns. The column headers are informational and relational. The row headers are familiar and unfamiliar. In all graphs, the horizontal axis represents the four repair types and the vertical axis represents the mean of repairs per 10 minutes conversational turns ranges from 0.0 to 0.6. In graph 1, the bars have the data as follows. SISR: 0.24. SIOR: 0.02. OISR: 0.05. OIOR: 0.04. The data points at SISR scattered between 0.0 and 0.45, SIOR between 0.0 and 0.15, OISR between 0.0 and 0.2, and OIOR between 0.0 and 0.2. In graph 2, the bars have the data as follows. SISR: 0.21. SIOR: 0.0. OISR: 0.03. OIOR: 0.0. The data points at SISR scattered between 0.0 and 0.5, SIOR scattered at 0, OISR between 0.0 and 0.2, and OIOR between 0.0 and 0.1. In graph 3, the bars have the data as follows. SISR: 0.35. SIOR: 0.02. OISR: 0.05. OIOR: 0.04. The data points at SISR scattered between 0.0 and 0.6, SIOR between 0.0 and 0.1, OISR between 0.0 and 0.2, and OIOR between 0.0 and 0.25. In graph 4, the bars have the data as follows. SISR: 0.3. SIOR: 0.0. OISR: 0.02. OIOR: 0.01. The data points at SISR scattered between 0.0 and 0.6, SIOR scattered at 0, OISR between 0.0 and 0.2, and OIOR between 0.0 and 0.1.

Open in a new tab

Distribution of conversational repairs in NT–NT conversations by context. NT = neurotypical; SISR = self-initiated self-repair; SIOR = self-initiated other-repair; OISR = other-initiated self-repair; OIOR = other-initiated other-repair.

Figure 7.

Four combinations of a bar chart and a scatter plot arranged 2 rows and 2 columns. The column headers are informational and relational. The row headers are familiar and unfamiliar. In all graphs, the horizontal axis represents the four repair types and the vertical axis represents the mean number of repairs per 10 minutes conversational turns ranges from 0.00 to 0.1. In graph 1, the bars have the data as follows. SISR: 0.30. SIOR: 0.02. OISR: 0.15. OIOR: 0.05. The data points at SISR scattered between 0.0 and 0.50, SIOR between 0.0 and 0.15, OISR between 0.0 and 0.50, and OIOR between 0.0 and 0.15. In graph 2, the bars have the data as follows. SISR: 0.27. SIOR: 0.0. OISR: 0.08. OIOR: 0.0. The data points at SISR scattered between 0.0 and 0.1, SIOR between 0.0 and 0.25, OISR between 0.0 and 0.30, and OIOR between 0.0 and 0.25. In graph 3, the bars have the data as follows. SISR: 0.35. SIOR: 0.02. OISR: 0.05. OIOR: 0.03. The data points at SISR scattered between 0.0 and 0.75, SIOR between 0.0 and 0.15, OISR between 0.0 and 0.30, and OIOR between 0.0 and 0.25. In graph 4, the bars have the data as follows. SISR: 0.30. SIOR: 0.0. OISR: 0.02. OIOR: 0.01. The data points at SISR scattered between 0.0 and 0.60, SIOR scattered at 0, OISR between 0.0 and 0.15, and OIOR between 0.0 and 0.10.

Open in a new tab

Distribution of conversational repairs in NT–PD conversations by context. NT = neurotypical; PD = Parkinson's disease; SISR = self-initiated self-repair; SIOR = self-initiated other-repair; OISR = other-initiated self-repair; OIOR = other-initiated other-repair.

Figure 8.

