988368

1. Introduction
honological processing and stuttering are mutually related in Children Who Stutter (CWS). Behavioral studies declared that children with persistent stuttering experience a longer delay in phonological development than recovered children [1]. Furthermore, phonological disorders are more prevalent in CWS, compared to the healthy population [2]. Electrophysiological investigations also found that the neural basis of phonological tasks in CWS differs from that of their fluent peers [3]. Using the same tasks in adults who stutter presented significant differences between adults who stutter and their healthy counterparts in electrophysiological assessments; however, they had similar functions in behavioral tasks [4]. Adults who stutter were reported to have a longer reaction time than their nonstutterer peers [5], which may be because they had slower language processing [6]. However, such evidence in children is scarce. The electrophysiological assessment of stuttering in children is very helpful; such data could provide an insight to the neural basis of stuttering in CWS before it changes by compensatory strategies and emerging secondary behaviors [7].
Stuttering is a speech disorder which can be affected by emotions [8, 9, 10]. It is suggested that stuttering behaviors change during an emotional situation, especially in CWS [9]. Furthermore, negative emotions could slow articulation rates in children with persisting stuttering; however, it is not an influential factor in recovered and Typically-Developing Children (TDC) [10]. It remains unclear how emotional content can change fluency, as there is no concise electrophysiological evidence in this respect. 
Additionally, emotion is not a well-defined concept [11]; it could be any mental experience, i.e. intense and hedonistic, and being pleasant is a key component in emotion [12]. Generally, the emotions based on affective actions might be best defined in two main aspects, including valence and arousal. Valence reflects how a pleasurable and arousal object refers to high physiological activity of the event or word which produce agitation [13]. Emotion in communication is a complicated phenomenon, i.e. expressed through two channels; verbal or emotional content and non-verbal or emotional prosody [14]. Emotional content, specifically valence impacts language processing [15, 16]. Besides, dipolar fronto-occipital activity in topography was detected in emotional word processing [17]. Although a large body of studies exists on emotional perception, this concept is conspicuously limited in emotion literature [18]. Emotional content words required further neural activity for reading aloud tasks, compared to neutral words. Besides, positive words with a high valence had an easier processing than negative words with a low valence [19]. 
According to Levelt’s model of speech production, phonological processing is one part of language processing, i.e. supposed to be followed by articulation [20]. Then, it can be a main part of every verbal task, like loud reading [21]. Therefore, loud reading was used to assess phonological processing in this study and the time (100 to 400 msec) before the onset of articulation was considered as the phonological processing level [22].
Event-Related Potentials (ETPs) recording is a noninvasive neuroimaging method with high temporal resolution. Accordingly, there is an extensive interest to explore the neural basis of language processing by this method [23]. ERP is widely used in emotional language processing and various relevant data have been reported [24, 25]. Kissler and Herbert suggested that cortical activation in emotional word processing differs from that of the neutral words. Their results indicated that N1 (110-140 ms), Early Posterior Negativity (EPN, 216-320 ms), and Late Positive Potentials (LPP, 432-500 ms) are observed in time window in emotional content words in silent reading task [26]. Kissler et al. illustrated that EPN is increased in high-arousal words in silent reading tasks [25]. Generally, there exist two main approaches to analyze ERPs recording, including response-locked and stimulus-locked time window. Laganaro proposed a different method for ERP analysis, which integrates response-locked and stimulus-locked time windows. Then, analysis is conducted from stimulus to response. This process can lead to different results in ERP analysis in language production tasks [27].
ERP studies to avoid possible artifact resulting from overt articulation, have used silent tasks, where respondents were requested to say the words in their mind. Interestingly, they presented the same planning processes [27]. We analyzed ERP/EEG before overt articulation level according to Levelt’s model to avoid noisy EEG results from articulation.
This study aimed to probe emotional content and neutral words in phonological processing (100-400 ms before articulation). The same was measured through a reading aloud task in CWS using ERP recording and comparing the data with their TD peers. 
2. methods
Ten CWS (3 females and 7 males; age range: 7-10 years (Mean±SD = 8.9±0.11 years) and 10 TDC who were matched in age (Mean±SD = 8.7±0.12 years) and gender participated in this study (Table 1). 


