Will Technological Innovations in Language Processing Models Threaten Translators?
The rapid advancements in language technologies today mark a new era with the introduction of the language models. However, do these technological developments pose a threat to the profession of translation, or do they present new opportunities and collaboration prospects? In this article, we will discuss the impact of these language models on the translation profession.
An AI-based language model that is specifically designed for language translation, has the capacity to understand, learn, and synthesize vast amounts of language data. This technology aims to overcome language barriers, enabling users to obtain more effective and meaningful translations in natural language. The objective of translation profession goes far beyond overcoming language barriers.
The emergence of such technologies naturally raised concerns among some translators. Let’s explore these concerns together:
However, translation is not merely a mechanical conversion of one language to another. This profession involves cross-cultural communication, attention to subtle details, creativity, and research. Language models may provide support in laying a foundation for translators, but it is unlikely to replace the human touch. Translation encompasses more than just translating words; it involves being a cultural ambassador, conveying emotions, meticulous work, and a process that demands creativity. While language models can perform mechanical translations, they cannot fully replace the deep understanding and human touch that translators provide. Translators can enhance their work by leveraging the technological advantages offered by language models. Such tools can indeed provide support to translators in achieving faster and more accurate translations. However, seeking professional translation assistance is crucial to overcome the limitations of language models and add a human touch to the texts.
While language models offer new possibilities to translators, collaboration remains paramount. By utilizing technology, translators can produce faster and more effective translations, but it is essential for the translations provided by language models to be revised by human hands.
As a result, the translation profession will evolve to become more valuable and effective with technological advancements, as services offered by translators who don’t translate like machines will be unique and of very high quality. Large language models have no intention of competing translators or putting an end to the profession.
We would like to conclude our article with the list of conditions that renowned figure in the field of translation, Andre Gile, believes an ideal translator should meet.
• Competence: Involves dedicating sufficient time and energy to meet the client’s requirements.
• Loyalty: Connected to translation norms and especially relationship norms. It involves both information extraction (omission) and the addition of information not present in the source text to the target text.
• Clarity of Ideas: Concerned with the understandability of messages.
• Linguistic Acceptability: Related to the linguistic accuracy of the target text and errors in word and sentence structure, as well as spelling mistakes.
• Accuracy of Terminology: Refers to the precision and appropriateness of terminology.
• Acceptability: Generally pertains to the linguistic and stylistic appropriateness of the target text in terms of writing.
• Professional Behavior/Behavioral Component: Associated with working conditions such as delivery dates and team spirit (Gile, 2013).
As evident, a translator is not merely a simple figure or step in a translation process. This list, based on Andre Gile’s specified criteria, emphasizes that translators should focus not only on language proficiency but also on factors such as customer satisfaction, loyalty, clarity of meaning, and terminology suitability.
According to the expert author Marcin Frąckiewicz in technical communication and artificial intelligence, these models tend to rely heavily on statistical methods and can be prone to biased results as they only reflect the data on which they were trained. Associated with their usage are potential risks, and they can generate unnatural or inaccurate texts. Looking at translation in this context, the sheer volume of such interfaces indicates that they cannot be specialized through translations done entirely by a human translator, pointing to a high error rate. Therefore, we should consider these programs as powerful but imperfect tools that support translation. Just as in the field of medicine, where highly successful machines and tools exist, we still prefer a doctor to make the diagnosis and a surgeon to perform the operation, in translation, we should trust the translation process and use technology as a tool that enhances the process, not replaces it.
We wish to reiterate the complexity and significance of our profession by noting that artificial intelligence still cannot replace the multifaceted roles of translators, and that the objectives and processes of NLP companies differ from those of translators.
