The effect of time pressure and semantic relatedness in spoken word production: A topographic ERP study

https://doi.org/10.1016/j.bbr.2020.112587Get rights and content

Abstract

Speaking involves information-universal and information-specific systems, and it remains unknown about the influence of two systems in spoken word production. We manipulated time pressure (with and without) as an information-universal factor and the semantic relatedness between target names and distractor words (semantically related and unrelated) as an information-specific factor in the picture-word interference (PWI) task, while electrophysiological signals were measured concurrently. Naming latencies showed a significant semantic interference effect in the condition without time pressure but none in the condition with time pressure. Spatio-temporal segmentation analysis indicated that time pressure only prolonged the duration of Map 4 ranging from about 250–350 ms post picture presentation, and meanwhile shortened the duration of Map 5 starting around 350–520 ms. Within the framework of serial models, the two time periods were presumably associated with lexical selection and phonological encoding in spoken word production. Cluster-based permutation analysis revealed that semantic relatedness modulated ERPs around 320−360 ms after picture onset in the condition without time pressure, while 530−570 ms in the condition with time pressure. Our findings reveal that time pressure prolongs lexical selection and shortens phonological encoding. Furthermore, time pressure dynamically modulates whether semantically related distractors exceed a competition threshold to interfere with target word selection.

Introduction

The time course of speech planning processes has been at the forefront of investigations in psycholinguistic research. Classical speech production models have agreed on successive stages based on chronometric evidence [1,2] and more recent electrophysiological evidence ([[3], [4], [5]]; for a review, see [6]). A meta-analysis by Indefrey and Levelt ([7], updated in [8]) concluded that speech production serially undergoes conceptual preparation (0−200 ms), lemma retrieval (200−275 ms), phonological encoding (275−355 ms), phonetic encoding (355−455 ms) and articulation (455−600 ms). The word planning from an abstract concept to a physical speech sound takes about 600 ms, an average calculated across different words and different speakers.

Speakers employ information-universal and information-specific systems in spoken word production. Hence, the factors influencing naming latencies among participants include information-universal and information-specific factors. The former refers to general cognitive abilities concerning processing speed, working memory, and inhibition ability [9,10], while the latter refers to linguistic variables concerning semantic, orthographic or phonological property for a language [[11], [12], [13]]. Typically, information-specific factors influence certain mental processes involved in spoken word production. However, whether information-universal factors influence all processes or certain processes (i.e., lexical selection) remains unclear, and the interplay between two systems is largely unknown. Here we investigated this topic by manipulating time pressure (as an information-universal factor) and semantic relatedness (as an information-specific factor) in a picture-word interference (PWI) task with electrophysiological measures.

In the literature, time pressure is one of the information-universal factors to impact cognitive process [[14], [15], [16]]. Speaking is an online activity occurring under time constraints, and speakers need to achieve a trade-off between speed and accuracy [17,18]. It is typical that time pressure not only compels speakers to increase production speed by shortening the planning time, but also induces more speech errors. Studies have addressed the influence of time pressure on speech errors [19,20]. Dhooge and Hartsuiker [19] found that speakers more frequently misnamed distractors instead of targets under time pressure, indicating decreased attention resources involved in error monitoring (see also [21]). An electrophysiological study reported that participants exhibited decreased error-related negativity under time pressure, suggesting low attention resources available for self-monitoring in speaking [20]. However, little is known about how time pressure (as an information-universal factor) influences speaking speed variabilities.

The PWI paradigm is widely used to examine processes involved in spoken word production [[22], [23], [24]]. In this task, participants are presented with pictures and superimposed distractor words simultaneously, and are asked to name the pictures while ignoring the distractors. A semantic relationship between a target picture and a distractor slows down picture naming relative to an unrelated condition, which is called the semantic interference effect. It has been demonstrated that this effect is localized at the stage of lexical selection in spoken word production by a bunch of behavioral [11,25,26] and ERP [3,4,27,28] studies. Semantic interference effect reflects the lexical selection competition among co-activated semantically related lexical representations during lemma retrieval [25,29,30,31] or arises from the response exclusion difficulty given that semantically related distractors are more difficult to exclude from the response buffer than unrelated ones due to response relevance [32,23].

