Chain of representations

The heterogeneous nature of the Barents Region as a research area raises questions regarding the comparative research setting, such as possible problems with variations in meanings between different languages and cultures. This question has to be considered carefully in such research, which involves respon-dents representing several different languages, heterogeneous living environments and cultural heritages (Cartmel 2003; Puuronen 2000, 252). In this research, the challenges of multilingual and -cultural research settings were confronted already in the very beginning. The survey questionnaire had to be designed in a way that it would be as suitable as possible for the purpose of collecting data from a heterogeneous research area. It was necessary to write the first draft versions of the questionnaire already in English, because only a few of the research partners who helped with practical and cultural issues were Finnish-speaking. The final version of questionnaire was translated by three different professional translators. Even though the translators were professional, the comparative research setting always includes some doubt about the use of coherent and suitable concepts in every language and culture (Cartmel 2003, 92).

In addition, even though there were discussions between the researcher and each translator about the meaning of the questions and concepts, some divergences in the questions may have occurred because of translation. Thus the respondents in each country may have understood the questions a bit differently and answered on the grounds of slightly different points of view; certainly upon different cultural heritage.

The possible deviations and/or variations in the data which evolved during the translation questionnaires and the answering the questions are only part of

consequences of the representation¹⁴ chain. The whole representation chain con-sists of different actor levels within different phases of the survey; all the way from the beginning of the research process – setting the research question, de-fining research population (see Hirsjärvi et al. 1997, 179; Moser & Kalton 1971, 53-54) and designing the questionnaire – to the moment when you, the reader, have read the last page of this thesis.

The original idea of the representation chain was inspired by Eco’s (1990, 1992) claim that text has always had many different meanings, including at least the following two: 1) intentio auctoris, the author’s meaning and 2) intentio lec-toris, meaning constructed by the reader. Many of the difficulties associated with comparative research are especially derived from the ‘problem of meaning’, i.e.

the equivalence in meaning in different languages, and the issues of interpreta-tion and representainterpreta-tions. It can be argued that all the actors involved in the re-search process, such as translators and respondents – with their different possible societal and cultural backgrounds – have their own meanings and representations of research issues, as their role may be (Cartmel 2003, 97-98; on equivalence in meaning: Eco 2001, 9-20). The chain of representation is thus marking those ar-ticulation points in which the angle and producer of representation has changed, and which are thus the most probable sources of fallacy in respect to the re-searcher’s original meaning, intentio auctoris, in this research. One must ac-knowledge that every stage of the research process, especially in survey research, is a potential source of inaccuracy (Moser & Kalton 1971, 45). The idea of the representation chain is illustrated as follows:

Figure 2: The chain of representations

36

14 Representation here refers to subjective and individual reflections, ideas and interpretations of the surrounding social life and world. Representations are thus personal, and as Duncan & Ley (1993, 4) have stated, “our representations of the world cannot be other than ‘partial truths’. The researcher’s representation, intentio auctoris, is also only a partial truth; “not an absolute con-cept” (Kuusisto-Arponen 2003) or representation of issues in question.

37

The first link in the representation chain is the researcher, who sets research questions, defines the sample population and designs¹⁵ the questionnaire¹⁶. Within this process the researcher is investing the questionnaire with her own representations of the research issue and setting. These representations are formed on the grounds of an understanding of the preliminary theoretical frame and research context. Second link consist of translators, who are translating the questionnaire¹⁷. In the translation process translators are using their personal understanding and knowledge of the research questions and concepts used, but also contextual information received from the researcher. Their representations are actualised on a conceptual and linguistic level.

The third level – and link in the chain – of representations is the respondent, who fills in the questionnaire and produces again a new representation of the research questions on the basis of his/her personal and local perspective and situation. The important thing here is that the respondent is able understand the words and the meanings of terms used in the questions, and to form an idea and representation (which is as similar as possible to the researcher’s original repre-sentation, i.e. intentio auctoris) of issue under investigation; and he/she is also able to formulate another, individual, representation in answer to the question, which will in turn be intelligible to the researcher (Bechhofer & Paterson 2000, 77).

The fourth level, and again another type of representation, is formed by re-searcher during the data analysis. All previous representations are cumulated during the analysing process. This accumulation is inescapable, since represen-tations that have existed in the minds of translators and respondents are embed-ded in the data, indistinguishable for the researcher. Thus it is impossible for the researcher to compare, reflect and separate her own representations from other peoples’ previous representations. The researcher no longer has any means of correcting possible previous misunderstandings at this point. Nor is there much which can be done at this level in order to verify equivalence between the repre-sentations of other actors in the chain of reprerepre-sentations, or to locate possible sources of misinterpretation.

On the last level of the chain of representations the reader introduces his/her own intentio lectoris – yet another fresh, personal viewpoint and context of un-derstanding – to the written research. This last link in the chain differs from for-mer links in that it does not process or affect the research process itself. The connection back to the first link exists only in the ideal situation where the reader is giving feedback to the researcher who has designed survey questionnaire and

15 Errors arising from the questionnaire itself are typically in relation to the length of the questionnaire; the order, structure and wording of questions; and the level of language used (for children or for specialist groups) (Moser & Kalton 1971, 390-391; Bechhofer & Paterson 2000, 81).

