Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2025 Sep 26.
Published before final editing as: Neuropsychology. 2025 Sep 22:10.1037/neu0001040. doi: 10.1037/neu0001040

Phonological networks remain intact in multiple sclerosis

Allison Link 1,*, Amy L Lebkuecher 2,3,*, Abigail L Cosgrove 1,4, Nichol Castro 5, Nancy D Chiaravalloti 6,7,8, Lauren B Strober 6,7, Michele T Diaz 1
PMCID: PMC12462895  NIHMSID: NIHMS2104436  PMID: 40991791

Abstract

Objective:

Multiple Sclerosis is a neurological condition characterized by white and gray matter decline that leads to slower motor function and cognitive impairment. Although language remains relatively intact, individuals with MS often have word retrieval difficulties. Previous research suggests that these difficulties may be related to vocabulary, the number of words an individual knows, and other semantic aspects of language. However, few studies have examined phonological aspects of speech.

Method:

We examined speech in 89 individuals with MS and 88 age-matched neurotypical adults using a phonemic verbal fluency task. We took a network science approach, building a phonological network from participants’ responses and their close phonological neighbors. We then examined the local network characteristics (degree, clustering coefficient) of participants’ responses to assess whether responses differed between the groups.

Results:

Although individuals with MS produced fewer responses during the task, the network characteristics of their responses were similar to neurotypical adults, the control group. Moreover, lexical characteristics such as word length and lexical frequency were also similar between groups (Model R2 values < 1%). Finally, a forward flow analysis, which quantifies the phonological similarity between adjacent responses and provides a metric of how people search phonemic space, did not differ between groups.

Conclusions:

Overall, these results suggest that phonological aspects of speech remain stable in individuals with MS. Word retrieval difficulties in MS may arise from neurological changes in semantic processes, in combination with other cognitive abilities such as speed of processing and executive function, which are common in MS.

Introduction

Individuals with multiple sclerosis (MS) often experience cognitive decline, most notably in speed of processing, learning and memory, and executive function (e.g., Chiaravalloti & DeLuca, 2008; Rao et al., 1991). Although language impairments are not considered to be a hallmark of the disease, individuals with MS often experience frustrating word finding difficulties (Brandstadter et al., 2020; El-Wahsh et al., 2020; Grezmak et al., 2023), lexical, syntactic, and morphosyntactic deficits (Grigoriadis et al., 2024; Martzoukou et al., 2025; Rahimifar et al., 2025; Šubert et al., 2023) and impairments recognizing speech in noise (Iva et al., 2020). However, the locus of these speech difficulties has been debated. Several recent studies have suggested that impairments in semantic processes may underlie these word retrieval difficulties (Lebkuecher et al., 2021, 2024), but phonological aspects of retrieval in MS have not been explored to the same extent (though see Dvorak et al., 2024) and theoretical models of language suggest that impairments in phonological processes may contribute to retrieval failures (Burke et al., 1991, 2000). Here we used network science to examine word retrieval during a phonemic verbal fluency task among individuals with MS and age-matched neurotypical adults. We hypothesized that if word retrieval difficulties are primarily based on semantic decline and impaired domain general abilities, phonological networks and phonological aspects of word retrieval may not be impaired.

Multiple Sclerosis and Language

MS is an auto-immune disorder that causes inflammation of the central nervous system (CNS), resulting in damage to the myelin sheath that insulates nerve fibers (Amato et al., 2001, 2008, 2010; Chiaravalloti & DeLuca, 2008; Rao et al., 1991) as well as both cortical and deep gray matter (Eijlers et al., 2019; Horakova et al., 2009; Rahnemayan et al., 2025; Tiberio et al., 2005). These neural changes typically lead to significant motor and cognitive impairments. Though MS can cause significant impairments in cognition, moderate to severe aphasia comparable to the impairment observed in many stroke survivors is rare even in those with severe symptoms (Lacour et al., 2004). However, subtler language deficits, such as lexical retrieval failures are frequently reported by individuals with MS (Brandstadter et al., 2020; Dvorak et al., 2024; El-Wahsh et al., 2020; Šubert et al., 2023). For example, individuals with MS have been shown to have deficits on common picture naming tasks, such as the Boston Naming Test (Galioto et al., 2021; Grezmak et al., 2023; Sepulcre et al., 2011) and have shown lexical, syntactic, and morphosyntactic declines during both neuropsychological assessment and in natural speech (Grigoriadis et al., 2024; Martzoukou et al., 2025; Rahimifar et al., 2025; Šubert et al., 2023). They have also shown declines recognizing speech in noise, even at early stages of the disease (Iva et al., 2020), and phonemic processing proficiency has been linked to self-reported word retrieval difficulties (Dvorak et al., 2024). These findings are consistent with the idea that language may be impaired among individuals with MS.

Similar language production impairments in individuals with MS have also been noted on tests of verbal fluency (Beatty et al., 1989; Galioto et al., 2021; Henry & Beatty, 2006; Rao et al., 1991; Sepulcre et al., 2011). Verbal fluency is frequently employed to assess language and executive abilities in a variety of clinical populations because of its brevity and simplicity in scoring (Benton et al., 1983; Ruff et al., 1996). During phonemic fluency participants produce words that begin with the same letter (typically F, A, and S), which requires them to consider the phonemic features of words and engage in phonological and articulatory processes. Semantic fluency requires participants to search within a semantic category (e.g., “animals” and “fruits & vegetables”) which requires foraging in semantic memory for appropriate words and linking semantic information to corresponding phonological representations (Shao et al., 2014). Given these task constraints, it has been proposed that both fluency tasks also rely on domain-general abilities including speed of processing and executive function (e.g., to inhibit responses that do not fit the category or ones that have already been produced, Amunts et al., 2020; Friesen et al., 2015; Pitteri et al., 2023; Sepulcre et al., 2011). However, semantic verbal fluency additionally relies on semantic knowledge and may incur additional semantic competition effects that are often found in other tasks that involve naming within semantic categories (e.g., picture-word interference, blocked cyclical naming, Belke et al., 2005; Howard et al., 2006; Kroll & Stewart, 1994; Schnur et al., 2006).

Clinical populations with language impairments, such as aphasia, and individuals with dementia typically exhibit decreased performance in both semantic and phonemic verbal fluency (Arroyo-Anlló et al., 2012; Monsch et al., 1992; Rinehardt et al., 2014). However, prior studies with clinical populations suggest that changes in phonemic fluency are smaller compared to differences in semantic fluency. Likewise, studies on healthy aging suggest that there are smaller age-related differences in phonemic fluency compared to semantic fluency across adulthood (Diaz et al., 2021; Kavé, 2005; Kavé & Knafo-Noam, 2015; Tombaugh et al., 1999; Troyer, 2000). These findings suggest that although phonemic fluency may be affected by clinical conditions and typical aging, the changes are generally smaller compared to semantic fluency.

So, what leads to the different patterns of results that have been observed between phonemic and semantic fluency? Some have suggested that the sharper decline of semantic fluency in both clinical populations and typical aging may be related to the cognitive demands of the task, rather than declines in semantic knowledge (e.g., Baciu et al., 2016; Gordon et al., 2018), as age-related stability in vocabulary is well-established (e.g., Kavé & Halamish, 2015; Verhaeghen, 2003). Measures of vocabulary provide an estimate of the number of words an individual knows, and performance on this measure is likely related to the depth and breadth of semantic knowledge. Semantic fluency involves searching within a specific sub-category of semantic memory, which could entail increased semantic competition among potential responses (Chen & Mirman, 2012; Mirman, 2011; Mirman & Magnuson, 2008; Shao et al., 2014). In contrast, phonemic fluency involves a less constrained search of a broader pool of lexical representations that are less likely to engage semantic competition due to lower semantic overlap.

