The effect of automatic speech recognition on Iranian interpreters' cognitive load: An fNIRS study

SKASE Journal of Translation and Interpretation, 2025; 18(2): 240–58 doi: 10.33542/JTI2025-S-15 240 The effect of automatic speech recognition on Iranian interpreters’ cognitive load: An fNIRS study Adeleh Mirzaee*, Mir Saeed Mousavi Razavi*, Fatemeh Parham*, Mehrdad Dadgostar§# *Allameh Tabataba’i University, §MGH Institute of Health Professions, #Athinoula A. Martinos Center for Biomedical Imaging Abstract In the dynamic landscape of language interpreting, the integration of technology has ushered in a new era, marked by the advent of computer-assisted interpreting (CAI) tools. However, while the promise of improved communication through these tools is evident, a crucial facet demanding scrutiny is the intricate relationship between computer-assisted interpreting and the cognitive load experienced by interpreters. The present study was an attempt to use functional Near-Infrared Spectroscopy (fNIRS) to compare the cognitive load experienced by interpreters when utilizing a CAI tool, in this case automatic speech recognition (ASR), versus interpreting without such a tool. To this end, 12 interpreters were asked to perform two tasks: 1) simultaneous interpreting with the help of ASR (ASRSI) and 2) interpreting without ASR (NoASR-SI). fNIRS records changes in the concentration of oxyhemoglobin [ΔHbO2] and deoxyhemoglobin [ΔHbR] as it is sensitive to hemodynamic changes in the blood. Therefore, the interpreters’ cognitive load in both tasks was measured using the analysis of the HbO2 signals, which are an indicator of brain activation, and a paired t test was used to determine if there was a significant difference between the means of concentration changes of the two tasks. The results showed that the left temporal cortex (LTC) was significantly activated (p<0.05) during simultaneous interpreting from English into Persian. Furthermore, the mean of changes in concentration of HbO2 revealed that more cognitive load was experienced in interpreting without ASR compared to interpreting with ASR, meaning cognitive load was reduced when using ASR. In addition, participants’ feedback regarding the integration of ASR into interpreting was investigated through a questionnaire. The findings showed that participants’ subjective perceptions of ASR did not fully correspond to the objective neural activity recorded during simultaneous interpreting with and without ASR support. Keywords: fNIRS; simultaneous interpreting; cognitive load; computer-assisted interpreting; automatic speech recognition 1. Introduction 1.1 Interpreting Interpreting is a complex and long-standing practice, distinct from translation, as it existed before the development of writing (Pöchhacker 2016: 9). According to Seleskovitch (1976: 96), interpreting involves both understanding and rendering ideas, requiring the interpreter to simultaneously handle two roles in language and communication. Unlike other forms of communication, interpreting requires the same individual to both express ideas and understand another speaker’s ideas at the same time. This very feature of interpreting makes it a demanding activity in terms of cognitive load and mental effort (Mousavi Razavi 2020). Building upon this understanding of the inherent cognitive complexity of interpreting, Gerver (1976) further Adeleh Mirzaee, Mir Saeed Mousavi Razavi, Fatemeh Parham, Mehrdad Dadgostar 241 refines the concept by defining interpreting from a cognitive psychology viewpoint. He conceptualizes it as a sophisticated form of human information processing, encompassing “perception, storage, retrieval, transformation, and transmission of verbal information,” and emphasizing its susceptibility to various linguistic, motivational, and situational factors (Gerver 1976: 167). Thus, both perspectives converge on the notion that interpreting is not merely a linguistic transfer, but a complex cognitive operation subject to a multitude of influencing variables. 1.2 Cognitive load in interpreting Over time, the concept of cognitive load in interpreting has attracted considerable academic interest. Scholars from both within the field of interpreting studies and from other disciplines have investigated this issue, recognizing its potential to deepen our understanding of the cognitive demands involved in interpreting and to contribute to the development of more effective training methods and performance strategies (Seeber 2013). This growing interest stems from two key motivations: the need to conceptualize and analyze the complex and demanding nature of interpreting, and the desire to explore how interpreters navigate these challenges (Chen 2017). Both Seeber (2013) and Chen (2017) define cognitive load in interpreting within the framework