In the lexical competition model, Roelofs and colleagues incorporate a selective attention mechanism to interpret the presence of semantic interference effect [[33], [34], [35]]. Distractors become potent competitors only when they receive enough activation to exceed the competition threshold (see also [36]). Thus, it presents a more attention-demanding context in the PWI task with semantically related words relative to the unrelated ones [33,34]. Recent studies combining the PWI task with the spatial attention paradigm found that semantically related distractors yielded interference on condition that sufficient attention was devoted to them [37,38]. Piai and colleagues [39] manipulated the visibility of distractor words in the PWI task, and found that when participants cannot detect masked distractors consciously, the distractors did not elicit semantic interference effect. However, it remains unknown whether time pressure influences semantic interference effect in spoken word production.

In order to examine whether time pressure and semantic relatedness influence all processes or certain processes involved in spoken word production, we select Chinese-English unbalanced bilinguals and the second language (L2) speaking as target process, because the role of attention is of varying importance in the first language (L1) and L2 production [40]. Levelt [41] suggested that linguistic encoding (i.e., lexical selection) is largely automatic and involves less attention in L1 production, whereas Declerck and Kormos [42] suggested that L2 speakers require more attention for searching the appropriate lemma linked to the concept (see also [40]), indicating that the influence of time pressure and semantic relatedness would be larger in L2 than in L1 spoken word production.

Naming latencies present large variations among studies, and it is controversial whether the variation is distributed in each process involved in picture naming [43], or alternatively, only a few of specific processes [44]. Schuhmann and colleagues [43] observed a sharp contrast between their production latency (470 ms) and the estimated latency (600 ms) in a meta-analysis study [7]. Schuhmann et al. [43] attributed shorter speaking latencies to simple target names and further hypothesized that all processes involved in production with shorter latencies were faster than those in production with longer latencies. However, Laganaro and colleagues [44] compared different groups with varying production latencies (fast, median and slow) in their study, and found that the difference between fast and slow speakers occurred in a specific topographic map associated with lexical selection in the time interval of 200−350 ms.

In the literature, most ERP studies mainly addressed the temporal course of one specified stage (i.e., lexical selection or phonological encoding) involved in speaking. A few of EEG studies have attempted to gain insight into the time course of the semantic interference effect, and most of these have located it at the lexical-semantic level. Studies found that lemma retrieval starts approximately from 200 ms and completes around 300 ms post pictures onset [7,8]. However, there is no general agreement on the onset latency or the duration of semantic interference effect, and this is partly due to the different groups of participants and different stimulus sets among studies.

In order to avoid these confounding factors, we adopted a within-participants and within-items design when examining the influence of time pressure and semantic relatedness. In other words, we used the same group of participants and the same set of stimuli for different conditions, thus the variations among participants and stimuli were excluded in the present study.

ERP studies demonstrated a serial pattern of spoken word production [2,7,8], and researchers adopted this time framework as a template to compare with specific findings. However, it should be noted that there are serial and non-serial models concerning how information flows in speech production. Serial models argue that semantic processing muse be completed before phonological processing [2], while non-serial models dispute that phonological processing begins on the basis of early partial information provided by semantic processes [1,45,46]. Both models predict that semantic processing occurs before phonological processing.

A few of researchers propose a parallel pattern of speech production, which assumes that different constituents of words (e.g., semantic, lexical and phonological knowledge) are grouped together into an integrated unit and get activated in the brain near simultaneously [[47], [48], [49]]. This assumption receives support from Strijkers et al. [28] who reported that lexical frequency, cognate and language status (variables related to phonological processing) all produced early ERPs modulations within 200 ms. Furthermore, Miozzo, Pulvermüller, and Hauk (2015) found that semantic variables (in left frontotemporal regions) and phonological variables (in left middle temporal gyrus) became manifest in the same latency range (around 150 ms), suggesting that phonological processing might begin in parallel with semantic processing. Importantly, the proponents of parallel hypothesis acknowledge that the proposal is still very tentative, and it requires further empirical support [50].

