Glossary
Bilingual concordancer: a tool that allows a user to enter a search term and to retrieve all examples of that search term from a parallel corpus. The source language sentences (containing the search term) are displayed in one column and the target language translations of those sentences are displayed in a parallel column.
Corpus (plural = corpora): a large collection of texts in electronic form that can be processed with the help of tools such as concordancers.
Interpretation: the transfer of a message from one spoken or signed language to another.
Locale: a combination of a language and region and their linguistic and cultural, conventions and preferences (e.g. French Canada, French Belgium).
Localization: Translation that goes beyond a given language (e.g. English, French) and takes into account a regional language variety and its cultural conventions and preferences (e.g. Canadian English, Canadian French).
Large Language Model (LLM): a type of artificial neural network designed for language processing tasks (e.g., question-answering, text generation, summarization, text simplification, and translation). These models contain a huge corpus of texts that a tool such as ChatGPT can consult to identify linguistic patterns and use as examples.
Machine learning: a branch of artificial intelligence which uses data and algorithms to imitate the way that people learn and which gradually improves its accuracy.
Machine translation (MT): a type of software that takes a source text written in one language and produces a target text in another language.
Neural machine translation (NMT): a corpus-based approach to machine translation that uses artificial neural networks and machine learning techniques. The software is provided with an extremely large parallel corpus of texts and their translations and it consults these previously translated texts and “learns” to translate new texts.
Parallel corpus (plural = parallel corpora): a bilingual corpus that contains source texts aligned with their translations. The texts are broken down into sentences and each sentence in the source text is linked to its translation in the target text. A bilingual concordancer can be used to retrieve the source language sentences containing a search term and to display these alongside the equivalent target language sentences.
Portal: a specially designed web-based platform that collects information from different sources and presents it in a uniform way.
Prompt: the input or query provided by a user to initiate conversation or request assistance from an AI tool such as ChatGPT. The prompt typically consists of a question or instruction.
Prompt engineering: the process of creating input (e.g., a question or an instruction) for a tool based on large language model (LLM) (e.g., ChatGPT). The more specific and clear the prompt, the more likely you are to get useful results.
Rule-based machine translation (RBMT): the earliest approach to machine translation that attempted to get computers to process language using large bilingual dictionaries and complex sets of grammar rules.
Source language: the language of the original message that needs to be translated (i.e., the start language).
Source text: the text containing the message that needs to be translated (i.e., the start text).
Target language: the language into which the translation takes place (i.e., the end language).
Target text: the translated text (i.e., the end text).
Term: a word or expression from a specialized field of knowledge.
Term bank: a dictionary-like database that contains terms along with information about the terms, such as definitions, synonyms, equivalents in other languages, and notes about how terms are used (e.g. whether they belong to a specific language variety, such as Canadian English or Canadian French).
Translation: the transfer of a written message from one language to another.
