Employee voice: a knowledge map to provide conceptual clarity and future research directions

The increasing globalisation, the multiculturality of workplaces and the current challenges for organisations generated the need for researchers to support them to manage the workforce. Although the relevance of employee voice for these themes, a lack of shared understanding about this topic results in fragmented literature across and within research fields that limits theoretical advancement and deep comprehension of the phenomenon. Our first aim is to offer a literature review of employee voice by combining systematic and bibliometric methods; the second aim is to understand voice ’ s main issues and implications by considering different research streams. The results present an integrated framework of the leading intellectual knowledge and reveal the main research focuses on voice in domestic contexts. The discussion underlines the cultural issue and context as critical elements for future research by proposing avenues for scholars and some implications for organisations to benefit from the contributions of their members


Introduction
There is a general consensus that allowing workers to speak out can bring relevant issues to light and thus contribute to problem-solving, organisational growth and performance improvement (Bashshur & Oc, 2015;Chamberlin, Newton, & Lepine, 2017;Kim, MacDuffie, & Pil, 2010). Employee voice, defined as any opportunity and way through which employees express ideas and opinions within the organisation Wilkinson & Fay, 2011), stands as a complex phenomenon, often investigated by disciplines that do not conversate with each other (Barry & Wilkinson, 2022). In the organisational behaviour (OB) literature, employee voice refers to discretionary behaviours, which manifests itself in face-to-face contexts through direct informal voice; in human resource management (HRM) literature, scholars refer to voice as formal expressions, distinguished in indirect and direct voice (Barry & Wilkinson, 2016;Mowbray, Wilkinson, & Tse, 2015); finally, in the industrial relations (IR) field, voice is seen from a collective point of view to represent employees' interests (Freeman & Medoff, 1984).
The growing interest in the study of employee voice has culminated in a compartmentalised and exponential growth of publications, themes, and debates (Barry & Wilkinson, 2022;Wilkinson, Barry, & Morrison, 2020). The fragmentation of the literature, based on different conceptualisations and related focus, makes the conversation between the disciplines difficult, failing thus to develop a broad research agenda and practical contributions due to the partial vision of the entire story on voice. Consequently, research fails to support organisations in hearing employee voices and benefit from these in decision improvement and organisational innovation (Barry & Wilkinson, 2016;Wilkinson, Gollan, Kalfa, & Xu, 2018). In addition, employee voice is critical for overcoming problems in organisations (Morrison, 2014), and could be vital to manage the current phenomena and the challenges for organisations (Cohen & Roeske-Zummer, 2021;Hodge, 2022), such as the great resignation, the trend of voluntary and mass employees' resignation from their jobs (Curtis, 2021), or the quiet quitting, where employees fulfil their job duties without subscribing to 'work is life' culture to guide their career and development (Warrick & Cady, 2022).
With our study we aim to answer the call for the need for systematisation of the bodies of literature that investigated employee voice, by adopting a holistic approach that integrates separate and diverse scholarly contributions (Wilkinson, Barry, & Morrison, 2020). To answer this call, the article offers a systematic review of the employee voice literature through the application of complementary bibliometric investigation, which supports the systematisation and integration of previous studies to advance our knowledge of employee voice. We differentiate from preceding reviews on the subject (Bashshur & Oc, 2015;Klaas, Olson-Buchanan, & Ward, 2012;Mowbray, Wilkinson, & Tse, 2015) by adopting the contemporary approach to reviewing literature through bibliometric methods, which have contributed to connecting multiple streams of research so far treated separately.
This study provides methodological, theoretical and practical contributions. First, the method adopted for reviewing the literature improves rigour and allows for in-depth review. Complementing the systematic literature review method with the quantitative bibliometric analysis enables mapping the field's knowledge structure (Dabić, Maley, Dana, Novak, Pellegrini, & Caputo, 2020;Zupic & Cater, 2015). Indeed, the bibliometric method offers a systematic and impartial way to structure literature reviews, allowing us to complement and update Bashshur & Oc (2015); Klaas, Olson-Buchanan, and Ward (2012) and Mowbray, Wilkinson, and Tse (2015) which focused on a broader field of voice. We thus contribute to the HRM, IR and OB literatures by bridging them together in a systematic way to offer a wider perspective of employee voice for research and practice. As concerns the second contribution of our research, we clarify and offer a systematisation of employee voice by identifying thematic clusters over three different research periods. Our work addresses the problem of fragmentation by sorting the phenomena into well-distinguished and manageable types or categories in support of an agenda for future research that calls for the integration of multiple disciplines to understand and investigate employee voice thoroughly. In this regard, based on the results of the systematic review, the third contribution offers a roadmap that shows the progressive development of the conceptualisation of employee voice. By doing this, we examine the paths of employee voice literature, identifying the connections between the different perspectives. Consequently, we contribute to the theoretical advancement of the field by developing a research agenda to bring forward the study of employee voice. Finally, our discussion leads us to propose relevant issues for scholars as well as some practical implications for any kind of organisations and, in particular, for international ones, to allow them to benefit from the contributions of their members. Next section introduces the method used for both the bibliometric and systematic literature review.

Method
We aimed to select papers based on the best relevance in terms of quality rating and influential capacity ( Figure 1) to map the current research. Following established practices (Baker, Pandey, Kumar, & Haldar, 2020;Sassetti, Marzi, Cavaliere, & Ciappei, 2018), the first step concerned a comprehensive search using the Web of Science (WoS) Social Science Citation Index database, business and management subject area. Then we intersected with the 2021 Academic Journal Guide chartered by the Association of Business Schools (ABS) and, to seek relevant documents, we focus on the quality of references by selecting only articles published in journals ranked 3 and above to assure the best relevance of the documents also in terms of quality rating and influential capacity (Alegre, Berbegal-Mirabent, Guerrero, & Mas-Machuca, 2018;George, Karna, & Sud, 2022). We adopted an iterative process to select the keywords for the search string in WoS platform, cross-validating results with Scopus. The best string resulted from having only the keyword 'voice' to obtain the broadest possible coverage. The first author performed the initial search at the end of February 2020 and the second search in March 2020. The search found 807 articles in WoS, compared with 779 from Scopus, ranging from 1985 until 2019.