16 In this case the original questionnaire was written in English. This phase of research may al-ready have been a possible source of error, since English is not researcher/author’s mother tongue. Some words and concept used in the questions may have had particular connotations or meanings which the researcher was not familiar with.

17 In this research the original questionnaire was written in English and it was translated into languages used in the research area: Finnish (the researcher’s own mother tongue) Swedish (the translator’s mother tongue), Norwegian (not the translator’s mother tongue) and Russian (the translator’s mother tongue).

38 analysed the received data. If this ideal situation materialises and the reader re-ports about possible failures in the analysis or new, contradictory representa-tions, it may result in the scientific principle of self-correction being carried out (Frankfort-Nachmias & Nachmias 1996, 21). This, of course, requires that re-searcher is willing to conduct further research and write new, more reflective texts about the research issue.

This idea of the chain of representation is an important point when thinking about the research process, the production of information and data analysis criti-cally, and trying to find possible sources of distorted information in research which involves respondents from several different countries, and where there is more than one language being used in the research setting and the data. In this research there have been many articulation points in which the representations produced (by translators, respondents and readers) may have diverged from the intentio auctoris – the researcher’s original, theoretically guided meaning, under-standing and representation of the research problem and issue.

It is important to note that intentio auctoris in this case is the researcher’s subjective notion of the research issue and the world around us. As such, repre-sentations corresponding to intentio auctoris are still subjective and only ‘partial truths’ about the world (Duncan & Ley 1993, 4) constructed on the grounds of personal experiences and notions of place. In the case of social sciences, re-searchers’ personal ways of thinking and theoretical perspectives are inevitably embedded in the formulation of research questions and questionnaire wording.

Intentio auctoris is partly situational and contemporary, because it is dependant on the author’s discipline and knowledge of contemporary scientific debates – even trends – and dominant theoretical concepts. We must also remember that aspects of the problem of meaning such as these are virtually impossible to avoid or to eliminate completely in this kind of comparative research setting.

The chain of representation shows us the articulation points in data collec-tion. It is fair to assert that in this research the chain of representation is not pro-ducing fundamentally false information, since the topic of the questionnaire is fairly common and certainly familiar to all respondents: they all have personal opinions about their living environment, their personal future and possible mi-gration plans. Translators as well, relatively speaking, share at least this much cultural context with both the researcher and respondents. In addition, translators are also familiar with local resources and their profession and educational back-ground makes them aware of the problems of meaning in translation.

Besides the consequences of the chain of representations, survey research is criticised for a lack of personal contact with the respondents. It has been argued that the survey researchers often end up with false conclusions, especially in quantitative analysis, because of a lack of personal contact with respondents and because the structured response options for each question prepared in advance by the researcher may not fully cover the opinions of respondents. These aspects of data have to be carefully considered during the analysis, especially when the

39

research issue and context is unfamiliar to researcher¹⁸ (Heikkilä 1999, 16). Also international comparative research in general is seen to have certain pitfalls.

Standard problems in these comparative research settings are, e.g., language skills, difficulties in making reliable comparisons, difficulties in formulating identical and precision measures, and differences in data collection methods between countries where secondary data is used (Cartmel 2003, 86, 90). In this research, the usual difficulties associated with comparative research are taken into account by keeping sampling compatible in each country. Data here has been collected – i.e. the questionnaire designed and the most important decisions made – by only one researcher (the author).

The nature of a comparative research setting usually requires the utilisation of extensive and comparative data. This in turn means that data can be regarded as superficial but broadly based (Heikkilä 1999, 15). The superficiality critique is usually posed because the researcher is not getting very deep into the lives of respondents using a survey method, especially if only structured questionnaires are used. A survey investigates all respondents according to the same pattern, thus potentially overlooking the subjective features and experiences in different life situations. This may also be the case with surveys used to collect data in youth studies, and survey research targeting young people has been criticised for this reason. It is said that standardised survey questionnaires are not suitable for studies dealing with young people, because it is difficult to focus on societal processes and chances that are relevant and important for young people through a survey. Also using a survey method to study behaviour (especially among young people) is not recommended because the reasons for certain behaviours are difficult or fully impossible to remember¹⁹, analyse and depict afterward (Frankfort-Nachmias & Nachmias 1996, 206). It is also said that surveys give a more conventional picture of young people than is seen in real life, because un-conventional young persons are not answering the questions and/or young people answer in the way they think researcher wants them to answer (Puuronen 1997, 195).

In order to avoid these problems, research should also look for a broader picture of existing social and living conditions than what structured questions can provide (Raunio 1999, 197). In this research this is done by using triangula-tion – utilising both quantitative and qualitative data together with existing lit-erature and documentation about the research issue. In addition to this, the researcher has used the understanding of young people’s worlds which she picked up during her earlier research projects (Soininen 1998; 1999) and through participation in various national and international seminars and conferences dealing with youth studies. Academic background knowledge is used here to complement a personal ability to read the surrounding environment and cultural signals of contemporary society.

18 The danger of ending up with fundamentally false conclusions for this reason in this study is marginal, since author became familiar with issues relating to this research topic already while writing her master’s thesis about young people living in the peripheries of eastern Finland (Soininen 1998).

19 Memory errors are one type of response error that can arise among respondents. Other such errors are, e.g., a lack of knowledge, misunderstanding, overstating or cheating (Moser & Kalton 1971, 387-388).

40