AI in ELT · Reference Guide
AI Glossary for English Language Teachers
Artificial intelligence is changing how English language teachers plan lessons, give feedback, and assess learners — but the terminology moves faster than most training keeps up with. This AI glossary for English language teachers brings together 100 essential terms, from foundational concepts like machine learning and natural language processing to classroom-specific applications like automated essay scoring, AI-assisted feedback, and adaptive learning. Each entry is written in plain language, with a short definition, a practical example grounded in real ELT and TESOL classroom use, and links to related terms — so you can look up one word or read the whole thing as a working reference for AI in ELT.
Browse the Full AI Glossary for English Language Teachers (A–Z)
Jump to any letter below, or use the search box above to find a specific term.
Adaptive Learning
DefinitionA method where digital systems adjust content, pace, and difficulty based on a learner's performance in real time. AI analyzes responses and modifies the learning path automatically, aiming to match instruction to individual needs rather than a fixed curriculum.
Example in ELTAn adaptive grammar app gives a student easier sentence-correction tasks after several incorrect answers.
AI-Assisted Feedback
DefinitionAutomated commentary generated by an AI system on student work, covering areas such as grammar, coherence, vocabulary use, or structure. It supplements, rather than replaces, teacher feedback and can be delivered instantly.
Example in ELTA writing platform flags subject-verb agreement errors in a student's essay before the teacher reviews it.
AI Bias
DefinitionSystematic errors in AI output caused by unrepresentative or skewed training data. This can lead to unfair or inaccurate results for certain accents, dialects, or cultural contexts, which is a key concern in language education tools.
Example in ELTA pronunciation app consistently misjudges non-native English accents as “incorrect” due to biased training data.
AI Content Detector
DefinitionA tool designed to estimate whether a piece of text was generated by AI rather than written by a human. It analyzes patterns such as word predictability and sentence structure, though results are probabilistic, not definitive proof.
Example in ELTA teacher runs a suspicious essay through an AI content detector before discussing it with the student.
AI Ethics
DefinitionThe study of moral principles guiding the responsible design and use of AI systems, including fairness, privacy, transparency, and accountability. In education, it addresses concerns like data protection and equitable access to AI tools.
Example in ELTA school creates an AI ethics policy outlining how student writing data may and may not be used.
AI Hallucination
DefinitionAn instance where an AI system generates false, fabricated, or nonsensical information while presenting it as factual. This commonly occurs with generative language models and requires human verification of outputs.
Example in ELTA chatbot invents a fake grammar rule when asked to explain the present perfect tense.
AI Literacy
DefinitionThe knowledge and skills needed to understand, use, and critically evaluate AI tools. For teachers, this includes recognizing AI capabilities and limitations and applying tools appropriately within pedagogical goals.
Example in ELTA teacher training session builds AI literacy by showing staff how to verify chatbot-generated lesson content.
AI Lesson Planner
DefinitionSoftware that uses AI to generate structured lesson plans based on inputs such as topic, level, and duration. It can suggest objectives, activities, and materials, which teachers then review and adapt.
Example in ELTA teacher uses an AI lesson planner to draft a 45-minute lesson on conditionals for B1 learners.
AI Rubric Generator
DefinitionA tool that automatically creates assessment rubrics based on task type, learning objectives, and proficiency level. It saves preparation time while allowing teachers to customize criteria and scoring bands.
Example in ELTA teacher generates a rubric for an IELTS-style speaking task in seconds using an AI rubric generator.
AI Tutor
DefinitionA software system that simulates one-on-one tutoring by responding to learner questions, providing explanations, and adjusting difficulty based on performance. It offers practice outside class but does not replace human instruction.
Example in ELTA student practices conversational English with an AI tutor between weekly lessons.
AI Writing Assistant
DefinitionA tool that supports writing tasks by suggesting corrections, rephrasing, or generating text. Common features include grammar checking, style suggestions, and vocabulary alternatives, aimed at improving clarity and accuracy.
Example in ELTA student uses an AI writing assistant to check tone before submitting a formal email assignment.
