Sketching the Landscape of Speech Perception Research (2000–2020): A Bibliometric Study

Based on 6,407 speech perception research articles published between 2000 and 2020, a bibliometric analysis was conducted to identify leading countries, research institutes, researchers, research collaboration networks, high impact research articles, central research themes and trends in speech perception research. Analysis of highly cited articles and researchers indicated three foundational theoretical approaches to speech perception, that is the motor theory, the direct realism and the computational approach as well as four non-native speech perception models, that is the Speech Learning Model, the Perceptual Assimilation Model, the Native Language Magnet model, and the Second Language Linguistic Perception model. Citation networks, term frequency analysis and co-word networks revealed several central research topics: audio-visual speech perception, spoken word recognition, bilingual and infant/child speech perception and learning. Two directions for future research were also identified: (1) speech perception by clinical populations, such as hearing loss children with cochlear implants and speech perception across lifespan, including infants and aged population; (2) application of neurocognitive techniques in investigating activation of different brain regions during speech perception. Our bibliometric analysis can facilitate research advancements and future collaborations among linguists, psychologists and brain scientists by offering a bird view of this interdisciplinary field.

Introduction

Speech perception is a vital means of human communication, which involves mapping speech inputs onto various levels of representation, for example phonetic/phonemic categories and words (Samuel, 2011). However, the underlying mechanism of this process is more complex than it appears to be. First, unlike a written sequence of words, the speech stream is continuous and thus requires segmentation before it can be further processed. Second, there is no simple or direct correspondence between phones/phonemes as perceived and the acoustic patterns generated by articulatory gestures due to variations induced by different phonetic contexts and talker vocal characteristics.¹ The fact that adult native speakers can effortlessly perceive speech from various speakers even in relatively noisy environments has intrigued researchers from multiple disciplines, such as audiology, experimental psychology, brain science, computer science, phonetics and linguistics.

Historically speaking, compared with other aspects of human perception, the study of speech perception began relatively late around the 1950s thanks to the accessibility of technologies that are essential for speech research, such as sound spectrographs and acoustic speech synthesisers. In one historical review, Hawkins divided speech perception research into three periods (Hawkins, 2004): the early period (1950–1965), the middle period (1965–1995) and recent developments (1995–2004). The early period saw attempts to examine some fundamental issues in speech perception, such as cerebral dominance in speech processing (Kimura, 1961), the role of memory and context, the separation of a speech signal from environment sounds or integration of different modalities (Sumby and Pollack, 1954). The second period was built on the first period and witnessed birth and growth of important research topics, for example categorical perception, that is listeners perceive categories when presented with continuous stimuli, and theories, for example the motor theory (Liberman et al., 1967) and the quantal theory (Stevens, 1972). Although the review provided insightful reflections on key issues and achievements of the first two stages, it did not cover much of the advancements in the twenty-first century. There exist several other recent qualitive reviews of speech perception (Diehl et al., 2004; Samuel, 2011; Antoniou, 2018); however, they focused more on general theoretical aspects of the field with consideration of only a limited number of research articles.

The first two decades of the twenty-first century have witnessed a surge of academic publications in all fields. It is, thus, increasingly challenging to keep oneself at the front line of and to have a bird view of research developments, especially in interdisciplinary fields, such as speech perception. Synthesising research findings effectively from different disciplines becomes an integral part of research advancements. Bibliometric analysis provides a systematic, transparent and objective approach to understanding a certain research field (Aria and Cuccurullo, 2017). It employs statistical and data visualisation methods to examine bibliographic big data, covering a much larger scope of research items compared with traditional qualitative reviews. So far, the field of speech perception has not been investigated via bibliometric analysis, although there exist such attempts in neighbouring fields, such as linguistics (Lei and Liao, 2017), applied linguistics (Lei and Liu, 2019), and in some interdisciplinary topics, such as visual word recognition (Fu et al., 2020), second language pronunciation (Demir and Kartal, 2022) and multilingualism (Lin and Lei, 2020).

In the present study, a bibliometric analysis was conducted to answer the following research questions:

1. What are the most active countries and research institutes, who are the leading researchers, and how do they collaborate in speech perception research?

2. What are the most impactful theories and models?

3. What are the important research themes/topics and future directions?

Materials and Methods

Data Collection

First, we did a topic search of the term ‘speech perception’ (in the quotation mark) using the Web of Science Core Collection² for research articles (Document type = article) published in journals indexed in Social Sciences Citation Index (SSCI), Science Citation Index Expanded (SCIE), Emerging Sources Citation Index (ESCI) on 24th of February 2021. The time span was set between 2000 and 2020 and non-English papers were excluded. Originally, 9,436 research articles were found. Given that the focus of the present bibliometric analysis is on speech perception from the perspectives of phonetics/linguistics, psychology, neuroscience and speech pathology, pure medical research under the Web of Science category of otorhinolaryngology (n = 2,731) was excluded.

The full records and cited references of these articles were downloaded and processed by the bibliometrix package (Aria and Cuccurullo, 2017) in R (R Core Team, 2018). We checked the data set and removed articles with missing data, which further reduced the number of research articles to 6,407 (see Figure 1). The bibliometrix package developed in the R language provides a variety of functions for comprehensive bibliometric analysis and can be integrated with other R packages seamlessly for more advanced data modelling and visualisation. We chose the bibliometrix package over other software because it provides a more open, flexible, customisable and reproducible workflow.

FIGURE 1

Figure 1. The workflow diagram.

Data Wrangling

The downloaded bibliometric information was converted to a data frame, in which each row represents one article and each column one field tag in the original export file. Bibliometrix automatically implemented a set of cleaning rules. First, all texts were transformed into uppercase. Second, non-alphanumeric characters, punctuation symbols and extra spaces were removed. Third, authors’ first and middle names were truncated to the initials.