Two bar charts with error bars for familiar equals familiar and familiar equals unfamiliar. In both graphs, the horizontal axis represents four repair types and the vertical axis represents the repair rate ranges from 0.0 to 0.5. The four repair types are SISR, SIOR, OISR, and OIOR. The dyad conditions: NT NT, NT mild, NT mild to moderate, and NT moderate are represented by different shades. Graph 1. The data are as follows. SISR. NT NT: 0.25. NT mild: 0.32. NT mild to moderate: 0.34. NT moderate: 0.25. SIOR. NT NT: 0.02. NT mild: 0.03. NT mild to moderate: 0.01. NT moderate: 0.01. OISR. NT NT: 0.07. NT mild: 0.08. NT mild to moderate: 0.09. NT moderate: 0.19. OIOR. NT NT: 0.04. NT mild: 0.045. NT mild to moderate: 0.035. NT moderate: 0.034. Graph 2. The data are as follows. SISR. NT NT: 0.34. NT mild: 0.39. NT mild to moderate: 0.32. NT moderate: 0.29. SIOR. NT NT: 0.01. NT mild: 0.012. NT mild to moderate: 0.0. NT moderate: 0.02. OISR. NT NT: 0.05. NT mild: 0.035. NT mild to moderate: 0.08. NT moderate: 0.09. OIOR. NT NT: 0.035. NT mild: 0.033. NT mild to moderate: 0.033. NT moderate: 0.033.

Open in a new tab

Plot showing the interaction between partner familiarity and repair type on rate of conversational repairs. NT = neurotypical; SISR = self-initiated self-repair; SIOR = self-initiated other-repair; OISR = other-initiated self-repair; OIOR = other-initiated other-repair.

Discussion

In the current study, we investigated use of conversational repairs in a large clinical corpus of conversations between people with PD and NT communication partners and a comparison corpus of conversations between two NT partners. In all conversations, repairs were a robust phenomenon; in the NT–NT conversations, repairs occurred once every 76.8 s, and in the NT–PD conversations, once every 64.0 s. However, this difference in overall number of repairs was not significant. When looking at the rate of repairs, repairs occurred at a higher rate in NT–PD versus NT–NT conversations; this finding supported our hypothesis that NT–PD dyads would engage in more repairs than NT–NT dyads. The fact that NT–PD conversations used repairs at a greater rate than NT–NT dyads appears to be a direct result of the impaired intelligibility of the PD participant due to their concomitant dysarthria: When intelligibility is reduced, more communication misunderstandings occur, and thus more repairs are needed to reestablish common ground.

Regarding the types of repairs used, across both NT–NT and NT–PD corpora, SISR was the most frequent type of repair used by interlocutors. This finding provides important empirical evidence for Schegloff et al.'s (1977) seminal claim that the prevalence of SISRs far outweighs the presence of other types of conversational repair. An example of this from an NT participant from our corpus is as follows:

P1: I see two pots … I mean plants near the door.

The participant immediately recognizes that “pots” might not be specific enough and self-identifies this as a trouble source, which they then immediately repair with a better description of what they see, “plants.” SISR is likely the most common type of repair because the opportunity for repair initiation and completion arises before it does for other types of repairs, as the self-partner is able to self-correct the trouble source before their conversational partner even has the chance to orient to the trouble source. Thus, the participant can preemptively assure conversational grounding before their recipient-partner has the chance to respond to the trouble source. It is also the only type of repair that can occur within a single conversational turn, making it the most efficient repair mechanism, which may also contribute to why it is so prevalent compared to other types of conversational repair (Sacks et al., 1974; Schegloff et al., 1977). With the single turn repair, SISR may also disrupt the flow of conversation less than other-initiated repairs, making them more favorable. Furthermore, SISRs may prevent bigger misunderstandings later in the conversation, so they can be seen as preventative of more costly repairs (i.e., other-initiated repairs that require multiple conversational turns; Clark, 1994).