Algorithm
DefinitionA defined sequence of steps or rules that a computer follows to complete a task or solve a problem. Algorithms underlie all AI systems, determining how input data is processed into output results.
Example in ELTA vocabulary app's algorithm decides which words to review next based on past mistakes.
Annotation
DefinitionThe process of labeling data with relevant tags so it can be used to train AI models. In language contexts, this might mean marking grammatical errors, parts of speech, or sentiment in text samples.
Example in ELTResearchers annotate a corpus of learner essays to train an automated error-detection tool.
API (Application Programming Interface)
DefinitionA set of rules that allows different software programs to communicate and exchange data. Developers use APIs to connect educational platforms with AI services such as language models or speech recognition.
Example in ELTA school's learning platform uses an API to connect with a speech recognition service for pronunciation practice.
Artificial General Intelligence (AGI)
DefinitionA theoretical form of AI capable of understanding, learning, and performing any intellectual task a human can, across multiple domains. AGI does not currently exist and differs from today's task-specific AI tools.
Example in ELTUnlike current chatbots, a true AGI system could design an entire curriculum and adapt it without human guidance.
Artificial Intelligence (AI)
DefinitionThe broad field of computer science focused on building systems that perform tasks typically requiring human intelligence, such as understanding language, recognizing patterns, or making decisions.
Example in ELTAI powers tools that automatically grade multiple-choice vocabulary quizzes and give instant results.
Augmented Reality (AI-Enhanced)
DefinitionTechnology that overlays digital content, often powered by AI, onto the physical world through a device's camera. In language learning, it can display translations or interactive labels on real objects.
Example in ELTStudents point a tablet at classroom objects to see AI-generated English vocabulary labels appear.
Automated Essay Scoring (AES)
DefinitionAI-based systems that evaluate written essays and assign scores based on criteria such as grammar, coherence, and vocabulary range. AES is widely used in large-scale testing to speed up grading.
Example in ELTA TOEFL-style practice platform uses automated essay scoring to give students an instant band estimate.
Automated Speech Recognition (ASR)
DefinitionTechnology that converts spoken language into written text using AI. ASR enables applications like pronunciation feedback, dictation tools, and voice-controlled learning platforms.
Example in ELTA pronunciation app uses ASR to compare a student's spoken sentence with the correct native pronunciation.
Automation
DefinitionThe use of technology to perform tasks with minimal human intervention. In education, automation reduces repetitive workload such as grading, attendance tracking, or generating practice materials.
Example in ELTA teacher automates weekly vocabulary quiz creation using an AI-powered worksheet generator.
Big Data
DefinitionExtremely large and complex data sets that traditional processing methods cannot handle efficiently. AI systems use big data to identify patterns, such as common learner errors across thousands of students.
Example in ELTA language platform analyzes big data from millions of exercises to identify the most common verb tense mistakes.
Bot
DefinitionA software program that performs automated tasks, often simulating conversation or interaction. Educational bots can answer questions, provide practice dialogue, or send reminders to students.
Example in ELTA messaging bot sends daily vocabulary reminders to students studying for an exam.
Chat Interface
DefinitionThe visual and interactive layer through which users communicate with an AI system, typically resembling a messaging app. It determines how users type input and receive AI-generated responses.
Example in ELTStudents type questions into a chat interface to get instant explanations of idioms.
Chatbot
DefinitionAn AI program designed to simulate conversation with users through text or voice. Chatbots range from simple rule-based systems to advanced models capable of open-ended dialogue.
Example in ELTA student roleplays a job interview in English with a chatbot to build speaking confidence.
ChatGPT
DefinitionA widely used generative AI chatbot developed by OpenAI, built on a large language model. It can answer questions, explain concepts, and generate text, making it a common tool referenced in education.
Example in ELTA teacher asks ChatGPT to generate five reading comprehension questions for a short story.
Classifier
DefinitionAn AI model trained to sort input data into predefined categories. In language tools, classifiers can identify proficiency level, sentiment, or error type within a piece of text.
Example in ELTA placement test tool uses a classifier to sort student writing samples into CEFR levels.