Though decreased verbal fluency performance is robust and one of the earliest detectible features in individuals with progressive MS (Henry & Beatty, 2006), the nature of these word finding difficulties has not been thoroughly addressed since they are often attributed to other cognitive deficits, such as slower speed of processing, which are hallmarks of the disease (Beatty et al., 1989; Jakimovski et al., 2019; Wachowius et al., 2007). In at least one study, changes in confrontation naming were noted in individuals with MS both with and without information processing speed deficits (Galioto et al., 2021), suggesting that naming and lexical access challenges may be independent of declines in speed.

Consistent with cognitive impairments in language specifically, Lebkuecher and colleagues (2021) found that vocabulary was the only significant predictor of semantic verbal fluency performance in individuals with MS, even when considering both speed of processing and working memory performance. For phonemic fluency, both vocabulary and information processing speed were found to be significant predictors of performance. Collectively, these findings suggest that linguistic abilities in addition to domain general abilities such as information processing speed may change during the course of MS and influence the extent of word finding difficulties in this population.

Network Science

While picture naming, verbal fluency, and vocabulary tasks provide ways of measuring language performance, network science approaches have become increasingly useful in cognitive and clinical sciences to examine the organization of language systems (e.g., Cosgrove et al., 2021; Lebkuecher et al., 2024; Steyvers & Tenenbaum, 2005; Vitevitch, 2008, 2022; Wulff et al., 2019). Networks can be constructed from a variety of tasks (e.g., auditory lexical decisions, Castro & Vitevitch, 2022; semantic verbal fluency, Cosgrove et al., 2021) and data (dictionary entries, Vitevitch, 2008). In language networks, ‘nodes’ represent words, while ‘edges’ or connections between nodes reflect semantic, phonological, or another type of linguistic similarity between words.

After a network has been formed, network measures can be calculated to quantify various aspects of the network’s structure (Barabási, 2016; Sporns, 2011). Network characteristics can describe individual nodes (microscopic), such as the number of connections to other nodes (i.e., degree) or how often neighbors of a node are connected to one another (i.e., clustering coefficient). Network characteristics can also describe communities within a network (mesoscopic), such as how a network breaks into sub-communities (i.e., modularity). Finally, we can use network science to characterize the network as a whole, looking at the global scale (macroscopic). Often node-level or microscopic network characteristics can be calculated both for individual nodes or averaged across all nodes to provide an average measure of that network characteristic for the entire network. For example, clustering coefficient can be calculated both for a single node (e.g., the clustering coefficient of the word ‘dog’) or averaged across all the nodes of the network to provide a more global index of interconnectedness.

In addition to network science furthering our understanding of semantic and phonological aspects of language in general, there has been a growing body of literature applying network science to special populations, for example examining how aging affects network structure (Cosgrove et al., 2021, 2023; Dubossarsky et al., 2017; Wulff et al., 2022) and how network structure is affected by clinical disorders such as aphasia (e.g., Castro, 2021; Vitevitch et al., 2023). In line with this latter aim, Lebkuecher and colleagues (2024) analyzed semantic verbal fluency responses of individuals with MS and aged-matched neurotypical adults, where networks represented words produced during the task and network edges were defined by the co-occurrence of responses. Consistent with previous studies on semantic verbal fluency, individuals with MS produced significantly fewer responses compared with neurotypical adults. In addition, individuals with MS had semantic networks that were less efficient, less interconnected, and more segmented than neurotypical peers (i.e., the networks were characterized by longer path lengths, less clustering, and higher modularity values). Moreover, simulations of spreading activation and a percolation analysis suggested that activation degraded faster among individuals with MS, and that their semantic networks were less flexible. These findings suggest that individuals with MS had weaker and less robust semantic networks which may contribute to their language retrieval difficulties.

While differences in semantic processes among individuals with MS have been documented through vocabulary, semantic verbal fluency, picture naming, and semantic network studies, other aspects of language have been less documented among individuals with MS. Indeed, word retrieval, while guided by semantic aspects of language, also involves phonology. Dvorak and colleagues (2024) documented phonological processing deficits in MS and suggested the deficits may stem from early cortical degeneration in the posterior superior temporal/supramarginal region. However, the authors do not provide a cognitive, theoretical account for the origin of these phonological processing deficits.

The Transmission Deficit Hypothesis (Burke et al., 1991, 2000) suggests that word retrieval difficulties, particularly among older adults, occur when semantic representations fail to activate corresponding phonological representations (e.g., tip-of-tongue states). This hypothesis may also be relevant for examining phonological processes in MS given the documented differences in semantic networks. Phonemic verbal fluency which is guided by phonological and articulatory processes while also imposing fewer semantic selection demands (Diaz et al., 2021; Kavé & Knafo-Noam, 2015), may be one way to investigate these processes.

The Present Study

The goal of the current study was to analyze phonemic verbal fluency responses from individuals with MS and age-matched neurotypical individuals using a phonological network approach. Phonological networks are generally constructed by placing edges between words that differ by the addition, subtraction, or deletion of a single phoneme. Moreover, phonological relationships between words can be further quantified by calculating the differences in phonemic features. In the present study three phonological networks were constructed based on the responses to phonemic verbal fluency tasks. Individual networks were constructed for each letter of the phonemic verbal fluency task (letters F, A, and S), incorporating all of the unique responses produced by all participants and their phonological neighbors (i.e., words that differ by the addition, deletion, or substitution of 1 phoneme). We chose to create the phonological network in this way to balance focusing on the responses that participants provided while also including words that are likely to be phonologically related to participant responses. Once a network was created, we examined the node-level or micro-scopic characteristics of individual responses within the network.1We focused on two network characteristics: degree, the number of nodes connected to a given node, which provides an index of how connected a given node is, that is, it reflects the node’s centrality or importance in the overall network, with higher degree values reflecting more integral nodes. The second network characteristic we examined was clustering coefficient, the proportion of a node’s neighbors that are also connected to one another, which reflects the node’s interconnectedness. We also examined the lexical frequency and number of phonemes of the individual responses to assess whether responses from individuals with MS were similar to those from neurotypical adults. Although previous work pointed to semantic network differences among individuals with MS compared to neurotypical adults, these have been attributed to differences in vocabulary. In contrast, phonological deficits have not been strongly noted among individuals with MS, despite the observed differences in phonemic fluency (though see Dvorak et al., 2024). If prior network differences were due primarily to semantics, then phonological networks may be similar among individuals with MS and neurotypical adults. These findings would support accounts that argue phonemic fluency deficits in MS can be partially attributed to observed impairments in domain general abilities such as information processing speed. Conversely, if phonological networks differ among individuals with MS and controls, then this may support recent findings of phonemic processing deficits in MS and the predictions of the Transmission Deficit Hypothesis, (Burke et al., 1991, 2000; Dvorak et al., 2024).

Methods

Participants

MS Sample

The data from individuals with MS were collected by researchers at the Kessler Foundation as part of a larger behavioral intervention project to improve learning and memory in individuals with MS (Chiaravalloti et al., 2013; see also Lebkuecher et al., 2021). Participants with MS (N = 89, 69 females; Mage= 49.4, SDage= 9.3) had a clinical diagnosis of MS from a neurologist and had symptoms for a minimum of 12 months (M = 224.5 months, SD = 139.0 months). We included everyone who had phonemic verbal fluency scores (89 out of 94 individuals). Participants were excluded if they reported history of a neurological condition other than MS, brain injury, learning disability, substance abuse disorder, severe psychiatric illness, or use of corticosteroids, anticonvulsants, benzodiazepines, and/or opiates. Anyone who experienced a significant worsening of symptoms within one month of enrolling in the original study was also excluded. Participants were compensated $10 per hour of participation. See Table 1 for additional demographic details of participants with MS.