SISR was the most common type of repair regardless of if dyad was NT–NT or NT–PD and, if NT–PD, the severity of the dysarthria of the PD participant. However, there were significantly fewer SISRs when the partner with PD presented with dysarthria of moderate severity. This is a notable finding. Specifically, it suggests that the severity of the speech degradation modulates the use of SISRs. We consider several plausible explanations. First, the participants with PD with moderate dysarthria often speak at a lower volume but do not perceive their own speech volume well (J. P. Clark et al., 2014; De Keyser et al., 2016). Thus, they may not realize that their quiet and/or distorted speech caused a trouble source for their partner and thus will not self-repair. Another explanation, not mutually exclusive from the first, is that participants with PD with moderate dysarthria take fewer and shorter turns, minimizing the opportunities for SISRs. Indeed, it has been previously shown that people with dysarthria tend to take shorter conversational turns than their NT peers (Comrie et al., 2001); our descriptive data (see Table 4) showed that the participants with PD with moderate dysarthria did take fewer conversational turns relative to the other participants with PD. However, even when number of conversational turns was accounted for, PD–moderate dyads continued to show a lower repair rate of SISRs than other dyad types. Thus, dysarthria severity impacts the distribution and number of repairs, as those with more degraded speech produced fewer SISRs.

The data also showed that OISRs were the second most frequent type of repair, with SIOR and OIOR occurring relatively infrequently. This finding was, again, robust regardless of the dyad type, and provides empirical support for Schegloff et al.'s (1977) claim that when trouble sources are acknowledged by one's partner (other-initiated), there is still a preference for self-repair (i.e., preference for OISR over OIOR). That this pattern holds in interactions involving people with PD is an important finding. In our corpus, the use of OISRs was also modulated by dysarthria severity of the participant with PD. Specifically, conversations with participants who have mild-to-moderate dysarthria from PD had more OISRs than NT–NT and NT–mild dyads, and conversations with participants with moderate dysarthria had even more OISRs. That is, the more impaired the dysarthria, the more OISRs were utilized. This is also a notable finding and makes intuitive sense: when conversing with a person with more degraded speech, the NT partners must ask their PD partner to repeat or clarify more times to ensure mutual understanding. This finding supports what Bloch and Wilkinson (2011) have previously claimed, that OISRs are relied upon more in conversations with people with dysarthria. Furthermore, this finding is supported by the reported cognitive data (see Table 3). Because participants with PD have slower processing speeds, they may be less likely to self-regulate trouble sources. Furthermore, processing speed may be critical in instigating repairs. This then leads to necessary OISRs initiated by the NT participant. It has been shown that other pragmatic abilities are also affected by processing speed in speakers with PD, such as interpreting ambiguity and making inferences (McKinlay et al., 2009); this suggests that conversational repairs are affected similar to other pragmatic abilities in the population with PD.

Finally, our data showed that conversational context impacted the rate and distribution of repairs in the conversations of both NT–NT and NT–PD dyads. First, there was a significantly greater repair rate in informational versus relational conversations, supporting our hypothesis that informational conversations would yield more repairs than relational ones. Given that clarifying common ground is essential in interactions where accurate information must be exchanged to achieve success (i.e., identify differences between picture sets), more repairs will be utilized. Furthermore, the higher rate of SISRs in informational conversations suggests that participants prioritize quickly resolving misunderstandings to maintain common ground, which is more critical in these contexts versus in relational conversations. OISRs were also more frequent in informational conversations likely because they can be face-threatening compared to SISRs; one's face can be defined as the desire to be a likable, normal, contributing, and supporting member of society (Scollon et al., 2012), and politeness refers to saving's one face. The goal of the informational conversations, which was to exchange accurate information to identify differences in two pictures, requires highly precise common ground, but this may come at the cost of the relational aspect of the conversation (i.e., face and politeness). However, the goal of relational conversations is to build rapport, and thus saving face and maintaining politeness may be prioritized over understanding all the details. Additionally, the increased time and cognitive demand necessary for other-initiated repairs may have contributed to the difference in their presence between informational and relational conversations. Participants may have been more likely to expend these additional resources to achieve their goal in informational conversations (Mertens & de Ruiter, 2021). In relational conversations, on the other hand, participants may have felt that this additional time and cognitive effort was not worthwhile, but rather the desire to keep the conversation flowing took precedence, resulting in fewer OISRs. Importantly, however, the presence of PD (and dysarthria severity) did not disrupt these tendencies for a higher rate of conversational repairs in informational versus relational conversations.