Results and Discussion

The present study surveyed a period of 21 years, and 6,407 journal articles that met our selection criteria were authored by 12,381 researchers, 438 of whom wrote single-authored articles and 11,943 of whom co-authored with others (see Table 1). 564 articles are single-authored and the number of authors for each paper on average is 1.93. The collaboration index, that is co-authors per article calculated only using the multi-authored article set, is 2.04, indicating that most speech perception research involves collaboration between at least two authors. As for research impact, the average number of citations per document is 27.91 times and that breaks down to 2.42 times per year.

TABLE 1

Table 1. Summary information on retrieved speech perception studies (2000–2020).

In the following sections, we first report major publication venues for speech perception research (section “The Most Popular Publication Venues”) and then investigate research productivity as indicated by publications and identify leading countries and researchers (section “Research Productivity”). Next, we demonstrate connections among universities/research institutes and countries via collaboration networks (section “Research Collaborations”). Last but not least, foundational and time-honoured theories of speech perception were revealed via citation analysis of publications and authors, whereas co-citation networks, bibliographic coupling networks, term frequency and co-word analysis based on keywords and abstracts were used to uncover more recent research themes/cohorts and future directions (section “Impactful Research Work and Key Research Themes”).

The Most Popular Publication Venues

Table 2 shows the top 20 journals that publish speech perception studies. These journals belong to different disciplines, indicating the interdisciplinary nature of the field. For example, Journal of the Acoustical Society of America published the largest number of speech perception research, and two subjects in its scope that are relevant to speech perception are speech, music and noise, as well as psychology and physiology of hearing. Developmental Science published 76 papers on this topic from the perspective of developmental psychology. Other publication channels range from linguistics/phonetics journals (e.g., Journal of Phonetics, Language and Speech), to psychology or psycholinguistic journals (e.g., Journal of Speech Language and Hearing Research, Frontiers in Psychology, PLOS one, Journal of Experimental Psychology), as well as neuroscience journals (e.g., Neuroimage, Brain and Language, Neuropsychologia, Journal of Neuroscience, Frontiers in Human Neuroscience, Cerebral Cortex, Journal of Cognitive Neuroscience, Neuroreport).

TABLE 2

Table 2. Top 20 journals publishing speech perception studies.

It should be noted that the journal rankings based on the total number of publications should be interpreted with caution as some new journals, that is PLOS one (2006), Frontiers in psychology (2007) and Frontiers in human neuroscience (2007), were launched after 2000. Thus, we also normalised the total publications by the years of publication during the period of survey, for example 14 years for PLOS one. The rankings for PLOS one and Frontiers in Psychology did not change but that for Frontiers in Human Neuroscience changed.

Research Productivity

There is a yearly and drastic increase in speech perception journal articles from 2000 to 2020 (see Figure 2), suggesting that it is an area of increasing productivity. A drop can be seen in 2020, reflecting the impact of the COVID-19 pandemic. Given that most speech perception experiments require face-to-face testing, social distancing requirements slowed down or interrupted some of the data collection activities and led to a short-term deceleration in research output.

FIGURE 2

Figure 2. Publications per year of speech perception studies (2000–2020).

To address this issue, it should be noted that many online testing environments and software had been developed before COVID-19 for psychological experiments and they yield generally comparable results to the data obtained in the lab (Crump et al., 2013) if users follow standardised and widely accepted guidelines (Grootswagers, 2020). It is expected that these online testing methods will be further developed, better recognised and more frequently applied than before the pandemic.

In addition, publication counts were calculated for each country and the top ten most productive countries were listed in Table 3. The United States is undoubtedly the leading country in this area of research, followed by the United Kingdom. It is also interesting to note that multiple country publications account for a relatively small portion of all articles published by US researchers, suggesting that a relatively large amount of research work was done through domestic collaboration. On the other hand, France, China and Spain have a higher proportion of multiple country publications, although their total number of international publications was not as high as the United States and the United Kingdom.

TABLE 3

Table 3. Top 10 most productive countries in speech perception research (2000–2020).

To examine variations in research productivity for each of the ten countries over the 21-year period, we plotted the growth curve for each of them in Figure 3. The United States is the leading country in the number of publications throughout the period with an overall higher increase rate than the other nine countries, which have relatively low increase rates.

FIGURE 3

Figure 3. Publication trends in speech perception research for the top 10 most productive countries.

The top 20 most productive authors are shown in Table 4. Mitterer H. came first in terms of number of journal articles, followed by McQueen J.M. Both authors also have a high H-index, a measure of research impact, that is the highest number of publications of a scientist that each received h or more citations (Schreiber, 2008). Pisoni D.B., Rosen S., Kraus N. and Werker J.F. also have published more than 40 papers with relatively high H-index.

TABLE 4

Table 4. Top 20 most productive authors of speech perception research (2000–2020).

To characterise the dynamic relations between publication and citation, the yearly publication and citation counts were plotted in Figure 4. The size of the circles is in proportion to the number of publications. The darkness of the circles equals to the total citations per year. Authors, such as Rosen S., Kraus N., Werker J.F., Hickok G., Poeppel D. and Stevenson R.A., produced research work that were highly cited over the period from 2000 to 2020.

FIGURE 4

Figure 4. Yearly publications and citations for the top 20 productive researchers of speech perception (2000–2020). The size and the darkness of the circles are in proportion to the number of publications and to the total citations per year, respectively.

Research Collaborations

Given that speech perception is an interdisciplinary topic that requires teamwork across traditional disciplines, and across universities/institutes and countries, collaboration networks were constructed here to demonstrate how authors, their affiliated institutions and countries related to each other.

Figure 5 illustrates the collaboration network of the top 20 active universities/institutes. The size of each node is proportional to its degree of collaboration. One prominent mega-cluster is the triangle collaboration pathway among Radboud University Nijmegen, the Max Planck Institute for Psycholinguistics and Western Sydney University. This was driven by the well-known psycholinguist, Anne Culter, former director of the Max Planck Institute for Psycholinguistics, who worked at both places over the period. Other clusters indicate research collaborations among universities in the United States, Canada and the United Kingdom.

FIGURE 5

Figure 5. Top 20 active universities/institutes in speech perception research and their collaboration network. Each node in the network represents a different university/institute and the node’s diameter corresponds to the strength of the institute’s collaboration with others. Lines represent collaboration pathways among these universities or institutes.