The study subjects were monolingual native Persian-speaking children recruited by parents’ report. All study participants needed to have a healthy or corrected-to-normal vision and normal hearing. According to the speech therapist’s opinion, the research participants had no history of speech or language disorders, except stuttering in the experimental group. Speech and language development was also reported as normal by the study subjects’ parents. Additionally, an expert pediatric neurologist confirmed the lack of neurological or psychological disorders. Besides, their psychomotor development was normal. Edinburg Handedness Inventory [28] was completed for each child; the left-handed children were excluded from this investigation. CWS who met these inclusion criteria were referred to us from private speech therapy clinics in Tehran City, Iran, or from pediatric neurology clinics in Children’s Medical Center of Tehran University of Medical Sciences. The control group members were selected based on the age and gender of the experimental group from the speech therapy student’s families, public schools, and private language classes in Tehran City, Iran. The Ethics Committee of Tehran University of Medical Sciences (TUMS) approved the present study. All parents provided written informed consent forms for their children’s participation in this research.
This study was conducted in the ERP lab in the Rehabilitation Faculty of Tehran University of Medical Sciences from June to November 2017. 
The emotional Persian words list has been prepared by Nazari et al., which included 180 words with their scores in arousal and valence [29]. These words were divided into positive (high-valence), negative (low-valence), and neutral ones based on their valence score. Each category included 40 words. Then, this 120-word list was assessed in children respecting being understandable. Finally, the emotional words list for children was prepared by Salehi et al. [16]. Ultimately, the words in the final list were matched concerning the length of word (syllable) and frequency (P>0.05) and were prepared for the electrophysiological test (Table 2). 


EEG recording was explained simply for children and their parents, then they visited the laboratory to acclimated to it. Later, the children’s parents completed the demographic information forms. Concurrently, Stuttering Severity Index-3 [30] was performed by an expert speech therapist. Then, the training phase was conducted by 10 words to prepare the explored children. It was instructed that the child should read the presented word at the soonest possible. 
Next, the main trial with 120 words 40 words per category (positive, negative, neutral) was presented to them in a pseudorandom sequence. Notably, a plus sign (+) was illustrated between words as a fixation for 1000 milliseconds. The stimulus, i.e. words, were presented in the center of the monitor in Btitr black font with size 64 on a grey background for 2000 milliseconds.
The recording was conducted in an acoustic room, while children were seated 40 cm from the PC monitor. Simultaneously, the research participants were requested to wear an ERP cap with 64 electrodes, i.e. personally selected per child based on their head circumference from 3 cap sizes. The Fpz was at 1/10 Nasion to Inion. Since the response was reading aloud, a microphone was placed 10 cm from the mouth for recording verbal response and reaction time. 
Recording EEG was performed by EB-Neuro system and Galileo Net software (Italy) with a sample rate of 256 Hz. It was 64 channels recording by 10-10 international system of electrode placement. All electrodes were referenced to left and right mastoids. The data bandwidth was equal to 0.1 Hz to 40 Hz. Verbal responses, i.e. recorded between 200-2000 ms after stimuli presentation, were analyzed.
The EEGLAB software was employed for ERP visualization and analysis. The obtained data were preprocessed with Artifact Subspace Reconstruction (ASR) and PREP pipeline [31, 32]. Time windows for ERP and the region of interest characterized by repeated measure were conducted by EEGLAB.
We performed the data analysis in SPSS. In this study, continuous variables were expressed as Mean±Standard Deviation (SD). Repeated-measures Analysis of Variance (ANOVA) and Independent Samples t-test were used for between-group comparisons. All statistical tests were two-tailed and P<0.05 was considered statistically significant.
3. Results
Reading accuracy and Reaction Time (RT) were obtained from accurate recorded signal and analyzed as behavioral results.
The obtained data indicated no significant interaction between CWS and TDC in the number of accurate words (F2, 36=0.69; P=0.506). Our analysis suggested a significant relationship between accuracy and the emotional category of words (F2, 36=9.68; P<0.001). As per Figure 1, in both research groups, the accuracy of positive emotional content words was significantly more than the accuracy of negative emotional content words (P=0.007) and neutral words (P=0.002).  Although the mean value of reaction time in the CWS group was more than the TDC, this difference was not statistically significant (F1, 18 =2.68; P=0.119). As shown in Figure 2, ANOVA data revealed significant differences in reaction time concerning negative, positive, and neutral words between the explored CWS and TDC (F2, 36=6.20; P=0.005).  The amplitude data analysis in 11 regions of the brain, for neutral, positive, and negative words were performed for the studied CWS and TDC by t-test, and the relevant results are explained as follows. As per Table 3, the mean minimium and maximum vaues of amplitude were compared between the explored CWS and TDC. 