The second dimension along which we examined the impact of conversational context was partner familiarity, where we had hypothesized that familiar partners would use fewer repairs than unfamiliar partners due to, presumably, some level of previously established common ground between the familiar dyad. This hypothesis was partially supported, as dyads involving a PD partner with moderate dysarthria showed significantly lower rate of SISRs and higher rate of OISRs in familiar conversations. The reduced rate of SISRs with familiar partners may in fact be evidence of some a priori common ground that familiar partners in NT–moderate PD dyads share. The fact that we observed an increased rate of OISRs with familiar partners in these conversations with partners with moderate dysarthria may suggest that familiar partners are attempting to support their conversational partner by initiating a repair, which then allows them to continue with common ground. This result differs from what was previously observed in NT–NT conversations, where repair types were not modulated by partner familiarity in informational or relational conversations (Colman & Healy, 2011).

There were a few limitations to the present study. First, a small number of participants were able to have all four conversation types. This was due to logistical scheduling difficulties, trying to coordinate two sets of familiar couples to be able to meet at the date and location. However, the size and diversity of our corpus (194 total conversations from 74 unique dyads) was sufficient to account for intra-individual and interindividual variability and should generalize to the studied population. Second, the manual coding of repair types was not blind to the participant's diagnosis (e.g., if they were an NT speaker or a speaker with PD) because of the marked speech features of dysarthria associated with PD. An alternative to this approach for a future study would be to code repairs based off orthographic transcripts so that the coder could be blind to patient diagnosis. Finally, the participant demographics were imbalanced in terms of participant gender: more males were included than females. However, this is in line with global incidence and prevalence data regarding PD, which indicates that men are diagnosed at a much higher rate than women (Dorsey et al., 2018; Yang et al., 2020).

The major contribution of this work is that it is the first large scale study of repairs in the conversations of people with PD and concomitant dysarthria, evaluating how the use of conversational repairs differs from that of NT speakers. Given the established need for improving communicative participation for people with PD, future research will explore the relationship between the use of repairs and communicative participation, with the eventual goal of identifying hypothesis-driven conversation-based treatment targets that support improved communicative participation for people with PD. From a clinical perspective, our findings have implications for communication rehabilitation for individuals with PD. By identifying the specific repair strategies used by speakers with PD and how this differs from NT speakers, clinicians may tailor interventions to improve conversational interactions by focusing on strategies that are most successful in resolving communication breakdowns (i.e., conversational repairs). Understanding the challenges individuals with PD face with regard to conversational repairs can inform the future development of targeted therapies as well. For example, this work may be used as starting ground to develop clinical assessments that incorporate conversational analysis as part of the speech and language evaluation. As traditional assessments often focus on isolated speech tasks, incorporating conversational data could provide a more comprehensive understanding of communicative participation and the communication challenges for individuals with PD, which could then inform personalized intervention strategies.

Conclusions

This large corpus of conversations between NT–PD and NT–NT dyads revealed that conversational repairs are a robust phenomenon in interactions, including those involving participants with PD. Furthermore, repairs occurred at a higher rate in NT–PD dyads relative to NT–NT dyads. SISRs are the most frequent type of repair for both NT–NT and NT–PD dyads, but their frequency decreases as dysarthria severity of the person with PD increases. Furthermore, when dysarthria severity increases, the occurrence of OISRs also increases, reflecting the increased communication breakdowns that occurs with reduced intelligibility of the participant with PD. Finally, conversational context modulated the use of repairs. These corpus data are important because they provide a clear picture of how conversational repairs are used in the interactions between NT people and people with PD, and how these differ from interactions between two NT conversational partners.

Data Availability Statement

De-identified numerical data and statistical code are available at https://osf.io/kvq7d.