Figure 6 shows the collaboration network of the top 20 collaborative countries. The United States has demonstrated a relatively strong collaboration capacity indicated by node size. Its collaboration pathways reached out to Canada, Spain, Japan and China. Another cluster connected some geographically close European countries, such as the United Kingdom, France, Germany, Belgium, Switzerland, Northway and Italy. A third pathway bridged the Netherlands and Australia, presumably for the same reason underlying the university-level collaboration as identified above between the two countries.

FIGURE 6

Figure 6. Top 20 countries’ collaboration on speech perception studies. Each node in the network represents a different nation and the node’s diameter corresponds to the strength of a nation’s collaboration with other countries. Lines represent collaboration pathways between countries.

Impactful Research Work and Key Research Themes

In the previous sections, we have identified key publication venues, leading researchers, universities and countries as well as their collaboration pathways. Next, we first uncover impactful research work and researchers via citation analysis and then identify major research themes/trends via citation network analysis and term frequency analysis of keywords, titles and abstracts as well as their co-occurrence patterns.

Table 5 shows high impact research as indicated by citations based on the articles in our data set as well as their cited references. The paper with the highest citation (McGurk and Macdonald, 1976), for example, reported the multisensory illusion in audio-visual speech perception, later also called the McGurk effect and the paradigm they used in the paper has a substantial influence on later research. Research item 3 (Sumby and Pollack, 1954) was a similar and even earlier attempt at examining visual contribution to speech perception, which found that the visual contribution to speech intelligibility negatively correlated with the speech-to-noise ratio. Both phenomena reflect the multi-modal nature of speech perception, which is an important theoretical issue.

TABLE 5

Table 5. Top 20 highly cited research articles.

On the other hand, Table 6 shows highly cited researchers as identified via citation analysis. For example, Naatanen R. (researcher 5) is well-recognised for his work on neurocognitive mechanisms underlying speech perception as revealed by the mismatch negativity (MMN), an auditory event-related brain potential that is elicited by discriminable changes in regular auditory input (Näätänen, 2001; Näätänen et al., 2005, 2007, 2014).

TABLE 6

Table 6. Top 20 highly cited authors in speech perception studies.

A close look at the articles in Table 5 and some articles of researchers identified in Table 6, for example Werker J.F., Hickok G., Liberman A.M., reveals several important theoretical approaches to speech perception. Thus, we further synthesised the two tables to provide a theory-driven and coherent account of the results and, in doing so, highlighted some work or researchers that originally ranked low in each table.

First of all, the motor theory (Liberman et al., 1967), whose founder is identified as researcher 4 in Table 6, argues that the perceptual primitives in speech perception are not acoustic cues, but neural commands to articulators, or more recently, intended gestures in one’s mind (Liberman and Mattingly, 1985, research item 5 in Table 5). Listeners reconstruct talkers’ intended gestures, which are not susceptible to phonetic variations, from the speech signal. In this way, the motor theory naturally links speech perception and production and resolves the problem of lack of invariants between phonemes and acoustic signals.

An alternative theoretical approach to speech perception was proposed by Carol Fowler (see Table 6, researcher 17), that is the direct realist approach (Fowler, 1989; Best, 1995). The theory shares with the motor theory the view that perceptual targets are gestural in nature but argues that the actual gestures rather than the intended gestures are directly perceived. The gestural information correlates with acoustic patterns via the principles of acoustic physics. This implies that there is no need for mental representation and consequently no need for perceptual normalisation when dealing with variability in speech. In contrast to the motor theory which argues for a domain-specific mechanism, the direct realist approach contends that speech perception is domain-general, just like perception of other events in the world.

Apart from fundamental theoretical constructs per se, one essential theoretical issue from a developmental perspective is to account for how our efficiency in perceiving native language is obtained at the cost of losing sensitivities to speech sounds in other languages (Werker and Tees, 1984, research item 2 in Table 5 and researcher 2 in Table 6). One line of research that examines this issue aims at identifying the time point when babies are attuned by their ambient linguistic environments and lose sensitivity to non-native speech categories. For example, in one of the pioneering studies, 6-month-old infants in the United States and Sweden showed native language effects on speech perception (Kuhl et al., 1992).

Another line of research focuses on native language influences on perceiving non-native speech by adults. Several theoretical models (that account for such influences) and/or their founders were identified in Table 5 and/or Table 6. The Perceptual Assimilation Model (PAM: Best, 1995), as indicated in research item 13 in Table 5 and researcher 13 in Table 6, posits that adult listeners perceive non-native segments in an unfamiliar language according to their similarities to or discrepancies from the native gestural constellations that are closest to them in native phonological space. The model is built on the direct realist approach to speech perception and has been validated via a number of studies on non-native perception of consonants (Best et al., 2001), vowels (Tyler et al., 2014; Faris et al., 2018) and lexical tones (Chen et al., 2020).

The Speech Learning Model (SLM: Flege, 1995, research item 15 in Table 5 and researcher 8 in Table 6), accounts for native-language-induced perceptual difficulties via two mechanisms: (1) equivalence classification, in which distinct non-native phones are categorised to a single native category; (2) native language filtering, where features of non-native phones that are important phonetically but not phonologically, that is forming no meaningful contrasts, are filtered out.

A third model, the Native Language Magnet model (NLM: Kuhl, 1994; Kuhl and Iverson, 1995), whose founder ranked first in Table 6, posits that native language experience alters perceived distances in the acoustic space underlying phonetic categories and consequently influences non-native perception and production. The phonetic prototype, that is the ideal instance of a phonetic category, developed via native language experience, acts as a ‘perceptual magnet’ for other tokens in the category. It attracts those tokens towards itself by reducing the discrimination sensitivity as compared with non-prototypic tokens of the same category.

