Table of Contents
Although Tonga is small and its impact on the global geo-political stage is limited, the way in which the country fits into the contemporary global system has attracted its share of attention. Since Marcus’ early and cogent observations on the fact of Tongan transnationalism (1981), a great deal of ethnography has been done both in Tonga and with Tongan communities overseas. In just the last 10 years there have been significant full length ethnographies of contemporary Tongan political economy. Evans (2001) and van der Grijp (1993, 2004), for example, have written extensively on the way that the current Tongan economy is shaped by the world economic system. While van der Grijp is interested in processes of globalisation, Evans’ work extends to transnationalism—but more to the manifestations on transnational practice in Tonga than a considered assessment of the Tongan transnational system as a whole (see Lee 2006). More detailed work on remittance practices has also been done (Brown 1994, 1995; Brown and Connell 1993). Others, including Besnier (2004), have written eloquently on the ways in which Tongan economic and linguistic practices have embraced a cosmopolitan view of both themselves and the world, and subtly on shifting engagements with tradition and exchange (Addo and Besnier, 2008).
A second trend in recent ethnographic work has been a focus on the Tongan migrant communities; work by Lee (2003) and Small (1997) has focused on the lives, practices and prospects of Tongan migrants. Most recently, we have seen work by Tongan ethnographers, like that of Ka’ili (2005), which focuses on the continuities in Tongan cultural practices in overseas communities.
Much contemporary work derives from concern about just how robust transnational ties are, especially over generations (see Lee 2003, 2004, 2006). Questions about the stability of migration streams, the potential for capital accumulation via migration and remittances, the stability of remittances from migrating and second-generation Tongans are key for Tongan individuals and families, and indeed, for the greater Tongan polity and economy. Whether and how remittances streams continue over time is, arguably, the core element to the stability of many Pacific Island economies, and certainly this is true of Tonga. In her 2004 article ‘“Second generation” Tongan transnationalism: Hope for the future’, Lee develops an argument in support of the ‘remittance decay hypothesis’ (from Brown and Foster 1995), based primarily on her discussions with second generation and younger migrants who indicated that they have no intention of remitting to people in Tonga in the future, in part because they had no one left in Tonga to whom to remit.
Remittances play a fundamental role not only in the Tongan economy, but also in the maintenance of the Tongan gift exchange system. Evans (1999, 2001) shows in detail how remittances of both cash and kind play an on-going role in the creation and maintenance of relationships between both individuals and groups in multiple locations within the transnational system. While it is true that the notion of ‘transnational corporations of kin’ developed by Bertram and Watters (1985; see also Marcus 1981), is problematic (see James 1993 and Munro 1990), the notion that remittances are then to be understood in terms of the actions of individuals (James 1991, 1993; Lee 2004) is also debatable (Evans 1999: 143-144). Gift exchange, of which remittances are a type, are inherently relationship producing. Some of the greatest threats to the continuity of resource streams into Tonga are probably those based on the attenuation of gift exchange relationships which include people in Tonga, because of potential loss of the capacity to use lands to produce materials (Evans 1999). One of the problems with all these assessments of trajectory of the Tongan transnational system however, is that there is not yet a demonstrable decay in the flows of material through what we are calling the Tongan Transnational System (TTS). This is not to assert that the resource flows that energise the TTS need remain stable, but rather to make the claim that in spite of some very obvious transformations in the social, political, and economic landscape in the TTS, a decline is still not demonstrable—this in the face of claims that the end is nigh that go all the way back to Christine Ward Gailey’s work on commodification in the 1980s (Gailey 1987).
While a good deal of attention has been paid to Tongan transnationalism, a contemporary anthropology of Tonga and the TTS is problematic because the system is geographically large, and the people operating within the system so diverse (Lee 2004, 2007a, 2007b; Small 1997). The anthropological conceit of a unified singular ‘culture’ that can be investigated and described has been under pressure for quite some time. This pressure came first from political economists who demanded that the embeddedness of economies and cultures within the world system be recognised (Asad 1973; Wolf 1982; Worsley 1984), and more recently from post-modernist authors suggesting, not unreasonably, that the experience of human social life is positioned by such things as age, gender, class and sexuality, and thus totalitarian images and representations of a culture were just that: totalitarian (see for example Haraway 1988). It is this later sensibility that shapes much recent work. After the little girl Dorothy in the Wizard of Oz, we might observe that no one, not even people actually living in Kansas, is in Kansas anymore.