In the second step, we defined the inclusion/exclusion criteria based on the relevance of the content of the articles. In our analysis, we decided to include all the documents that respect the definition of employee voice provided by Kwon and Farndale (2020): 'an employee behaviour aimed at suggesting organisational improvement and/or raising complaints or dissatisfaction about work-related issues through either formal or informal voice channels'. The whole team of researchers then manually selected documents to include in the analysis following the agreed-upon criteria. To ensure the best fit with the definition proposed above and preserve the process from individual biases or errors, the researchers confronted the different versions of the data set, discussing and solving any disagreements about the composition of the final data set. Some borderline cases required the researchers to intensely read the article because there was no definitive decision about inclusion/exclusion. More in detail, the assessment of the title, abstract and keywords was not enough for some articles to make a decision; other cases required a confrontation between researchers about the decision since there was no final agreement. By reading the full text for these cases to determine their relevance, the researchers thus assessed the extent to which the full text dedicated attention to voice conceptualisation and implications discussion for research and practice.
In addition, the primary and most common motivations for exclusion were related to the lack of coherence according to the definition adopted and the absence of focus on our topic of interest in the text. More in detail, as we noted above, the keyword adopted allowed the best coverage in search results leading, at the same time, led to the presence in the data set of some articles focused on different topics, such as organisational, political and strategic issues (Humphrey & Ashforth, 2000;Piderit, 2000), and discrimination debate (Goltz, 2005), which involved a different concept of voice than our research. Other articles adopted the analysis of the human voice for various research purposes (Backhaus, Meyer, & Stockert, 1985;DeGroot & Motowidlo, 1999); others indicated voice as a dimension of organisational justice, but from a deeper reading of the text emerged a lack of focus and contribution to voice literature (Husted & Folger, 2004;van den Bos, 2002). However, we excluded those articles that, based on the in-depth reading of the text content, revealed the absence of focus on the concept of voice and exploration of the related implications. In the end, the final data set was composed of 470 documents.
The third step regarded the pre-processing phase with a data set treatment for data cleaning (e.g., solving inconsistencies, such as homogenising names, titles and keywords,) using Notepad++ and Bibexcel (Persson, Danell, & Schneider, 2009). This activity was essential to ensure consistency in the data used to perform the bibliometric analysis (Andersen, 2019). More in detail, authors, journals and references appeared in various forms in the articles examined. For example, for the references by the author LePine JA, two different forms of the name were present (LePine, Jeffrey A. and LePine, J. A.), which caused two cited references; therefore, they were corrected and homogenised. In a similar way, abbreviations of titles in the cited references and different labels of cited journals were corrected. Finally, missing and not captured values by WoS were manually added.
The fourth step concerned the identification of the most important articles to include in the review by a complete bibliometric analysis of the field. The bibliometric analysis entailed: first, the study of the trends in publication, which led us to identify the three periods of evolution based on the criteria described in the following section; second, the keyword co-occurrence analysis, which offers an overview of the employee voice field, with topics of interest located with others; third, the bibliographic coupling of three equal periods to identify the knowledge structure of the field and the most important papers for the systematic review. In this research, bibliographic coupling and keyword co-occurrence complement each other: the former approach identifies the recent topics investigated; the latter approach enriches the thematic interpretation of the results derived from the bibliographic coupling of each period (Andersen, 2019;Donthu, Kumar, Mukherjee, Pandey, & Lim, 2021). At the end of this process, 165 articles were selected for the systematic literature based on the visualisation maps and the link strength indexes. The fifth step concerned the systematic review of the essential articles previously identified (Elsbach & van Knippenberg, 2020;Rojon, Okupe, & McDowall, 2021), which led to the determination of thematic clusters, extensively discussed in the results section. Table 1 summarises the number of articles included per historical period and for each cluster first in the bibliometric analysis and then in the systematic review.
At the end of the process, in step 6, we systematised the results regarding the progressive development of the conceptualisation of employee voice, and we fleshed out a knowledge map to provide conceptual clarity and future research directions. Based on this map shown in Figure 6, we examine the paths of employee voice literature within and over the evolution of periods, identifying the connections between the different perspectives and the dominance and influence which occurred in the research on employee voice. Then, we propose relevant issues for scholars as well as some practical implications for organisations. Finally, we summarised the main topics that emerged from our results and the consequent implications for future research based on Gervasi et al. (2022) and Pellegrini, Ciampi, Marzi, and Orlando (2020).

Description of the bibliometric analyses
Bibliographic coupling is a bibliometric technique that finds links of strength according to a similarity algorithm. In doing this, it allows mapping the current research on field-based references shared between two documents (Van Eck & Waltman, 2014), according to which the more reference lists are similar, the stronger the connection between two documents (Zupic & Cater, 2015). It reveals the best results when it concerns articles referred to a limited period, as citation patterns change over time (Andersen, 2019;Sassetti et al., 2018). A first analysis of the evolution of research over time by the number of publications per year ( Figure 2) indicated the existence of different periods of development. A deeper investigation led us to identify the year 1998 as the first turning point, where research on voice starts to grow significantly compared to previous years ( Figure 2). The year 1998 coincides with the publication of Van Dyne and LePine (1998), which is the most cited paper in our data set, and that several studies are considered as the foundation study for the field (Barnett, Henriques, & Husted, 2020;Danvila-del-Valle, Estévez-Mendoza, & Lara, 2019). Having identified the first turning point, we followed the established methodological prescription of dividing in equal timeframes of 10 years each (Barnett, Henriques, & Husted, 2020;Castillo-Vergara, Alvarez-Marin, & Placencio-Hidalgo, 2018;Wilden, Hohberger, Devinney, & Lumineau, 2019). The bibliometric analyses were then performed separately for each period with the software VOSviewer following Van Eck and Waltman (2014). VOSviewer uses a specific algorithm to determine the clusters ensuring no overlapping articles in more than one cluster. To construct the map, starting from a co-occurrence matrix, VOSviewer calculates a similarity matrix. The process locates items with high similarity close to each other. Then the solution map is translated, rotated and reflected to ensure consistent results (Van Eck & Waltman, 2010).
VOSviewer makes it possible to create and visualise maps working on network data and provides distance-based visualisations of bibliometric networks indicating the relatedness of the nodes. Each node is a circle of the map and is related to the content of the analysis; for instance, in keyword co-occurrence analysis, each circle of the visualisation map represents a keyword, and its size is related to the number of occurrences in the database. Similarly, in the bibliographic coupling visualisation map, each circle is a document of the database, and the size of the circle is related to the strengths of the link based on shared references that it has with other circles: the more links a document has, the larger the size of the circle. The distance between the two circles shows the similarity of references (Sassetti et al., 2018;Van Eck & Waltman, 2014).
We also ran the keyword co-occurrence analysis to identify the topics of research and their relatedness to complement the identification of the conceptual network structure of each period (Van Eck & Waltman, 2010). The visualisation map of the keywords proposes different conceptual clusters that are useful to support the researcher in the systematic review; nevertheless, there can be some overlaps between the keyword clusters with different coupling clusters. This is because keywords and themes are related to each other in the literature, for example, articles in two different clusters might correspond to the same keyword cluster. Similarly, a coupling cluster could be associated with keywords belonging to different clusters (Andersen, 2019).
One of the weaknesses of keyword co-occurrence analysis is related to multiple meanings that a word may be given in a paper (Van Eck & Waltman, 2014;Zupic & Cater, 2015). Therefore, we adopted WoS KeyWords Plus®, which ensures consistency in classifying keywords . Analysing through multiple bibliometric tools makes it possible to reduce the individual bias of the single method while ensuring a complete picture (Andersen, 2019;Zha, Melewar, Foroudi, & Jin, 2020).

Results
This section presents the results of the bibliometric analysis and the systematic literature review for each of the three time periods.
Analysing the publications trends allows us to obtain a picture of the evolution of employee voice studies ( Figure 2). According to the criteria described above, the first period starts from 1986 to 1997 and is characterised by limited research activity. The second period refers to 1998-2008; research activity increases with some continuity. The third period begins in 2009 and is characterised by an acceleration of scientific activity, indicating an increasing development of employee voice research.
1986-1997, the foundation: the IR perspective The first period includes 28 articles. Our review suggests that there was not yet a clear definition of employee voice, rather, it is embedded in IR debating, essentially the labour line of institutional economics and focused on covering all the sides of employment relationship (Kaufman, 2006). With its roots in trade unionism and collective bargaining, voice and representation mostly coincide within IR literature, which strives to bring basic democratic practices within organisations through the law, ensuring opportunities for participation and representation for workplace decisions, and the employees' interest protection in the resolution of workplace disputes and organisational justice (Kaufman, 2006;Wilkinson, Dundon, Donaghey, & Freeman, 2020). It is possible to identify this dominant focus from Figure 3a, where the main terms frequently occurring are 'participation', 'voice', 'unions', 'justice' and 'performance', along with other keywords such as 'psychological contract', 'employee representation' and 'conflict'. In addition, the keyword co-occurrence results demonstrated that the term 'voice' is not the most connected. Indeed, it is interesting to note that the centre of the figure is dominated by the term 'participation' that is connected to almost all keywords of the map, differently by 'voice' that lacks connections with the terms positioned on the right of the figure, such as 'boards', 'unions' and 'employee representation'.
Based on the above, we deepen how the IR purposes trace the direction of employee voice research in exploring these arguments.
The workers' representation and labour-management relationship is the red cluster of Figure 3b, which we associated to the red cluster of keywords co-occurrence, including terms such as 'boards', 'unions' and 'employee representation'. The articles of this group follow the collective conception of voice by focusing on the institutions for worker representation in the workplace decisions and their impact on organisational outcomes (Kleiner & Lee, 1997;Miller & Mulvey, 1991), and the labour-management relationship (Godard, 1992;Kennedy, Drago, Sloan, & Wooden, 1994).
The most relevant article of this group in terms of the highest link strength to other studies is Boroff and Lewin (1997). Their results strongly conclude that loyal employees who experienced unfair workplace treatment 'suffer in silence' (Boroff & Lewin, 1997: 60) rather than exercise voice. The empirical validity of the exit/voice model in different national labour markets with special attention to unionism is the focus of most of the articles of this group. For instance, Leigh (1986) focused on the USA, Miller and Mulvey (1991) and Kennedy et al. (1994) focused on the Australian context, and Kleiner and Lee (1997) reported evidence from South Korea. Keeping the focus on collective voice, Kennedy et al. (1994) paid close attention to the effective activity of unions in the workplace in terms of bargaining with management over training. Finally, Kleiner and Lee (1997) showed that voice in work councils and unions impacts HRM practices such as wages, turnover, employee satisfaction and labour productivity. The study of Olson-Buchanan (1996) assumed a central position. It is among the first to link distinct pieces of literature, going through research and theory of industrial-organisational psychology with social psychology and procedural justice.
Employee voice as an organisational fairness instrument cluster (Figure 3b, green) denotes a fundamental and common research interest in decision-making, which frames voice as a form of subordinate participation (Korsgaard & Roberson, 1995: 658) and instrument that ensures the fairness of the process (Goodwin & Ross, 1992;Magner, Welker, & Johnson, 1996). Indeed, the 'justice' keyword stands out on the map. We adopted this term in the co-occurrence analysis as an umbrella term covering related words, such as perceived fairness, equity and specific types of organisational justice, e.g., procedural justice, which relate to the other keywords of the map.
Specifically, Korsgaard, Schweiger, and Sapienza (1995) emerge in this group for the highest power of link strength, denoting the relevance of this study for the contribution to research in this first period: voice can be a useful instrument to enhance perceptions of the fairness of the decision-making process. In this line, Korsgaard and Roberson (1995) propose two perspectives of voice conception: one considers voice regardless of the effective influence over the decision, indicated as non-instrumental voice; the other evaluates employee voice related to the effective impact on a decision via indirect influence, i.e., instrumental voice. This perspective of analysis shows some initial results of employee voice about the possible consideration of it as a multidimensional construct in terms of suitability to different purposes of analysis, as voice impacts employees' justice perception of fairness and also influences employees' attitudes, such as the intention to quit (Daly & Geyer, 1994).
The Exit-Voice-Loyalty (EVL) framework cluster (Figure 3b, yellow) is motivated by a strong focus on the EVL framework proposed by Hirschman (1970) and the following authors, as demonstrated by keywords such as 'satisfaction', 'exit' and 'neglect' (Figure 3a). Despite Hirschman (1970) being a source not included in our data set, the papers discussed in our review reveal and confirm the critical role of this book for employee voice research.