Cognitive Computing
DefinitionSystems designed to simulate human thought processes, such as reasoning and problem-solving, in complex and unstructured situations. It overlaps with AI but emphasizes mimicking human-like understanding.
Example in ELTA cognitive computing system interprets ambiguous student phrasing to infer intended meaning during feedback.
Computer-Assisted Language Learning (CALL)
DefinitionThe broader field of using computer technology, including AI, to support language teaching and learning. CALL encompasses everything from basic vocabulary software to advanced adaptive platforms.
Example in ELTA CALL program combines listening exercises, quizzes, and AI feedback in one platform.
Content Generation
DefinitionThe use of AI to automatically produce text, images, or other materials based on a prompt. In education, this includes generating reading passages, quizzes, or lesson materials.
Example in ELTA teacher uses content generation to quickly create three versions of a reading passage at different levels.
Content Moderation
DefinitionThe process, often AI-assisted, of reviewing and filtering user-generated content to ensure it meets safety and appropriateness standards. This is relevant for platforms where students interact with AI or each other.
Example in ELTAn online classroom's chat tool uses content moderation to block inappropriate language automatically.
Contextual Understanding
DefinitionAn AI system's ability to interpret meaning based on surrounding words, prior conversation, or situational cues, rather than processing language literally or in isolation.
Example in ELTAn AI tutor uses contextual understanding to correctly interpret “book” as a verb in “I need to book a flight.”
Conversational AI
DefinitionAI technology specifically designed to engage in natural, back-and-forth dialogue with users. It combines speech or text recognition with language generation to hold coherent conversations.
Example in ELTStudents use a conversational AI app to practice ordering food in English before a class roleplay.
Corpus (Language Corpus)
DefinitionA large, structured collection of authentic texts used to study language patterns or train AI language models. Corpora help identify common word usage, collocations, and grammar structures.
Example in ELTA teacher checks a learner corpus to see which prepositions students misuse most often.
Corpus Linguistics
DefinitionThe academic study of language using large text corpora, often supported by computational tools. It informs how AI language models learn grammar, vocabulary, and usage patterns.
Example in ELTCorpus linguistics research shows that “make a decision” is far more common than “do a decision.”
Data Privacy
DefinitionData privacy in AI education refers to protecting personal information — such as student writing, voice recordings, or performance data — from misuse when processed by AI systems. It underpins responsible technology adoption in schools and requires clear policies on storage and sharing.
Example in ELTA school ensures its AI writing tool does not store or share students' personal essays without consent.
Data Set
DefinitionA structured collection of data used to train, test, or evaluate an AI model. In language education, data sets often consist of text samples, audio recordings, or learner responses that shape how accurately a tool performs.
Example in ELTA pronunciation app is trained on a data set of thousands of native and non-native speech recordings.
Deep Learning
DefinitionA subset of machine learning that uses multi-layered neural networks to identify complex patterns in large amounts of data. Deep learning powers many modern language tools, including translation systems and speech recognition software.
Example in ELTDeep learning enables a translation app to produce increasingly natural-sounding English sentences.
Dialogue System
DefinitionAn AI application designed to conduct structured or open-ended conversations with users, often used for language practice. Dialogue systems can follow scripted scenarios or generate flexible responses based on user input.
Example in ELTA dialogue system simulates a hotel check-in conversation for intermediate English learners.
Digital Literacy
DefinitionThe ability to effectively and critically use digital tools, including AI applications, for learning and communication. For teachers, digital literacy includes evaluating AI outputs and integrating tools appropriately into lessons.
Example in ELTDigital literacy training helps teachers assess whether an AI-generated worksheet is pedagogically sound.
Discourse Analysis (AI-Assisted)
DefinitionThe computational study of language use beyond the sentence level, examining coherence, cohesion, and conversational structure. AI tools can automatically analyze discourse features in student writing or speech to support feedback.
Example in ELTAn AI tool highlights weak paragraph transitions in a student's argumentative essay.
Edtech (Educational Technology)
DefinitionTechnology, including AI-powered tools, designed specifically to support teaching and learning. Edtech spans learning management systems, adaptive platforms, and classroom apps used to enhance instruction and assessment.