So far, most ERP studies have confirmed an onset of lexical access around 150–200 ms post picture onset [3,4,[51], [52], [53]], and a later onset of phonological encoding around 300–400 ms [[54], [55], [56]]. Using intracranial recordings, Sahin et al. [57] also found temporally separated effects for lexical (around 200 ms), grammatical (around 320 ms), and phonological (around 450 ms) processing within Broca area. In the present study, we thus followed the temporal courses of spoken production provided by ERP studies, which was in line with the serial models.

The current study aims to investigate the influence of time pressure and semantic relatedness in spoken word production. Specifically, we are interested in whether the two factors influence all or certain processes before articulation, and whether there is an interaction between time pressure and semantic relatedness, in spoken word production. We manipulated time pressure (with and without) and the semantic relatedness between target names and distractor words (semantically related and unrelated) in the PWI task, while electrophysiological signals were measured concurrently.

We employed a spatio-temporal segmentation analysis to compare the topographic maps among different experimental conditions (for similar procedures, see [43,58,59]). This method utilizes cluster-analysis to segment continual electrical field variations into several functional microstates (also referred to as maps). Each microstate is represented by a topography and remains stable for several tens of milliseconds and then abruptly switches to a new microstate. It has been suggested that the sequence of microstates reflects successive stages of cognitive processing [60,61], thus we can compare the durations of similar brain processes in different experimental conditions. We predict that the stage of lexical selection is longer in the semantically related condition than the unrelated condition, and there is an interaction between time pressure and semantic relatedness.

Section snippets

Participants

Twenty students from Renmin university of China were recruited and paid for their participation. They were Chinese-English bilinguals (10 males, mean age = 24 ± 4 years) who received their schooling in China and were exposed to English after the critical period [62]. Their English proficiency was assessed by the College English Test for Band 4 (CET4), which is an examination for the students of non-English majors, and is held by the Chinese ministry of education. To ensure that participants

Behavioral results

A paired t-test on the self-report question showed that participants felt more pressured in the block with time pressure than without (1.63 versus 4.26, t(18) = -16.77, p < .001), which suggested a valid manipulation of time pressure. A participant whose response rate was lower than 50 % in the block with time pressure was excluded. Only correct responses were included in further statistical analyses. Thus, data were removed if: (1) participants stuttered or hesitated while naming the target;

Discussion

In a PWI task, we investigated how time pressure and semantic relatedness influence the planning processes underlying word production, and whether the two factors interact with each other using spatio-temporal segmentation and cluster-based permutation analyses. Critically, the spatio-temporal segmentation data provided clear evidence that time pressure prolonged the duration of Map 4 (around 250−350 ms) but shortened the duration of Map 5 (around 350−530 ms). Another noteworthy finding was

Conclusion

To conclude, the present study addresses how time pressure and semantic relatedness influence the processes in spoken word production and how the two factors interact with each other. The findings indicate that in face of time pressure, speakers may increase alertness to lemma selection but reduce phonological checking to gain speed, and selectively allocate more attention to picture naming but little attention to distractor processing.

Author statement

Qingfang Zhang and Xiao Cai designed the experiment. Xiao Cai and Yulong Yin performed data analysis. Qingfang Zhang, Xiao Cai and Mingkun Ouyang wrote the manuscript.

Declaration of Competing Interest

None.

Acknowledgements

This work was supported by the Key Project of Beijing Social Science Foundation in China (16YYA006), and the Fundamental Research Funds for the Central Universities, and the Research Funds of Renmin University of China (18XNLG28) to Qingfang Zhang.

References (107)

  • K. Strijkers et al.

    The temporal dynamics of first versus second language production

    Brain Lang.

    (2013)
  • D. Koester et al.