Acknowledgments

This research was supported by the National Institute on Deafness and Other Communication Disorders Grant R01DC020713, awarded to Stephanie A. Borrie. We would like to acknowledge Amber Abrams who supported the coding of conversational repairs for this project. We would also like to thank all of our participants who made this study possible.

Funding Statement

This research was supported by the National Institute on Deafness and Other Communication Disorders Grant R01DC020713, awarded to Stephanie A. Borrie.

Footnotes

1

Due to logistical reasons, a small number of participants were unable to participate in all four conversations.

2

In formula syntax (Bates et al., 2015), the model was of the form: number of repairs ~ type of repair × dyad type + (1|conversation).

3

The model was of the form: number of repairs ~ type of repair × dyad severity + (1|conversation).

4

The model was of the form: repair rate ~ type of repair × conversational goal + type of repair × familiarity + type of repair × dyad severity + type of repair × familiarity × dyad severity + (1|conversation).

References

  1. Albert, S., & De Ruiter, J. P. (2018). Repair: The interface between interaction and cognition. Topics in Cognitive Science, 10(2), 279–313. 10.1111/tops.12339 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Altmann, L. J., & Troche, M. S. (2011). High‐level language production in Parkinson's disease: A review. Parkinson's Disease, 2011(1), Article 238956. 10.4061/2011/238956 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Baker, R., & Hazan, V. (2011). DiapixUK: Task materials for the elicitation of multiple spontaneous speech dialogs. Behavior Research Methods, 43(3), 761–770. 10.3758/s13428-011-0075-y [DOI] [PubMed] [Google Scholar]
  4. Bates, D., Mächler, M., Bolker, B., & Walker, S. (2015). Fitting linear mixed-effects models using lme4. Journal of Statistical Software, 67(1), 1–48. 10.18637/jss.v067.i01 [DOI] [Google Scholar]
  5. Baylor, C., & Darling-White, M. (2020). Achieving participation-focused intervention through shared decision making: Proposal of an age-and disorder-generic framework. American Journal of Speech-Language Pathology, 29(3), 1335–1360. 10.1044/2020_AJSLP-19-00043 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Baylor, C. R., Yorkston, K. M., Eadie, T. L., Miller, R. M., & Amtmann, D. (2009). Developing the Communicative Participation Item Bank: Rasch analysis results from a spasmodic dysphonia sample. Journal of Speech, Language, and Hearing Research, 52(5), 1302–1320. 10.1044/1092-4388(2009/07-0275) [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Bloch, S. (2006). Trouble sources and repair in acquired dysarthria and communication aid use: A conversation analysis study [Doctoral dissertation, University of London]. UCL Discovery. https://discovery.ucl.ac.uk/id/eprint/66888/ [Google Scholar]
  8. Bloch, S., & Barnes, S. (2020). Dysarthria and other-initiated repair in everyday conversation. Clinical Linguistics & Phonetics, 34(10–11), 977–997. 10.1080/02699206.2019.1705915 [DOI] [PubMed] [Google Scholar]
  9. Bloch, S., & Wilkinson, R. (2004). The understandability of AAC: A conversation analysis study of acquired dysarthria. Augmentative and Alternative Communication, 20(4), 272–282. 10.1080/07434610400005614 [DOI] [Google Scholar]
  10. Bloch, S., & Wilkinson, R. (2009). Acquired dysarthria in conversation: Identifying sources of understandability problems. International Journal of Language & Communication Disorders, 44(5), 769–783. 10.1080/13682820802546969 [DOI] [PubMed] [Google Scholar]