Example in ELTA school adopts an edtech platform that combines AI feedback with traditional grammar lessons.
Error Detection
DefinitionThe process by which AI systems identify mistakes in language use, such as grammar, spelling, or word choice errors, within written or spoken text. Error detection underlies most automated feedback and correction tools.
Example in ELTAn error detection tool underlines incorrect article usage in a student's paragraph.
Explainable AI
DefinitionAI systems designed so their decisions and outputs can be understood and interpreted by humans, rather than functioning as an opaque “black box.” Explainability builds trust when AI is used for grading or feedback.
Example in ELTAn explainable AI grading tool shows exactly which rubric criteria lowered a student's score.
Feedback Loop
DefinitionA cyclical process in which an AI system's output informs future input, allowing continuous improvement. In education, feedback loops help adaptive platforms refine their recommendations based on learner performance over time.
Example in ELTA vocabulary app's feedback loop increases repetition of words a student consistently mistranslates.
Fine-Tuning
DefinitionThe process of further training a pre-existing AI model on a smaller, specialized data set to improve performance on a specific task. Fine-tuning allows general language models to become more accurate for tasks like ELT feedback.
Example in ELTDevelopers fine-tune a language model on learner essays to improve its grammar correction accuracy.
Fluency Scoring (AI)
DefinitionAn automated evaluation of how smoothly and naturally a learner speaks or writes, based on features like pace, hesitation, and word choice. AI fluency scoring is commonly used in speaking assessments.
Example in ELTA speaking test app gives a fluency score after analyzing a student's one-minute response.
Formative Assessment
DefinitionOngoing evaluation used to monitor student learning and provide feedback for improvement, rather than assigning a final grade. AI tools increasingly support formative assessment through instant, low-stakes feedback.
Example in ELTAn AI quiz generator creates weekly formative assessments to track vocabulary retention.
Generative AI
DefinitionA category of AI capable of producing new content, such as text, images, or audio, based on patterns learned from training data. Generative AI powers tools like chatbots and content generators widely used in education.
Example in ELTA teacher uses generative AI to create three reading passages on the same topic at different levels.
Grammar Checker
DefinitionA tool that automatically identifies and suggests corrections for grammatical errors in written text. Modern grammar checkers use AI to detect context-dependent mistakes beyond simple spelling errors.
Example in ELTA student runs their essay through a grammar checker before submitting it for peer review.
GPT (Generative Pre-trained Transformer)
DefinitionA type of large language model architecture trained on vast text data to generate human-like language. GPT models, including those behind tools like ChatGPT, are commonly referenced in AI-assisted education.
Example in ELTA teacher explains to colleagues that GPT-based tools can draft lesson materials but still require review.
Human-in-the-Loop
DefinitionAn approach where human judgment remains part of an AI-driven process, reviewing or correcting outputs rather than relying on full automation. This is central to responsible AI use in education.
Example in ELTA human-in-the-loop grading system lets teachers approve or adjust AI-suggested essay scores.
Hyperparameter
DefinitionA configuration setting chosen before training an AI model that controls how the model learns, such as learning speed or complexity. Hyperparameters affect a model's accuracy and behavior but are set externally, not learned from data.
Example in ELTDevelopers adjust hyperparameters to make a grammar-checking model more sensitive to minor errors.
Image Recognition
DefinitionAn AI capability that identifies objects, text, or scenes within images. In language learning, image recognition can support vocabulary practice by labeling real-world objects in a learner's target language.
Example in ELTA vocabulary app uses image recognition to name classroom objects a student photographs.
Immersive Learning
DefinitionAn instructional approach that places learners in realistic or simulated environments, often enhanced by AI, to practice language in context. Immersive learning increases engagement and contextual understanding.
Example in ELTStudents use an immersive learning app to order food in a simulated AI-powered restaurant scenario.
Instructional Design
DefinitionThe systematic process of planning and structuring lessons or courses to achieve specific learning outcomes. AI tools increasingly support instructional design by suggesting objectives, sequencing, and activities.
Example in ELTA curriculum coordinator uses AI to speed up instructional design for a new exam-preparation course.