    Morphological priming in overt language production: electrophysiological evidence from Dutch

    NeuroImage

    (2008)
  • M. Laganaro et al.

    Dynamics of phonological–phonetic encoding in word production: evidence from diverging ERPs between stroke patients and controls

    Brain Lang.

    (2013)
  • C. Perret et al.

    Exploring the multiple-level hypothesis of AoA effects in spoken and written object naming using a topographic ERP analysis

    Brain Lang.

    (2014)
  • E. Maris et al.

    Nonparametric statistical testing of EEG‐ and MEG‐data

    J. Neurosci. Methods

    (2007)
  • A.M. Dale et al.

    Dynamic statistical parametric mapping: combining fMRI and MEG for high-resolution imaging of cortical activity

    Neuron

    (2000)
  • R. Oostenveld et al.

    Brain symmetry and topographic analysis of lateralized event-related potentials

    Clin. Neurophysiol.

    (2003)
  • A.M. De Groot et al.

    Lexical representation of cognates and noncognates in compound bilinguals

    J. Mem. Lang.

    (1991)
  • T. Guo et al.

    An ERP study of inhibition of non-target languages in trilingual word production

    Brain Lang.

    (2013)
  • V. Piai et al.

    Event-related potentials and oscillatory brain responses associated with semantic and Stroop-like interference effects in overt naming

    Brain Res.

    (2012)
  • M. Laganaro et al.

    Electrophysiological correlates of different anomic patterns in comparison with normal word production

    Cortex

    (2009)
  • A. Roelofs

    The WEAVER model of word-form encoding in speech production

    Cognition

    (1997)
  • L.Y. Ganushchak et al.

    Motivation and semantic context affect brain error-monitoring activity: an event-related brain potentials study

    NeuroImage

    (2008)
  • G.S. Dell

    A spreading activation theory of retrieval in sentence production

    Psychol. Rev.

    (1986)
  • W.J.M. Levelt et al.

    A theory of lexical access in speech production

    Behav. Brain Sci.

    (1999)
  • S. Aristei et al.

    Electrophysiological chronometry of semantic context effects in language production

    J. Cogn. Neurosci.

    (2011)
  • A. Costa et al.

    The time course of word retrieval revealed by event-related brain potentials during overt speech

    Proc. Natl. Acad. Sci. U.S.A.

    (2009)
  • M. Laganaro

    ERP topographic analyses from concept to articulation in word production studies

    Front. Psychol.

    (2014)
  • L.Y. Ganushchak et al.

    The use of electroencephalography in language production research: a review

    Front. Psychol.

    (2011)
  • P. Indefrey

    The spatial and temporal signatures of word production components: a critical update

    Front. Psychol.

    (2011)
  • M. Baciu et al.

    Functional MRI evidence for the decline of word retrieval and generation during normal aging

    Age

    (2016)
  • N. Boudiaf et al.

    Behavioral evidence for a differential modulation of semantic processing and lexical production by aging: a full linear mixed-effects modeling approach

    Aging Neuropsychol. Cogn.

    (2018)
  • S.J. Lupker

    The semantic nature of response competition in the picture-word interference task

    J. Mem. Cogn.

    (1979)
  • P.A. Starreveld et al.

    Time-course analysis of semantic and orthographic context effects in picture naming

    J. Exp. Psychol. Learn. Mem. Cogn.

    (1996)
  • D.M. Burke et al.

    Theoretical approaches to language and aging

  • J.T. King et al.

    Influence of time pressure on inhibitory brain control during emergency driving

    IEEE Trans. Syst. Man Cybern. Syst.

    (2018)
  • T.A. Salthouse

    A Theory of Cognitive Aging

    (1985)
  • J. Kormos

    Speech production and the cognition hypothesis

  • P. Robinson

    Cognitive complexity and task sequencing: studies in a componential framework for second language task design

    Int. Rev. Appl. Lingu.

    (2005)
  • P.A. Starreveld et al.

    Word substitution errors in a speeded picture-word task

    Am. J. Psychol.