Table 1.

Descriptive statistics and percentage of participants with MS with cognitive impairment (CI) on measures of vocabulary and cognitive control.

Assessment Mean SD Min Max % CI
WASI Vocabulary 59.70 11.20 25.00 80.00 11.24
SDMT 45.10 13.00 11.00 79.00 39.33
D-KEFS Trails 122.00 64.20 23.00 271.00 34.83
D-KEFS Color-Word Interference 73.60 32.20 25.00 180.00 17.98
Open in a new tab

Note. CI = cognitive impairment.; WASI = Wechsler Abbreviated Scale of Intelligence; SDMT = Symbol Digit Modalities Test; D-KEFS = Delis-Kaplan Executive Function System.

Neurotypical Individuals

The neurotypical adults (N = 88; females = 61; Mage = 52.0, SDage = 11.4) were initially recruited as part a larger research project examining language and aging at The Pennsylvania State University (Diaz et al., 2021, 2022). Neurotypical participants reported no history of neurological disorders or significant health conditions (e.g., diabetes, heart disease, cancer, etc.) and performance on a range of cognitive assessments suggested that they were neurotypical, healthy older adults. Independent samples t-tests and chi-squared tests were used to assess potential group differences between participant groups. Participants included in the present analyses did not differ significantly from the MS sample in age, t(175) = 1.38, p = .17 or gender χ2(1) = 1.51, p = .22. Additionally, neurotypical adults had significantly more years of education on average (t(173.39) = 3.67, p < .001, M = 17.15, SD = 2.65) relative to individuals with MS (M = 15.74, SD = 2.43), and higher vocabulary scores (t(133.72) = 7.44, p < .001, M = 13.42, SD = 1.82) than individuals with MS (M = 10.39, SD = 3.37). Because different vocabulary assessments were used for the two groups, we ran an independent samples t-test on the scaled scores instead of the raw scores. Neurotypical adults were compensated $10 per hour for their participation. All participants provided written informed consent and all procedures were approved by an institutional review board. See Table 2 for additional demographic details of the neurotypical adults.

Table 2.

Descriptive statistics for healthy older participants on measures of cognition

Assessment Mean SD Min Max
MMSE 28.88 1.03 27 30
WAIS-III Vocabulary 55.28 5.42 32 69
Digit Symbol (RT) 1618.7 326.19 983.96 3000.56
Digit Symbol (ACC) 97.95% 1.92% 91.67% 100%
Verbal Working Memory (PU) 0.74 0.15 0.30 0.99
Stroop Effect (ms) 64.60 79.54 −54.93 419.95
Open in a new tab

MMSE = Mini Mental State Exam; WAIS = Wechsler Adult Intelligence Scale; RT = reaction time; ACC = accuracy; PU = partial-credit unit scoring; ms = milliseconds

Procedure

All participants completed a phonemic verbal fluency task (Benton et al., 1983; Ruff et al., 1996) in person and data were manually transcribed or collected using an audio recording device. Participants were prompted with a letter of the alphabet and be asked to list as many words as they could starting with that letter in 60 seconds. They were instructed to avoid proper names for people or places (e.g., “Felix”, “Florida”), words that share the same root word with different suffixes (e.g., “fake”, “faking”), and sequential numbers (e.g., “fifty-one”, “fifty-two”, etc.). For scoring the verbal fluency task, such items were considered incorrect. In addition, repetitions, non-words, or responses outside of the letter category (i.e., cinnamon for ‘S’) were also considered to be incorrect. Participants completed the task a total of three times with the letters F, A, and S.

Network Analyses

Participants’ verbal fluency responses were examined using a phonological network approach (Vitevitch, 2008). A separate network was constructed for each letter of the phonemic verbal fluency task. Nodes included all responses produced by participants and their phonological neighbors that differed by a single phoneme addition (e.g., “fake” - “flake”), substitution (e.g., “fake” - “bake”), or deletion (e.g., “fake” - “ache”). Participants’ responses from the phonemic verbal fluency task were preprocessed and cleaned using Python v3.12.2 (Rossum & Drake, 2009) to remove incorrect responses such as compound words (e.g., “a cappella”, “alley cat”) and onomatopes (e.g., “ah”, “achoo”). Loan words from other languages (e.g., “saké”) and different variations of the same root word (e.g., fail/failing) that were produced by different participants were included in the network. Transcriptions of participant responses were passed through the Irving Phonotactic Online Dictionary (IPhOD; Vaden et al., 2009) calculator tool to retrieve a list of their phonological neighbors. All potential neighbors were included in the network, regardless of their initial phoneme, since they may be activated during language production due to their phonological similarity. Undirected-unweighted edges were placed between neighbors (see Table 3 for descriptive network characteristics). Finally, phonological networks were estimated in Python for each letter (F-A-S, see Figure 1).

Table 3:

Characteristics of the phonological networks by letter.

Letter Number of
Responses		 Number of
Nodes		 Number of
Edges		 Number of
Isolates
F 513 3352 5012 87
A 636 2628 3100 175
S 920 5132 9768 127
Open in a new tab

Figure 1:

Figure 1:

Open in a new tab

Network visualizations for each letter of the phonemic verbal fluency task (F, A, S; Healthy controls, N = 88; Individuals with MS, N = 89).

The local node properties degree (i.e., the number of edges connected to node) and Clustering Coefficient (local CC, i.e., the extent to which neighbors of a node are also neighbors of each other) were calculated for the network analyses. We also examined the lexical characteristics of participant responses (number of phonemes and word frequency). The CMU Pronouncing Dictionary was used to generate phonological transcriptions and calculate the number of phonemes for each response. Word frequency was also calculated using data from the Exquisite Corpus which includes text from Wikipedia, subtitles, the news, books, Twitter Reddit, and other web text. Prior to conducting statistical analyses, the data were visually inspected, and we confirmed that variances did not differ between the groups. Additionally, we quantitatively examined the distributions using the describe function from the Psych package in R (Revelle, 2025). This revealed that two variables, clustering coefficient and word frequency were highly skewed (skew >3). A log10 transformation reduced the skews to < 1 and we therefore performed the statistical analyses on the log-transformed values for these variables. Group differences in local node properties and lexical characteristics were assessed using simple linear regressions in which we examined the relationship between participant group and each predictor. Because the groups differed in years of education, we included education as a covariate in the regressions.

We also conducted a phonemic forward flow analysis to compare the phonemic variability of responses across the groups. Forward flow is an analysis method that compares the similarity between subsequent responses (Gray et al., 2019). This has primarily been used to assess semantic similarity between responses and higher semantic diversity has frequently been linked to increased creativity. Here we adapted this approach to assess phonemic similarity between items. Phonemic forward flow was determined by calculating the phonemic edit distance between each response. For example, the average phonemic edit distance between subsequent responses (e.g., an addition, deletion, or substitution of a single phoneme would be an edit distance of 1). Then these phonemic similarities were averaged separately for each letter (“F”, “A”, “S”) for each participant. We then compared the phonemic forward flow values for individuals with MS and neurotypical adults using simple linear regression.

Transparency and Openness

We report all manipulations and measures in the study. All data, analysis code, and research materials are available through the Open Science Framework: https://osf.io/e5x94. This study’s design and its analysis were not pre-registered.