In this paper we argue, by way of demonstration, that there is some untapped and certainly under-used potential in quantitative methods for assessing and describing contemporary transnational systems. Currently most quantitative analyses regarding Tonga are devoted to economic data—remittances and the like—but in other areas of the social sciences quantitative methods are used over a much wider range of phenomena. For several reasons, statistical analysis of sociological data is virtually non-existent in contemporary anthropology. While we recognise that there are real and immediate limitations and constraints to quantitative questionnaire style research, in this paper we discuss the results of just such a quantitative survey undertaken in Tonga and in the Tongan community in New Zealand. We offer this as a small contribution to conceptualising and describing the Tongan Transnational system—one of the striking elements of this way of describing the TTS is that fragmentation of attitudes and commitments, and therefore behaviour, seems less pervasive than some ethnographic accounts might suggest.
Developing a prognosis for Tongan migration and remittance behaviour on the basis of attitudinal assessments is not uncommon (James 1991, Lee 2004, Small 1997). Indeed we might make a cautious claim that in the absence of economic data demonstrating the expected decay in remittance behaviour, attitudinal assessments are the meat of the argument. But one key limitation to our current understanding of how people throughout the TTS think about the values and practices that underpin the system, is a lack of systematic assessment or reporting. That is, because qualitative methods are generally employed, it is difficult to determine the relative intensity or distribution of attitudes within the TTS. Though it has become quite uncommon to use quantitative surveys in anthropology, one of the advantages of such techniques is that analysis of attitudes, and most importantly the differences in attitudes within a sample, can be conducted. It is important to note here that difference within a sample is important in terms of how it is positioned—that is, for example, in terms of how gender, age, and location (i.e. subject position!) are related to variations in attitude. Neither the critique above, nor the analysis offered below, are intended to suggest that ethnographic representation is somehow flawed; no quantitative analysis can replace the impact of nuanced ethnographic practice. Nonetheless quantitative analyses do have something to offer, and this something has been all but lost to anthropology over the last couple of decades.
To develop such a quantitative description of the TTS, a questionnaire-style survey assessing the attitudes of people regarding key elements of Tongan identity was conducted in Tongatapu and Auckland in 2005. Participants were Tongan adults (n=691) aged 16 years and older living in Tonga (n=504) and in the Tongan community in Auckland (n=187).[2]
The survey used was a questionnaire developed collaboratively and administered to each of the respondents.[3] It was comprised of four sections relating to: 1) demographics; 2) elements of Tongan identity; 3) rank order of important characteristics of a Tongan; and 4) characteristics of a good woman and a good man. This paper makes use of the first two sections: demographics and elements of Tongan identity.
In the ‘demographic’ section of the survey, people were asked basic questions such as gender, place of birth, place of residence, year of birth, time spent away from Tonga, level of education, occupation and church denomination membership. The ‘elements of Tongan identity’ section consisted of a series of items in which people were asked to assess their responses to particular statements along a seven-point Likert scale.[4] For example:
Over the course of a lifetime, these things are the responsibilities of all Tongans:
Teaching daughters to weave and make ngatu (i.e. barkcloth)
Seven dependent variables (or factors) were derived from the 64 items in the ‘elements of Tongan identity’ section of the questionnaire using a factor analysis.[5] This technique is one that groups statistically related individual items into larger and more robust meta-items (i.e. factors) that can then be used for subsequent analysis. In lay terms the technique establishes which items relate to the same underlying attitudes and groups them together. That is, the items in a factor are grouped together because they follow the same patterns of variation (are statistically similar to one another). Only those factors that include a minimum number of items (in this case, three) and meet certain statistical tests for reliability and robustness are used. The items included within each factor were then assessed in order to identify the common theme unifying the items. The items that form the factors are interesting in and of themselves as they are indicative of a collective view of the inter-relation between societal values.[6]
[1] At the request of the editors we present this analysis in a colloquial style. We do so out of a desire to make the paper accessible to the anticipated readership of this volume and in the knowledge that this may disconcert those more familiar with the conventions of quantitative sociological work. Such is the current schism between qualitative and quantitative traditions in the contemporary social sciences that positioning this paper is somewhat problematic—we have done our best to balance the sensibilities of both broad traditions here.