of limited cognitive capacity, but they emphasize different aspects of the concept. Seeber (2013) characterizes cognitive load as the proportion of an individual’s finite cognitive capacity that is occupied by a given task, highlighting the inherent constraints of the cognitive system. In contrast, Chen (2017: 643) defines cognitive load more specifically in the context of interpreting, describing it as “that portion of an interpreter’s limited cognitive capacity devoted to performing an interpreting task in a certain environment”. While both definitions acknowledge the finite nature of cognitive capacity, Chen’s perspective introduces an environmental dimension, suggesting that cognitive load is not only task-dependent but also influenced by external factors within the interpreting setting. Gile’s Effort Model (2009) is a prominent model in interpreting studies which emphasizes the cognitive aspects of language. According to this model, simultaneous interpreting involves managing multiple cognitive demands, including listening, translating and speaking in real-time. Therefore, managing cognitive load is critical for interpreters, as excessive cognitive load could lead to omissions, substitutions, and other errors (Pöchhacker 2016). Error analysis in interpreting could then be insightful when speaking of cognitive load and performance quality (Barik 1971; Altman 1994; Anazawa et al. 2012; Mirzaee & Mousavi Razavi 2021). Different models and methods have been used by various scholars to measure cognitive load (cf. DeLeeuw & Mayer 2008; Ayres et al. 2021; Ouwehand et al. 2022). Cognitive load measurement methods are essential tools in understanding how much mental effort individuals expend during various tasks. Paas et al. (2003) and Schultheis and Jameson (2004) refer to a taxonomy of different methods ranging from subjective and analytical methods to performance and psycho-physiological methods. Among these methods, the psycho-physiological method offers a significant advantage in measuring cognitive load by directly assessing physiological responses that naturally fluctuate with cognitive changes. This direct assessment bypasses the subjective biases inherent in self-reported data, offering a more objective measure since these responses are involuntary (Seeber 2013). Among these methods neuroimaging techniques were the most widely used SKASE Journal of Translation and Interpretation 242 ones, enabling researchers to find a way to the interpreters’ “black box” (Seeber 2013). To date, different studies have been conducted on translation and interpreting based on neuroimaging techniques such as positron emission tomography (PET) (Price et al. 1999; Rinne et al. 2000), electroencephalography (EEG) (Petsche et al. 1993; Kurz 1995; Szarkowska et al. 2016), functional magnetic resonance imaging (fMRI) (Ahrens et al. 2010; Hervais-Adelman et al. 2014), and functional near-infrared spectroscopy (fNIRS) (Lin et al. 2018; Ren et al. 2019; He & Hu 2022; Yan et al. 2024). In this study, fNIRS was employed to assess the cognitive load of interpreters. As a non-invasive neuroimaging technique, fNIRS monitors fluctuations in oxygenated and deoxygenated hemoglobin concentrations, which reflect neural activation in the brain. Compared to EEG, fNIRS offers better spatial resolution, while it surpasses PET and fMRI in temporal resolution, making it a versatile tool for studying brain activity (Ren et al. 2019; Zhuang et al. 2022). Its robustness to motion artifacts and environmental noise, along with minimal body constraints, enables high ecological validity, particularly in naturalistic settings such as bilingual reading, translation, and interpretation (Ren et al. 2019; Yan et al. 2024). 1.3 Related neuroimaging studies Kurz (1995) used electroencephalography (EEG) to explore the neural correlates of directionality during shadowing (repeating the speech word for word) and simultaneous interpreting (SI) tasks. However, the tasks were not performed verbally rather mentally (without actual speaking). The findings highlighted the critical involvement of the temporal regions, particularly the left temporal lobe, in language processing, especially when interpreting into one’s L2. This aligns with Petsche et al. (1993), who used EEG to demonstrate that interpreting into one’s second language (L2) demands greater cognitive load compared to interpreting into one’s first language (L1). Further supporting this, Szarkowska et al. (2016) employed EEG and self-report measures to examine cognitive load during intralingual and interlingual interpreting. Their results indicated that interlingual respeaking imposed a higher cognitive load, though interpreters reported lower mental effort, suggesting a connection between interpreting proficiency and respeaking competence. Price