  11. Bloch, S., & Wilkinson, R. (2011). Acquired dysarthria in conversation: Methods of resolving understandability problems. International Journal of Language & Communication Disorders, 46(5), 510–523. 10.1111/j.1460-6984.2011.00076.x [DOI] [PubMed] [Google Scholar]
  12. Borrie, S. A., Wynn, C. J., Berisha, V., & Barrett, T. S. (2022). From speech acoustics to communicative participation in dysarthria: Toward a causal framework. Journal of Speech, Language, and Hearing Research, 65(2), 405–418. 10.1044/2021_JSLHR-21-00306 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Clark, H. H. (1994). Managing problems in speaking. Speech Communication, 15(3–4), 243–250. 10.1016/0167-6393(94)90075-2 [DOI] [Google Scholar]
  14. Clark, J. P., Adams, S. G., Dykstra, A. D., Moodie, S., & Jog, M. (2014). Loudness perception and speech intensity control in Parkinson's disease. Journal of Communication Disorders, 51, 1–12. 10.1016/j.jcomdis.2014.08.001 [DOI] [PubMed] [Google Scholar]
  15. Clift, R. (2016). Conversation analysis. Cambridge University Press. [Google Scholar]
  16. Collis, J., & Bloch, S. (2012). Survey of UK speech and language therapists' assessment and treatment practices for people with progressive dysarthria. International Journal of Language & Communication Disorders, 47(6), 725–737. 10.1111/j.1460-6984.2012.00183.x [DOI] [PubMed] [Google Scholar]
  17. Colman, M., & Healy, P. (2011). The distribution of repair in dialogue. Proceedings of the Annual Meeting of the Cognitive Science Society, 33, 1563–1568. [Google Scholar]
  18. Comrie, P., MacKenzie, C., & McCall, J. (2001). The influence of acquired dysarthria on conversational turn-taking. Clinical Linguistics & Phonetics, 15(5), 383–398. 10.1080/02699200110036380 [DOI] [Google Scholar]
  19. De Keyser, K., Santens, P., Bockstael, A., Botteldooren, D., Talsma, D., De Vos, S., Van Cauwenberghe, M., Verheugen, F., Corthals, P., & De Letter, M. (2016). The relationship between speech production and speech perception deficits in Parkinson's disease. Journal of Speech, Language, and Hearing Research, 59(5), 915–931. 10.1044/2016_JSLHR-S-15-0197 [DOI] [PubMed] [Google Scholar]
  20. Dideriksen, C., Christensen, M. H., Tylén, K., Dingemanse, M., & Fusaroli, R. (2023). Quantifying the interplay of conversational devices in building mutual understanding. Journal of Experimental Psychology: General, 152(3), 864–889. 10.1037/xge0001301 [DOI] [PubMed] [Google Scholar]
  21. Dingemanse, M., Roberts, S. G., Baranova, J., Blythe, J., Drew, P., Floyd, S., Gisladottir, R. S., Kendrick, K. H., Levinson, S. C., Manrique, E., Rossi, G., & Enfield, N. J. (2015). Universal principles in the repair of communication problems. PLOS ONE, 10(9), Article e0136100. 10.1371/journal.pone.0136100 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Dorsey, E., Sherer, T., Okun, M. S., & Bloem, B. R. (2018). The emerging evidence of the Parkinson pandemic. Journal of Parkinson's Disease, 8(s1), S3–S8. 10.3233/JPD-181474 [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Kendrick, K. H. (2015). The intersection of turn-taking and repair: The timing of other-initiations of repair in conversation. Frontiers in Psychology, 6, Article 250. 10.3389/fpsyg.2015.00250 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Kitzinger, C. (2012). Repair. In Sidnell J. & Stivers T. (Eds.), The handbook of conversation analysis (pp. 229–256). Wiley. 10.1002/9781118325001.ch12 [DOI] [Google Scholar]
  25. Kuzentsova, A., Brockhoff, P. B., & Christensen, R. H. B. (2017). lmerTest package: Tests in linear mixed effects models. Journal of Statistical Software, 82(13), 1–26. 10.18637/jss.v082.i13 [DOI] [Google Scholar]
  26. Landis, J. R., & Koch, G. G. (1977). The measurement of observer agreement for categorical data. Biometrics, 33(1), 159–174. 10.2307/2529310 [DOI] [PubMed] [Google Scholar]