Intelligent Tutoring System
DefinitionAn AI-based platform that provides personalized instruction and feedback similar to a human tutor, adapting to a learner's pace and needs. These systems combine adaptive learning with targeted practice.
Example in ELTAn intelligent tutoring system gives a struggling student extra practice on modal verbs.
Interactive Learning
DefinitionA teaching approach that actively engages students through participation, often supported by AI-driven tools such as quizzes, simulations, or conversational practice.
Example in ELTAn interactive learning app lets students role-play a job interview with instant AI feedback.
Knowledge Base
DefinitionA structured repository of information that an AI system references to answer questions or generate accurate responses. A well-maintained knowledge base improves the reliability of AI tutoring and chatbot tools.
Example in ELTA school's AI chatbot draws from a knowledge base of approved grammar rules to avoid incorrect answers.
Language Model
DefinitionAn AI system trained to understand and generate human language by predicting likely word sequences. Language models underpin most modern writing assistants, chatbots, and translation tools.
Example in ELTA language model suggests natural phrasing alternatives for an awkward sentence in a student's essay.
Large Language Model (LLM)
DefinitionA language model trained on massive amounts of text data, enabling it to perform diverse tasks such as answering questions, summarizing, and generating essays. LLMs are the technology behind tools like ChatGPT.
Example in ELTA teacher uses an LLM to generate discussion questions based on a short story.
Learning Analytics
DefinitionThe collection and analysis of data about learners' progress and behavior, used to improve teaching decisions and personalize instruction. AI enhances learning analytics by identifying patterns across large groups of students.
Example in ELTLearning analytics reveals that most students struggle with reported speech, prompting a review lesson.
Learning Management System (LMS)
DefinitionA software platform used to deliver, track, and manage educational courses and content. Many LMS platforms now integrate AI features for grading, feedback, and content recommendations.
Example in ELTTeachers upload AI-generated quizzes directly into the school's learning management system.
Lemmatization
DefinitionA natural language processing technique that reduces words to their base or dictionary form, known as a lemma. Lemmatization helps AI tools analyze vocabulary accurately regardless of word inflection.
Example in ELTA vocabulary analysis tool uses lemmatization to count “run,” “ran,” and “running” as the same word.
Lexical Analysis
DefinitionThe process of examining vocabulary use within a text, including word frequency, range, and complexity. AI-powered lexical analysis helps assess a learner's vocabulary proficiency automatically.
Example in ELTA writing tool's lexical analysis shows a student is overusing basic adjectives like “good” and “nice.”
Machine Learning (ML)
DefinitionA branch of AI in which systems learn patterns from data rather than following explicit programmed rules. Machine learning underlies most modern language tools, from spell checkers to speech recognition.
Example in ELTMachine learning allows a grammar tool to improve its accuracy as it processes more student essays.
Machine Translation
DefinitionThe use of AI to automatically translate text or speech from one language to another. While useful for quick comprehension, machine translation output often requires human review for nuance and accuracy.
Example in ELTA student uses machine translation to check the meaning of an unfamiliar English idiom.
Metadata
DefinitionDescriptive information about data, such as a text's author, date, or topic, that helps AI systems organize and retrieve content efficiently. Metadata supports search and recommendation features in educational platforms.
Example in ELTA resource library uses metadata tags like “B1” and “grammar” to help teachers find relevant materials.
Model Training
DefinitionThe process of teaching an AI system to perform a task by exposing it to large amounts of example data. Model training determines how accurately a tool performs tasks like grammar correction or scoring.
Example in ELTModel training on thousands of graded essays helps an AI scoring tool align with human raters.
Multimodal AI
DefinitionAI systems capable of processing and combining multiple types of input, such as text, images, and audio, within a single task. Multimodal AI enables richer, more interactive learning experiences.
Example in ELTA multimodal AI tool listens to a student's speech and displays corrected text with pronunciation tips.
Named Entity Recognition
DefinitionA natural language processing technique that identifies and classifies proper nouns in text, such as names, places, or organizations. This supports tasks like automated reading comprehension question generation.
Example in ELTAn AI tool uses named entity recognition to automatically generate “who” and “where” questions from a text.