The most recent model is the Second Language Linguistic Perception model, or L2LP (Escudero, 2005), though relevant papers of which were not identified in Table 5 nor Table 6 because the two tables based on citations were biased towards earlier theories. Its founder, Paola Escudero, was among the top 20 most productive authors in Table 4. L2LP accounts for second language speech learning from the initial state to ultimate attainment. For the initial state, L2LP claims that non-native listeners rely on optimal perception, a perception grammar attuned by their native language. In other words, listeners initially perceive non-native phones in line with the acoustic features of their native language, called the Full Copying hypothesis. The model also contends that there is a direct link between perception and production and that perception of both native and non-native phones should match the acoustic properties of phones in participants’ native language/dialects (Escudero and Vasiliev, 2011).

These models all aim at explaining native language influences on non-native speech perception but with different theoretical foci. PAM, SLM and L2LP propose ways that non-native phones can be assimilated into native categories but only PAM and L2LP make explicit predictions concerning discrimination of contrasts based on assimilation patterns. SLM focuses more on perception and production of individual non-native phones, whereas NLM is concerned more about non-native phones that are assimilated as prototypical or non-prototypical tokens within a given native category and thus does not include predictions on contrasts of non-native phones that cross native phonological boundaries.

The evaluation of research impact based on total citations biases towards senior researchers and early publications. However, this fits our aim to find the theoretically foundational research and/or their founders. The identified theories are generally in line with findings of previous qualitative reviews (Diehl et al., 2004; Samuel, 2011; Antoniou, 2018), supporting the validity of our approach. Nonetheless, to compensate for the bias, we used citation network plots to visualise different citation relations, for example co-citation or bibliographic coupling, the latter of which reveals more about impactful emerging research.

When two articles are both cited in a third article, they are co-cited, forming a co-citation relation (Aria and Cuccurullo, 2017). Articles are co-cited when they contribute to similar or different aspects of a certain topic or paradigm. For example, in cross-language speech perception research, SLM, PAM, NLM and L2LP-related papers are often co-cited to lay out theoretical foundations and to form competing predictions. We constructed a co-citation network in which each node represented a research article and only the top 30 research articles were shown (Figure 7). The size of each node is proportional to the degree of co-citation. Clusters were found with the default algorithm in the bibliometrix package (Aria and Cuccurullo, 2017). Centrality statistics of each node can be found in the Supplementary Material (Supplementary Table A).

FIGURE 7

Figure 7. Co-citation network of speech perception research. Nodes stand for research articles. For each node, the label colour is the same as its cluster and its size is proportional to its co-citation degree.

Figure 7 illustrates co-citation relations in speech perception research. Four clusters were identified. First, research in the blue cluster was centred around the cortical organisation of speech processing. For example, the authors (Hickok and Poeppel, 2007, also identified as research item 4 in Table 5 and Hickok was identified as researcher 3 in Table 6), argued for a dual-stream model of speech processing, in which a ventral stream processes speech signals for comprehension and a dorsal stream maps acoustic speech signals to frontal lobe articulatory networks. The ventral stream is largely bilaterally organised whereas the dorsal stream is strongly left-hemisphere dominant.

Second, research in the green cluster was grouped around the TRACE model (McClelland and Elman, 1986), in which speech perception is simulated through excitatory and inhibitory interactions of processing units or ‘the Trace’ at feature, phoneme and word levels. In terms of perceptual targets, the model used acoustic dimensions of speech as its inputs, which is different from the motor theory and the direct realist approach. Unlike the two pure theoretical/psychological approaches, the TRACE model, representing a computation approach to the issue, considered both computational and psychological adequacy. That is the model can be programmed to recognise real speech and account for human performance in speech perception.

In addition, the red cluster revolved around audio-visual perception research (e.g., McGurk and Macdonald, 1976). The purple cluster of research focused on how native language development attunes speech perception (e.g., Werker and Tees, 1984).

Another citation relation is bibliographic coupling. Two articles are bibliographically coupled if at least one cited source appears in the references of both articles (Kessler, 1963). Some researchers argued that bibliographic coupling analysis better reflects unique information than co-citation when constructing citation networks (Kleminski et al., 2020). While co-citation analysis reveals more about relationships among older papers, bibliographic coupling analysis reflects more about the current research front. For this reason, the network (see Figure 8) based on bibliographic coupling relations was constructed here. Similar to the co-citation network, each node represented a research article and only the top 30 research articles were shown. The size of each node is proportional to the degree of bibliographic coupling (Aria and Cuccurullo, 2017). Clusters were found with the default algorithm in the bibliometrix package. Centrality statistics of each node can also be found in the Supplementary Material (Supplementary Table B).

FIGURE 8

Figure 8. Bibliographic coupling network of speech perception research. Nodes stand for research articles. For each node, the label colour is the same as its cluster and its size is proportional to its bibliographic coupling degree.

Three clusters of research were identified. First, research in the red cluster focused on developmental aspects of speech perception (Werker and Curtin, 2005; Choi et al., 2018; Werker, 2018). In the two review papers (Choi et al., 2018; Werker, 2018), the authors examined how infants’ universal sensitivities to human speech are attuned to their native language and specifically emphasised the multisensory nature of this process.

Second, research in the blue cohort concentrated on studies of cognitive processes in speech perception, such as audio-visual speech perception with application of the state-of-art brain science technologies (Sato et al., 2010; Irwin et al., 2017). For example, Irwin et al. (2017) tested typical developing children’s ability to detect the missing segment in spoken syllables with and without visual information and measured their behaviour and neural activities. They found that visual information attenuated the brain responses, suggesting that children were less sensitive to the missing segment when visual information of a face producing that segment was presented.

Third, research in the green cluster examined general speech perception issues, such as the lack of invariance problem (Kleinschmidt and Jaeger, 2015) or the perception–production link in speech learning (Baese-Berk, 2019).

Although co-citation and bibliographic coupling networks can reveal central research themes or topics, they require subjective interpretation of the clusters. To compensate for this disadvantage, term frequency analysis and co-word analysis were employed, the results of which can be directly interpreted according to the meanings of the words/phrases.

Assuming that terms with high frequency indicate important themes in the data set, term frequency can be used to identify important themes of a field (Khasseh et al., 2017; Zhao et al., 2018). We first analysed the frequency of author-defined keywords and Web of Science (WOS)-generated keywords.