Results

Participants phonemic verbal fluency responses for each letter were compared between groups, individuals with MS produced significantly fewer items for each letter and fewer unique responses overall (see Table 4). However, the lower performance of individuals with MS may not be considered a clinical deficit (i.e., it was not more than 1.5 SD below the control group mean). Simple linear regressions were conducted in R to examine the influence of Group (individuals with MS, neurotypical adults) on the degree, local CC, number of phonemes, and word frequency of each participants response. Because there were significant group differences in years of education, we included education as a covariate. In most of the analyses, Group was not a significant predictor and Education was not a significant covariate. However, there were significant group differences in word frequency for F and A, in which individuals with MS produced responses that were less frequent than the control group. Additionally, Education was also a significant covariate in the A word analyses (β = −.025, t(2090) = −2.482, p < .05). See Table 5 for group means and standard deviations and Table 6 for regression results).

Table 4:

Phonemic Verbal Fluency Responses

Letter Group Total Unique
Responses		 M SD t p
F MS 367 11.98 4.6 −3.92 <.001
HC 390 14.83 5.0
A MS 393 10.66 4.65 −3.91 <.001
HC 463 13.28 4.2
S MS 576 12.98 5.27 −4.21 <.001
HC 664 16.16 4.74
Open in a new tab

Note. MS = multiple sclerosis; HC = healthy control.

Table 5:

Means and SDs of local network characteristics and lexical properties for F-A-S by group.

Property F A S
Node degree
MS 17.91 (18.06) 9.82 (14.97) 19.01 (21.32)
HC 18.63 (18.23) 10.87 (16.27) 18.11 (19.56)
Local CC
MS .08 (.12) .06 (.13) .10 (.12)
HC .08 (.12) .06 (.13) .09 (.11)
Phoneme length
MS 4.65 (1.68) 5.03 (1.99) 4.66 (1.74)
HC 4.66 (1.72) 5.07 (2.18) 4.70 (1.85)
Word frequency
MS 2.05e-04 (.001) 5.18e-04 (.003) 1.12e-05 (.00027)
HC 2.27e-04 (.001) 1.08e-03 (.004) 9.27e-05 (.00032)
Open in a new tab

Note. MS = multiple sclerosis; HC = healthy control.

Table 6:

Results from simple linear regressions for the effect of group on local network characteristics, lexical properties, and forward flow for F-A-S responses

Property β SE t-value p-value R2
F Responses
   Degree −0.719 0.753 −0.955 .340 .0004
   CC 0.017 0.017 0.991 .322 .0006
   # Phonemes −0.010 0.070 −0.150 .880 .000006
   Frequency −0.086 0.041 −2.084 .0373* .002
   FF 0.030 0.125 0.240 .811 .0003
A Responses
   Degree −1.047 0.687 −1.523 .128 .001
   CC 0.006 0.030 0.200 .841 .00004
   # Phonemes −0.032 0.091 −0.349 .727 .00005
   Frequency −0.122 0.052 −2.332 .019* .004
   FF −0.037 0.135 −0.274 .785 .0004
S Responses
   Degree 0.897 0.809 1.109 .268 .0004
   CC 0.019 0.015 1.228 .220 .0008
   # Phonemes −0.042 0.071 −0.601 .548 .0001
   Frequency −0.032 0.035 −0.895 .371 .0003
   FF −0.008 0.126 −0.070 .944 .00005
Open in a new tab

CC = clustering coefficient; Frequency = word frequency; FF = forward flow

We also calculated the phonemic forward flow values for each letter for each participant to assess the phonemic similarity in how the two groups searched phonemic space. Individuals with MS and neurotypical adults produced responses phonemic forward flow values that were not statistically different across all cues: (F: F(1, 173) = 0.058, p =.81, R2 < .001; A: F(1, 173) = 0.075, p =.78, R2 < .001; S: F(1, 173) = 0.004, p =.94, R2 < .001). See Figure 2 for visual display of the data.

Figure 2: Phonemic Forward Flow Distributions.

Figure 2:

Open in a new tab

Phonemic forward flow (FF) values for individuals with MS (N = 89) and neurotypical adults (N = 88), which reflects the average phonemic edit distance between subsequent responses. There were no statistical differences in phonemic FF between the groups.

Discussion

Individuals with MS often report word finding difficulties and previous research has linked these difficulties, particularly in semantic verbal fluency, to differences in vocabulary and semantic memory. But phonological aspects of speech have not been extensively examined. Here we focus on phonological aspects of spoken language using a phonological network approach based on a phonemic verbal fluency task in individuals with MS compared to aged-matched neurotypical adults. Individuals with MS produced significantly fewer items during the phonemic verbal fluency task than neurotypical adults. However, their responses did not differ on phonological network measures of node degree (i.e., the number of connections to a node) or local clustering coefficient (i.e., the extent to which neighbors of words are also neighbors of each other). Moreover, the lexical characteristics of responses did not differ between the groups in terms of number of phonemes. Word frequency differed between the groups for the F and A categories, with individuals with MS producing lower frequency words, which if anything supports the idea that phonological features did not interfere with word retrieval. Finally, we used a phonemic forward flow analysis to assess the phonemic similarity of adjacent responses and found no differences between the groups. Collectively, these findings suggest that phonological aspects of speech remain intact in individuals with MS.

Phonemic Verbal Fluency

While our findings suggest that individuals with MS produce items that are similar to neurotypical adults, the effect of MS diagnosis on phonemic verbal fluency performance is robustly reported in the clinical literature (Henry & Beatty, 2006; Pitteri et al., 2023). Our results are consistent with these previous observations. For all three phonemic prompts, individuals with MS produced fewer items. However, it should be noted that their lower performance may not have fallen to the level to be considered a clinical deficit, that is, their performance was on average ~0.6 SD lower than the control group not > 1 SD, see Table 4 for full details. Verbal fluency is a timed task, with performance most commonly assessed as the number of correct responses produced within one minute. One of the cognitive and neurological hallmarks of MS, is slower speed of information processing (Beatty et al., 1989; Galioto et al., 2021; Jakimovski et al., 2019; Wachowius et al., 2007). Thus, it is unsurprising that individuals with MS produce fewer items, especially given the time constraints of the task. Moreover, individuals with MS often experience increased difficulty with articulation, this oral-motor slowing explains a small, but significant amount of variance in both phonemic and semantic verbal fluency (Arnett et al., 2008) as well as other tasks that require speeded oral responses (Smith & Arnett, 2007). In addition to motor slowing, individuals with MS also often experience declines in executive function (e.g., Chiaravalloti & DeLuca, 2008; Pitteri et al., 2023; Rao et al., 1991) which could further contribute to word selection difficulty, a higher number of repeated words, or more intrusions, words which do not conform to the task instructions (i.e., proper names, words that start with a different letter). Repetitions and intrusions are typically considered errors and not counted and thus would also lower the total fluency score. Given the multiple ways in which speed and executive function may negatively contribute to phonemic verbal fluency performance, it is notable that the responses provided by individuals with MS largely did not statistically differ from unaffected adults in terms of lexical characteristics which suggests that phonemic aspects of retrieval were not impaired among individuals with MS.

In addition to the lexical characteristics of the items, our primary focus was on phonemic network characteristics: degree and clustering coefficient. In the present analyses, degree refers to the number of items that are phonemically related to a given response, while clustering coefficient reflects the number of response neighbors that are also related to one another (i.e., interconnectivity). One of the strengths of the network science approach that we adopted here, is that it allowed us to quantitatively examine the types of words individuals with MS produce irrespective of the number of items that they produced. The phonemic network was created out of all participant responses, and all closely-related phonological neighbors (i.e., differing by one phoneme). This created a model of a phonological network within which we could examine the characteristics of specific items. Individuals with MS and matched neurotypical adults produced items with similar degree values and clustering coefficients, suggesting that both groups recall phonologically similar items, and that phonological aspects of language remain unimpaired in MS.