  27. Lenth, R. V. (2024). emmeans: Estimated marginal means, aka least-squares means (Version 1.9.0) [R package]. CRAN. https://CRAN.R-project.org/package=emmeans [Google Scholar]
  28. Levelt, W. J. (1983). Monitoring and self-repair in speech. Cognition, 14(1), 41–104. 10.1016/0010-0277(83)90026-4 [DOI] [PubMed] [Google Scholar]
  29. Liss, J. M., Spitzer, S., Caviness, J. N., Adler, C., & Edwards, B. (1998). Syllabic strength and lexical boundary decisions in the perception of hypokinetic dysarthric speech. The Journal of the Acoustical Society of America, 104(4), 2457–2466. 10.1121/1.423753 [DOI] [PubMed] [Google Scholar]
  30. McAuliffe, M. J., Baylor, C. R., & Yorkston, K. M. (2017). Variables associated with communicative participation in Parkinson's disease and its relationship to measures of health-related quality-of-life. International Journal of Speech-Language Pathology, 19(4), 407–417. 10.1080/17549507.2016.1193900 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. McCabe, R. (2007). Conversation analysis. In Slade M. & Priebe S. (Eds.), Choosing methods in mental health research: Mental health research from theory to practice (pp. 40–62). Routledge. [Google Scholar]
  32. McKinlay, A., Dalrymple-Alford, J., Grace, R., & Roger, D. (2009). The effect of attentional set-shifting, working memory, and processing speed on pragmatic language functioning in Parkinson's disease. European Journal of Cognitive Psychology, 21(2–3), 330–346. 10.1080/09541440802281266 [DOI] [Google Scholar]
  33. Mertens, J. B., & de Ruiter, J. P. (2021). Cognitive and social delays in the initiation of conversational repair. Dialogue & Discourse, 12(1), 21–44. 10.5210/dad.2021.102 [DOI] [Google Scholar]
  34. Miller, N., Deane, K. H., Jones, D., Noble, E., & Gibb, C. (2011). National survey of speech and language therapy provision for people with Parkinson's disease in the United Kingdom: Therapists' practices. International Journal of Language & Communication Disorders, 46(2), 189–201. 10.3109/13682822.2010.484849 [DOI] [PubMed] [Google Scholar]
  35. Miller, N., Noble, E., Jones, D., & Burn, D. (2006). Life with communication changes in Parkinson's disease. Age and Ageing, 35(3), 235–239. 10.1093/ageing/afj053 [DOI] [PubMed] [Google Scholar]
  36. Moya-Galé, G., & Levy, E. S. (2019). Parkinson's disease-associated dysarthria: Prevalence, impact and management strategies. Research and Reviews in Parkinsonism, 9, 9–16. 10.2147/JPRLS.S168090 [DOI] [Google Scholar]
  37. Planalp, S., & Benson, A. (1992). Friends' and acquaintances' conversations I: Perceived differences. Journal of Social and Personal Relationships, 9(4), 483–506. 10.1177/0265407592094002 [DOI] [Google Scholar]
  38. R Core Team. (2020). R: A language and environment for statistical computing. R Foundation for Statistical Computing. https://www.R-project.org/ [Google Scholar]
  39. Sacks, H., Schegloff, E. A., & Jefferson, G. (1974). A simplest systematics for the organization of turn taking for conversation. Language, 50(4), 696–735. 10.2307/412243 [DOI] [Google Scholar]
  40. Sapir, S. (2014). Multiple factors are involved in the dysarthria associated with Parkinson's disease: A review with implications for clinical practice and research. Journal of Speech, Language, and Hearing Research, 57(4), 1330–1343. 10.1044/2014_JSLHR-S-13-0039 [DOI] [PubMed] [Google Scholar]
  41. Schegloff, E. A. (2007). Sequence organization in interaction: A primer in conversation analysis (Vol. 1). Cambridge University Press. 10.1017/CBO9780511791208 [DOI] [Google Scholar]
  42. Schegloff, E. A., Jefferson, G., & Sacks, H. (1977). The preference for self-correction in the organization of repair in conversation. Language, 53(2), 361–382. 10.2307/413107 [DOI] [Google Scholar]