In addition, bigram analysis, that is two-word combinations, was applied to the titles and abstracts of the articles. When extracting bigrams, we used the unnest_tokens function in the tidytext package (Silge and Robinson, 2016) in R and filtered package-defined stop words as they either occur in the first or second part of the bigram.

Table 7 shows the top 20 frequent terms. The most prominent research theme concerns speech perception in the clinical context with participants, such as children suffering from hearing loss or (developmental) dyslexia and the effects of cochlear implantation. Another line of research focuses on examining psycholinguistic issues, for example working memory, attention, spoken word recognition, as well as neurocognitive mechanisms underlying speech perception using brain science techniques, such as functional magnetic resonance imaging (fMRI) or electroencephalogram (EEG, including event-related brain potentials or ERPs).

TABLE 7

Table 7. Top 20 terms from author-defined keywords, machine-generated keywords, title-based and abstract-based two-word phrases.

In addition to term frequency, co-word networks were used to reveal both frequency and co-occurrence relationships among keywords. Co-word networks were constructed based on author-defined keywords (Figure 9). Keywords, such as cochlear implants or hearing loss, co-occur frequently with children, suggesting that this line of research focuses on hearing loss children in particular and examines the effect of cochlear implantation.

FIGURE 9

Figure 9. Author-defined keywords co-occurrence network.

Another group of frequently co-occurring keywords again indicates the application of neuroscience technologies, such as EEG, fMRI or magnetoencephalography, in investigating brain activities during speech processing.

General Discussion

In this section, we compare the findings of the present bibliometric review with previous qualitative reviews to provide a balanced discussion. First, our analysis revealed some important aspects of the field that were ignored by traditional qualitative reviews, such as major publication venues, research productivity and research collaboration.

Second, our analysis successfully detected major theoretical approaches in speech perception that were selected in previous qualitative reviews. These reviews featured in detailed descriptions of the theories with in-depth discussions of relevant empirical studies.

At the sublexical level, three theoretical approaches were introduced in two qualitative reviews (Diehl et al., 2004; Samuel, 2011), that is the motor theory, the direct realism and the general auditory approach, the former two of which were identified in our analysis. The third approach was not identified, presumably because it encompasses a number of theories in different publications (Diehl and Kluender, 1989; Ohala, 1996; Sussman et al., 1998). All of these theories shared that speech is perceived via the same auditory mechanisms involved in perceiving non-speech sounds and listeners recover messages from the acoustic signal without referencing it to articulatory gestures. We acknowledge that compared with experienced researchers in the field, the bibliometric analysis lacks the power of identifying certain research cohorts that used different terminologies for a similar theoretical approach.

In addition, Samuel (2011) discussed five phenomena/research paradigms relevant to foundational speech perception theories, that is categorical perception (Liberman et al., 1967), the right ear advantage (i.e., an advantage for perceiving speech played to the right ear, Shankweiler and Studdert-Kennedy, 1967), trading relations (i.e., multiple acoustic cues can trade off against each other in speech perception, Dorman et al., 1977), duplex perception (i.e., listeners perceive both non-speech and speech percept when acoustic cues of a speech sound are decomposed and presented separately to each ear, Rand, 1974) and compensation for coarticulation (i.e., perceptual boundaries are shifted to compensate for the coarticulation induced phonetic variations, Mann and Repp, 1981). These phenomena provided support or tests in the development of different theoretical approaches. Our analysis detected the categorical perception paradigm as it is closely related to the motor theory.

At the lexical and higher level, Samuel compared several models that account for how perceived phonetic features result in recognition of spoken words and/or utterances. All the models shared that speech perception involves activation of lexical representations via sublexical features, which was originally proposed by the Cohort model (Marslen-Wilson, 1975, 1987). However, they differed in whether the process is entirely bottom-up, as stated in the Merge model (Norris et al., 2000) and the fuzzy logical model (Massaro, 1989), or interactive, as maintained by the TRACE model (McClelland and Elman, 1986) and the Adaptive Resonance Theory (Grossberg, 1980). The TRACE model was identified in our co-citation analysis as an important research cohort and Norris D., who is the founder of a series of models including the Merge model, was identified in the citation analysis as an impactful researcher.

Three lexical-level phenomena were also discussed in Samuel’s review: the phonemic restoration effect (i.e., listeners’ ability to perceptually restore the missing phoneme, Warren, 1970), the McGurk effect (McGurk and Macdonald, 1976) and the Ganong effect (i.e., the bias of perceiving an ambiguous segment as a sound that forms a word, Ganong, 1980). In our analysis, we identified the original paper that reported the McGurk effect as an impactful research work and this topic as a research cohort.

Although Samuel’s review covered statistical learning and perceptual learning by infants and adults, the author did not discuss second language speech learning models. Another review with a focus on bilingual speech perception (Antoniou, 2018) discussed two speech learning models, that is SLM and PAM/PAM-L2. Both models were identified in our analysis, and we detected two additional models, NLM and L2LP.

It is important to reiterate here that our bibliometric review aims to complement, but not to replace, the existing qualitative reviews. Therefore, qualitative and bibliometric reviews should be combined to provide a comprehensive view of a field.

Conclusion

In this study, a bibliometric analysis of journal articles published between 2000 and 2020 was conducted to map speech perception research. Major publication venues, general trends in research productivity, leading countries, research institutes and researchers were identified. Given that speech perception is an interdisciplinary field that requires research collaboration across traditional disciplinary boundaries, we did a research collaboration analysis, which indicated that researchers in the United States tended to collaborate more with researchers based in Canada, Spain, Japan and China, whereas geographically close European countries, such as the United Kingdom, France, Germany, Belgium, Switzerland, Northway and Italy, tended to collaborate with each other.

Analysis of research articles and researchers with high citations revealed three impactful theoretical approaches to speech perception, that is the motor theory, the direct realist approach and the computational approach, and four influential models that address key issues in non-native speech perception, that is SLM, PAM, NLM, and L2LP. Citation networks and term frequency analyses of keywords, titles and abstracts identified important research issues, such as cortical organisation of speech processing, audio-visual speech perception, spoken word recognition and infant/child speech learning.

Two directions for future research are (1) speech perception across the lifespan, for example infants and aging populations, or clinical populations, such as children with hearing loss or developmental dyslexia, and (2) application of neurocognitive techniques in understanding brain regions involved in speech perception and the time course of brain activities in speech processing.

Data Availability Statement

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found at: https://osf.io/5hxg2/?view_only=42ccbe057ad84faca0e2560e4c0e36c3.

Author Contributions

JC and HC: conceptualisation and writing—review and editing. JC: methodology, software, writing—original draft preparation, and visualisation. HC: supervision and funding acquisition. All authors contributed to the article and approved the submitted version.

Conflict of Interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Publisher’s Note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary Material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpsyg. 2022.822241/full#supplementary-material

Footnotes

1. ^Some research that we analysed examined processing talker variability in speech to perceive linguistic categories but talker voice recognition, which is a related but different topic, is beyond the scope of this study.

2. ^The Web of Science Core Collection provides full records, including cited references, which are essential for some bibliometric analysis, such as co-citation analysis. It has been widely used in conducting bibliometric analysis research. Other databases, which do not include full records or are not supported by the bibliometrix package, are not considered.

References

Antoniou, M. (2018). “Speech Perception,” in The Listening Bilingual. eds. F. Grosjean and K. Byers-Heinlein (Hoboken, NJ, USA: John Wiley & Sons, Inc.), 41–64.

Google Scholar

Aria, M., and Cuccurullo, C. (2017). Bibliometrix: An R-tool for comprehensive science mapping analysis. J. Informet. 11, 959–975. doi: 10.1016/j.joi.2017.08.007

CrossRef Full Text | Google Scholar

Baese-Berk, M. M. (2019). Interactions between speech perception and production during learning of novel phonemic categories. Atten. Percept. Psychophys. 81, 981–1005. doi: 10.3758/s13414-019-01725-4

PubMed Abstract | CrossRef Full Text | Google Scholar

Best, C. T. (1995). “A Direct Realist Perspective on Cross-Language Speech Perception,” in Speech Perception and Linguistic Experience: Theoretical and Methodological Issues in Cross-language Speech Research. ed. W. Strange (Timonium, MD: York Press), 167–200.

Google Scholar

Best, C., McRoberts, G. W., and Goodell, E. (2001). Discrimination of non-native consonant contrasts varying in perceptual assimilation to the listener’s native phonological system. J. Acoust. Soc. Am. 109, 775–794. doi: 10.1121/1.1332378

PubMed Abstract | CrossRef Full Text | Google Scholar

Chen, J., Best, C. T., and Antoniou, M. (2020). Native phonological and phonetic influences in perceptual assimilation of monosyllabic Thai lexical tones by mandarin and Vietnamese listeners. J. Phon. 83:101013. doi: 10.1016/j.wocn.2020.101013

CrossRef Full Text | Google Scholar

Choi, D., Black, A. K., and Werker, J. F. (2018). Cascading and multisensory influences on speech perception development. Mind Brain Educ. 12, 212–223. doi: 10.1111/mbe.12162

CrossRef Full Text | Google Scholar

Crump, M. J. C., McDonnell, J. V., and Gureckis, T. M. (2013). Evaluating Amazon’s mechanical Turk as a tool for experimental behavioral research. PLoS One 8:e57410. doi: 10.1371/journal.pone.0057410

PubMed Abstract | CrossRef Full Text | Google Scholar

Demir, Y., and Kartal, G. (2022). Mapping research on L2 pronunciation: a bibliometric analysis. Stud. Second. Lang. Acquis., 1–29. doi: 10.1017/S0272263121000966

CrossRef Full Text | Google Scholar

Diehl, R. L., and Kluender, K. R. (1989). On the objects of speech perception. Ecol. Psychol. 1, 121–144. doi: 10.1207/s15326969eco0102_2

CrossRef Full Text | Google Scholar

Diehl, R. L., Lotto, A. J., and Holt, L. L. (2004). Speech perception. Annu. Rev. Psychol. 55, 149–179. doi: 10.1146/annurev.psych.55.090902.142028

CrossRef Full Text | Google Scholar

Dorman, M. F., Studdert-Kennedy, M., and Raphael, L. J. (1977). Stop-consonant recognition: release bursts and formant transitions as functionally equivalent, context-dependent cues. Percept. Psychophys. 22, 109–122. doi: 10.3758/BF03198744

CrossRef Full Text | Google Scholar

Escudero, P. (2005). Linguistic Perception and second Language Acquisition: Explaining the Attainment of Optimal Phonological Categorization. The Netherlands: Netherlands Graduate School of Linguistics.

Google Scholar

Escudero, P., and Vasiliev, P. (2011). Cross-language acoustic similarity predicts perceptual assimilation of Canadian English and Canadian French vowels. J. Acoust. Soc. Am. 130, EL277–EL283. doi: 10.1121/1.3632043

PubMed Abstract | CrossRef Full Text | Google Scholar

Faris, M. M., Best, C., and Tyler, M. D. (2018). Discrimination of uncategorised non-native vowel contrasts is modulated by perceived overlap with native phonological categories. J. Phon. 70, 1–19. doi: 10.1016/j.wocn.2018.05.003

CrossRef Full Text | Google Scholar

Flege, J. E. (1995). “Second-language speech learning: theory, findings, and problems,” in Speech Perception and Linguistic Experience: Issues in Cross-Language Research. ed. W. Strange (Timonium, MD: York Press), 229–273.

Google Scholar

Fowler, C. A. (1989). Real objects of speech perception: A commentary on Diehl and Kluender. Ecol. Psychol. 1, 145–160. doi: 10.1207/s15326969eco0102_3

CrossRef Full Text | Google Scholar

Fu, Y., Wang, H., Guo, H., Bermúdez-Margaretto, B., and Domínguez Martínez, A. (2020). What, where, when and how of visual word recognition: A Bibliometrics review. Lang. Speech 64, 900–929. doi: 10.1177/0023830920974710

PubMed Abstract | CrossRef Full Text | Google Scholar

Ganong, W. F. (1980). Phonetic categorization in auditory word perception. J. Exp. Psychol. Hum. Percept. Perform. 6, 110–125. doi: 10.1037//0096-1523.6.1.110

CrossRef Full Text | Google Scholar

Grootswagers, T. (2020). A primer on running human behavioural experiments online. Behav. Res. 52, 2283–2286. doi: 10.3758/s13428-020-01395-3

PubMed Abstract | CrossRef Full Text | Google Scholar

Grossberg, S. (1980). How does a brain build a cognitive code? Psychol. Rev. 87, 1–51. doi: 10.1037/0033-295X.87.1.1

PubMed Abstract | CrossRef Full Text | Google Scholar

Hawkins, S. (2004). “Puzzles and Patterns in 50 Years of Research on Speech Perception,” in From Sound to Sense: 50+ Years of Discoveries in Speech Communication. eds. J. Slifka, S. Manuel, J. Perkell, and S. Hufnagel (MA, USA: MIT, Cambridge), B223–B246.

Google Scholar

Hickok, G., and Poeppel, D. (2007). The cortical organization of speech processing. Nat. Rev. Neurosci. 8, 393–402. doi: 10.1038/nrn2113

CrossRef Full Text | Google Scholar

Irwin, J., Avery, T., Turcios, J., Brancazio, L., Cook, B., and Landi, N. (2017). Electrophysiological indices of audiovisual speech perception in the broader autism phenotype. Brain Sci. 7:60. doi: 10.3390/brainsci7060060

PubMed Abstract | CrossRef Full Text | Google Scholar

Kessler, M. M. (1963). Bibliographic coupling between scientific papers. Am. Doc. 14, 10–25. doi: 10.1002/asi.5090140103

CrossRef Full Text | Google Scholar

Khasseh, A. A., Soheili, F., Moghaddam, H. S., and Chelak, A. M. (2017). Intellectual structure of knowledge in iMetrics: A co-word analysis. Inf. Process. Manag. 53, 705–720. doi: 10.1016/j.ipm.2017.02.001

CrossRef Full Text | Google Scholar

Kimura, D. (1961). Cerebral dominance and the perception of verbal stimuli. Canadian J. Psychol. 15, 166–171. doi: 10.1037/h0083219

CrossRef Full Text | Google Scholar

Kleinschmidt, D. F., and Jaeger, T. F. (2015). Robust speech perception: recognize the familiar, generalize to the similar, and adapt to the novel. Psychol. Rev. 122, 148–203. doi: 10.1037/a0038695

PubMed Abstract | CrossRef Full Text | Google Scholar

Kleminski, R., Kazienko, P., and Kajdanowicz, T. (2020). Analysis of direct citation, co-citation and bibliographic coupling in scientific topic identification. J. Inf. Sci. 48, 349–373. doi: 10.1177/0165551520962775

CrossRef Full Text | Google Scholar

Kuhl, P. K. (1994). Learning and representation in speech and language. Curr. Opin. Neurobiol. 4, 812–822. doi: 10.1016/0959-4388(94)90128-7

CrossRef Full Text | Google Scholar

Kuhl, P. K., and Iverson, P. (1995). “Linguistic experiencce and the ‘Perceptual Magnet Effect’,” in Speech Perception and Linguistic Experience: Issues in Cross-language Research. ed. W. Strange (Timonium, MD: York Press), 121–154.

Google Scholar

Kuhl, P., Williams, K., Lacerda, F., Stevens, K., and Lindblom, B. (1992). Linguistic experience alters phonetic perception in infants by 6 months of age. Science 255:606, –608. doi: 10.1126/science.1736364

PubMed Abstract | CrossRef Full Text | Google Scholar

Lei, L., and Liao, S. (2017). Publications in linguistics journals from mainland China, Hong Kong, Taiwan, and Macau (2003–2012): A Bibliometric analysis. J. Quant. Ling. 24, 54–64. doi: 10.1080/09296174.2016.1260274

CrossRef Full Text | Google Scholar

Lei, L., and Liu, D. (2019). Research trends in applied linguistics from 2005 to 2016: A Bibliometric analysis and its implications. Appl. Linguis. 40, 540–561. doi: 10.1093/applin/amy003

CrossRef Full Text | Google Scholar

Liberman, A. M., Cooper, F. S., Shankweiler, D. P., and Studdert-Kennedy, M. (1967). Perception of the speech code. Psychol. Rev. 74, 431–461. doi: 10.1037/h0020279

CrossRef Full Text | Google Scholar

Liberman, A. M., and Mattingly, I. G. (1985). The motor theory of speech perception revised. Cognition 21, 1–36. doi: 10.1016/0010-0277(85)90021-6

PubMed Abstract | CrossRef Full Text | Google Scholar

Lin, Z., and Lei, L. (2020). The research trends of multilingualism in applied linguistics and education (2000–2019): A Bibliometric analysis. Sustainability 12:6058. doi: 10.3390/su12156058

CrossRef Full Text | Google Scholar

Mann, V. A., and Repp, B. H. (1981). Influence of preceding fricative on stop consonant perception. J. Acoust. Soc. Am. 69, 548–558. doi: 10.1121/1.385483

PubMed Abstract | CrossRef Full Text | Google Scholar

Marslen-Wilson, W. D. (1975). Sentence perception as an interactive parallel process. Science 189, 226–228. doi: 10.1126/science.189.4198.226

PubMed Abstract | CrossRef Full Text | Google Scholar

Marslen-Wilson, W. D. (1987). Functional parallelism in spoken word-recognition. Cognition 25, 71–102. doi: 10.1016/0010-0277(87)90005-9

PubMed Abstract | CrossRef Full Text | Google Scholar

Massaro, D. W. (1989). Testing between the TRACE model and the fuzzy logical model of speech perception. Cogn. Psychol. 21, 398–421. doi: 10.1016/0010-0285(89)90014-5