Phonological and Semantic Aspects of Speech

In contrast to the present results on phonological verbal fluency, previous investigations of semantic verbal fluency found significant differences between individuals with MS and neurotypical adults. Using a similar network science approach with semantic verbal fluency, Lebkuecher and colleagues found that the semantic networks of individuals with MS were less efficient, less interconnected, and contained more sub-communities (2024). Moreover, spreading activation simulations suggested that activation degrades faster for individuals with MS compared to age-matched neurotypical adults. In addition to differences in semantic network structure, decline in semantic verbal fluency performance among individuals with MS has also been linked to vocabulary. In a separate study, Lebkuecher and colleagues found that vocabulary was a significant predictor of semantic verbal fluency performance in individuals with MS (2021). Collectively, these studies suggest that semantic aspects of language contribute to language production difficulties in individuals with MS greater than phonological aspects.

So why might these differences in phonological and semantic networks between individuals with MS and healthy older adults arise? Both phonemic and semantic verbal fluency tasks rely on information processing speed to recall words, as well as executive function ability to inhibit responses that do not match the category or words that have already been produced (e.g., Amunts et al., 2020). However, semantic searches may entail additional semantic interference that comes with naming within a category. For example, in picture-word interference tasks, naming pictures in the presence of a word from the same category (e.g., naming a picture of a cat with the word ‘dog’ printed over it) leads to slower naming times compared to picture naming with an unrelated or phonologically-related word. Naming words within a letter category (e.g., ‘F’), will lead to words that share similar sounds, but may not come from the same semantic category. Thus, the differences between individuals with MS and healthy older adults in semantic networks may arise from declines in semantic aspects of executive function.

These findings have both theoretical and clinical implications for language processing and cognition. First, the similarity in phonological responses among individuals with MS and the age- and gender-matched control group, and the common finding of similar performance on phonemic verbal fluency among older and younger adults suggests that basic phonological processes remain intact both in MS and in typical aging. This is contrary to what the Transmission Deficit Hypothesis (TDH) might suggest (e.g., Burke et al., 1991). However, much of the evidence in support of the TDH comes from studies that examined the tip-of-the-tongue phenomenon which, by definition, is an explicit phonological retrieval failure, and older adults tend to have more of these experiences. It could be the case that while phonological processes are impaired during Tip of the Tongue states, they are less impaired, phonologically, in other lexical retrieval situations (e.g., phonemic verbal fluency, conversations). If older adults and individuals with MS are not strongly impaired in phonological processes, then clinically, this would suggest that language interventions on other areas of language (e.g., semantic processes) or cognition (e.g., executive function) might have more impact.

Constraints on Generality

Although our results are promising, there are several limitations to the current experiment. The symptomatology of individuals with MS is highly variable and although all individuals with MS had a clinical diagnosis of MS with symptoms for a minimum of 1 year, some had been experiencing symptoms for over a decade. We were not able to examine disease severity in the current design, and it is possible that this may influence phonological processes, particularly among individuals who experience more severe MS symptoms. Related to this, we did not have an explicit measure of oral motor slowing, which would certainly be related to tasks like verbal fluency. Differences in this dimension may have contributed to the overall lower phonemic verbal fluency scores that we observed for the individuals with MS.

One benefit of the network science approach that we used is that having different numbers of responses did not directly influence degree and clustering coefficient values because the same phonological network was used for both groups. However, network science is a modeling tool that relies on the assumption that words are related in the way that the edges are defined (i.e., in this case phonological edit distances). Thus while degree, local clustering coefficient, and phonological edit distance capture aspects of phonological neighborhood density, it could be that there are phonological differences which were not revealed through the particular metrics we chose to examine in this study. Indeed, it is possible that the limited speech sample from a brief task (3 minutes of production) and a restricted range of phonemes (F, A, S) may not have fully characterized participants’ phonological abilities. At least one other study has found that phonemic processing is lower among individuals with MS, and this was related to increased self-reports of word finding difficulties (Dvorak et al., 2024). Their particular measure of phonemic processing, the Wechsler Individual Achievement Test, specifically measures how sounds are manipulated within words, and we did not examine phonemic errors, like substitutions or deletions, which would be an interesting avenue for future research.

Conclusions

Individuals with MS often experience cognitive slowing that includes increased word finding difficulties. While prior work has linked these behavioral observations to differences in semantic networks and vocabulary, the present study examined phonological aspects of production. We found that individuals with MS produce phonologically related items that have similar network characteristics (degree, local clustering coefficient) and similar lexical characteristics (number of phonemes, lexical frequency), and that individuals with MS search phonemic space (forward flow) similarly compared to neurotypical adults. These findings suggest that although individuals with MS experience increased word retrieval difficulties, phonological aspects of retrieval are similar to age-matched neurotypical adults and that language treatment options that focus on semantic or cognitive processes may be more effective options for individuals with MS.

Key Points.

Question:

This paper examines how the sounds of a language (i.e., its phonology) influences speaking in people who have Multiple Sclerosis (MS).

Findings:

Although people with MS spoke less than people who did not have MS, the phonological aspects of what they produced were similar. This suggests that declines in speaking for individuals with MS is not caused by differences in the parts of the brain that process the sounds of the words.

Importance:

These findings suggest that when people with MS have difficulty speaking, it may not be due to the sounds of the words. It may be due to other factors, such as the meaning of the words or how quickly people with MS can respond in general.

Next steps:

Word finding difficulty is an extremely frustrating experience that happens more often when we grow older and for individuals with some neurological conditions. Future research should continue to examine the causes that lead to speaking difficulties in both clinical conditions and healthy aging.

Acknowledgments

This publication was supported by funding from the National Institutes of Health (NIH), National Institute on Aging (Grant R01 AG034138, awarded to Michele T. Diaz) and from the National Institutes of Health, Eunice Kennedy Shriver National Institute of Child Health and Human Development (Grant R01 HD045798, awarded to Nancy D. Chiaravalloti). The content is solely the responsibility of the authors and does not necessarily represent the official views of the funding agencies. A draft of this article and all data and analysis scripts for this project are available on the Open Science Framework at https://osf.io/e5x94/. This research has not been previously reported.

Footnotes

CRediT Statement: AL: Formal Analysis, Writing – Original Draft; ALL: Conceptualization, Formal Analysis, Investigation, Writing – Original Draft; ALC: Conceptualization, Data Curation, Formal Analysis, Investigation, Writing – Review & Editing; NC: Conceptualization, Formal Analysis, Writing – Review & Editing; LBS: Data Curation, Investigation, Writing – Review & Editing; NDC: Data Curation, Investigation, Writing – Review & Editing, Funding Acquisition; MTD: Conceptualization, Data Curation, Formal Analysis, Investigation, Funding Acquisition, Supervision, Writing – Original Draft, Writing – Review & Editing. Correspondence concerning this article should be addressed to Amy L. Lebkuecher, Cognition and Action Lab, Moss Rehabilitation Research Institute, 50 Township Line Road, Elkins Park, PA 19027, Amy.Lebkuecher@jefferson.edu or Michele T. Diaz, Department of Psychology, The Pennsylvania State University, 140 Moore Building, University Park, PA 16802, mtd143@psu.edu

1

In contrast to our prior work which largely focuses on meso- and macro-scopic network properties, we focused on microscopic network properties here for pragmatic reasons. There was a single phonological network for each prompt (i.e., ‘F’, ‘A’, and ‘S’) which would have yielded identical meso- and macro-scopic properties for both groups.