  43. Scollon, R., Scollon, S. W., & Jones, R. H. (2012). Intercultural communication: A discourse approach (3rd ed.). Wiley. [Google Scholar]
  44. Sedikides, C., Campbell, W. K., Reader, G. D., & Elliot, A. J. (1999). The relationship closeness induction task. Representative Research in Social Psychology, 23, 1–4. [Google Scholar]
  45. Sixt Börjesson, M., Hartelius, L., & Laakso, K. (2021). Communicative participation in people with amyotrophic lateral sclerosis. Folia Phoniatrica et Logopaedica, 73(2), 101–108. 10.1159/000505022 [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Speer, S. P., Mwilambwe-Tshilobo, L., Tsoi, L., Burns, S. M., Falk, E. B., & Tamir, D. I. (2024). Hyperscanning shows friends explore and strangers converge in conversation. Nature Communications, 15(1), Article 7781. 10.1038/s41467-024-51990-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Spencer, K., Friedlander, C., & Brown, K. (2020). Predictors of health-related quality of life and communicative participation in individuals with dysarthria from Parkinson's disease. International Journal of Neurodegenerative Disorders, 3(1), 14. 10.23937/2643-4539/1710014 [DOI] [Google Scholar]
  48. Templeton, E. M., Chang, L. J., Reynolds, E. A., Cone LeBeaumont, M. D., & Wheatley, T. (2023). Long gaps between turns are awkward for strangers but not for friends. Philosophical Transactions of the Royal Society B: Biological Sciences, 378(1875), Article 20210471. 10.1098/rstb.2021.0471 [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Torrence, J. M., Baylor, C. R., Yorkston, K. M., & Spencer, K. A. (2016). Addressing communicative participation in treatment planning for adults: A survey of U.S. speech-language pathologists. American Journal of Speech-Language Pathology, 25(3), 355–370. 10.1044/2015_AJSLP-15-0049 [DOI] [PubMed] [Google Scholar]
  50. Van Engen, K. J., Baese-Berk, M., Baker, R. E., Choi, A., Kim, M., & Bradlow, A. R. (2010). The Wildcat Corpus of native-and foreign-accented English: Communicative efficiency across conversational dyads with varying language alignment profiles. Language and Speech, 53(4), 510–540. 10.1177/0023830910372495 [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Weintraub, S., Dikmen, S. S., Heaton, R. K., Tulsky, D. S., Zelazo, P. D., Bauer, P. J., Carlozzi, N. E., Slotkin, J., Blitz, D., Wallner-Allen, K., Fox, N. A., Beaumont, J. L., Mungas, D., Nowinski, C. J., Richler, J., Deocampo, J. A., Anderson, J. E., Manly, J. J., Borosh, B., … Gershon, R. C. (2013). Cognition assessment using the NIH Toolbox. Neurology, 80(11, Suppl. 3), S54–S64. 10.1212/WNL.0b013e3182872ded [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Yang, W., Hamilton, J. L., Kopil, C., Beck, J. C., Tanner, C. M., Albin, R. L., Ray Dorsey, E., Dahodwala, N., Cintina, I., & Hogan, P. (2020). Current and projected future economic burden of Parkinson's disease in the U.S. NPJ Parkinson's Disease, 6(1), Article 15. 10.1038/s41531-020-0117-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Yorkston, K. M., Baylor, C. R., Klasner, E. R., Deitz, J., Dudgeon, B. J., Eadie, T., Miller, R. M., & Amtmann, D. (2007). Satisfaction with communicative participation as defined by adults with multiple sclerosis: A qualitative study. Journal of Communication Disorders, 40(6), 433–451. 10.1016/j.jcomdis.2006.10.006 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

De-identified numerical data and statistical code are available at https://osf.io/kvq7d.

Articles from Journal of Speech, Language, and Hearing Research : JSLHR are provided here courtesy of American Speech-Language-Hearing Association

RESOURCES