et al. (1999) utilized positron emission tomography (PET) to investigate brain activation during translation and language switching. Similarly, Rinne et al. (2000) conducted a PET study to examine the cognitive demands of SI between L1 and L2. Their findings revealed that interpreting into L1 primarily activated the left frontal region, while interpreting into L2 elicited more extensive activation across the left fronto-temporal area. Recent research has continued to utilize brain imaging methods, such as functional magnetic resonance imaging (fMRI) to gain a better understanding of interpreting. Ahrens et al. (2010) conducted a preliminary fMRI study involving student interpreters to compare brain activity during simultaneous interpreting and free speech production. The results revealed significant differences in neural activation, emphasizing the heightened cognitive demands of SI. Unlike free speech, which primarily engages language production areas, SI activates additional regions responsible for dual-language processing and rapid information transfer, particularly the left superior temporal sulcus. Complementing this, Hervais-Adelman et al. (2014) used fMRI to compare brain activity during SI and shadowing in multilingual participants. Their findings indicated that both tasks modulated activity in the superior temporal lobe, with overlapping neural activation patterns, suggesting shared cognitive mechanisms between SI and shadowing. Adeleh Mirzaee, Mir Saeed Mousavi Razavi, Fatemeh Parham, Mehrdad Dadgostar 243 Functional near-infrared spectroscopy (fNIRS) has been employed in a limited number of studies within Translation and Interpreting Studies. Lin et al. (2018) combined behavioral measures and fNIRS to assess cognitive effort during pairing (linking translation-equivalent structures between the source language (SL) and target language (TL) stored in long-term memory during SI), transphrasing (explaining what/where the item is rather than giving its direct equivalent in the TL), and non-translation (producing the sound of the SL item rather than giving its direct equivalent in the TL) tasks. The study revealed cognitive overload in the left prefrontal cortex (PFC) during SI, though the ecological validity of the findings was limited due to the use of word-level stimuli and a narrow focus on one brain region. More recently, He and Hu (2022) used fNIRS to investigate the neural mechanisms of simultaneous interpreting, comparing professional interpreters and non-interpreter bilinguals. Their results demonstrated distinct brain activation patterns and functional connectivity in interpreters, highlighting the impact of expertise on cognitive processes. Both groups, however, relied on the right dorsolateral prefrontal hub during interpreting, suggesting a shared neural resource for managing the task’s demands. Another study utilizing fNIRS was conducted by Yan et al. (2024) who monitored the hemodynamic response in participants’ brains during consecutive interpreting tasks. By using fNIRS, the researchers could effectively capture the neural correlates of mental workload (MWL) and identify specific brain regions involved in the interpreting process, such as the inferior frontal gyrus, middle temporal gyrus, and inferior temporal gyrus. 1.4 Computer-assisted interpreting tools In the dynamic landscape of language interpreting, the integration of technology has ushered in a new era, marked by the advent of computer-assisted interpreting (CAI) tools. These sophisticated tools, with their focus on enhancing efficiency and quality in interpreting, contribute to the ongoing transformation of language-related professions in the face of technological advancements (Prandi 2018: 29). However, as of now, interpreters have access to a restricted variety of CAI tools, and their features may not comprehensively address every stage of the interpreting process (Prandi 2018: 30). In the similar vein, Tripepi Winteringham (2011: 89 ) notes that in general, the progress of technology in the field of interpreting has been notably slow, especially when contrasted with the rapid pace of technological integration observed in written translation. However, while the promise of improved communication through these tools is evident, a crucial facet demanding scrutiny is the intricate relationship between computer-assisted interpreting and the cognitive load experienced by interpreters (cf. Prandi 2018). It is well established that interpreters face inherently high cognitive demands, as they must seamlessly process linguistic nuances, cultural contexts, and real-time information. With the introduction of CAI tools into this demanding domain, questions arise regarding their influence on the cognitive load experienced by interpreters. However, despite the breakthrough these tools are making, only a limited number of studies has been dedicated to the use of computer-assisted interpreting tools in interpreting, particularly in an Iranian context (Costa et al. 2014; Fantinuoli 2017b, 2017a, 2018; Prandi 2018). The first exploratory study conducted to examine whether the use of CAI tools leads to an increase or a decrease in cognitive load was by Prandi (2018). Some years later, Mellinger (2023) adopted a socio-cognitive lens to explore the interplay between technology and interpreter cognition, emphasizing key constructs such as embedded and embodied cognition, SKASE Journal of Translation and Interpretation 244 extended cognition, and distributed cognition. In her doctoral dissertation, Frittella (2024) provides a comprehensive examination of the cognitive implications of CAI tools in interpreting and offers practical, evidence-based recommendations for integrating these tools into interpreter training programs. Advancements in automatic speech recognition (ASR) and, more recently, artificial intelligence (AI) have opened new possibilities for providing interpreters with fully automated support during SI (Fantinuoli 2017b). Several studies have explored the impact of ASR in SI of numbers (Desmet et al. 2018; Defrancq & Fantinuoli 2021; Pisani & Fantinuoli 2021). As Pöchhacker (2016: 188) observed, ASR is widely recognized as a technology “with considerable potential for changing the way interpreting is practiced”. Similarly, Fantinuoli (2023: 65) suggests that “the use of raw speech recognition or speech translation could prove to be an effective means to decrease interpreters cognitive load and improve performances”. However, this issue needs to be experimentally evaluated. The present study thus aimed to explore the impact of ASR as an instance of CAI tool on Iranian interpreters’ cognitive load and sought to find answer to the following question: How is the interpreters’ cognitive load different when they use ASR while interpreting compared to when they do not use it? 2. Methods 2.1 Participants The study involved 12 participants including 10 men and 2 women (mean age = 39.08 ± 3.77 years). Two key criteria were employed for participant selection. First, all participants were required to complete an online English proficiency test provided by the British Council, with a minimum required level of C1 to ensure advanced language proficiency. Second, participants needed to have at least two years of experience in the interpreting market and be actively earning a living through this profession. All participants were physically and mentally healthy, with no reported history of neurological or psychiatric conditions, nor were they using any medications. Furthermore, all individuals had normal or corrected-to-normal vision and exhibited normal color perception. All participants had either a Master’s or a PhD degree and took part in the experiment voluntarily. They were native Persian speakers and had English as their second language (L2). Prior to participation, each subject provided informed consent, and as a token of appreciation, they received a small gift. The research was approved by the Research Ethics Committee of Allameh Tabataba’i University in Tehran (IR.ATU.REC.1403.041), and the protocol was carried out according to the relevant guidelines. 2.2 Task description and procedure To conduct the fNIRS study, the National Brain Mapping Laboratory (NBML) of the University of Tehran was chosen as the setting of the experiment as it offers a wide range of services for cognitive studies including EEG, fMRI, fNIRS, EMG, TMS, tDCS, and eyetracker. This study involved two tasks: (1) simultaneous interpreting with the assistance of ASR (ASR-SI) and (2) simultaneous interpreting without ASR (NoASR-SI). To create fNIRScompatible tasks, a preparatory process was undertaken to select and modify the materials. A TED Talk video by Al Gore on climate change (available at Adeleh Mirzaee, Mir Saeed Mousavi Razavi, Fatemeh Parham, Mehrdad Dadgostar 245 https://www.youtube.com/watch?v=rUO8bdrXghs, accessed 2024-05-27) was selected. An eight-minute segment from the beginning of the 29-minute video was chosen for interpretation into Persian. The video was divided into eight one-minute segments to facilitate task design. For the ASR component, the SpeechTexter system (accessible via https://www.speechtexter.com/) was used. The accuracy of the ASR system was assessed by comparing its output to the existing transcriptions of the video, revealing a 98% accuracy rate. However, since this system does not support the Persian language, it was only usable for speech transcription of the English version of the video while being interpreted into Persian. As a result, the reverse direction of interpreting—Persian to English—was not explored in this study. Four alternating video segments (1, 3, 5, and 7) were designated for ASR-assisted interpreting, while the other four (2, 4, 6, and 8) were designated for interpreting without ASR. To demonstrate this setup, the video and the ASR webpage were displayed side by side using cascaded windows, ensuring both were visible simultaneously (see Figure 1). To avoid potential internet connectivity issues on the experiment day, the videos were pre-played alongside ASR, and the sessions were screen-recorded. These pre-recorded sessions ensured seamless playback during the experiment without requiring a live internet connection. Figure 1: Preview of the cascaded windows The finalized videos were sent to the lab expert for programming and integration into the experimental task design using MATLAB software. The experimental task was structured as follows: 1. An initial 60-second pre-rest period. 2. A task block of simultaneous interpreting with ASR (ASR-SI), followed by a 60-second rest period. 3. A task block of simultaneous interpreting without ASR (NoASR-SI), followed by a 60second rest period. 4. A final 60-second post-rest period. This cycle was repeated four times. At the start of each task block, a 2-second red fixation cross appeared at the center of the screen as a cue. The eight blocks (four for each task) were presented consecutively, with a 60-second rest period after each. After the final block, a post- SKASE Journal of Translation and Interpretation 246 rest period of 60 seconds concluded the session. This structured design ensured consistent timing and allowed for the measurement of hemodynamic responses during both task and rest periods. Figure 2 shows the process of the experimental task. Figure 2: Schematic representation of the experimental task design. The session included four cycles, each consisting of interpreting with ASR (ASR-SI) and without ASR (NoASR-SI) blocks, separated by 60-second rest periods 2.3 fNIRS data acquisition The present study employed a 48-channel fNIRS system (OxyMon fNIRS, Artinis) at the National Brain Mapping Laboratory in Tehran. The fNIRS system was used to record concentration changes in oxyhemoglobin [ΔHbO2] and deoxyhemoglobin [ΔHbR] as it is sensitive to hemodynamic changes in the blood. The device transmits two wavelengths of nearinfrared light (730 nm and 850 nm), with a sampling frequency of 10 Hz, allowing for the measurement of hemodynamic changes in the cortical brain regions. These hemodynamic responses were analyzed using the Modified Beer-Lambert law which describes the relationship between light absorption and concentration changes of hemoglobin in tissue (Orbig et al. 2000; Tornov et al. 2000). The fNIRS signals were collected from 24 channels, comprising 10 transmitters and 10 detectors. The placement of these channels was informed by prior neuroimaging studies which identified active brain regions involved in translating and interpreting process, such as the inferior and dorsolateral frontal region (Rinne et al. 2000; Hervais-Adelman et al. 2015; He et al. 2021), prefrontal regions (Rinne et al. 2000; Quaresima et al. 2002; Hervais-Adelman et al. 2015; He et al. 2017), Broca’s area (Tommola et al. 2000; He et al. 2017), and the left temporal area (Kurz 1995; Hervais-Adelman et al. 2015). However, due to the fixed format of the channel patches, as shown in Figure 3, as well as the limitations in the lab, the channels were positioned to correspond with only two of the above-mentioned regions. Specifically, the channels were strategically placed across two primary regions of the brain: the medial prefrontal cortex (MPFC) and the temporal cortex (TC). Each of these regions was further subdivided into the right and left hemispheres, with the MPFC additionally partitioned into a central region, resulting in the following subdivisions: left MPFC (LMPFC), central MPFC (CMPFC), right MPFC (RMPFC), left temporal cortex (LTC), and right temporal cortex (RTC). To ensure comprehensive coverage of the target brain regions, three optode probe patches were employed, including one patch with 20 channels and two patches, each containing 4 channels. A fixed inter-channel distance of 3 cm was maintained between each transmitter and detector, resulting in an approximate cortical penetration depth of 1.5 cm. Figure 3 illustrates the spatial distribution of the channels. This configuration was designed to optimize spatial resolution while maintaining a standardized inter-optode distance, ensuring robust and reliable hemodynamic measurements. The blue circles represent receivers, the yellow circles ASR-SI Four Repetitions Pre-Rest + Rest + Post-Rest 60 Sec 60-70 Sec 60 Sec 2 Sec 60 Sec 60-70 Sec 2 Sec 60 Sec NoASR-SI Rest t Cue Cue Adeleh Mirzaee, Mir Saeed Mousavi Razavi, Fatemeh Parham, Mehrdad Dadgostar 247 denote transmitters, and the white ones indicate the 24 channels utilized in the fNIRS setup. The location of the channels in each region is as follows: Channels 1, 2, 3, and 4 are in RTC, channels 5, 6, 7, and 8 are in LTC, channels 9, 10, 11, 12, 13, 14, and 15 are in RMPFC, channels 16 and 17 are in the center (CMPFC) and channels 18, 19, 20, 21, 22, 23 and 24 are in LMPFC. Figure 3: fNIRS channels configuration. The blue circles indicate receivers, the yellow circles denote transmitters, and the white circles represent the 24 channels. Channels are distributed across the Right Temporal Cortex (RTC: 1–4), Left Temporal Cortex (LTC: 5–8), Right Medial Prefrontal Cortex (RMPFC: 9–15), Central Medial Prefrontal Cortex (CMPFC: 16–17), and Left Medial Prefrontal Cortex (LMPFC: 18–24) SKASE Journal of Translation and Interpretation 248 2.4 fNIRS data analysis Functional near-infrared spectroscopy (fNIRS) is a powerful tool for monitoring neuronal activity, but its signals are often contaminated by physiological noise and motion artifacts. A critical step in this process involves identifying and removing physiological noise and motion artifacts, which are inherent to fNIRS signals. Physiological noise is primarily attributed to hemodynamic fluctuations, such as variations in cardiac pulsations (0.8–1.2 Hz), respiration (0.1–0.5 Hz), and blood pressure including Mayer waves (~0.1 Hz). Motion artifacts, on the other hand, are primarily caused by body movements, particularly head motion (Dadgostar et al. 2013). fNIRS data typically requires preprocessing to ensure accurate analysis. The best cognitive signal band was extracted using the discrete wavelet transform (DWT) to overcome these difficulties and successfully filter signals in the 0.003–0.08 Hz frequency range. By removing motion artifacts and physiological noise, which mostly appear above 0.08 Hz, this method creates a clean fNIRS-HbO2 dataset for further examination. This preprocessing pipeline enhances signal reliability by mitigating systemic noise and artifacts, ensuring robust insights into cerebral neural activity (Einalou et al. 2015; Dadgostar et al. 2016; Einalou et al. 2017; Dadgostar et al. 2018; Shirzadi et al. 2020; Shirzadi et al. 2024; Asadi et al. 2025). As established in fNIRS-related literature, this specific neuroimaging technique is effective in detecting changes in blood oxygenation. Therefore, in order to assess the activation of a specific area in the brain, one can measure the regional concentration changes in oxyhemoglobin and deoxyhemoglobin. Simply put, when a specific brain region is activated, HbO2 increases whereas HbR decreases. Given the fact that HbO2 is the most sensitive indicator of blood flow changes, only HbO2 signals were analyzed in this study. Thus, the concentration changes in oxyhemoglobin were computed across all the 24 channels for each participant and for both tasks. These HbO2 signals were then averaged in each of the five determined brain regions to be considered as an indicator of activation. Finally, a paired t test was performed to determine significant differences in brain activation between ASR-SI and NoASR-SI in the afore-mentioned brain regions (e.g., LMPFC, CMPFC, RMPFC, LTC, and RTC). The process of fNIRS data analysis is depicted in Figure 4. Figure 4: Block diagram of the fNIRS data analysis pipeline 3. Results 3.1 fNIRS results The fNIRS results indicated that, in most channels, the average concentration changes of HbO2 during ASR-SI were lower than those observed in NoASR-SI. These findings suggest that interpreting with the help of ASR is associated with a reduced cognitive load, while interpreting without ASR appears to increase cognitive demand. Consequently, it can be inferred that ASR functioning as a CAI tool, potentially alleviates the cognitive burden on interpreters. f R ignals Removing hysiological oise and ass iltering using T alculating the ean of oncentration hanges in HbO t Test Adeleh Mirzaee, Mir Saeed Mousavi Razavi, Fatemeh Parham, Mehrdad Dadgostar 255 DeLeeuw, Krista E. & Mayer, Richard E. 2008. A comparison of three measures of cognitive load: Evidence for separable measures of intrinsic, extraneous, and germane load. Journal of Educational Psychology 100: 223–34. Desmet, Bart & Vandierendonck, Mieke & Defrancq, Bart. 2018. Simultaneous interpretation of numbers and the impact of technological support. In Fantinuoli, Claudio (ed.), Interpreting and technology. Berlin: Language Science Press. 13–27. Einalou, Zahra & Maghooli, Keivan & Setarehdan, Seyed Kamaledin & Akin, Ata. 2015. Effective channels in classification and functional connectivity patternof prefrontal cortex by functional near infrared spectroscopy signals. Optik 127(6): 3271–75. Einalou, Zahra & Maghooli, Keivan & Setarehdan, Seyed Kamaledin & Akin, Ata. 2017. Graph theoretical approach to functional connectivity in prefrontal cortex via fNIRS. Neurophotonics 4(4): 041407. Fantinuoli, Claudio. 2017a. Computer-assisted preparation in conference interpreting. Translation and Interpreting 9(2): 24–37. Fantinuoli, Claudio. 2017b. Speech recognition in the interpreter workstation. In, Proceedings of the translating and the computer 39 conference. London: Editions Tradulex. 25–34. Fantinuoli, Claudio. 2018. Computer-assisted interpreting: Challenges and future perspectives. In Corpas Pastor, Gloria & Durán Muñoz, Isabel (eds.), Trends in e-tools and resources for translators and interpreters. Leiden: Brill. 153–76. Fantinuoli, Claudio. 2023. Towards AI-enhanced computer-assisted interpreting. In Corpas Pastor, Gloria & Defrancq, Bart (eds.), Interpreting technologies – current and future trends. Amsterdam/Philadelphia: John Benjamins. 46–71. Frittella, Francesca Maria. 2024. Computer-assisted interpreting: Cognitive task analysis and evidenceinformed instructional design recommendations. Surrey: University of Surrey. (Doctoral dissertation.) doi: 10.15126/thesis.901410. Gerver, David. 1976. Empirical studies of simultaneous interpretation: A review and a model. In Brislin, R. W. (ed.), Translation: Applications and research. New York: Gardiner. 165–207. Gile, Daniel. 2009. Basic concepts and models for interpreter and translator training. Amsterdam: John Benjamins Publishing. He, Yan & Hu, Yinying. 2022. Functional connectivity signatures underlying simultaneous language translation in interpreters and non-interpreters of Mandarin and English: An fNIRS study. Brain Sciences 12(2): 273. He, Yan & Hu, Yinying & Yang, Yaxi & Li, Defeng & Hu, Yi. 2021. Optical mapping of brain activity underlying directionality and its modulation by expertise in Mandarin/English interpreting. Frontiers in Human Neuroscience, 15: 649578. doi: 10.3389/fnhum.2021.649578. He, Yan & Wang, Meng-Yun & Li, Defeng & Yuan, Zhen. 2017. Optical mapping of brain activation during the English to Chinese and Chinese to English sight translation. Biomedical Optode Express 8: 5399–5411. doi: 10.1364/BOE.8.005399. Hervais-Adelman, Alexis G. & Moser-Mercer, Barabara & Golestani, Narly. 2015. Brain functional plasticity associated with the emergence of expertise in extreme language control. NeuroImage 14: 264–74. SKASE Journal of Translation and Interpretation 256 Hervais-Adelman, Alexis G. & Moser-Mercer, Barabara & Michel, Christoph M & Golestani, Narly. 2014. fMRI of simultaneous interpretation reveals the neural basis of extreme language control. Cerebral Cortex 25: 4727–39. doi: 10.1093/cercor/bhu158. Kurz, Ingrid. 1995. Watching the brain at work – an exploratory study of EEG changes during simultaneous interpreting (SI). Interpreters’ Newsletter 6: 6–13. Lin, Xiaohong & Lei, Victoria Lai Cheng & Li, Defeng & Yuan, Zhen. 2018. Which is more costly in hinese to English simultaneous interpreting, ‘pairing’ or ‘transphrasing’? Evidence from an fNIRS neuroimaging study. Neurophotonics 5(2): 025010. Mellinger, Christopher D. 2023. Embedding, extending, and distributing interpreter cognition with technology. In Corpas Pastor, Gloria & Defrancq, Bart (eds.), Interpreting technologies – current and future trends. Amsterdam/Philadelphia: John Benjamins. 195–216. Mirzaee, Adeleh & Mousavi Razavi, Mir Saeed. 2021. Directionality and error typology in EnglishPersian simultanous interpreting: A descriptive-analytic corpus-based study. New Voices in Translation Studies 25: 54–80. doi: 10.14456/nvts.2021.13. ousavi Razavi, ir aeed. 0 0. The relationship between iranian simultaneous interpreting trainees’ progress rate and their multiple intelligences. In Carsten, Sinner & Paasch-Kasier, Christine & Härtel, Johannes (eds.), Translation in the digital age. Translation 4.0. Newcastle: Cambridge Scholars Publishing. 26–40. Orbig, Hellmuth & Wenzel, Rüdiger & Kohl, Matthias & Horst, Susanne & Wobst, Petra & Steinbrink, Jens & Thomas, Florian & Villringer, Arno. 2000. Near-infrared spectroscopy: Does it function in functional activationstudies of the adult brain? International Journal of Psychophysiology 35: 125–42. Ouwehand, Kim & van der Kroef, Avalon & Wong, Jacqueline & Pass, Fred. 2022. Measuring cognitive load: Are there more valid alternatives to likert rating scales? Frontiers in Psychology 6: 146– 58. Pass, Fred & Tuovinen, Juhani & Tabbers, Huib & Van Gerven, P. W. 2003. Cognitive load measurement as a means to advance cognitive load theory. Educational Psychologist 38: 63– 71. doi: 10.1207/S15326985EP3801_8. Petsche, Hellmuth & Etlinger, Susan C. & Filz, Oliver. 1993. Brain electrical mechanisms of bilingual speech management: An initial investigation. Electroencephalography and Clinical Neurophysiology 86: 385–94. Pisani, Elisabetta & Fantinuoli, Claudio. 2021. Measuring the impact of automatic speech recognition on interpreter’s performances in simultaneous interpreting. n aiwen, ang & inghan, Zheng (eds.), Empirical studies of translation and interpreting: The post-structuralist approach. New York: Routledge. 181–97. Pöchhacker, Franz. 2016. Introducing interpreting studies. New York: Routledge. Prandi, Bianca. 2018. An exploratory study on CAI tools in simultaneous interpreting: Theoretical framework and stimulus validation. In Fantinuoli, Claudio (ed.), Interpreting and technology Berlin: Language Science Press. 29–60. Price, Cathy J. & Green, David W. & von Studnitz, Rosewitha. 1999. A functional imaging study of translation and language switching. Brain 122: 2221–35. Quaresima, Valentina & Ferrari, Marco & van der Sluijs, Marco C & Menssen, Jan & Colier, Willy N. 2002. Lateral frontal cortex oxygenation changes during translation and language switching Adeleh Mirzaee, Mir Saeed Mousavi Razavi, Fatemeh Parham, Mehrdad Dadgostar 257 revealed by non-invasive near-infrared multi point measurements. Brain Res Bull 59: 235–43. doi: 10.1016/s0361-9230(02)00871-7. Ren, Houhua & Wang, Meng-Yun & He, Yan & Du, Zhengcong & Zhang, Jiang & Zhang, Jing & Li, Defeng & Yuan, Zhen. 2019. A novel phase analysis method for examining fNIRS neuroimaging data associated with Chinese/English sight translation. Behavioural Brain Research 361: 151–58. Rinne, Juha O. & Tommola, Jorma & Laine, Matti & Krause, Bernando J. & Schmidt, Dirk & Kaasinen, Valtteri & Teräs, Mika & Sipila, Hannu & Sunnari, Marianna. 2000. The translating brain: Cerebral activation patterns during simultaneous interpreting. Neuroscience Letters 294: 85– 88. Schultheis, Holger & Jameson, Anthony. 2004. Assessing cognitive load in adaptive hypermedia systems: Physiological and behavioral methods. In Bra, Paul & Brusilovsky, Peter & Conejo, Ricardo (eds.), Adaptive hypermedia and adaptive web-based systems. Berlin: Springer. 225– 34. Seeber, Kilian G. 2013. Cognitive load in simultaneous interpreting: Measures and methods. Target 25: 18–32. Seleskovitch, Danica. 1976. Interpretation, a psychological approach to translation. In Brislin, Richard W. & Anderson, R. Bruce W. (eds.), Translation: Applications and research. New York: Gardner Press. 92–116. Shirzadi, Sima & Dadgostar, Mehrdad & Einalou, Zahra & Erdogan, Sinem Burcu & Akin, Ata. 2024. Sex based differences in functional connectivity during a working memory task: An fNIRS study. Frontiers in Psychology 15: 1207202. doi:10.3389/fpsyg.2024.1207202. Shirzadi, Sima & Einalou, Zahra & Dadgostar, Mehrdad. 2020. Investigation of functional connectivity during working memory task and hemispheric lateralization in leftand righthanders measured by fNIRS. Optik 221: 165347. doi: 10.1016/j.ijleo.2020.165347. zarkowska, Agnieszka & Krejtz, Krzysztof & utka, Łukasz & ilipczuk, Olga. 016. ognitive load in intralingual and interlingual respeaking – a preliminary study. Poznan Studies in Contemporary Linguistics 52: 209–33. Tommola, Jorma & Laine, Matti & Sunnari, Marianna & Rinne, Juha O. 2000. Images of shadowing and interpreting. Interpreting 5: 147–67. Tornov, Vlad & Franceschini, Maria Angela & Filiaci, Mattia & Fantini, Sergio & Wolf, Martin & Michalos, Antionios & Gratton, Enrico. 2000. Near-infrared study of fluctuations in cerebral hemodynamics during rest and motor stimulation: Temporal analysis and spatial mapping. Medical Physics 27: 801–15. Tripepi Winteringham, Sarah. 2011. The usefulness of icts in interpreting practice. Computer Science Linguistics: 87–99. Yan, Hao & Zhang, Yi & Feng, Yanqin & Li, Yang & Zhang, Yueting & Lee, Yujun & Chen, Maoqing & Shi, Zijuan & Liang, Yuan & Hei, Yuqin & Duan, Xu. 2024. Assessing mental demand in consecutive interpreting: Insights from an fNIRS study. Acta Psychologica 243: 104132. doi: 10.1016/j.actpsy.2024.104132. Zhuang, Chu & Meidenbauer, Kimberly L. & Kardan, Omid & Stier, Andrew J. & Choe, Kyoung Whan & Cardenas-Iniguez, Carlos & Huppert, Theodore J. & Berman, Marc G. 2022. Scale invariance in fNIRS as a measurement of cognitive load. Cortex 154: 62–76. SKASE Journal of Translation and Interpretation 258 Adeleh Mirzaee PhD Candidate, Allameh Tabataba’i University Faculty of English Translation Studies Tehran, Iran e-mail: [email protected] Mir Saeed Mousavi Razavi Allameh Tabataba’i University Faculty of English Translation Studies Tehran, Iran e-mail: [email protected] Fatemeh Parham Allameh Tabataba’i University Faculty of English Translation Studies Tehran, Iran e-mail: Parham.a[email protected] Mehrdad Dadgostar MGH Institute of Health Professions Charlestown, MA, USA Athinoula A. Martinos Center for Biomedical Imaging, Harvard-MIT Health Sciences and Technology Charlestown, MA, USA e-mail: mdadgos[email protected]