References

  1. Amato MP, Goretti B, Ghezzi A, Lori S, Zipoli V, Moiola L, Falautano M, De Caro MF, Viterbo R, Patti F, Vecchio R, Pozzilli C, Bianchi V, Roscio M, Martinelli V, Comi G, Portaccio E, Trojano M, & Multiple Sclerosis Study Group of the Italian Neurological Society. (2010). Cognitive and psychosocial features in childhood and juvenile MS: Two-year follow-up. Neurology, 75(13), 1134–1140. 10.1212/WNL.0b013e3181f4d821 [DOI] [PubMed] [Google Scholar]
  2. Amato MP, Goretti B, Ghezzi A, Lori S, Zipoli V, Portaccio E, Moiola L, Falautano M, De Caro MF, Lopez M, Patti F, Vecchio R, Pozzilli C, Bianchi V, Roscio M, Comi G, Trojano M, & Multiple Sclerosis Study Group of the Italian Neurological Society. (2008). Cognitive and psychosocial features of childhood and juvenile MS. Neurology, 70(20), 1891–1897. 10.1212/01.wnl.0000312276.23177.fa [DOI] [PubMed] [Google Scholar]
  3. Amato MP, Ponziani G, Siracusa G, & Sorbi S (2001). Cognitive dysfunction in early-onset multiple sclerosis: A reappraisal after 10 years. Archives of Neurology, 58(10), 1602–1606. 10.1001/archneur.58.10.1602 [DOI] [PubMed] [Google Scholar]
  4. Amunts J, Camilleri JA, Eickhoff SB, Heim S, & Weis S (2020). Executive functions predict verbal fluency scores in healthy participants. Scientific Reports, 10(1), Article 1. 10.1038/s41598-020-65525-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Arnett PA, Smith MM, Barwick FH, Benedict RHB, & Ahlstrom BP (2008). Oralmotor slowing in multiple sclerosis: Relationship to neuropsychological tasks requiring an oral response. Journal of the International Neuropsychological Society: JINS, 14(3), 454–462. 10.1017/S1355617708080508 [DOI] [PubMed] [Google Scholar]
  6. Arroyo-Anlló EM, Lorber M, Rigaleau F, & Gil R (2012). Verbal fluency in Alzheimer’s disease and Aphasia. Dementia, 11(1), 5–18. 10.1177/1471301211416609 [DOI] [Google Scholar]
  7. Baciu M, Boudiaf N, Cousin E, Perrone-Bertolotti M, Pichat C, Fournet N, Chainay H, Lamalle L, & Krainik A (2016). Functional MRI evidence for the decline of word retrieval and generation during normal aging. In Age (Vol. 38, Issue 1, p. 3). 10.1007/s11357-015-9857-y [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Barabási A-L (2016). Network Science. Cambridge university press. [Google Scholar]
  9. Beatty WW, Goodkin DE, Beatty PA, & Monson N (1989). Frontal lobe dysfunction and memory impairment in patients with chronic progressive multiple sclerosis. Brain and Cognition, 11(1), 73–86. 10.1016/0278-2626(89)90006-7 [DOI] [PubMed] [Google Scholar]
  10. Belke E, Meyer AS, & Damian MF (2005). Refractory effects in picture naming as assessed in a semantic blocking paradigm. Quarterly Journal of Experimental Psychology, Section A, Human Experimental Psychology, 58, 667–692. 10.1080/02724980443000142 [DOI] [PubMed] [Google Scholar]
  11. Benton AL, Hamsher K, & Sivan AB (1983). Controlled oral word association test (COWAT). Multilingual Aphasia Examination. [Google Scholar]
  12. Brandstadter R, Fabian M, Leavitt VM, Krieger S, Yeshokumar A, Katz Sand I, Klineova S, Riley CS, Lewis C, Pelle G, Lublin FD, Miller AE, & Sumowski JF (2020). Word-finding difficulty is a prevalent disease-related deficit in early multiple sclerosis. Multiple Sclerosis Journal, 26(13), 1752–1764. 10.1177/1352458519881760 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Burke DM, MacKay DG, & James LE (2000). Theoretical approaches to language and aging. In Perfect TJ & Maylor EA (Eds.), Models of cognitive aging (pp. 204–237). Oxford University Press. [Google Scholar]
  14. Burke DM, MacKay DG, Worthley JS, & Wade E (1991). On the tip of the tongue: What causes word finding failures in young and older adults? Journal of Memory and Language, 30(5), Article 5. 10.1016/0749-596X(91)90026-G [DOI] [Google Scholar]
  15. Castro N. (2021). Methodological Considerations for Incorporating Clinical Data Into a Network Model of Retrieval Failures. Topics in Cognitive Science, tops.12531. 10.1111/tops.12531 [DOI] [PubMed] [Google Scholar]
  16. Castro N, & Vitevitch MS (2022). Using Network Science and Psycholinguistic Megastudies to Examine the Dimensions of Phonological Similarity. Language and Speech, 32. [DOI] [PubMed] [Google Scholar]
  17. Chen Q, & Mirman D (2012). Competition and cooperation among similar representations: Toward a unified account of facilitative and inhibitory effects of lexical neighbors. Psychological Review, 119(2), 417–430. 10.1037/a0027175 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Chiaravalloti ND, & DeLuca J (2008). Cognitive impairment in multiple sclerosis. The Lancet Neurology, 7(12), 1139–1151. 10.1016/S1474-4422(08)70259-X [DOI] [PubMed] [Google Scholar]
  19. Chiaravalloti ND, Moore NB, Nikelshpur OM, & DeLuca J (2013). An RCT to treat learning impairment in multiple sclerosis. Neurology, 81(24), 2066–2072. 10.1212/01.WNL.0000437295.97946.A8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Cosgrove AL, Beaty RE, Diaz MT, & Kenett YN (2023). Age differences in semantic network structure: Acquiring knowledge shapes semantic memory. Psychology and Aging. 10.1037/pag0000721 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Cosgrove AL, Kenett YN, Beaty RE, & Diaz MT (2021). Quantifying flexibility in thought: The resiliency of semantic networks differs across the lifespan. Cognition, 211, 104631. 10.1016/J.COGNITION.2021.104631 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Diaz MT, Karimi H, Troutman SBW, Gertel VH, Cosgrove AL, & Zhang H (2021). Neural sensitivity to phonological characteristics is stable across the lifespan. NeuroImage, 225, 117511. 10.1016/j.neuroimage.2020.117511 [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Diaz MT, Zhang H, Cosgrove AL, Gertel VH, Troutman SBW, & Karimi H (2022). Neural sensitivity to semantic neighbors is stable across the adult lifespan. Neuropsychologia, 171, 108237. 10.1016/j.neuropsychologia.2022.108237 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Dubossarsky H, Deyne SD, & Hills TT (2017). Quantifying the structure of free association networks across the life span. Developmental Psychology, 53(8), 1560–1570. 10.1037/DEV0000347 [DOI] [PubMed] [Google Scholar]
  25. Dvorak E, Levy S, Anderson JR, & Sumowski JF (2024). Phonemic processing is below expectations and linked to word-finding difficulty in multiple sclerosis. Multiple Sclerosis (Houndmills, Basingstoke, England: ), 30(10), 1374–1378. 10.1177/13524585241259648 [DOI] [PubMed] [Google Scholar]