[2] The sample was selected by Paul Harms; this sample was serendipitous. Most of the data was collected within church congregations, church choirs, and other institutional contexts both in Tonga and New Zealand; these were places where Harms was more and less known to the people involved. The best return rate on forms handed out was in contexts where he was most familiar with the greatest number of people and they were about as familiar with him (see Evans 2001, 10–11 on the significance of social relationship and survey response in Tonga). The church minister, the choir director, or the school principal was asked to introduce the researcher (Harms) and then the research was explained. The endorsement from a trusted leader was significant; survey forms were only handed out in places where such a leader was known to the researcher. In Auckland, in various congregations of the Siasi ‘o Tonga (Church of Tonga) a church official introduced Harms. Initial contact happened at a regular Monday night faikava (kava drinking party) which involved young men in the Siasi ‘o Tonga from all over Auckland and often also their church ministers. In some places, people filled in the forms immediately, but usually people would take them home and return them later. Some forms were also completed by Tongan students at the University of Auckland. Those were mostly filled in on the spot, although some were returned later. Similarly, a small number of forms were given to friends who handed them out to friends and colleagues; these were returned filled in. Though this sample was not random, we have every reason to believe it representative; further, the techniques for analysis have been selected to control for variation in gender, age, location, and level of education.
[3] The survey was developed by Paul Harms in both English and Tongan (of which he is a fairly fluent speaker). Half of the questions were written in each language, and then translated to the other. The questions and their organisation were based on a combination of ideas about the key issues in Tongan identity. This first draft was revised (and reduced in size) by Evans and then the questions were revised again in Tongan by Harms. Harms’ questions were then evaluated by a number of Tongan first language speakers, including Siaosi Kavapalu, Seini Laungā, Suli Liava'a, and Fatai Vave, who all made important comments and suggestions. Tongan demographer Viliami Liava'a’s input was especially helpful. There was a final revision of the content and wording of the Tongan form further according to these comments, and then a final revision to the English version to match it. Distribution of the survey in Tonga was assisted by Siaosi Kavapalu and Sione Koloamātangi.
[4] Because of concern over the impact of having a seven–point scale but only five written cues, the seven–point scale was reduced to a five–point scale by collapsing the values for 2 & 3 and 5 & 6.
[5] A factor analysis using principal components analysis and varimax orthogonal rotation was used for scale construction. The distribution for each item was examined to identify any non-discriminating items (those for which 90 per cent or more chose the same response) for deletion. The criteria for retention of a factor for the purpose of scale construction were: 1) at least three items retained per factor; 2) eigenvalue of at least 1.0; 3) interpretability of the factor. Individual items were retained for each factor analysis derived scale when associated factor loadings were 0.40 or higher (Pett, Lackey and Sullivan, 2003). Pett et al. (2003) also suggest 10 to 15 subjects per item for factor analyses, a standard met in this study. The retained items identified in each factor were assessed for internal reliability, based on a Cronbach’s alpha of at least 0.65, and then summed to create each respective scale. Among the 17 factors having eigenvalues of 1.0 or greater, seven met the criteria for inclusion laid out above. Further statistical detail is beyond the scope of this paper.
[6] Herein lies one of the key problems with quantitative techniques—this statement is true only insofar as the questions asked reflect societal values in reasonable ways. In other words, asking sensible questions is a pre-requisite for getting sensible answers! In this study, the creation of good questions was achieved via the process outlined in footnote #3.