The minimum score of amplitude (negative) of the CWS in parietal (P=0.003), left posterior (P=0.002), left temporal (P=0.001), and occipital (P=0.001) aspects were significantly smaller than those of the TDC (P<0.005). The maximum score of amplitude of CWS in prefrontal, right frontal, and left frontal dimensions were significantly more negative than those of the TDC (P=0.008, P=0.009, and P=0.01, respectively). There were no significant differences in other regions (P>0.05).
According to Table 3, the mean maximum amplitude in the prefrontal (P=0.02) and left frontal (P=0.018) aspects of the studied CWS was significantly higher than that of the TDC. There were significant differences between the explored CWS and TDC groups concerning the right posterior in the mean minimum score of amplitude (P=0.031). The mean value of the minimum amplitude was also smaller in the left frontal, left temporal, parietal, and left posterior of the CWS group; however, these scores were not statistically significant (P=0.067, P=0.070, P=0.052, P=0.63 respectively). There were also no significant differences in other regions (P>0.05). 
The mean maximum value of amplitude in the CWS group was significantly higher than those of the TDC in the prefrontal and left frontal regions (P=0.020 P=0.018, respectively). The mean maximum score of amplitude in the CWS group concerning the central, right temporal, parietal, and left posterior regions was higher than those of the TDC group; however, these values were not statistically significant (P=0.080, P=0.086, P=0.057, P=0.068, respectively). Besides, there were no significant differences in other regions (P>0.05). 
As per Table 3, the mean minimum value of amplitude in the frontal, left temporal, parietal and occipital regions of the CWS group was lower (negative) than that of the TDC; however, these values were not statistically significant (P=0.074, P=0.050, P=0.070, P=0.060, respectively). There were also no significant differences in other regions (P>0.05). 
The mean maximum value of amplitude in the prefrontal and right frontal of the explored CWS was significantly greater than that of the TDC group (P=0.036 P=0.016, respectively). There were no significant differences in other regions (P>0.05). 
As per Figure 3, Figure 4, and Figure 5 (Appendix 1), comparing CWS and TDC by global field method [27] suggested that the peak to peak distance in CWS were greater than those of the TDC in neutral words in all brain areas.  The peak to peak distance in the CWS group was significantly more than those of the TDC, except in occipital lobe for positive words. For negative words, the peak to peak distances were greater in the CWS than the TDC group in the left temporal, right frontal, prefrontal, and parietal areas. 
As shown in Figure 6, there existed a very different pattern of energy distribution in the explored CWS and TDC.  Accordingly, more activity for neutral word reading task in the studied CWS was a limited area in occipital; however, in TDC, there was a very wide area in occipital lobe. A diffusive pattern was observed in the studied CWS, compared to the TDC group. 
As per Figure 7, the most similar pattern was detected in positive words. More activity in the occipital lobe was detected in both study groups for the positive word reading task. However, there was an irregular pattern of energy for positive words in the investigated CWS, compared to the TDC group. Besides, more activity was observed in the central areas of the CWS group. Similarly, there exists a diffuse and irregular pattern for negative words in CWS. A higher activity level was recorded in the central areas for the explored CWS, compared to the TDC group. However, the highest activity level was detected in the occipital lobe for both study groups (Figure 8).  4. Discussion
The main purpose of this study was to compare emotional word processing between CWS and TDC. Our results are discussed in two main parts, including behavioral and electrophysiological results. 