Narrow AI
DefinitionAI systems designed to perform a specific, limited task rather than general reasoning across domains. Nearly all current educational AI tools, including grammar checkers and chatbots, are examples of narrow AI.
Example in ELTA pronunciation-scoring app is a narrow AI tool that only evaluates speech, not general knowledge.
Natural Language Generation (NLG)
DefinitionThe AI process of producing coherent written or spoken language from data or instructions. NLG enables tools to write essays, explanations, or feedback automatically.
Example in ELTAn NLG system generates a personalized progress summary for each student at the end of term.
Natural Language Processing (NLP)
DefinitionThe broad field of AI focused on enabling computers to understand, interpret, and generate human language. NLP is the foundation for grammar checkers, chatbots, and translation tools used in ELT.
Example in ELTNLP allows a writing tool to detect that a sentence is grammatically correct but stylistically awkward.
Natural Language Understanding (NLU)
DefinitionA subfield of NLP focused specifically on interpreting the meaning and intent behind human language input, rather than generating responses. NLU allows chatbots to respond appropriately to varied phrasing.
Example in ELTThanks to NLU, a chatbot understands both “What's the weather?” and “Is it raining?” as the same request.
Neural Network
DefinitionA computing system loosely inspired by the human brain, made up of interconnected layers that process data to recognize patterns. Neural networks are the foundation of deep learning and most modern AI language tools.
Example in ELTA neural network powers the speech recognition engine behind a pronunciation practice app.
Open-Source AI Model
DefinitionAn AI model whose underlying code and, often, training details are made publicly available for use, study, or modification. Open-source models offer transparency but may require technical expertise to deploy.
Example in ELTA university research team fine-tunes an open-source AI model for automated ELT error analysis.
Overfitting
DefinitionA training problem where an AI model learns the training data too precisely, including its noise and errors, reducing its accuracy on new, unseen data. Overfitting can make a tool unreliable in real classrooms.
Example in ELTAn overfitted essay-scoring model performs well on sample essays but scores real student work inaccurately.
Parsing
DefinitionThe process of analyzing a sentence's grammatical structure to identify parts of speech and relationships between words. Parsing enables AI tools to detect grammar errors accurately.
Example in ELTA grammar checker parses a sentence to identify that the subject and verb do not agree.
Personalized Learning
DefinitionAn educational approach that tailors content, pace, and support to each learner's individual needs, often powered by AI analysis of performance data.
Example in ELTPersonalized learning software gives a stronger student more challenging reading passages automatically.
Plagiarism Detection
DefinitionAI-powered tools that compare submitted text against large databases to identify copied or unoriginal content. These tools support academic integrity but require careful interpretation of results.
Example in ELTA teacher uses plagiarism detection software before grading a suspicious student essay.
Predictive Text
DefinitionAn AI feature that suggests likely next words or phrases as a user types, based on patterns in language use. Predictive text can support writing fluency but may also reduce independent word choice.
Example in ELTA student's typing app suggests “have been” after they type “I,” reflecting common English usage.
Prompt
DefinitionThe input, typically a question or instruction, given to an AI system to generate a specific response. The quality and clarity of a prompt significantly affects the usefulness of AI output.
Example in ELTA teacher writes a detailed prompt asking an AI tool for five B1-level reading questions.
Prompt Engineering
DefinitionThe skill of crafting clear, effective prompts to guide AI systems toward accurate and useful outputs. Strong prompt engineering improves the reliability of AI-generated lesson materials.
Example in ELTBetter prompt engineering helps a teacher get grade-appropriate vocabulary from an AI content generator.
Question Generation (AI)
DefinitionThe automated creation of comprehension, discussion, or assessment questions based on a given text, using AI to identify key information and phrasing.
Example in ELTAn AI tool generates ten comprehension questions from a news article for a reading class.
Reinforcement Learning
DefinitionA machine learning method where an AI system learns by receiving rewards or penalties based on its actions, gradually improving performance through trial and error.
Example in ELTA gamified vocabulary app uses reinforcement learning principles to increase difficulty as students answer correctly.
