Translation in HCI: Formal Representations for Work Analysis and Collaboration

ABSTRACT

In a previous paper, I provided a foundation argument for translation as a pivotal activity in analysis work in the fields of HCI, with profound epistemological and ethical consequences. This technical note extends the argument with a formal notation for translation work in HCI, with application to work analysis and collaboration.

Keywords

Task analysis, systems analysis, translation, translator, ethnocriticism, ethnocritical heuristics, positionality

© Copyright ACM 1997

INTRODUCTION

In [5], I offered an analysis of the work of an analyst in Human-Computer Interaction (HCI), based on the concept of ethnocriticism [4]. In brief, ethnocriticism helps us to view the role of the analyst in a workplace organization as bridging across dimensions of difference and power:

• Difference: The analyst often inhabits a different work culture, background, and social class from the people whose work s/he is analyzing. The analyst studies the work of people whose perspective is different from her or his own perspective.
  
• Power: The workplace organization often accords the analyst greater authority (verbal privilege) than the workers. The analyst's interpretation of the work carries greater weight in the organization than the interpretation of the people who actually do the work.

The analyst is usually concerned to develop a representation of the work domain that can be understood and acted upon by other software professionals. In this way, the analyst is involved in the task of translating (out of the work domain and into the software domain), and the analyst's report (or representation) of the work becomes a type of translation.

BACKGROUND

Williams and Begg discussed one type of translating, in which a single individual (an analyst) has professional knowledge of both the software domain and an application-relevant work domain [8]. In [5], I took up the more general case of an analyst who is initially ignorant of the work domain, and who gathers sufficient information to write a translation (e.g., a requirements document or a specification). Guided by contemporary translation theory [3, 7], I attempted to show the intensely mediated, processed, and interpreted nature of such translations. The analyst's practices in translation invoke important issues in epistemology and ethics, with far-reaching consequences for users and other system stakeholders [5].

This paper is intended as a step from the qualitative discussions of [5] toward a more formal description of translation processes in HCI, with particular attention to the Dialogic theory of Bakhtin [2]. Future papers will apply this formalism to describe and assess translation practices in collaborative and participatory work analyses in HCI.

FORMAL APPROACHES TO TRANSLATION IN HCI

The notations in this paper are due in part to Barnstone's discussion of equivalence in translation [3]. Barnstone argues that the concept of equivalence is based on (a) the recognition of difference, and (b) the discovery of relationships across differences. Thus, the naive view of translation -- a perfect transportation of meaning from one language to another -- must give way to a mediated mapping of equivalences. Formally, the naive view,

[E1]

(perfect equivalence), is replaced by the more realistic view:

[E2]

Within HCI, it is helpful for us to consider the language context of such an equivalence. For word W₁ in the users' work language L₁ and word W₂ in the software professionals' language L₂, expression [E2] becomes

[E3]

It is important to note that translation is often not symmetrical. For example, most translators work between what is, for them, their native language (L_N) and one or more foreign languages (L_F), with unequal fluencies in L_N and L_F. In the case of retranslation from L_N to L_F and back to L_N,

[E4]

the outcome, W_?, of the retranslation into L_N, may not be the same as the initial word, W₁, in L_N (and conversely).

This abstract discussion of native and foreign languages begs the question of the identity of the translator: For whom are certain languages native or foreign? Thus, the translator, T, should appear in the expression:

[E5]

The translator is, of course, an active and transformative agent in the process of translating. As Venuti argues, the process of translation involves particular choices [7]. The variety of choices available to translators is reflected in the large number of specific theories of translation cited by Barnstone [3]. The remainder of this paper concerns the active agency of the translator or translators. However, two more issues of translation process must first be presented.