    (1999)
  • A. Costa et al.

    On the categorical nature of the semantic interference effect in the picture-word interference paradigm

    Psychon. Bull. Rev.

    (2005)
  • B.Z. Mahon et al.

    Lexical selection is not by competition: a reinterpretation of semantic interference and facilitation effects in the picture-word interference paradigm

    J. Exp. Psychol. Learn. Mem. Cogn.

    (2007)
  • M. Ouyang et al.

    The effect of lexical cohort size is independent of semantic context effects in a picture-word interference task: a combined ERP and sLORETA study

    Front. Hum. Neurosci.

    (2019)
  • M. Miozzo et al.

    When more is less: a counterintuitive effect of distractor frequency in the picture-word interference paradigm

    J. Exp. Psychol. Gen.

    (2003)
  • R. Dell’Acqua et al.

    ERP evidence for ultra-fast semantic processing in the picture-word interference paradigm

    Front. Psychol.

    (2010)
  • K. Strijkers et al.

    Tracking lexical access in speech production: electrophysiological correlates of word frequency and cognate effects

    Cereb. Cortex

    (2010)
  • R. Abdel Rahman et al.

    Semantic context effects in language production: a swinging lexical network proposal and a review

    Lang. Cogn. Process.

    (2009)
  • R. Abdel Rahman et al.

    Dismissing lexical competition does not make speaking any easier: a rejoinder to Mahon and Caramazza

    Lang. Cogn. Process.

    (2009)
  • A. Roelofs

    Goal-referenced selection of verbal action: modeling attentional control in the Stroop task

    Psychol. Rev.

    (2003)
  • A. Roelofs et al.

    Selective attention and distractor frequency in naming performance: comment on Dhooge and Hartsuiker (2010)

    J. Exp. Psychol. Learn. Mem. Cogn.

    (2011)
  • Cited by (6)

    • Language dominance influences bilingual lexical selection quantitatively but not qualitatively in language production: Evidence from Chinese-English bilinguals

      2022, Lingua
      Citation Excerpt :

      Based on the empirical findings, they provided a computational model which is known as bilingual WEAVER++ (Word Encoding by Activation and VERification + + model; see Roelofs, 1992, 1993, 1998, 2003, 2007, 2008a, 2008b, 2008c, 2014) to describe the lexical selection in a within-language competition way and simulate both the semantic interference effect and translation facilitation effect, further demonstrating the effectiveness of the within-language competition view in explaining the two effects. WEAVER++ is a symbolic computational model of monolingual lexical selection originally designed to account for typical RT effects in monolingual PWI paradigm, such as the semantic interference effect (Cai et al., 2020; Roelofs and Piai, 2017) and the identity facilitation effect (Blackford et al., 2012; Piai et al., 2016). However, the between-language semantic interference effect and the translation facilitation effect were mainly elicited in bilinguals whose L1 and L2 are both alphabetical languages, such as Catalan-Spanish bilinguals (Costa et al., 1999), Dutch-English bilinguals (Roelofs et al., 2016), English-French bilinguals (Dylman and Barry, 2018), French-English bilinguals (Sudarshan and Baum, 2019), Spanish–English bilinguals (Dylman and Barry, 2018) and Swedish-English bilinguals (Dylman and Barry, 2018).

    • Language proficiency moderates the effect of L2 semantically related distractors in L2 spoken word production

      2021, Brain Research
      Citation Excerpt :

      For electrophysiological measures, studies on monolingual semantic interference effects have reported early effects starting between 150 and 250 ms (e.g., Aristei et al., 2011; Costa et al., 2009; Rose et al., 2018; for meta-analysis estimation, see also Indefrey, 2011; Indefrey and Levelt, 2004). In the context of L2 speech production, bilinguals probably need a longer time to consciously search for the appropriate lemma matching the activated concept (Cai et al., 2020; Declerck and Kormos, 2012). Thus, we expected to observe later ERP modulations corresponding to lexical competition in L2 production tasks than in L1 production tasks.

    View full text