  26. Eijlers AJC, Dekker I, Steenwijk MD, Meijer KA, Hulst HE, Pouwels PJW, Uitdehaag BMJ, Barkhof F, Vrenken H, Schoonheim MM, & Geurts JJG (2019). Cortical atrophy accelerates as cognitive decline worsens in multiple sclerosis. Neurology, 93(14), e1348–e1359. 10.1212/WNL.0000000000008198 [DOI] [PubMed] [Google Scholar]
  27. El-Wahsh S, Bogaardt H, Kumfor F, & Ballard K (2020). Development and validation of the communication and language assessment questionnaire for persons with multiple sclerosis (CLAMS). Multiple Sclerosis and Related Disorders, 43, 102206. 10.1016/j.msard.2020.102206 [DOI] [PubMed] [Google Scholar]
  28. Friesen DC, Luo L, Luk G, & Bialystok E (2015). Proficiency and Control in Verbal Fluency Performance across the Lifespan for Monolinguals and Bilinguals. Language, Cognition and Neuroscience, 30(3), 238–250. 10.1080/23273798.2014.918630 [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Galioto R, Macaron G, Lace JW, Ontaneda D, & Rao SM (2021). Is computerized screening for processing speed impairment sufficient for identifying MS-related cognitive impairment in a clinical setting? Multiple Sclerosis and Related Disorders, 54, 103106. 10.1016/j.msard.2021.103106 [DOI] [PubMed] [Google Scholar]
  30. Gordon JK, Young M, & Garcia C (2018). Why do older adults have difficulty with semantic fluency? Neuropsychology, Development, and Cognition. Section B, Aging, Neuropsychology and Cognition, 25(6), 803–828. 10.1080/13825585.2017.1374328 [DOI] [PubMed] [Google Scholar]
  31. Gray K, Anderson S, Chen EE, Kelly JM, Christian MS, Patrick J, Huang L, Kenett Y, & Lewis K (2019). “Forward Flow”: A New Measure to Quantify Free Thought and Predict Creativity. American Psychologist, 74(5), Article 5. 10.1037/amp0000391 [DOI] [PubMed] [Google Scholar]
  32. Grezmak T, Lace JW, Nakamura K, Ontaneda D, & Galioto R (2023). “It’s on the tip of my tongue!” exploring confrontation naming difficulties in patients with multiple sclerosis. Multiple Sclerosis and Related Disorders, 71, 104579. 10.1016/j.msard.2023.104579 [DOI] [PubMed] [Google Scholar]
  33. Grigoriadis P, Bakirtzis C, Nteli E, Boziki M-K, Kotoumpa M, Theotokis P, Kesidou E, & Stavrakaki S (2024). Morphosyntactic Abilities and Cognitive Performance in Multiple Sclerosis. Brain Sciences, 14(3), 237. 10.3390/brainsci14030237 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Henry JD, & Beatty WW (2006). Verbal fluency deficits in multiple sclerosis. Neuropsychologia, 44(7), 1166–1174. 10.1016/j.neuropsychologia.2005.10.006 [DOI] [PubMed] [Google Scholar]
  35. Horakova D, Dwyer MG, Havrdova E, Cox JL, Dolezal O, Bergsland N, Rimes B, Seidl Z, Vaneckova M, & Zivadinov R (2009). Gray matter atrophy and disability progression in patients with early relapsing-remitting multiple sclerosis: A 5-year longitudinal study. Journal of the Neurological Sciences, 282(1–2), 112–119. 10.1016/j.jns.2008.12.005 [DOI] [PubMed] [Google Scholar]
  36. Howard D, Nickels L, Coltheart M, & Cole-Virtue J (2006). Cumulative semantic inhibition in picture naming: Experimental and computational studies. Cognition, 100(3), 464–482. 10.1016/j.cognition.2005.02.006 [DOI] [PubMed] [Google Scholar]
  37. Iva P, Fielding J, Clough M, White O, Godic B, Martin R, & Rajan R (2020). Speech Discrimination Tasks: A Sensitive Sensory and Cognitive Measure in Early and Mild Multiple Sclerosis. Frontiers in Neuroscience, 14, 604991. 10.3389/fnins.2020.604991 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Jakimovski D, Weinstock-Guttman B, Roy S, Jaworski M, Hancock L, Nizinski A, Srinivasan P, Fuchs TA, Szigeti K, Zivadinov R, & Benedict RHB (2019). Cognitive profiles of aging in multiple sclerosis. Frontiers in Aging Neuroscience, 11(MAY), 457830. 10.3389/FNAGI.2019.00105/BIBTEX [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Kavé G. (2005). Phonemic fluency, semantic fluency, and difference scores: Normative data for adult Hebrew speakers. Journal of Clinical and Experimental Neuropsychology, 27(6), 690–699. 10.1080/13803390490918499 [DOI] [PubMed] [Google Scholar]
  40. Kavé G, & Halamish V (2015). Doubly blessed: Older adults know more vocabulary and know better what they know. Psychology and Aging, 30(1), Article 1. 10.1037/a0038669 [DOI] [PubMed] [Google Scholar]
  41. Kavé G, & Knafo-Noam A (2015). Lifespan development of phonemic and semantic fluency: Universal increase, differential decrease. Journal of Clinical and Experimental Neuropsychology, 37(7), 751–763. 10.1080/13803395.2015.1065958 [DOI] [PubMed] [Google Scholar]
  42. Kroll JF, & Stewart E (1994). Category interference in translation and picture naming: Evidence for asymmetric connection between bilingual memory representations. Journal of Memory and Language, 33(2), 149–174. 10.1006/jmla.1994.1008 [DOI] [Google Scholar]
  43. Lacour A, Seze JD, Revenco E, Lebrun C, Masmoudi K, Vidry E, Rumbach L, Chatel M, Verier A, & Vermersch P (2004). Acute aphasia in multiple sclerosis. Neurology, 62(6), 974–977. 10.1212/01.WNL.0000115169.23421.5D [DOI] [PubMed] [Google Scholar]
  44. Lebkuecher AL, Chiaravalloti ND, & Strober LB (2021). The role of language ability in verbal fluency of individuals with multiple sclerosis. Multiple Sclerosis and Related Disorders, 50, 102846. 10.1016/J.MSARD.2021.102846 [DOI] [PubMed] [Google Scholar]
  45. Lebkuecher AL, Cosgrove AL, Strober LB, Chiaravalloti ND, & Diaz MT (2024). Multiple sclerosis is associated with differences in semantic memory structure. Neuropsychology, 38(1), 42–57. 10.1037/NEU0000924 [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Martzoukou M, Nousia A, Messinis L, Konstantopoulos K, & Nasios G (2025). Language and Cognitive Impairments in Multiple Sclerosis: A Comparative Study of RRMS and SPMS Patients. Archives of Clinical Neuropsychology: The Official Journal of the National Academy of Neuropsychologists, 40(4), 775–782. 10.1093/arclin/acae110 [DOI] [PubMed] [Google Scholar]
  47. Mirman D. (2011). Effects of near and distant semantic neighbors on word production. Cognitive, Affective, Behavioral Neuroscience, 11(1), 32–43. 10.3758/s13415-010-0009-7 [DOI] [PubMed] [Google Scholar]
  48. Mirman D, & Magnuson JS (2008). Attractor dynamics and semantic neighborhood density: Processing is slowed by near neighbors and speeded by distant neighbors. Journal of Experimental Psychology: Learning, Memory, & Cognition, 34(1), 65–79. 10.1037/0278-7393.34.1.65 [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Monsch AU, Bondi MW, Butters N, Salmon DP, Katzman R, & Thal LJ (1992). Comparisons of verbal fluency tasks in the detection of dementia of the Alzheimer type. Archives of Neurology, 49(12), 1253–1258. 10.1001/archneur.1992.00530360051017 [DOI] [PubMed] [Google Scholar]