Semantic Analysis
DefinitionThe AI process of interpreting the meaning of words and sentences, beyond their literal structure, to understand intent and context. Semantic analysis supports more accurate feedback on writing tasks.
Example in ELTSemantic analysis helps a writing tool recognize that “make an homework” is a meaning error, not just grammar.
Sentiment Analysis
DefinitionAn AI technique that identifies the emotional tone of a piece of text, such as positive, negative, or neutral. In education, it can help assess persuasive or opinion-based writing.
Example in ELTA teacher uses sentiment analysis to check whether a student's review is convincingly negative or positive.
Speech Synthesis
DefinitionAI technology that converts written text into natural-sounding spoken audio. Speech synthesis supports listening practice and accessibility for learners with different needs.
Example in ELTA reading app uses speech synthesis to read a story aloud in a natural British accent.
Supervised Learning
DefinitionA machine learning approach where a model is trained on labeled data, learning to map inputs to known correct outputs. Most grammar and essay-scoring tools rely on supervised learning.
Example in ELTSupervised learning trains an essay-scoring model using thousands of essays already graded by human teachers.
Text-to-Speech (TTS)
DefinitionA specific application of speech synthesis that converts written text into audible speech, commonly used in listening exercises and accessibility tools.
Example in ELTA vocabulary app uses text-to-speech so students can hear correct pronunciation of new words.
Token
DefinitionA basic unit of text, such as a word or part of a word, that an AI language model processes individually. Tokens are the building blocks language models use to read and generate text.
Example in ELTThe word “unbelievable” may be split into multiple tokens depending on the model's tokenization method.
Tokenization
DefinitionThe process of breaking text into smaller units, or tokens, so an AI system can analyze or generate language. Tokenization is a foundational step in nearly all natural language processing tasks.
Example in ELTTokenization allows a grammar tool to examine each word in a sentence separately for errors.
Training Data
DefinitionThe data set used to teach an AI model how to perform a task, shaping its accuracy and behavior. Poor-quality training data can lead to biased or unreliable AI outputs in education.
Example in ELTA pronunciation app's training data includes speech samples from learners of many language backgrounds.
Transfer Learning
DefinitionA machine learning technique where a model trained on one task is adapted for a related task, saving time and data requirements. Transfer learning allows general AI models to be customized for ELT purposes.
Example in ELTDevelopers use transfer learning to adapt a general language model for grading IELTS-style essays.
Transformer
DefinitionA neural network architecture that processes entire sequences of text at once, enabling models to understand context more effectively. Transformers are the foundation of most modern large language models.
Example in ELTThe transformer architecture allows a chatbot to keep track of context across a long conversation.
Unsupervised Learning
DefinitionA machine learning approach where a model identifies patterns in data without labeled examples or predefined correct answers. It is often used to discover hidden structures in large text collections.
Example in ELTUnsupervised learning helps researchers group learner essays by writing style without manual labeling.
Virtual Assistant
DefinitionAn AI-powered application that performs tasks or answers questions through voice or text interaction, often used for scheduling, reminders, or basic Q&A support.
Example in ELTA virtual assistant reminds students of upcoming assignment deadlines through a messaging app.
Voice Recognition
DefinitionAI technology that identifies and processes spoken words, often used interchangeably with automated speech recognition. It enables hands-free interaction and pronunciation assessment tools.
Example in ELTVoice recognition software allows a student to complete a speaking exercise without typing.
Word Embedding
DefinitionA technique that represents words as numerical vectors, capturing meaning and relationships based on context. Word embeddings allow AI models to recognize that “happy” and “joyful” are semantically similar.
Example in ELTWord embeddings help a synonym-suggestion tool recommend “content” as an alternative to “happy.”
Zero-Shot Learning
DefinitionAn AI model's ability to correctly perform a task it was not explicitly trained on, using general knowledge instead of task-specific examples. Zero-shot capability allows flexible use of large language models in new teaching contexts.
Example in ELTA teacher successfully asks an AI model to generate a debate rubric, a task it was never specifically trained for, using zero-shot learning.
Continue Learning at NIR Academy
Put these terms into practice
Sources & Further Reading