PubMed Abstract | CrossRef Full Text | Google Scholar

McClelland, J. L., and Elman, J. L. (1986). The TRACE model of speech perception. Cogn. Psychol. 18, 1–86. doi: 10.1016/0010-0285(86)90015-0

CrossRef Full Text | Google Scholar

McGurk, H., and Macdonald, J. (1976). Hearing lips and seeing voices. Nature 264, 746–748. doi: 10.1038/264746a0

PubMed Abstract | CrossRef Full Text | Google Scholar

Näätänen, R. (2001). The perception of speech sounds by the human brain as reflected by the mismatch negativity (MMN) and its magnetic equivalent (MMNm). Psychophysiology 38, 1–21. doi: 10.1111/1469-8986.3810001

PubMed Abstract | CrossRef Full Text | Google Scholar

Näätänen, R., Jacobsen, T., and Winkler, I. (2005). Memory-based or afferent processes in mismatch negativity (MMN): A review of the evidence. Psychophysiology 42, 25–32. doi: 10.1111/j.1469-8986.2005.00256.x

PubMed Abstract | CrossRef Full Text | Google Scholar

Näätänen, R., Paavilainen, P., Rinne, T., and Alho, K. (2007). The mismatch negativity (MMN) in basic research of central auditory processing: A review. Clin. Neurophysiol. 118, 2544–2590. doi: 10.1016/j.clinph.2007.04.026

PubMed Abstract | CrossRef Full Text | Google Scholar

Näätänen, R., Sussman, E. S., Salisbury, D., and Shafer, V. L. (2014). Mismatch negativity (MMN) as an index of cognitive dysfunction. Brain Topogr. 27, 451–466. doi: 10.1007/s10548-014-0374-6

PubMed Abstract | CrossRef Full Text | Google Scholar

Norris, D., McQueen, J. M., and Cutler, A. (2000). Merging information in speech recognition: feedback is never necessary. Behav. Brain Sci. 23, 299–325. doi: 10.1017/S0140525X00003241

PubMed Abstract | CrossRef Full Text | Google Scholar

Ohala, J. J. (1996). Speech perception is hearing sounds, not tongues. J. Acoust. Soc. Am. 99, 1718–1725. doi: 10.1121/1.414696

CrossRef Full Text | Google Scholar

R Core Team (2018). R: A language and environment for statistical computing. Vienna, Austria Available at: https://www.R-project.org/.

Google Scholar

Rand, T. C. (1974). Dichotic release from masking for speech. J. Acoust. Soc. Am. 55, 678–680. doi: 10.1121/1.1914584

PubMed Abstract | CrossRef Full Text | Google Scholar

Samuel, A. G. (2011). Speech perception. Annu. Rev. Psychol. 62, 49–72. doi: 10.1146/annurev.psych.121208.131643

CrossRef Full Text | Google Scholar

Sato, M., Buccino, G., Gentilucci, M., and Cattaneo, L. (2010). On the tip of the tongue: modulation of the primary motor cortex during audiovisual speech perception. Speech Comm. 52, 533–541. doi: 10.1016/j.specom.2009.12.004

CrossRef Full Text | Google Scholar

Schreiber, M. (2008). An empirical investigation of the g-index for 26 physicists in comparison with the h-index, the A-index, and the R-index. J. Am. Soc. Inf. Sci. Technol. 59, 1513–1522. doi: 10.1002/asi.20856

CrossRef Full Text | Google Scholar

Shankweiler, D., and Studdert-Kennedy, M. (1967). Identification of consonants and vowels presented to left and right ears*. Q. J. Exp. Psychol. 19, 59–63. doi: 10.1080/14640746708400069

PubMed Abstract | CrossRef Full Text | Google Scholar

Silge, J., and Robinson, D. (2016). Tidytext: text mining and analysis using tidy data principles in r. JOSS 1:37. doi: 10.21105/joss.00037

CrossRef Full Text | Google Scholar

Stevens, K. N. (1972). “The Quantal Nature of Speech: Evidence from Articulatory-Acoustic Data,” in Human Communication: A Unified View. eds. E. E. David and P. B. Denes (New York: McGraw-Hill), 51–66.

Google Scholar

Sumby, W. H., and Pollack, I. (1954). Visual contribution to speech intelligibility in noise. J. Acoust. Soc. Am. 26, 212–215. doi: 10.1121/1.1907309

CrossRef Full Text | Google Scholar

Sussman, H. M., Fruchter, D., Hilbert, J., and Sirosh, J. (1998). Linear correlates in the speech signal: The orderly output constraint. Behav. Brain Sci. 21, 241–259. doi: 10.1017/S0140525X98001174

PubMed Abstract | CrossRef Full Text | Google Scholar

Tyler, M. D., Best, C., Faber, A., and Levitt, A. G. (2014). Perceptual assimilation and discrimination of non-native vowel contrasts. PHO 71, 4–21. doi: 10.1159/000356237

PubMed Abstract | CrossRef Full Text | Google Scholar

Warren, R. M. (1970). Perceptual restoration of missing speech sounds. Science 167, 392–393. doi: 10.1126/science.167.3917.392

PubMed Abstract | CrossRef Full Text | Google Scholar

Werker, J. F. (2018). Perceptual beginnings to language acquisition. Appl. Psycholinguist. 39, 703–728. doi: 10.1017/S0142716418000152

CrossRef Full Text | Google Scholar

Werker, J. F., and Curtin, S. (2005). PRIMIR: A developmental framework of infant speech processing. Lang. Learn. Dev. 1, 197–234. doi: 10.1080/15475441.2005.9684216

CrossRef Full Text | Google Scholar

Werker, J. F., and Tees, R. C. (1984). Cross-language speech perception: evidence for perceptual reorganization during the first year of life. Infant Behav. Dev. 7, 49–63. doi: 10.1016/S0163-6383(84)80022-3

CrossRef Full Text | Google Scholar

Zhao, W., Mao, J., and Lu, K. (2018). Ranking themes on co-word networks: exploring the relationships among different metrics. Inf. Process. Manag. 54, 203–218. doi: 10.1016/j.ipm.2017.11.005

CrossRef Full Text | Google Scholar