Implications of computer code translation for translation studies

ABSTRACT

This article explores the intersection of the translation of machine code and translation of natural languages. Starting with the deep-seated metaphor the brain is a computer, this study demonstrates the extent to which computer science, cognitive science, linguistics and translation are intertwined. The parallels between difficulties of translating computer code and natural languages point to the failure to find a workable interlingua and the importance of complexity studies and emergent properties for both fields. Thus translation studies would do well to examine more carefully the extent to which a computational understanding of the brain has shaped basic concepts and approaches to translation. At the same time, homologies between the market for translation of computer languages and natural languages and the need in programming to be more sensitive to the needs of non-English speakers suggests that translation studies has much to offer to computer science.

KEYWORDS:

• Machine translation
• computer programming
• reverse engineering
• brain as computer
• cognitive science
• emergent properties

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Computers have up to four types: register memory, cache memory, main memory, and disk memory, but the basic distinction remains between immediately available (register) and retrievable (the other three).

2 Reading the more recent Bermúdez (2020), cognitive scientists today would probably argue that this task is passed to a sub-unit in the brain. Learning certain types of skills may thus involve the brain moving some routines from conscious CPU to more specialized areas.

3 Python indexes left to right starting from the number zero, and right to left starting from the number one.

4 Dialects of BASIC number over 400 (https://en.wikipedia.org/wiki/List_of_BASIC_dialects); Wikipedia gives 15 examples of esoteric languages (https://en.wikipedia.org/wiki/Esoteric_programming_language), 22 main markup languages (https://en.wikipedia.org/wiki/List_of_markup_languages), 72 document markup languages (https://en.wikipedia.org/wiki/List_of_document_markup_languages), 215 XML markup languages (https://en.wikipedia.org/wiki/List_of_XML_markup_languages), 8 general-purpose markup languages (https://en.wikipedia.org/wiki/General-purpose_markup_language), 28 content syndication markup languages (https://en.wikipedia.org/wiki/List_of_content_syndication_markup_languages), 23 lightweight markup languages (https://en.wikipedia.org/wiki/Lightweight_markup_language), and 19 user interface markup languages (https://en.wikipedia.org/wiki/List_of_user_interface_markup_languages). All Wikipedia pages were accessed on 28 August 2022.

5 They therefore differ from footnotes or other types of explanatory text in natural languages, because such material is meant to be read by the same audience that reads the main text, whereas comments in computer code are not meant to be read by the computer and are written in a different (natural) language.

6 Here I mean specifically obligatory redundancy, such as is found in the sentence “I am hot”, where the verb form provides the same information as the first-person pronoun – that the sentence is in the first person singular. Optional redundancy can be built into computer programs, especially to make the program more robust and less likely to fail.

7 This is true of all conventional computers. However, as noted above, recent attempts to apply Bayesian logic to the brain suggests that this may be changing, and quantum computing may offer an opportunity to shift this position radically, as such computers depend upon quantum states of matter, which exist in an indeterminate, either/or state. To date, however, quantum computing has had little impact on the way in which people think about computers generally, and I have not seen any references in translation studies to Bayesian logic.

8 Nida began with the rather cumbersome phrase “closest natural equivalent to the message of the source language” (1959, 19), then coined the term “dynamic equivalence” to describe what is essentially sense-for-sense translation, focusing on achieving an equivalent effect on the reader rather than on the text (Nida 1964). However, he later felt that the term “dynamic equivalence” led to misunderstandings and so preferred the term “functional equivalence”. (Nida and de Waard 1986), although dynamic equivalence is still widely used in the literature. It seems his choice was motivated by functionalism in anthropology and sociology (Statham 2005, 29–30), but “functional equivalence” also resonates well with the world of computer programming, which is all about functions and functionality.

9 See Chikofsky and Cross (1990) and Nelson (1996) for overviews. Yet another example of how the fields of cognition and artificial intelligence bleed into each other through metaphor is the monograph by Neukart, Reverse Engineering the Mind: Consciously Acting Machines and Accelerated Evolution (2017), which applies the concept of reverse engineering to the brain and cognitive processes.

Additional information

Notes on contributors

James St. André

James St. André is Associate Professor of Translation and Director of the Center for Translation Technology at The Chinese University of Hong Kong. His research focuses on translation theory and the history of Chinese-English translation in the 18th–20th centuries, recently with a focus on digital humanities approaches. His monograph Conceptualising China through Translation is due out in August 2023; other works include Translating China as Cross-Identity Performance (2018) and three edited volumes: Translation and Time: Migration, Culture, and Identity (2020), China and Its Others: Knowledge Transfer through Translation (co-edited with Peng Hsiao-yen, 2012), and Thinking through Translation with Metaphors (2010).