Two articles present the highest total link strength: Withey and Cooper (1989) and Spencer (1986). Withey and Cooper (1989) are central to the figure, and their study investigates employees' dissatisfaction responses in terms of exit, voice, loyalty and neglect. The study of Spencer (1986) focused on the relationship between the availability and quality of a certain number of formal voice mechanisms and voluntary turnover as an exit variable. In doing so, the author considers the presence of unions in the workplace and demonstrates that the availability of multiple voice mechanisms influences employees' retention, suggesting some additional insights into the Exit-Voice-Loyalty-Neglect (EVLN) framework.
This line of research is also found in Mayes and Ganster (1988), who found that two of the constructs of Hirschman's framework (voice and exit) are independent reactions not mutually exclusive and impact individually on organisational turnover rates. Keeley and Graham (1991) also proposed normative implications of Hirschman's framework EVL, providing clearer guidance concerning how much voice or exit is appropriate under different circumstances.
Employee voice in HRM (the beginning) cluster (Figure 3b, blue) appears more detached from the previous ones, and this could be because it introduces employee voice in organisations from another point of view, that is HRM practices applied to the workplace to increase the organisational performance. Indeed, another important keyword in this period that denotes the emerging of HRM perspective is 'performance' (Figure 3a).
In a non-union context (Feuille & Chachere, 1995), different mechanisms need to be implemented to allow employees to voice their opinion and participate in organisational decisionmaking about their work. Lewin and Mitchell (1992) is the study that connects this stream with the others. In their article, the authors examined the different voice systems available for organisations. Taking into account unionised and non-unionised settings, the characteristics of mandated voice systems and 'voluntary' voice systems affect employee turnover and increase intermittent employment. Feuille and Chachere (1995) tried to understand the conditions under which employers make formal mechanisms available in organisations for those employees not represented by unions, referring to the justice function of employee voice. The study finds that the more value attributed to human resource in the workplace and the greater the number of employee-oriented HR practices, the more formal non-union grievance procedures are implemented. In this vein, McCabe and  conceptual paper shows that the importance of employee voice for non-unionised organisations aside from the primary purpose of providing this voice mechanism to employees is related to avoiding unions; and this impacts the decision of some organisations to make grievance procedures available. The contribution of Aaron (1992), at the edge of the cluster, provides the legal perspective of employee voice when manifested through a non-union mechanism. The article highlights a limited function of non-union voice in a system legally based on union protection.

1998-2008: the raising of OB and the development of HRM perspective
The beginning of the second period, 1998, coincides with the publication of the most relevant article on employee voice, Van Dyne and LePine (1998), with more than 170 citations among all the documents included in the data set. These authors introduced the first definition of voice to the OB literature and developed an operational measure to investigate it. They defined employee voice as 'a promotive behaviour that emphasises the expression of constructive challenge intended to improve rather than merely criticise' (p. 110). Most recent authors of voice research consider this formal definition of voice and operational measure provided by Van Dyne and LePine (1998) as the starting point of voice research in OB literature (Morrison, 2011;Mowbray, Wilkinson, & Tse, 2015;Wilkinson, Barry, & Morrison, 2020). OB represents the dominant perspective during this second analysis period, which considers voice as a discretionary and improvement-oriented behaviour (Detert & Burris, 2007) and focuses on micro-level factors affecting voice behaviours by investigating it at the individual level. The individual level of analysis in the OB-centric perspective is also the focus of voice studies in the HRM field, which in this second period starts to profoundly investigate the interplay between non-union and union representation structures and the impact on employee attitudes towards organisations (Batt, Colvin, & Keefe, 2002;Bryson, 2004;Gollan, 2006). Moreover, this period includes other articles in the top 10 most cited documents of the entire database (Detert & Burris, 2007;Milliken, Morrison, & Hewlin, 2003). Therefore, we argue that this period is the most important in terms of research because it contains the core foundations of what employee voice research is nowadays.
Employee voice as a distinct OCB behaviour cluster of studies (Figure 4b, green) catches the eye for the central position and the presence of the most important articles of the period. It shows the central importance of OB, which found evidence in cluster green of Figure 4a, revealing keywords such as 'OCB' and 'extra-role behaviour'. Moreover, other keywords of this cluster, such as 'personality', 'SET' (social exchange theory), illuminate about the aspects investigated by the articles included in this group of documents. Studies in this cluster focus on identifying and validating different types of OBs and seek to identify antecedents of speaking up in the workplace (Burris,  Detert, & Chiaburu, 2008;Detert & Burris, 2007;. Exploring these research interests, we can identify an initial leading line of inquiry that explores voice along with other types of OCB (Organisational Citizenship Behaviour) affiliative and proactive behaviours, such as helping behaviours Stamper & Dyne, 2001;Zhou & George, 2001) and taking charge (Morrison & Phelps, 1999). It assumes an integrative explanation that takes into consideration situational factors involving broader feelings of employees towards the organisation (Stamper & Dyne, 2001), leadership antecedents to voice, such as openness (Burris, Detert, & Chiaburu, 2008;Detert & Burris, 2007;Morrison & Phelps, 1999) and person-centred antecedents, such as self-esteem , and personality traits (LePine & Van Dyne, 2001). Detert and Burris (2007) emerge among the most cited documents within the data set documents. These authors consider differences in employees in terms of proactivity and satisfaction, highlighting the importance of positive leader behaviours associated with creating a safe workplace that impact on employees' voice. In this sense, Morrison and Phelps (1999), reviewing OCB literature and extra-role, differentiate taking charge from other proactive behaviours, including voice. Individual personality traits are additionally relevant. This line of research is followed by LePine and Van Dyne (2001), assessing the relationship between individual differences (personality and cognitive ability) and specific behaviours such as voice, cooperative behaviour and task performance. Burris, Detert, and Chiaburu (2008) focused on the influence of employee psychological state, related to attachment to the organisation, on employees' motivation to speak to the supervisor. The study reveals the importance of leadership style and the quality of the relationship between leader and follower for employee perceptions, and finds that exit responses related to the EVLN framework can occur formally and through a detached state that inhibits voice behaviours. This framework represents the main topic investigated in the closest cluster (yellow).