  50. Pitteri M, Vannucci M, Dapor C, Guandalini M, Daffinà A, Marastoni D, & Calabrese M (2023). Prominent role of executive functioning on the Phonemic Fluency Test in people with multiple sclerosis. Journal of the International Neuropsychological Society: JINS, 29(9), 902–906. 10.1017/S1355617723000139 [DOI] [PubMed] [Google Scholar]
  51. Rahimifar P, Isazadeh R, Soltani M, Ghobadi R, Boazar A, Abaeian G, Aliabdi L, Majdinasab N, & Amini P (2025). Examination of High-Level Language Skills in 2 Phases of Multiple Sclerosis (Relapsing-Remitting & Secondary Progressive) in Comparison With Healthy Counterparts. Medical Journal of the Islamic Republic of Iran, 39, 22. 10.47176/mjiri.39.22 [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Rahnemayan S, Fathalizadeh A, Behroozi M, Talebi M, Naseri A, & Mehdizadehfar E (2025). FMRI insights into the neural alterations and clinical correlates in multiple sclerosis: A comprehensive overview of systematic reviews and meta-analyses. Brain Research Bulletin, 223, 111278. 10.1016/j.brainresbull.2025.111278 [DOI] [PubMed] [Google Scholar]
  53. Rao SM, Leo GJ, Bernardin L, & Unverzagt F (1991). Cognitive dysfunction in multiple sclerosis. Neurology, 41(5), 685–691. 10.1212/WNL.41.5.685 [DOI] [PubMed] [Google Scholar]
  54. Revelle W. (2025). psych: Procedures for Psychological, Psychometric, and Personality Research (Version 2.5.3) [R package]. https://CRAN.R-project.org/package=psych [Google Scholar]
  55. Rinehardt E, Eichstaedt K, Schinka JA, Loewenstein DA, Mattingly M, Fils J, Duara R, & Schoenberg MR (2014). Verbal Fluency Patterns in Mild Cognitive Impairment and Alzheimer’s Disease. Dementia and Geriatric Cognitive Disorders, 38(1–2), 1–9. 10.1159/000355558 [DOI] [PubMed] [Google Scholar]
  56. Rossum GV, & Drake FL (2009). Python 3 Reference Manual. CreateSpace. [Google Scholar]
  57. Ruff RM, Light RH, Parker SB, & Levin HS (1996). Benton controlled oral word association test: Reliability and updated norms. Archives of Clinical Neuropsychology, 11(4), 329–338. 10.1093/ARCLIN/11.4.329 [DOI] [PubMed] [Google Scholar]
  58. Schnur T, Schwartz M, Brecher A, & Hodgson C (2006). Semantic interference during blocked-cyclic naming: Evidence from aphasia. Journal of Memory and Language, 54(2), 199–227. 10.1016/j.jml.2005.10.002 [DOI] [Google Scholar]
  59. Sepulcre J, Peraita H, Goni J, Arrondo G, Martincorena I, Duque B, Velez de Mendizabal N, Masdeu JC, & Villoslada P (2011). Lexical access changes in patients with multiple sclerosis: A two-year follow-up study. Journal of Clinical and Experimental Neuropsychology, 33(2), 169–175. 10.1080/13803395.2010.499354 [DOI] [PubMed] [Google Scholar]
  60. Shao Z, Janse E, Visser K, & Meyer AS (2014). What do verbal fluency tasks measure? Predictors of verbal fluency performance in older adults. Frontiers in Psychology, 5(JUL). 10.3389/FPSYG.2014.00772 [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Smith MM, & Arnett PA (2007). Dysarthria predicts poorer performance on cognitive tasks requiring a speeded oral response in an MS population. Journal of Clinical and Experimental Neuropsychology, 29(8), 804–812. 10.1080/13803390601064493 [DOI] [PubMed] [Google Scholar]
  62. Sporns O. (2011). Networks of the brain. MIT Press. [Google Scholar]
  63. Steyvers M, & Tenenbaum JB (2005). The Large-Scale Structure of Semantic Networks: Statistical Analyses and a Model of Semantic Growth. Cognitive Science, 29(1), 41–78. 10.1207/s15516709cog2901_3 [DOI] [PubMed] [Google Scholar]
  64. Šubert M, Novotný M, Tykalová T, Srpová B, Friedová L, Uher T, Horáková D, & Rusz J (2023). Lexical and syntactic deficits analyzed via automated natural language processing: The new monitoring tool in multiple sclerosis. Therapeutic Advances in Neurological Disorders, 16, 17562864231180719. 10.1177/17562864231180719 [DOI] [PMC free article] [PubMed] [Google Scholar]
  65. Tiberio M, Chard DT, Altmann DR, Davies G, Griffin CM, Rashid W, Sastre-Garriga J, Thompson AJ, & Miller DH (2005). Gray and white matter volume changes in early RRMS: A 2-year longitudinal study. Neurology, 64(6), 1001–1007. 10.1212/01.WNL.0000154526.22878.30 [DOI] [PubMed] [Google Scholar]
  66. Tombaugh TN, Kozak J, & Rees L (1999). Normative Data Stratified by Age and Education for Two Measures of Verbal Fluency: FAS and Animal Naming. Archives of Clinical Neuropsychology, 14(2), 167–177. 10.1093/arclin/14.2.167 [DOI] [PubMed] [Google Scholar]
  67. Troyer AK (2000). Normative Data for Clustering and Switching on Verbal Fluency Tasks. Journal of Clinical and Experimental Neuropsychology, 22(3), Article 3. 10.1076/1380-3395(200006)22:3;1-V;FT370 [DOI] [PubMed] [Google Scholar]
  68. Vaden KI, Halpin HR, & Hickok GS (2009). Irvine Phonotactic Online Dictionary, Version 2.0 [Dataset]. http://www.iphod.com [Google Scholar]
  69. Verhaeghen P. (2003). Aging and vocabulary score: A meta-analysis. Psychology and Aging, 18(2), Article 2. 10.1037/0882-7974.18.2.332 [DOI] [PubMed] [Google Scholar]
  70. Vitevitch MS (2008). What Can Graph Theory Tell Us About Word Learning and Lexical Retrieval? Journal of Speech, Language, and Hearing Research, 51(2), 408–422. 10.1044/1092-4388(2008/030) [DOI] [PMC free article] [PubMed] [Google Scholar]
  71. Vitevitch MS (2022). What Can Network Science Tell Us About Phonology and Language Processing? Topics in Cognitive Science, 14(1), 127–142. 10.1111/tops.12532 [DOI] [PMC free article] [PubMed] [Google Scholar]
  72. Vitevitch MS, Castro N, Mullin GJD, & Kulphongpatana Z (2023). The Resilience of the Phonological Network May Have Implications for Developmental and Acquired Disorders. Brain Sciences, 13(2), 188. 10.3390/brainsci13020188 [DOI] [PMC free article] [PubMed] [Google Scholar]
  73. Wachowius U, Talley M, Silver N, Heinze HJ, & Sailer M (2007). Cognitive Impairment in Primary and Secondary Progressive Multiple Sclerosis. 10.1080/138033990513645, 27(1), 65–77. https://doi.org/10.1080/138033990513645 [DOI] [PubMed] [Google Scholar]
  74. Wulff DU, De Deyne S, Jones MN, Mata R, & The Aging Lexicon Consortium. (2019). New perspectives on the aging lexicon. Trends in Cognitive Sciences, 23(8), 686–698. 10.1016/j.tics.2019.05.003 [DOI] [PubMed] [Google Scholar]
  75. Wulff DU, Hills TT, & Mata R (2022). Structural differences in the semantic networks of younger and older adults. Scientific Reports 2022 12:1, 12(1), 1–13. 10.1038/s41598-022-11698-4 [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES