Voice in non-traditional employment relationships: a review and future research directions

Abstract Research on employee voice has been widely documented for workers in traditional employment relationships (TERs) and has offered a broad understanding of how they express their ideas and complaints at work. However, an under-explored area concerns how workers express voice in non-traditional employment relationships (NTERs) characterised by flexibility, temporality, instability, and insecurity. Studying voice in NTERs is of high importance due to its increased potential proliferation and associated precariousness. In this paper, we expand the knowledge frontiers in the voice literature by conducting an integrative review of empirical studies that explore voice among workers in NTERs. We identify the forms of voice available to non-traditional workers, the issues they are interested in voicing, how effective their voice is in influencing management decisions, determinants, and outcomes of their voice. Future research agenda offered concerning how the neglected area of voice among non-traditional workers can be addressed.


Introduction
Furthering ideas from the seminal works of Hirschman (1970) and Freeman and Medoff (1984), most of the voice research from both organisational behaviour (OB) and employment relations and human resource (ER/HR) perspectives have studied workers in traditional employment relationships (TERs) neglecting those in non-traditional employment relationships (NTERs). TERs, otherwise known as standard employment, are characterised by contract permanency, specified work hours, working on employer's premises, continuity, long-term career expectations, and income security (Ashford et al., 2007). Conversely, NTERs, otherwise known as alternative work arrangements, non-standard employment, flexible staffing arrangements, contingency work, vulnerable work, or precarious employment (Kalleberg, 2000), are characterised by flexibility, temporality, instability, and insecurity. The rise of this employment type has been widely reported, and there are forecasts that the number of workers in such employment arrangements will continuously increase (ILO, 2016). Recent events have also disrupted working arrangements and occasioned an increased rate of remote working and non-traditional employment patterns (Spurk & Straub, 2020). However, in most cases, these employment patterns enable precarity and exploitation of workers without recourse to discourse.
Extant voice literature assumes homogeneity of workers with less consideration for the peculiarities of non-traditional workers, making the theoretical paradigm of diversity in voice research missing (Syed, 2014;Wilkinson et al., 2021). For instance, while there is a plethora of studies and reviews on the operationalisation, dimensions, determinants/ inhibitors, and outcomes of voice of traditional workers (for example, see Morrison 2011Morrison , 2014Mowbray et al., 2015), there is a limited attempt for non-traditional workers' voice. Hence, deep and integrated understanding of workers' voice has been downscaled by the lack of theorization of voice offered to these workers. Corroborating these claims, Kaufman (2014) acknowledged the underdevelopment of voice theory, and Wilkinson et al. (2018) call for further theorization of the factors that shape workers' voice amidst workplace diversities.
In this article, we therefore conduct a meta-synthesis (see Cooke et al., 2012) of findings around voice among workers in NTERs. We use a systematic approach in reviewing employee voice studies within the NTER context to address the following research questions: (a) what forms of voice are available to workers in NTERs? (b) what types of issues do they voice? (c) how do they influence management decisions and actions with their voice? (d) what are the determinants and outcomes of voice for workers in NTERs? The paper addresses these questions by identifying, extracting, and evaluating previous empirical studies on employee voice in any form of NTER since the adaptation of voice to an organisational context by Freeman and Medoff (1984). The paper also offers future research directions for the theoretical advancement of the voice literature. Additionally, it contributes to the voice literature in the following ways. Firstly, it maps out from previous studies the voice forms available to non-traditional workers, the issues they express, and the level of influence they have on management actions and decisions. Secondly, the review looks at the determining factors, at the individual, firm, external, mediating, and moderating levels of analysis, that influence non-traditional workers' voice form, agenda, and influence. Thirdly, our review identifies the outcomes of non-traditional workers' voices as making an impact is predicated on whether the issues are integrative or distributive, creating a win-win or win-lose situation for employer and workers, respectively (Walton & McKersie, 1965).
Our understanding of employee voice will follow the dimensions espoused by Kaufman (2014). The choice of Kaufman's voice menu in this review is its all-encompassing and integrative nature. He illustrated voice movement from direct, shared, and upward communication, as argued from the OB viewpoint, to indirect and contested from the ER/ HR viewpoint. This will allow reviewing findings from the NTER voice literature through the lens of the forms of voice available to non-traditional workers, what issues dominate their concerns, and whether their voice depicts mere communication, or they can influence workplace decisions.

Traditional vs. non-traditional employment relationships
For this paper, integrating the classification criteria from previous scholars (Cappelli & Keller, 2013;Pfeffer & Baron, 1988;Spreitzer et al. 2017), we define non-traditional workers as workers with limited temporal, physical, and administrative attachments to the organisations they work for, restricted directive control from the employer, total control of their work scheduling and process; and highly flexible in their timing and location of work. Concisely, NTERs are characterised by limited degrees of attachment, low employer's control, and high flexibility, without necessarily implying vulnerability or precarity (Haapakorpi, 2021;Standing, 2018). For example, highly skilled and sought-after professionals in non-precarious jobs have limited temporal, physical, and administrative attachment to the different organisations they work for (Flinchbaugh et al., 2020), are not under the direct control of the employer (Cappelli & Keller, 2013), and are flexible with their timing and location of work (Menger, 2017;Spreitzer et al., 2017). NTERs include temporary employment, part-time employment, on-call/zero-hour, multi-party employment, and disguised employment or dependent self-employment (ILO, 2016). By temporary employment, we mean fixed-term employment. It covers workers in seasonal and casual work and those working on a project-basis for different organisations. Fixed-term contracts are terminable by date, end of an event, or completion of a task/project (ILO, 2016). While part-time work involves reduced work hours (usually less than 30-hours/ week) with some relative degree of permanency compared to other temporary workers (Zeytinoglu & Cooke, 2008), on-call/zero-hour work is characterised by a high level of unpredictability in the determination of work hours. Multi-party employment involves more than two parties (e.g. temporary agency work), while disguised work involves workers whose legal status is 'hidden' due to the nature of their engagement (e.g. freelancers, gig workers).
Authors (e.g. Ashford et al., 2007;Schoukens & Barrio, 2017) identified the core elements of TERs to include contract permanency, specified working hours, work done on employers' premises, continuous employment, long term career expectation, and income security. Extant review of the NTER literature pointed to the differential, triangular and alternative viewpoints to conceptualizing these employment arrangements. While the differential perspective, using a sui generis approach, views NTERs as an apparent deviation from the attributes and practices of TERs (hence, tagged 'bad jobs') (see Kalleberg, 2000;McGovern, Smeaton & Hill, 2004), the triangular viewpoint conceptualises NTERs from the lens of tripartite employment relationship patterns, arguing that a significant chunk of non-traditional workers can be found in temporary agency work (see Cranford et al., 2003;Feldman, 2006). The alternative perspective views NTERs as an alternative to TERs, which can be a choice for employers and employees (Katz & Krueger, 2019;Spreitzer et al., 2017). Integrating the three viewpoints, scholars (e.g. Cappelli & Keller, 2013;Spreitzer et al., 2017) argued that efforts at classifying jobs as TERs vs. NTERs are eroding as features attributed to each job category are permeating and intersecting. For example, traditional workers now work remotely with increasingly flexible work schedules. Hence, based on identified attributes, while TERs are not necessarily good, NTERs are also not necessarily bad.
Hence, for clarification, we draw on the recent works of ILO (2016) and Flinchbaugh et al. (2020) to enumerate the different work arrangements' features. The criteria used for comparison are derived from the employment relationship elements drawn from Chadwick and Flinchbaugh (2016) and Schoukens and Barrio (2017). Table 1 gives a view of the distinctions.
From the table, it is apparent that traditional workers have a great sense of personal subordination (i.e. total employer control), experience bilateral employment relations, have mutuality of obligations with employers, receive their pay directly from employers, economically dependent on their employers, indefinitely work on the employer's premises, have continuous employment and long-term career expectations, and mostly protected by employment legislation (see Chadwick & Flinchbaugh, 2016;Schoukens & Barrio, 2017). Conversely, the degree of presence or absence of all the mentioned attributes varies for non-traditional workers. For instance, while temporary workers, part-timers, and on-call/zero-hour workers have similar attributes with traditional workers in the areas of employers' control, bilateral relations, and direct remuneration; they have limited or no employment legislation protection, opportunity to work for many employers, economically independent, short-term career spell, and sometimes definite work duration. Non-traditional workers in temporary agency work and disguised employment have more evident employment conditions different from that of traditional workers. Table 1 indicates lack of personal subordination, multi-party relations, absence of mutuality of obligations with employers, indirect remuneration, economic independence, on-demand work schedule, indefinite and unpredictable work location, short-term career expectation, and lack of legislative protection.
In a nutshell, the employment relationship experiences of traditional and non-traditional workers differ. Hence, how and through what structure non-traditional workers experience voice, representation, involvement, and participation at work is vital to understanding their employment relationships' psychological composition. Studying the voice behaviour of these workers in their variations and different level of dissections; addresses call from scholars (e.g. Wilkinson et al., 2020) to explore the experiences of workers in NTERs. Note. eP-employers' premises, fT-full-time, neP-non-employer premises, ID-indefinite, aa-always available, uPD-unpredictable. source: adapted by the authors from flinchbaugh et al. (2020) and Ilo (2016). a This is dependent on the context as some countries, e.g. spain, offer some protection for the dependent self-employed.

Literature search
A major constraint in reviewing previous voice studies in the NTER context is its under-exploration. However, doing a systematic review in an under-studied field is not unusual (see Kelan, 2018;Kilby et al., 2018;Semrad et al., 2019). The voice construct's multidisciplinary nature necessitates the need to capture studies from varying schools of thought and methodological approaches. The search covered articles across OB, ER/ HR, and Labour Economics disciplines. It was limited to empirical studies published since 1984 when the voice construct was adapted to an organisational context by Freeman and Medoff (1984). For a thorough search, databases of online libraries such as EBSCO, Emerald, JSTOR, SAGE, Scopus, Springer, Taylor and Francis, Web of Science, and Wiley were combed for studies of interest. The main keywords used in the search were 'employee voic*' , 'worker voic*' , 'non-traditional employ*' , and 'non-standard employ*' . Included in the search were synonymous terms for these concepts as identified in the literature. For instance, 'employee participation' , 'employee involvement' , 'employee communication' , 'employee consultation' , 'labour disputes' , 'grievance studies' are synonyms of employee voice as identified by previous scholars (Wilkinson & Fay, 2011). Similarly, terms such as 'contingent work' , 'flexible employment' , 'alternative work arrangements' , 'on-demand work' , 'on-call work' , 'zero-hour contract' , 'casual work' , 'part-time employment' , 'temporary employment' , 'temporary agency work' , 'triangular employment' , 'gig work' , 'platform work' , 'independent contractors' which are used for NTERs (Collins et al., 2019;Flinchbaugh et al., 2020) were also used in the search.

Inclusion and exclusion criteria
The criteria to choose articles for review were: (a) the papers must be either quantitative, qualitative, or mixed-method empirical studies on any of the voice dimensions (form, agenda, and influence) among workers in NTERs; (b) the papers must consider either the determinants or outcomes of voice/non-voice among non-traditional workers; (c) employee voice or any of its identified synonyms must occur in relevant parts of the article; (d) the papers must be published in peer-reviewed scholarly journals; (e) the papers must be written in English. Grey literature that did not meet the above criteria was excluded (Adams et al., 2017).

Selection of relevant studies
The search produced 54 articles. After the first round of search, we decided to look for constructs with employee voice as a dimension  (Bagdadli & Gianecchini, 2019). We found that some scales of the 'Organisational Citizenship Behaviour' (OCB) and HR practices constructs have voice or participation as a dimension (e.g. Moorman & Blakely, 1995). Hence studies on OCB and HR practices within the NTER context that used scales with voice/participation as a dimension were included for review. Studies in this category were thoroughly checked for a strong presence of voice/participation in their analysis and discussion. This gave us an additional four articles. Using the reverse process (Kelan, 2018) to capture articles not found in the keyword search, we checked the references of the relevant articles and articles citing our previously selected articles. The reverse process also generated an additional two articles, leaving us with a total of 60 articles. After considering the inclusion and exclusion criteria, we were left with 33 articles to be reviewed. Figure 1 illustrates the article selection and review process.

Review procedure for selected studies
With NVivo 12, we analysed the texts using a manual content analysis method (Vaismoradi et al., 2013) to do a narrative synthesis (Bailey et al., 2017). The narrative synthesis method is used to identify a fragmented body of evidence to bring coherence (Bailey et al., 2017). It provides an opportunity to detail the story beneath a contrasting body of evidence, allowing reviewers to synthesise the findings for data coherence. The process involves thorough reading and line-by-line coding of the issues in each of the articles to identify, categorise and analyse concepts, find links between concepts, and provide insights for future research. Using a deductive approach, we coded text from the articles with a pre-determined coding template in mind (Finfgeld-Connett, 2014). We selected key phrases or sentences to sort the issues identified into different categories and collate related or similar categories (Xiao et al., 2020). The pre-determined coding themes considered include-employee voice dimensions (form, agenda, and influence), determinants, and outcomes of voice/non-voice.

Descriptive analysis
Descriptive analysis of the included studies reveals that the majority (n = 27) were published in ER/HR-related journals. Only a handful (n = 6) of studies were published in OB and related disciplines' journals. Many of the included studies investigated the voice forms available to workers in NTERs, with 19 studies each enquiring about direct/individual and indirect/collective voice, respectively. While some studies (e.g. Maffie, 2020) investigated both forms of voice, others investigated direct (e.g. Rybnikova, 2016)  It was observed that the term 'non-traditional employment' was not used by any of the articles. However, synonymous terms such as 'non-standard employment' , 'flexible employment' , 'temporary employment' , and 'contingent work' were employed. Most of the studies cover part-time employment (e.g. Gray & Laidlaw, 2002). Thirty of the studies were conducted in developed economies while only 3 in developing countries. Six were comparative studies using data sets from surveys conducted across European Union (EU) countries. Most studies (n = 24) employed a cross-sectional research design as compared to other designs (longitudinal-2, panel data-6, case studies-7). Quantitative methods for data collection and analysis were used in 13 studies, while qualitative methods were used in 16 and mixed methods in 4.

Workers' voice in non-traditional employment relationships: forms, agenda, and influence
Results from the review depict that most studies operationalise voice from the two dominant perspectives. Only a few (n = 4) (e.g. McDonnell et al., 2014) consider a combination of prosocial voice behaviour and voice structure/mechanism for an integrated view of non-traditional workers' voice experiences. Most studies (n = 29) focus distinctly on either voice mechanisms (e.g. Rybnikova, 2016) or voice behaviour (e.g. Qian et al., 2020), with the operationalisation of voice in terms of structure/mechanism dominating these studies. Hence, we synthesise findings on the voice dimensions (form, agenda, and influence) below and depicted in Table 2. The table indicates the dimensions of voice available to the different categories of non-traditional workers.

Forms of employee voice
Direct voice was the most common form of voice among non-traditional workers (see Goñi-Legaz & Ollo-López, 2017;McDonnell et al., 2014). Among professionals in disguised or dependent self-employment, an informal pattern of worker participation was reported (Ruiner et al., 2020). Daily walk-around conversations, in-depth conversations between individual employees and managers, suggestion schemes, electronic mails, staff surveys, social networking sites, installed support software are the direct forms of voice reported (see Gegenhuber et al., 2021;Good & Cooper, 2014). However, formal meetings, workgroups, quality circles, joint consultative committees (JCCs), short-term task force, professional associations/networks, alternative grassroots groups, and virtual networks are the collective forms of voice found (see Borghi et al., 2021;Rybnikova, 2016). Participative management through organised training and workshops, trade unions, and structured grievance procedures were also forms of participation afforded to non-traditional workers (e.g. Wood, 2016). While one study found work councils and JCCs as typical voice mechanisms (Markey et al., 2002), there are more reports of a weak collective voice for NTER workers (Soltani et al., 2018). Even where workers join unions, the impact on their voice experience is minimal (McDonnell et al., 2014).
Further analysis shows variations in voice forms for the types of NTER and study context. For example, among workers in multi-party employment, two issues were identified concerning individual and collective voice. Individually, the question is whom they voice their ideas and concerns to-the agency or the client organisation. There is contention among collective representation of temporary agency employees being domiciled within the agency or the client organisation. Håkansson  (2020) found that temporary agency workers are afforded collective representation both at the client and agency levels. Conversely, Rybnikova (2016) found temporary agency workers' representation rare due to their workplaces' transient nature and the relatively small size of the temporary work agency business. For workers in 'disguised employment' , our review's findings revealed that their dispositions to collective voice are changing and that their employers are increasingly enabling voice opportunities. In examining voice mechanisms among digital platforms, Gegenhuber et al. (2021) found that platforms enable voice mechanisms to bind workers to the platform in the face of increased competition and ensure workers' job quality and fair working conditions. Borghi et al. (2021) also found platform workers collective representation can either be through trade union or alternative grassroots group. They further argued that the complementary, alliances and otherwise of the union and alternative grassroot groups depend on the context of discourse/practice. Additionally, Saundry et al. (2006Saundry et al. ( , 2007 and Umney (2016) found freelancers to be increasingly interested in collective representation to voice their dissenting opinions through professional networks. These networks are not affiliated to trade unions as the workers believed strong advocacy of unions could impede their career growth (Wood et al., 2018). This need for collective representation of these workers may be due to evidence that they are less likely to be directly consulted (Hoque & Kirkpatrick, 2003).

Employee voice agenda
Studies (n = 8) (e.g. Ruiner et al., 2020) identified shared issues as dominant. This is due to the shared issues' integrative nature against the contested concerns' distributive nature. Markey et al. (2003) reported that part-timers have more input in improving shared decisions than their full-time counterparts. Similarly, Hoque and Kirkpatrick (2003) found professionals having higher input in staffing decisions. For gig workers, communication and feedback channels provided by platforms are mostly restricted to task-based processes. This may likely be a result of the job nature of these workers which largely depends on their availability and increased workplace processes and performance (Hoque & Kirkpatrick, 2003) and, the desire of management to control voice patterns and mute dissenting voices on contested issues (Gegenhuber et al., 2021). Hence, employers are more likely to be receptive to workers' input in decisions over shared concerns. On the one hand, future workplace plans, staffing issues, change in work practices, and health and safety are the identified shared decisions workers in NTERs have input.
On the other hand, grievance, equal opportunity, compensation, and holiday/work timing are distributive issues identified.

Employee voice influence
Our review found a varying degree of influence afforded to workers in the different variants of NTERs. Some studies (n = 5) (e.g. Soltani et al., 2018;Yang, 2012) reported low level of upward communication among temporary employees, part-timers, temporary agency workers, even where they have access to collective voice (Wood, 2016). In contrast, Donnelly et al. (2012), Ruiner et al. (2020), andMcDonnell et al. (2014) reported that temporary employees, independent contractors, and part-timers could offer suggestions, air their ideas about workplace issues, and complain about issues that affect them. Identified issues include-holiday/work timing, work tasks, pay and benefits, work conditions, and training needs (Donnelly et al., 2012).
Most studies reviewed found workers in NTERs having little or no influence over workplace decisions due to power imbalance (Wood, 2016), nature of their contract/job and industry (Ruiner et al., 2020). However, McDonnell et al. (2014) argued that part-time employees influence lower-level workplace decisions and accept wittingly their inability to influence higher-level management decisions. In comparison with workers in TERs, Markey et al. (2002) reported little or no difference in the degree of influence both part-timers and full-timers have on decisions that affect them at work. They argued that both workers in TERs and NTERs are less likely to display a reasonable degree level of influence in management decisions and actions. Surprisingly, Ruiner et al. (2020) found independent consultants in the health sector having a higher degree of influence than permanent physicians. They argued that the independent consultants could influence management actions by voicing due to expertise, freedom, and loyalty to professional values. They further reported that the independent consultants' voice influences management actions for favourable work conditions, even for permanent doctors.

Determinants
We present an integrated framework of the individual-firm-external determinants of voice in NTERs. These arise from our review in Figure  2 and further explain these determinants in this section.

Individual-level determinants.
A few of the reviewed studies focused on individual-level factors. Dominant among these individual determinants is work status (e.g. see Al-Amin & Islam, 2020;Johanson & Cho, 2009;Stamper & Van Dyne, 2001). Work status is a subset of the hierarchical configuration among organisational members, and workers who perceive themselves at the lower end of the cadre are likely to eschew upward voice behaviour (Morrison, 2011) with the belief that their voice input may be ignored. Reviewed studies found non-traditional workers to be highly cognizant of their status amidst other organisational members and felt least entitled to voice discontent or grievance (Sluiter et al., 2020). For example, casual retail workers facing sexual harassment rarely speak up due to their job's temporary nature (Good & Cooper, 2014). Also, interpersonal risks assessment (Qian et al., 2020), job insecurity (Benassi & Vlandas, 2016), perceived unemployment uncertainty, and degree of replaceability among freelancers and temporary agency workers (Sluiter et al., 2020) have been reported to impede voice. For temporary agency workers, workers' unionism orientation and ideology (Hakansson et al., 2017;Heery et al., 2004), power status and fear (Rybnikova, 2016) were found as determinants.
In addition to Sluiter and colleagues' findings for workers in disguised employment, Ruiner et al. (2020) and Maffie (2020) also found freedom, expertise, autonomy and exit, and conflict with customers/clients as determinants of freelancers and gig workers' voice. Findings from these studies revealed that independent freelancers with a high level of expertise, freedom, and autonomy would voice their discontent and ideas compared to standard workers (Ruiner et al., 2020). Similarly, for digital platform workers, conflict with clients/customers propels information sharing and discussion among colleagues, a recipe for gig workers' collective representation (Maffie, 2020).

Moderators and mediators.
Our review revealed that intervening variables influence the linkage between some individual-level determinants and non-traditional workers' voice. For moderators, preferred work status and organisational culture, and political savvy were identified (e.g. Qian et al., 2020;Stamper & Van Dyne, 2001). For mediators, few studies examined factors explaining how some individual-level determinants influence nontraditional workers' voice. Self-efficacy (Qian et al., 2020), social connection, and social media interaction (Maffie, 2020) were identified. Gender was also found to explain how work status influence non-traditional workers' voice (Booth & Francesconi, 2003;Hoque & Kirkpatrick, 2003). These studies reported less collective voice for both men and women, with the degree of union-coverage for women being lesser. Similarly, non-traditional female workers in non-professional jobs or the public sector are likely to be covered by a union compared to professionals and those in the private sector.

Firm-level determinants.
Studies examining the influence of firm-level factors on non-traditional workers' voice identified size and workers' composition, management ideology and control, structural and social factors, and cost as determinants. Concerning size, studies reviewed found a significant relationship between firm size and workers' composition and availability of workers' participation. Both Markey et al. (2002) and Markey et al. (2003) found small workplaces (<50 workers) to have a low incidence of workplace participative practices compared to larger firms. Gegenhuber et al. (2021) also found that platform size impacts how they perceive voice and the need to give workers voice opportunity. Also, they established that digital platforms consider crowd workers' composition (heterogeneity/homogeneity) before installing voice mechanisms to balance diverse perspectives and backgrounds of the workers. Management ideology and control were also a significant determinant of non-traditional workers' voice (Gegenhuber et al., 2021;Rybnikova, 2016). Among platform organisations, Gegenhuber et al. (2021) found the ability of digital platforms to shape workers' voice direction as a prerequisite for installing voice mechanisms. They argued that digital platforms prefer direct voice mechanisms in a bid to retain control as they refrain workers from taking collective action. Similarly, Rybnikova (2016) found that management leadership behaviour and control influence temporary agency workers' voice, arguing that authoritative leadership behaviour and control of temporary agency workers by supervisors in the client organisation stifle their voice. In terms of structural and social factors, Rybnikova (2016) suggests that workers' duration of assignment, working arrangements, designation of specific tasks, and the role of agency influence temporary agency workers' voice. In addition to firms' structural factors, we identified cost and resources as an influencer of an organisation's willingness to install voice mechanisms for non-traditional workers. Findings demonstrate that the decision of firms/platforms to design voice structures is highly influenced by the resources required to install and maintain the voice system (Gegenhuber et al., 2021).
External determinants. Our review found that researchers have extensively studied a few external determinants-industry attributes, institutional context, national legislation, and professional network. First, previous studies found that the industry where non-traditional workers work determines the level of voice opportunity available to them and the extent to which they can influence management decisions (Ruiner et al., 2020). Research demonstrates that the industry/sector of firms influences the level of flexibility they can adopt (Kalleberg, 2000) and impact how the degree of flexibility can influence employee outcomes (Guest, 2004). For instance, independent professional workers in the health industry were found to have more voice influence (Ruiner et al., 2020) compared to professional workers in the gig economy with a low level of influence over management actions and decisions (Gegenhuber et al., 2021;Maffie, 2020).
Second, Benassi and Vlandas (2016) and Pulignano and Signoretti (2016) considered institutional context. It was argued that national employment institutions and collective bargaining structures available in any country impact the use of non-traditional labour. The more structured the national employment, labour market, and collective bargaining institutions are, the more likely it is for all forms of workers (traditional and non-traditional) to have similar pay structures, employment terms and conditions, and be covered in collective agreements (Pulignano & Signoretti, 2016). This is in line with the argument of Oliver (2011) that with strong bargaining institutions and wider bargaining coverage, 'outsiders' can also benefit from 'insiders' institutions. Third, researchers have also explored the importance of national labour legislation for non-traditional workers' voice. In contexts of stricter employment legislation and protection for traditional workers and with little or no protection for their non-traditional counterparts, firms are more likely to exploit the vulnerability of these workers and increase their usage, except where legislations are specifically targeted to protect these workers (Pulignano & Signoretti, 2016).
Lastly, reviewed studies (Saundry et al., 2006(Saundry et al., , 2007(Saundry et al., , 2012Umney, 2016) found professional networks to significantly influence freelancers' collective voice. They argued that such worker networks serve as a springboard to a collective determination of their terms and conditions. Professional networks serve as advocates for freelance and project-based workers and as an avenue for the expression of discontent and bonding with colleagues. Hence, professional networks are likely to offer disguised workers some form of collective or network voice (Prouska & Kapsali, 2021).
Further results from our review indicate that, contrary to previous research which argues that low employment protection and regulation motivates union inclusiveness and collective voice for temporary agency workers, Benassi & Vlandas (2016) examined and affirmed high bargaining coverage, high union authority, and workers' movement ideology as determinants of collective voice for temporary agency workers.

Outcomes
Our analysis also points to the outcomes and consequences of voice. Unlike the many individual and organisational level outcomes reported for voice among traditional workers, few studies (n = 7) (e.g. Kroon & Freese, 2013;Piasna et al., 2013) examined how voice impacts turnover intention, job quality, job satisfaction, and sexual harassment of temporary workers, temporary agency workers, and part-timers. Besides, perception of autonomy, quality standard, workplace flexibility, and workers' perception of having fair working conditions was found as consequences of non-traditional workers' voice. Further findings reveal interactions between the outcomes of voice among non-traditional workers. For example, Good and Cooper (2014) found that temporary workers seldom exercise voice behaviour when they face sexual harassment and highly dissatisfied with their employers' responses.

Discussion and directions for future research
As one of the first systematic reviews to synthesize empirical evidence on non-traditional workers' voice, we departed from the previous voice reviews (Morrison, 2011(Morrison, , 2014Mowbray et al., 2015), which identified the antecedents and decision-making process of traditional workers' voice, to develop an integrated understanding of the voice form, agenda, and influence within the non-traditional work domain. Following calls from scholars (Wilkinson et al., 2020), we adopted the integrative perspective of voice and argued following Kaufman (2015) that employment contextual factors may likely influence workers' voice. Kaufman argues that the labour market's state concerning opportunities available and the employment climate determines employee voice. Similarly, Wilkinson et al. (2018) acknowledged that workers' diversity, concerning personal and institutional factors, shapes voice. Wilkinson et al. (2021) further argued that social, economic, and technological developments such as alternative/flexible work arrangements and the presence of marginalised/minority groups at work may threaten workers' voice. Therefore, non-traditional workers' voice concerning their labour market experience and employment climate may differ from those of traditional workers. We rested on this reasoning and extended it to have a cursory look at empirical studies on non-traditional workers' voice. Hence, this synthesis's objective was to conceptually map the evidence to date of traditional workers' voice determinants and outcomes.
We reviewed empirical studies on voice among non-traditional workers published since 1984 and identified factors that determine such workers' voice behaviour and outcomes. Thirty-three studies met our inclusion criteria, and findings from these studies indicated that a wide variety of voice channels and mechanisms are available to non-traditional workers. These variations are determined by the aims of the studies, type of NTER studied, contextual differences, varying predicting factors, and methodological differences. For instance, while voice among part-timers, temporary and disguised workers has received considerable attention, studies on temporary agency workers' voice remain limited. Similarly, individual-level determinants of voice have been considerably explored, and many factors have been reported as determinants, mediators, and moderators of NTER workers' voice. However, firm-level, and external factors influencing voice among NTER workers remain largely understudied. Corroborating our findings is the call from Morrison (2014) for the urgent need to consider more macro-level external factors in the determination of workers' voice.
Overall, we found that voice research among NTER workers is growing and receiving increasing attention. However, the studies remain fragmented. We address this need for integration by presenting a compilation of findings of voice research in NTERs in a framework (see Figure 2) and add to voice theory and previous work at conceptualising voice (Kaufman, 2015;Mowbray et al., 2015). Our review paves the way for more research into voice in NTERs. We set directions for future research opportunities following the identification of research gaps across the variants of NTERs studied, voice dimensions, context, determinants, and outcomes of voice. Based on the identified strands, we recommend five significant directions for future studies.
First, most of the studies we reviewed are domiciled among part-timers. However, with the tremendous growth in triangular employment and gig work (Hudson-Sharp & Runge, 2017), more research is needed to study the experience of temporary agency, project-based, and gig workers. For instance, both temporary agency and project-based workers are seconded to work with different organisations and on different projects at different times by the employing agencies or firms (Rybnikova, 2016;Turner, Huemann & Keegan, 2008). Future research can investigate who is responsible for initiating voice practices in triangular employment and how frequent change in organisational settings influences these workers' voice experiences.
Second, more research is needed on the individual-level determinants of voice in NTERs. Beyond gender, future studies should consider investigating how other socio-demographic attributes of workers in NTERs (e.g. education, employment duration, work experience, financial status, social status) influence their voice behaviour. For instance, there is the possibility that a non-traditional worker with good financial status who decide to work for personal fulfilment, identity, care for others, or service (Budd, 2014) would exercise more voice compared to a financially unstable worker who views work as a commodity or disutility (Budd, 2014). More studies on organisational level determinants are needed to understand non-traditional workers' voice, such as the influence of a narcissistic leadership style on voice.
Third, future research is highly pertinent to the outcomes of voice, or lack of voice, among non-traditional workers. Previous scholars (e.g. Wilkinson, Townsend, Graham & Muurlink, 2015) have argued that workers' voice, irrespective of employment contract, impacts the worker-firm relationship. While traditional workers' voice outcomes have been widely studied (see Holland, Allen & Cooper, 2013;, implications of voice or lack of it for non-traditional workers have been largely understudied (Wilkinson et al., 2018). This is relevant to the call for employee voice inclusion in workplace practices as favourable consequences of employee voice (e.g. wellbeing, engagement) have been widely verified in voice research, while the absence of voice has been linked to adverse outcomes (e.g. insubordination, absenteeism).
Fourth, despite reports from international labour and human rights organisations on the abuse of labour and the high level of precariousness experienced by workers in NTERs in the developing and underdeveloped economies (ITUC, 2019), little research has been completed on voice experiences of workers in NTERs in these climes. Studies from low and middle-income countries and other emerging nations would provide a broader understanding of the state of non-traditional workers' access to individual and collective voice.
Fifth, our review showed the dominant usage of cross-sectional research design, survey, and panel data collection method in previous studies. More longitudinal studies are needed exploring non-traditional workers' work duration within or across firms, labour market conditions, and terms and conditions of work. In addition, more qualitative studies are needed to gain deeper insights into these workers' voice experiences.

Conclusion
This review article explored the voice experience of workers in NTERs, established the research gaps, and suggested avenues for future research directions. We identified individual, firm, and external determinants of non-traditional workers' voice and the consequences of their voice and non-voice and presented these in an integrated framework that acts as a guide for future research. Besides Prouska and Kapsali's (2021) effort at modelling the determinants of project worker voice in the temporary employment context, previous integrative voice models (e.g. Kaufman, 2015;Mowbray et al., 2015;Nechanska, Hughes & Dundon, 2020) captured determinants of traditional workers' voice. These past models are broad, without special consideration for employment situation peculiarities. Joining Prouska and Kapsali (2021), our framework expands voice theory to non-traditional employment by capturing the peculiar elements of the NTER context. Specifically, we add to the voice theory by identifying individual-level and intervening (mediating and moderating) factors that influence non-traditional workers' voice, and their voice outcomes.
The limitations of our study are twofold. First, only empirical studies in peer-reviewed journals were included, leaving out conceptual papers, books, book chapters, working papers, theses, and dissertations. This was a conscious decision as the purpose of the review was to capture empirical work published in peer-reviewed academic journals to ensure the quality of the analysis. Second, the relatively small sample of articles reviewed meant that we could not fully account for the heterogeneous nature of the composition of workers, which may create varying experiences of voice. Neither were we able to account for differences in the methodological approaches of the studies we reviewed. However, the future directions we proposed in the paper include avenues to overcome these shortcomings in future research.

Data availability statement
We confirm that the data supporting the findings of this study are available within the article [and/or] its supplementary materials.