Employee voice and EVL theory cluster (Figure 4b, yellow) contains 13 articles denoting the predominant theoretical perspective that stands out in most of the voice publications about the EVL framework by Hirschman (1970) and following development (Farrell, 1983). Similarly to the keyword co-occurrence map of the previous period, in the yellow cluster of Figure 4a keywords emerge such as 'satisfaction', 'responses', 'loyalty', 'exit' and 'neglect', which are the main elements of the EVL framework. More in detail, the systematic review of articles of this cluster suggests that 'satisfaction' is addressed involving multiple aspects (Hagedoorn, Van Yperen, Van De Vliert, & Buunk, 1999), such as psychological contract violation (Turnley & Feldman, 1999, 2000, job insecurity (Sverke & Goslinga, 2003;Sverke & Hellgren, 2001) and specific dimensions of commitment (Luchak, 2003;Sverke & Hellgren, 2001). In this vein, two lines of research appear. The first line includes articles that consider situational factors, such as job-related, organisational attributes and individual differences, to explore possible responses to dissatisfaction based on the theoretical framework (Hagedoorn et al., 1999;Mishra & Spreitzer, 1998;Turnley & Feldman, 1999). Mishra and Spreitzer (1998) is this cluster's most coupled article. These authors adopt the EVLN framework to outline the different responses that employees manifest to downsizing according to their stress-based model, integrating the influence of fairness and justice on survivor responses. Following this line, the second relevant paper of this subgroup is Turnley and Feldman (1999), which shows an empirical study of situational factors and justice dimensions that explain different employee responses to violation of the psychological contract. The second line of research includes articles mainly published in IR journals that consider forms of collective voice to cope with uncertain employment situations and explore the consequences for organisations regarding turnover and exit (Sverke & Goslinga, 2003;Sverke & Hellgren, 2001). In particular, Sverke and Goslinga (2003) focused on job insecurity and found that dissatisfying employment conditions are primarily related to exit and loyalty reactions, but not to the use of voice in terms of union participation.
Employee voice as an organisational and human resource practices cluster of studies (Figure 4b, red) starts the integration between IR and HRM, particularly examining the interplay between non-union and union representation arrangements and the impact on employee perceptions, attitudes and performance. This cluster's line of research seems to focus on analysing the impact of alternative IR institutions, along with more direct forms of voice and HRM theories and practices. We associated 'commitment', 'job aspects' and 'turnover' keywords to this cluster (Figure 4a). Gollan (2006) is the most relevant article regarding connecting strength. Focusing on the effectiveness of union and non-union voice arrangements in the British context, the author highlighted that the complementary presence of both trade unions and direct voice mechanisms benefits employees by providing multiple channels for voice. The second relevant paper of this cluster is by Dundon and Gollan (2007), which focused on the conceptual analysis of the literature on voice mechanism and the conditions that influence the forms of voice present in the workplace. Findings revealed a tendency in British workplaces to replace traditional single union voice with multiple non-union voice arrangements. The study of the unions-HRM link is the research interest of Wood and Wall (2007), who frame employee voice among high commitment management practices, and of Batt, Colvin, and Keefe (2002) and Bryson (2004), who are focused on the effectiveness of voice mechanisms in benefitting employees and on the impact that voice has on HRM practices for organisational and employee outcomes such as quit rates and intentions to exit.
Employee voice as a perception of organisational fairness cluster of studies (Figure 4b, blue) completes the picture of the period, and focuses on organisational justice literature as shown by the related keyword 'justice' (Figure 4a). Voice influence individuals' evaluations of the fairness of processes and practices implemented by organisations (De Cremer, 2006;Kernan & Hanges, 2002). The most linked document of this cluster is De Cremer (2006), about the influence of procedural justice with leadership styles on two affective responses. The authors employed voice as an instrument to manipulate procedural justice perceptions, which refers to having a voice in decision procedures or not, highlighting that it impacts followers' emotional reactions. Kernan and Hanges (2002) is the second relevant article of this group; it finds voice to be a way to enhance perceptions of procedural fairness, developing a comprehensive model of survivor reactions to a reorganisation. Another relevant study is by Lind and van den Bos (2002), who developed the uncertainty management theory of fairness indicating that 'Voice, when included as an element of decision-making procedures, is one of the most powerful ways to engender judgments that the decision making process is fair' (Lind & van den Bos, 2002: 186).

2009-2019: the integration of IR, OB and HRM perspective
The last identified period involves 340 contributions in the bibliometric analysis. There are relevant attempts to integrate different management fields during this period and thus to overcome the 'silo' approach to studying voice. A more integrative conceptualisation of voice emerges and covers IR, HRM and OB perspectives. It defines voice as 'all of the ways and means through which employees attempt to have a say about, and influence, their work and the functioning of their organisation' (Wilkinson, Barry, & Morrison, 2020). Keyword co-occurrence results are in line with this integrative approach of voice research. The articles included in this period have produced above 1500 keywords. From Figure 5a, it is possible to note an increased integration of the previously disparate topics, which confirms the establishment of the development of research that looks at employee voice from 360 degrees.
The integration of voice perspectives cluster of studies (Figure 5b, red) is the most numerous in terms of articles, which are mainly oriented towards the study of employee engagement (Kwon, Farndale, & Park, 2016;Marchington & Suter, 2013;Rees, Alfes, & Gatenby, 2013) and satisfaction (Holland, Pyman, Cooper, & Teicher, 2011;Laroche, 2017). The primary focus is rooted in the HRM field, and many studies denote more or less direct integrations with other fields of literature (Avgar, Sadler, Clark, & Chung, 2016; Laroche, 2017; Marsden, 2013). The most connected article of this cluster for total link strength is Pohler and Luchak (2014), who integrate IR with HRM and OB. Specifically, focusing on high-involvement work practices and unions, this study highlights the importance of both individual and collective voice mechanisms in organisations, which balance the potential effects of the interests of management and unions to create more efficiency, equity and voice. Mowbray Wilkinson, and Tse (2015) and Mowbray (2018) represent the second and third relevant articles of this cluster resulting from the bibliographic coupling and present two integrative studies about literature on employee voice. Mowbray, Wilkinson, and Tse (2015) propose an integrative review that considers IR literature and the labour economics literature relevant to the HRM literature, thus connecting them as a unique discipline labelled HRM/ER. Mowbray (2018) extends the attention on managers and presents a qualitative case study that integrates HRM/ER and OB definitions and shows that the lessened power of role influences the creation of informal voice channels.