The behavioral analysis results suggested no significant differences between the studied TDC and CWS in response accuracy; however, the CWS group provided less accurate responses, compared to the TDC group. This difference cannot be attributed to reading disability, because both research groups had no history of reading disability. There was limited time for each word to be read; accordingly, the CWS group seemed to require further time for the emotional words reading aloud task. Apparently, temporal constraint made reading task different from routine reading. However, achieving less scores in accuracy for the aloud reading task in the explored CWS can be explained by a subtle and subclinical deficit in the phonological processing system [33]. however, the observed difference was not significant. Reaction times in the CWS group was longer than those of the TDC; however, the differences were not significant. It was predictable that stuttering would cause longer reaction time [5], because of slower language processing [6]. Despite this deficit in language processing system in CWS, the difference was not significant in the aloud reading task of single emotional words. Perhaps this simple task failed to challenge the phonological processing system. Thus, all investigated CWS were fluent in the aloud reading of single emotional words. It might reduce phonological processing load. 
Additionally, there were no significant differences between the CWS and TDC groups in behavioral results for phonological processing in emotional content words. This finding was in line with a previous study reporting that CWS and TDC were similar in behavioral analysis for phonological processing without considering emotional content [3, 4]. 
The collected electrophysiological data were analyzed by global field approach in 100 to 400 millisecond before articulation [27]. The relevant results illustrated significant differences concerning amplitude between the explored CWS and TDC in the prefrontal, right frontal, left frontal, left frontal, parietal, left posterior, and occipital regions in neutral words reading. Reportedly, there were greater amplitude for CWS than TDC. It is suggested that CWS requires greater neural activity than TDC in the motor and visual areas [34] for neutral words. There were significant differences between the explored CWS and TDC in the prefrontal, left frontal, and right posterior regions for positive word processing; the same was only true for the prefrontal and right frontal respecting negative word processing. These findings suggested that motor and visual regions and the areas related to emotional processing in the brain had higher neural activity in the CWS group than the TDC group for emotional and neutral words reading. In other words, different neural activity was recorded in CWS, in spite of fluent production and similar behavioral responses.
The greatest difference between the explored CWS and TDC regarded neutral words; the highest similarity concerned negative words. Emotional content words were more similarly processed by the studied CWS and TDC, compared to neutral words. Therefore, emotional content facilitates processing in CWS [18].
Replicating a previous study, we have detected dipolar fronto-occipital activity in topography for emotional words in TDC [17]. On the other hand, the topographical patterns of ERPs were different between the CWS and TDC groups. Accordingly, the recorded topography for the CWS group was not similar to a normal processing. Additionally, the most similar pattern in topography was recorded for high-valence words; subsequently, we concluded that valence has facilitated processing in the studied CWS and made it similar to a normal processing. These findings were consistent with those of previous studies [16, 19]. 
The severity of stuttering was considered as a contextual variable in the present study, which can be a dependent variable. Additionally, various tasks can be an appropriate representative for phonological processing. The reported results are comparable to the present study findings.
5. Conclusion 
The current research data suggested that high-valence emotion presented a normalizing effect on distribution; low-valence emotion provided facilitating effect on amplitude. Accordingly, the phonological processing of emotional content words was more similar to normal phonological processing, compared to neutral words in the studied CWS considering the electrophysiological results; however, behavioral results indicated no differences between the CWS and TDC groups. The same was reported for adults who stutter [4]. Adults and children who stutter seem to be vulnerable in phonological processing [3, 4]. 
Thus, differences between the explored CWS and TDC in electrophysiological data and similarities in behavioral data can attributed to subtle deficits as the neural basis of phonological processing level in speech production for the aloud reading of emotional and neutral words. 
These results can be helpful for clinical practice considering emotional words as a level of language processing in language hierarchy. 

Ethical Considerations
Compliance with ethical guidelines
This study was registered and approved by the Research Council, School of Rehabilitation, Tehran University of Medical Sciences (TUMS) on 2/28/2018. All study participant’s parents provided a written signed informed consent form.

Funding
This study was extracted from the PhD. dissertation of the firs author at the Department of Speech Therapy, School of Rehabilitation Sciences, Tehran University of Medical Sciences.

Authors' contributions
Manuscript preparation and edition, supervision, concept, design and literature review: All authors; Data gathering: Sousan Salehi, Ahmadreza Khatoonabadi; Statistical analysis: Sousan Salehi and Saman Maroufizadeh. 

Conflict of interest
The authors declared no conflicts of interest.

Acknowledgments
The authors would like to thenk the signal processing section in National Brain Mapping Lab (NBML) for helping us to analyze the recorded ERP. 


References

1. Paden EP, Yairi E, Ambrose NG. Early childhood stuttering II: initial status of phonological abilities. Journal of Speech, Language, and Hearing Research. 1999; 42(5):1113-24. [DOI:10.1044/jslhr.4205.1113] [PMID]
2. Wolk L. Intervention strategies for children who exhibit coexisting phonological and fluency disorders: a clinical note. Child Language Teaching and Therapy. 1998; 14(1):69-82. [DOI:10.1177/026565909801400104]
3. Weber-Fox C, Spruill JE, Spencer R, Smith A. Atypical neural functions underlying phonological processing and silent rehearsal in children who stutter. Developmental science. 2008; 11(2):321-37. [DOI:10.1111/j.1467-7687.2008.00678.x] [PMID] [PMCID]
4. Weber-Fox C, Spencer RMC, Spruill 3rd JE, Smith A. Phonologic processing in adults who stutter: electrophysiological and behavioral evidence. Journal of Speech, Language and Hearing Research. 2004; 47(6):1244-58. [DOI:10.1044/1092-4388(2004/094)]
5. Hand CR, Haynes WO. Linguistic processing and reaction time differences in stutterers and nonstutterers. Journal of Speech, Language and Hearing Research. 1983; 26(2):181-5. [DOI:10.1044/jshr.2602.181] [PMID]
6. Montgomery JW. Real-time language processing in school-age children with specific language impairment. International Journal of Language and Communication Disorders. 2006; 41(3):275-91. [DOI:10.1080/13682820500227987] [PMID]
7. Kaganovich N, Wray AH, Weber-Fox C. Non-linguistic auditory processing and working memory update in pre-school children who stutter: An electrophysiological study. Developmental neuropsychology. 2010; 35(6):712-36. [DOI:10.1080/87565641.2010.508549] [PMID] [PMCID]
8. Ward D. Stuttering and cluttering: Frameworks for understanding and treatment. United Kingdom: Psychology Press; 2006. https://psycnet.apa.org/record/2006-20272-000
9. Jones R, Choi D, Conture E, Walden T. Temperament, emotion and childhood stuttering. Seminars in Speech and language. 2014; 35(2):114-31. [DOI:10.1055/s-0034-1371755] [PMID] [PMCID]
10. Erdemir A, Walden TA, Jefferson CM, Choi D, Jones RM. The effect of emotion on articulation rate in persistence and recovery of childhood stuttering. Journal of Fluency Disorders. 2018; 56:1-17. [DOI:10.1016/j.jfludis.2017.11.003] [PMID] [PMCID]
11. Kleinginna PR, Kleinginna AM. A categorized list of emotion definitions, with suggestions for a consensual definition. Motivation and Emotion. 1981; 5(4):345-79. [DOI:10.1007/BF00992553]
12. Cabanac M. what is emotion? Behavioral Processes. 2002; 60(2):69-83. [DOI:10.1016/S0376-6357(02)00078-5]
13. Kensinger EA. Remembering emotional experiences: the contribution of valence and arousal. Reviews in Neurosciences. 2004; 15(4):241-51. [DOI:10.1515/REVNEURO.2004.15.4.241] [PMID]
14. Kotz SA, Paulmann S. Emotion, language, and the brain. Language and Linguistics Compass. 2011; 5(3):108-25. [DOI:10.1111/j.1749-818X.2010.00267.x]
15. Kuchinke L, Jacobs AM, Grubich C, Võ ML, Conrad M, Herrmann M. Incidental effects of emotional valence in single word processing: An fMRI study. NeuroImage. 2005; 28(4):1022-32. [DOI:10.1016/j.neuroimage.2005.06.050] [PMID]
16. Salehi S, Khatoonabadi AR, Ashrafi MR, Mohammadkhani Gh, Maroufizadeh S, Majdinasab F. The relationship between emotional content and word processing in normal Persian speaking children. Iranian Journal of Child Neurology. 2018; 12(4):140-52. [PMCID]
17. Imbir KK, Spustek T, Żygierewicz J. Effects of valence and origin of emotions in word processing evidenced by event related potential correlates in a lexical decision task. Frontiers in psychology. 2016; 7:271. [DOI:10.3389/fpsyg.2016.00271]
18. Vinson D, Ponari M, Vigliocco G. How does emotional content affect lexical processing? Cognition Emotion. 2014; 28(4):737-46. [DOI:10.1080/02699931.2013.851068] [PMID] [PMCID]
19. Salehi S, Khatoonabadi AR, Ashrafi MR, Mohammadkhani Gh, Maroufizadeh S. Valence effects on phonological processing in normal Persian speaking children: A study by ERP. Modern Journal of Language Teaching Methods. 2018; 8(2):49-76. [DOI:10.26655/mjltm.2018.2.2]
20. Peters HF, Hulstijn W, Van Lieshout PH. Recent developments in speech motor research into stuttering. Folia Phoniatrica et Logopaedica. 2000; 52(1-3):103-19. [DOI:10.1159/000021518] [PMID]
21. Wagner RK, Torgesen JK. The nature of phonological processing and its causal role in the acquisition of reading skills. Psychological Bulletin. 1987; 101(2):192-212. [DOI:10.1037/0033-2909.101.2.192]
22. Laganaro M, Perret C. Comparing electrophysiological correlates of word production in immediate and delayed naming through the analysis of word age of acquisition effects. Brain Topography. 2011; 24(1):19-29. [DOI:10.1007/s10548-010-0162-x] [PMID]
23. Banich MT, Mack M. Mind, brain, and language: Multidisciplinary perspectives. United Kingdom: Taylor Francis; 2003. [DOI:10.4324/9781410609182]
24. Conrad M, Recio G, Jacobs AM. The time course of emotion effects in first and second language processing: A cross cultural ERP study with German-Spanish bilinguals. Frontiers in Psychology. 2011; 2:351. [DOI:10.3389/fpsyg.2011.00351] [PMID] [PMCID]
25. Kissler J, Herbert C, Winkler I, Junghofer M. Emotion and attention in visual word processing-An ERP study. Biological Psychology. 2009; 80(1):75-83. [DOI:10.1016/j.biopsycho.2008.03.004] [PMID]
26. Kissler J, Herbert C. Emotion, etmnooi, or emitoon? - Faster lexical access to emotional than to neutral words during reading. Biological Psychology. 2013; 92(3):464-79. [DOI:10.1016/j.biopsycho.2012.09.004] [PMID]
27. Laganaro M. ERP topographic analyses from concept to articulation in word production studies. Frontiers in Psychology. 2014; 5:493. [DOI:10.3389/fpsyg.2014.00493]
28. Oldfield RC. The assessment and analysis of handedness: The Edinburgh inventory. Neuropsychologia. 1971; 9(1):97-113. [DOI:10.1016/0028-3932(71)90067-4]
29. Nazari MA, Khayat F, Poursharifi H, Hakimi M, Shojai Z. [Primary nomalization of emotional Farsi words (Persian)]. Applied psychological research Quarterly. 2014; 4(4):41-71. [DOI: 10.22059/JAPR.2014.52578]
30. Bakhtiar M, Seifpanahi S, Ansari H, Ghanadzade M, Packman A. Investigation of the reliability of the SSI-3 for preschool Persian-speaking children who stutter. Journal of Fluency Disorders. 2010; 35(2):87-91. [DOI:10.1016/j.jfludis.2010.02.003] [PMID]
31. Hunt MJ. Spectral signal processing for ASR. Keystone: Automatic Speech Recognition and Understanding (ASRU-99); 1999. http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.34.5207
32. Bigdely-Shamlo N, Mullen T, Kothe C, Su KM, Robbins KA. The PREP pipeline: standardized preprocessing for large-scale EEG analysis. Frontiers in Neuroinformatics. 2015; 9:16. [DOI:10.3389/fninf.2015.00016] [PMID] [PMCID]
33. Smith SL, Scott KA, Roberts J, Locke JL. Disabled readers’ performance on tasks of phonological processing, rapid naming, and letter knowledge before and after kindergarten. Learning Disabilities Research Practice. 2008; 23(3):113-24. [DOI:10.1111/j.1540-5826.2008.00269.x]
34. Hickok G. the functional neuroanatomy of language. physics of life reviews. 2009; 6(3):121-43. [DOI:10.1016/j.plrev.2009.06.001] [PMID] [PMCID]