First, we should note that the appealing notion of word-to-word equivalence (expressions [E2-E3]) is often insufficient. Translation frequently operates on a concept-to-concept basis, if words have different relationships in different languages. Translators search through several overlapping layers of equivalence to find the best expression. Baker, for example, devotes an entire chapter to each of the following classes of equivalence: word-level, above word level (collocation, idiom), grammatical, textual (theme and structure), textural (cohesion), and pragmatic [1]. In HCI, we might consider equivalence at the level of keystrokes or mouseclicks, lexical units, syntactic units, semantic units, tasks, and goals; or objects and attributes. Thus, the notation of words (W_i) in expressions [E3-E5] is an oversimplification, and should be replaced by a focus on concepts (C_i):

[E6]

The second issue is that each language is used within its own discourse community, whose language usage is shaped in part by that community's perspective (standpoint). This issue is particularly important when we consider the languages of workplaces and the languages of software development, because all of these languages undergo rapid evolution. In Bakhtinian theory [2], meaning is created dynamically by partners in an active discourse process called "the Dialogue." It may be important to emphasize the double, evolving, dialogic context of translation by replacing the apparently static language terms (L_i) with terms indicating two active, distinct discourse communities (D_i):

[E7]

Each discourse community exists in the larger context of workplaces. The dimensions of difference and power that I explored in [5] often result in different experiences and different enfranchisements of different discourse communities.

I now return to the role of the translator. The intentions, position, and practices of the translator may be characterized in several ways. In addition to the question of equivalence level (above), the translator is guided by her or his understanding of the purpose of the translation, and by derivative (or reactive) sets of goals and intentions.

The role of the translator becomes more complex when we consider that analysis work in HCI is increasingly performed by interdisciplinary teams, who are composed of members from several discourse communities. In simple terms, we may establish a mapping within the team of translator T₁ from discourse community D₁, T₂ from D₂, and so on.

We have also learned to consider the methods used by the team. These methods can be conveniently characterized in terms of three models: the objects to be used; the group process involved in working with the objects and with one another; and the decisions related to who participates in the process [6]. These factors may be modeled as:

[E8]

As analyzed in [5], the creation of an interdisciplinary team creates a new discourse community with its own Dialogic process to form within the team. The team's discourse may bridge between the translated-from discourse community, D₁, and the translated-to discourse community, D₂.

CONCLUSION

The formal model in expression [E8] may provide multiple points of access for characterizing and assessing translation processes in HCI. Cognitive studies may consider equivalence levels, goals, and concept mappings. Activity theorists may consider translation in terms of mediation, and equivalence levels in terms of activities, actions, and operations. Participatory and collaborative studies may consider methods, and the combinations of standpoints represented among translators. Sociological studies may consider discourse communities and their standpoints. Translation in HCI appears to require descriptions along all of these dimensions, and analysis through these multiple disciplines.

ACKNOWLEDGMENTS

I thank Marita Franzke for a very helpful discussion. I thank Marita Franzke and Scott Robertson for helpful comments on drafts of this technical note.

REFERENCES

[1] Baker, M. (1992). In other words: A coursebook on translation. London: Routledge.

[2] Bakhtin, M.M. (1981). The dialogic imagination: Four essays (M. Holquist, ed.; C. Emerson and M Holquist, trans). Austin: University of Texas Press.

[3] Barnstone, W. (1993). The poetics of translation: History, theory, practice. New Haven: Yale.

[4] Krupat, A. (1992). Ethnocriticism: Ethnography, history, literature. Berkeley: University of California Press.

[5] Muller, M.J. (1995). Ethnocritical heuristics for HCI work with users and other stakeholders. In Proceedings of Computers in Context: Joining Forces in Design. Århus Denmark: Århus University, 10-19.

[6] Muller, M.J., Hallewell Haslwanter, J.D., and Dayton, T. (1996). Participatory practices in the software lifecycle. In M. Helander et al. (eds.), Handbook of computer-human interaction, in press.

[7] Venuti, L. (1995). The translator's invisibility: A history of translation. London: Routledge.

[8] Williams, M.G., and Begg, V. (1992). Translation in participatory design. In INTERCHI'93 Adjunct Proceedings. Amsterdam: ACM, 141-142.