Employee voice and performance: the role of leaders cluster (Figure 5b, blue) is rooted in the OB research field and follows the line drawn by the most cited article, that of Van , about the investigation of the impact of employee voice and multiple extra-role proactive behaviours on performance.
The most relevant study of this cluster is the meta-analysis provided by Chamberlin, Newton, and Lepine (2017), which clarifies the concept of voice and its promotive and prohibitive forms and develops a theoretical framework considering the antecedents and consequences of these forms. The most important results of this study are identifying variables recognised as antecedents of voice, e.g., personal attributes, leadership and contextual factors; and psychological states of felt responsibility and engagement as the underlying mechanisms through which these variables impact performance (Chamberlin, Newton, & Lepine, 2017). The second relevant article was published 5 years earlier: Tornau and Frese (2013) developed a meta-analysis to clarify the similarities and differences between four proactivity constructs (proactive personality, personal initiative, voice and taking charge), identifying those related to personality and those related to behaviour. One of the main results shows that the proactive behaviours investigated predicted job performance much better than personality, representing also the main basis of the judgment rendered in supervisor reports. Regarding this, a specific trend of this cluster concerns the research attention paid to supervisors (Burris, 2012), not only for performance ratings rendered but also for other leadership factors associated with employee voice (Burris, Detert, & Romney, 2013;Howell, Harrison, Burris, & Detert, 2015;Lloyd, Boer, Keller, & Voelpel, 2015;Weiss, Kolbe, Grote, Spahn, & Grande, 2018). These studies move the attention of voice research from how employees perceive leaders to how leaders communicate with their followers and manage judgments about who speaks up (Weiss et al., 2018). Among the most recent and relevant, Howell et al. (2015) showed that managers systematically evaluate employee voice contributions based on heuristic and biased processing, influenced by employees' membership in minority groups or majority groups. Following this line, the empirical study of Fast, Burris, and Bartel (2014) showed that low managerial self-efficacy triggers a defensive stance towards employee voice, reducing voice solicitation and implementation of suggestions and increasing the tendency to discriminate against employees who speak up.
Moreover, Burris (2012) showed that managerial responses to followers' voice depended on the type of voice exhibited: less proactive forms of voice that support the status quo and don't challenge it generate more favourable reactions from managers. Finally, Weiss et al. (2018) show that inclusive leader language stimulates followers' voice, depending on their professional group membership.
The importance of leadership style and culture for employee voice cluster (Figure 5b, green) keeps the focus on OB and mainly on leadership styles and social relations, focusing on the culture and national context. Many of the articles in this group concern China-based field studies (Bai, Lin, & Liu, 2019;Kong, Huang, Liu, & Zhao, 2016;Xu, Qin, Dust, & DiRenzo, 2019), where it emerges either implicitly or more explicitly Confucian-based values that affect leadership. Indeed, leadership keywords refer to concepts such as authority, characterised by moral integrity and fatherly benevolence, influencing organisational characteristics and leadership traits (Zhang, Huai, & Xie, 2015;Zhu & Akhtar, 2017).
The most relevant article is by Xu et al. (2019), which highlights the influence of congruence of supervisor-follower as a contextual factor associated with voice and the critical role of leaders in creating a psychological safety environment. Yiwen Zhang, LePine, Buckman, and Wei's (2014) study presents additional results about contextual factors, showing the influence of specific leadership behaviours (e.g., transformational leadership). A general theme among these studies is the call of attention to the context in which data are gathered, which may influence the results, as the Chinese context tends to be characterised by high power distance. In a similar approach, moving to the European context, Guenter, Schreurs, van Emmerik, and Sun (2017) investigated the influence of specific types of leaders' behaviours, such as authentic leadership, to stimulate voice behaviour in individuals with low proactivity.
The contextual and social factors affecting voice behaviours cluster of studies (Figure 5b, yellow) is focused on the study of employee voice antecedents that refer to contextual factors and the social exchange relationships of the workplace. This cluster differs from the above in its specific focus on boundary conditions related to leadership affecting voice behaviours rather than on the main effects these elements have on employee voice. Moreover, the focus here is on the unit-level outcomes rather than on the individual level. Starting from the assumption that not all leaders have the resource and the motivation needed to implement suggestions from their employees, McClean, Burris, and Detert (2012) provide an empirical analysis to understand the conditions under which employee voice impacts exit, highlighting the importance of managers' responsiveness to voice among groups of employees.
The cluster interestingly includes most of the articles by Thomas Ng. Ng and Feldman (2015) show that when organisations allow employees a greater degree of control over employment arrangements based on personal preferences and needs (i.e., 'idiosyncratic deals', p. 893), they motivate employees to participate in organisational decisions through voicing their suggestions. Moreover, voice is studied in Ng and Feldman (2013) as a behavioural variable to understand the process through which people attribute a sense to social signals, highlighting that employees' perceptions of supervisor embeddedness signal and influence employees' perceptions of organisational trustworthiness and, in turn, affect their relationship with the organisation.
The different forms of employee voice cluster of studies (Figure 5b, purple) represents a new nuance in employee voice research, whereby studies focus on employee voice from a broader perspective. Alternative forms related to voice ranged from actual voice behaviours to whistleblowing (Kirrane, O'Shea, Buckley, Grazi, & Prout, 2017;Skivenes & Trygstad, 2010) and silence (Knoll, Hall, & Weigelt, 2019;Prouska & Psychogios, 2016, 2019. Klaas, Olson-Buchanan, and Ward (2012) provided the first relevant contribution, presenting an integrative review of alternative forms of voice determinants to understand how different processes and factors impact voice behaviour. Considering both unionised and non-unionised organisations, the authors investigate why employees manifest different forms of voice and how leaders' behaviours and organisational practices and policies influence these forms. The second and third relevant articles are by Prouska and Psychogios (2016) and Prouska and Psychogios (2019), concerning possible responses of employees in a long-term turbulent economic environment. Aiming to capture employee perceptions of workplace voice during the crisis period, these authors study the other side of the voice coin: silence (Prouska & Psychogios, 2016). These authors then proposed a new study of line managers' experience of employee voice and silence, considering their perspective and organisational position (Prouska & Psychogios, 2019).
Discussion: an agenda for future research The state of the art of employee voice literature shows that studies are developed in 'silos', reducing the possibility of cross-fertilisation and theoretical contamination among researchers (Barry & Wilkinson, 2022;Bashshur & Oc, 2015;Kaufman, 2015;Wilkinson, Barry, & Morrison, 2020). A need thus arises to assume a broader perspective, continuing the integration of different fields and approaching voice from a broad perspective that considers multiple forms for employees to communicate within the organisation their dissatisfaction about work-related issues or their suggestions for organisational improvement (Wilkinson, Barry, & Morrison, 2020). Therefore, considering a broader analysis of management perspectives, IR, OB and HRM help delve deeper into employee voice research's specific characteristics. This review shows that recent works adopting this new line of research enlarge the boundaries of interest towards alternative forms of voice such as whistle-blowing (Kirrane et al., 2017;Skivenes & Trygstad, 2010) and silence (Knoll, Hall, & Weigelt, 2019;Prouska & Psychogios, 2016, 2019. Moreover, the systematisation of results allows, on the one hand, proposing an integrated framework of the primary intellectual knowledge about the development of employee voice considering the individual perspective (Figure 6), on the other hand, to offer future research directions and the related potential implications for practice (Table 2). In this regard, the next step should be to understand how to translate this critical knowledge of voice into actual improvements for organisational effectiveness. In addition, the findings show that the research about voice mainly focuses on domestic contexts. The relevance suggested by recent research about cultural aspects and the research fronts identified in this work might represent an important starting point for future research on employee voice to support organisations in effectively managing employees in international contexts (Wang & Varma, 2019). The following paragraphs discuss in detail these main findings.

From industrial relations to the individual level of voice: what next?
Our analysis underlines that employee voice research experienced a progressive development of the conceptualisation of employee voice ( Figure 6) that starts with the predominance of a collective perspective. At the early stage, based on the results of our first period of analysis and as shown in the first quadrant of Figure 6, voice research reveals embedded in the IR perspective regarding broad employment relationships. Through its roots in trade unionism and collective bargaining, IR aims to bring into organisations democratic practices that ensure employees' opportunity for participation and representation in work decisions.
Accordingly, researchers introduced voice in this collective perspective associated with unions as the most powerful form of voice in representing employees' interests (Freeman & Medoff, 1984). The IR focus dominates and influences most studies composing the early stage of research on voice. For instance, as shown in Figure 6 central quadrant of second columninfluenced by the more broad intent of protecting employees' interests and giving them the legitimation to participate in decisions, scholars within organisational justice literature refer to voice as the instrument that ensures the fairness of the process through the right to participation (Korsgaard & Roberson, 1995;Korsgaard, Schweiger, & Sapienza, 1995). The justice perspective of voice opens new research directions on employee voice, which starts from the evidence that voice impacts employees' justice perception of fairness and influences employees' attitudes (Daly & Geyer, 1994). This last line of enquiry encouraged voice research over the years, and scholars have moved from a collective perspective of voice to a more individual level, considering is challenging yet promotive has become widely adopted by OB research. Over the years, scholars have provided relevant contributions about antecedents and consequences of employee voice at the individual level of analysis within this research field. For instance, studies highlighted the influence of individual characteristics and self-beliefs (Kong et al., 2016;LePine & Van Dyne, 2001), the contextual factors affecting the workplace climate and the employee's confidence to speak (Chen & Hou, 2016;Qin, DiRenzo, Xu, & Duan, 2014;Song, Wu, & Gu, 2017). Moreover, focusing on employee involvement, studies rooted in the HRM field investigate the possible forms that employee voice behaviour can assume within the organisation by considering voice as a practice to foster commitment, participation and organisational outcomes (Batt, Colvin, & Keefe, 2002;Marchington & Suter, 2013;Wood & Wall, 2007). As shown in the third column of Figure 6, the relevant results reached on the individual level of analysis have gradually highlighted the critical relationship between employee voice and the leadership dimension (Bai, Lin, & Liu, 2019;Detert & Burris, 2007;Fast, Burris, & Bartel, 2014;Guenter et al., 2017;McClean, Burris, & Detert, 2012). As shown in the third period of analysis, the research initially considered leaders the main target of voice behaviour. In this view, studies revealed that since leaders exercise an important influence on the implementation of practices and the value of employees' performance (Lebel, 2016;Morrison, 2011), their behaviours and styles influence the relationship with their employees and their perceptions, conditioning also the confidence to speak (Tangirala & Ramanujam, 2012;Wu, Zhaoli, Xian, & Zhenyu, 2017). Then, the attention to leaders shifted to characteristics and feelings related to their roles as boundary conditions influencing employee voice behaviour because leaders also affect the workplace where employee voice might occur (Howell et al., 2015;Tangirala & Ramanujam, 2012).
Based on the above, the present article offers new directions for this research field, highlighting the need to go beyond traditional research interests. Indeed, this systematic literature review results confirm that much has been done regarding the antecedents, boundary conditions and consequences of employee voice at the individual level of analysis. As detailed in Table 2, to move forward, it is necessary to translate this critical knowledge of voice into actual improvements in organisational effectiveness. It emerges that employee voice is an element that HRM needs to know how to manage (Barry & Wilkinson, 2022). At the practical level, the focus on formal and informal voice-based mechanisms is insufficient to encourage voice behaviours; rather, potential interaction effects occur between different approaches for voice (Pohler & Luchak, 2014), such as collective/individual, direct/indirect, union and non-union avenues for voice; it is critical, especially in turbulent moments and crises (Prouska & Psychogios, 2019). It follows that a shift to a broader perspective is required, from the individual dimension to the organisational level of analysis, which in turn requires an investigation into the processes and structures consistent with the contextual, individual and cultural factors, which can support employee voice within the organisation that hence benefits from employees' input. As such, the analysis of how micro and macro domains are connected is an important challenge for future studies as described in Table 2.
The context matters: how to manage voice in nowadays workplaces The organisational context assumes relevance along with factors related to cultural dimensions, as confirmed by the results of this study. Especially during the analysis of the third period, results suggest that the attention of future research should focus more on the role of cultural aspects in affecting organisations and employee voice (Kwon & Farndale, 2020;Wang, Baba, Hackett, & Hong, 2019;Xu et al., 2019). These reflections are related to the impossibility of finding global solutions for organisation and management problems, which face common basic issues within the organisations due to national cultural differences but respond differently according to each country (Hofstede, 2001).
In this regard, we believe there are opportunities for different streams of literature to crossfertilise each other and collaborate to explore the cultural and contextual issues in studying employee voice as proposed in table 2. For instance, the different orientation that management can have on employees' input affects the organisational context where employees have to decide to manifest their opinion or not (Kwon & Farndale, 2020). It is a critical issue for organisations nowadays, characterised by the increasingly relevant presence of individuals from various cultural contexts. In addition, as noted, the potential contribution of each member expressed through their voice can represent a competitive advantage in acquiring additional knowledge by leveraging the capabilities of employees, perspectives, insights and experiences. These, undoubtedly, are influenced by different cultural values and professional backgrounds (Brewster, Sparrow, & Harris, 2005;Tröster & van Knippenberg, 2012). Indeed, at a practical level, employees coming from different countries with different cultural and professional backgrounds can provide disparate perspectives, insights and experiences that enrich workplaces yet at the same time complicate the texture of relationships, management and decisions, impacting the firm's success (Lam & Mayer, 2014;Mackenzie, Podsakoff, & Podsakoff, 2011). A related critical topic for employee voice literature is cultural distance, affecting the willingness of employees to manifest their opinions in the workplace (Jiang, Le, & Gollan, 2018;Kwon & Farndale, 2020). Individuals are often reluctant to express their voice in the workplace due to the possible consequences that even a constructive suggestion might imply, as recipients of the communication, who are often on a higher level, could perceive it as a challenge to the status quo (Milliken, Morrison, & Hewlin, 2003;Morrison, 2011) or misunderstand the intention because of the cultural distance in an international context (Jeung & Yoon, 2018).
The above issues are relevant topics for future research on employee voice, which needs to go beyond a country-level focus to understand the main points and implications when the employee voice topic is discussed in the context of international HRM, as proposed in Table 2. In our opinion, the research fronts of voice literature identified in the third period represent a critical avenue for future research that could help organisations manage individuals from different backgrounds and benefit from employee suggestions by implementing effective enhancements.
For this purpose, the global leadership is critical to influence employees' behaviours. Consequently, we need to understand the leaders' managerial communication with subordinates, how leaders evaluate employees' contributions and the impact of leaders' behaviours on follower outcomes in the international context (Jeung & Yoon, 2018). Our findings have demonstrated that, over the years, a stable element that attracted research interest is leadership, influencing both employees and their willingness to voice opinions and the context where voice might occur (Jeung & Yoon, 2018). Consequently, leadership in nowadays workplace might foster voice by promoting extra-role behaviours and instilling a sense of participation and shared communication in employees, putting them central to the organisation. Therefore, future studies might deeply investigate shared bottom-up leadership approaches (Cheng, Bai, & Hu, 2022;Zeier, Plimmer, & Franken, 2021), which consider employee voice as an antecedent, and the impact of this kind of leadership on organisational performance and development, while taking into account the issue related to the current cross-cultural contexts.
Moreover, the leadership dimension represents a buffer to move from the individual dimension of employee voice to the contribution of individuals to the overall organisational performance and strategy. Future research should focus on how the advances in employee voice literature can translate into actual organisational effectiveness improvements. Researchers might focus on organisational aspects to shed light on the impact of different HR practices on employee voice nowadays, considering multiple cultural backgrounds and knowledge that characterise workplaces. Moreover, since HR practices might present different levels of cultural specificity, research might adopt a comparative approach that considers this aspect and cultural differences across countries to ensure the structures and channels within organisational processes encourage individuals' voice behaviours (Kwon & Farndale, 2020;Prouska, McKearney, Opute, Tungtakanpoung, & Brewster, 2022).
From the above considerations, Table 2 summarises the main topics that emerged from our results. We, then propose exploitative and explorative research avenues (Gervasi et al., 2022;Pellegrini et al., 2020). More in detail, the former refers to avenues already studied but may still present interesting further development; the latter refers to those new avenues that need investigation or are now explored in a very initial stage. Based on these, we propose the consequent implications for future research.

Conclusion
The study of employee voice has been approached by multiple disciplines, which stemmed from various starting issues and theoretical premises converged in the three main management domains: IR, HRM and OB. Nevertheless, the fragmented literature in disciplines that hardly conversate with each other has hindered the needed cross-fertilisation necessary to advance our comprehension of voice phenomenon. By presenting a systematic literature review, rigorously supported by bibliometric analyses, that considered the multiple approaches and disciplines, this paper has identified three time periods of employee voice research, providing order about how the concept and its study have developed. Overall, this study's unique contribution to the theory of employee voice, which also is a methodological contribution in complementing previous literature reviews adopting both a systematic and a bibliometric approach, is sorting and ordering knowledge to pave future research avenues (Sandberg & Alvesson, 2021).
Despite adopting a relatively innovative and rigorous bibliometric and systematic literature review approach, some limitations remain. In particular, one limitation may lie in focusing on a data set of articles published in journals ranked three and above in the 2021 Academic Journal Guide chartered by ABS, which may result in a lower representation in the data set of younger research streams. However, the dimension of the data set being considered allowed the identification of well-established themes. Moreover, to respond to the need for homogeneity of sources, our analysis has not considered books, conference proceedings or practice reports. Although our data set is still extensive, future studies could examine other further publications and scientific subjects to complement our results.
Similarly to previous bibliometric and systematic literature review studies (e.g., Sassetti et al., 2018), one of the main limitations of this research lies in the trade-off between the need to provide a broader panoramic view of the field paying attention to the in-depth analysis of the content. Although this method offers a systematic way to structure literature reviews, it can't substitute for extensive reading and knowledge of the field. Research needs to consider bibliometric methods to integrate traditional reviews and a way to adopt, adapt and summarise findings in an overview that shows the relationship between articles and related topics. The objective of our study was to provide a science map of employee voice research through multiple bibliometric analyses. It is an abstraction derived from reality . Science maps provide a starting point for analytical examination, but the interpretation strategy depends on the study focus and the authors' perspective (Zupic & Cater, 2015).
Conflict of interest. None.