5 exercises — the same word can mean very different things depending on context. Non-native speakers often know the everyday meaning but miss the technical one — or vice versa.
Words with multiple meanings in this set
thread — string (everyday) / email replied chain / unit of concurrent execution
fork — utensil (everyday) / fork() system call / copy a repository independently
commit — promise (everyday) / save a snapshot to Git history
token — coin substitute / NLP text unit / auth credential (JWT, OAuth)
model — fashion (everyday) / 3D mesh / MVC layer / trained ML algorithm
0 / 5 completed
1 / 5
A senior engineer writes in a code review:
"This computation is CPU-heavy — move it to a worker thread so the main thread stays responsive."
What does "thread" mean in this context?
Thread (computing) — a unit of concurrent execution within a process. A process can have multiple threads running simultaneously, sharing the same memory space. Main thread in browser JavaScript: the single thread that handles UI rendering and user events — blocking it freezes the page. Worker threads (Node.js) and Web Workers (browser): separate threads for CPU-intensive work. Thread safety: code that works correctly when accessed by multiple threads without data corruption.
The same word, different contexts: • Thread (computing) = unit of execution (this exercise) • Thread (email/Slack) = a chain of replies in one conversation ("reply in thread", "start a new thread") • Thread (everyday English) = string, fibre
All three are common in a software team's daily language. Context makes the meaning clear.
2 / 5
A team discusses an open-source project in Slack:
"The upstream project hasn't responded to the security PR in 3 months. We should fork it and maintain our own patched version."
What does "fork" mean here?
Fork (Git / open-source) — creating a personal or organizational copy of a repository that becomes independent from the original. You can make changes to your fork without affecting the original ("upstream") project. Common uses: contributing to open-source (you fork → make changes → open a pull request back to the original), or maintaining your own divergent version when upstream is inactive or doesn't meet your needs.
The same word, different contexts: • Fork (Git/GitHub) = copy a repository independently (this exercise) • Fork (Unix/Linux) = fork() system call that creates a child process by cloning the parent process — used in systems programming • Fork (project) = a project that started from another codebase and evolved independently (LibreOffice is a fork of OpenOffice) • Fork (everyday English) = the utensil; also a road junction
In team conversations: "fork the repo" = always the Git meaning.
3 / 5
In a daily standup:
"I'll commit the fixes tonight and push the branch — should be ready for review tomorrow."
What does "commit" mean here?
Commit (Git) — a snapshot of changes saved to the repository's history. Each commit has: a unique hash ID, an author, a timestamp, a commit message, and a diff. git commit -m "fix: resolve null check in payment handler" creates a new commit. Commits are the fundamental unit of change in Git — you review them in PRs, revert them if something goes wrong, cherry-pick them onto other branches, and squash them to clean up history.
The same word, different contexts: • Commit (Git) = save a snapshot of changes (this exercise — by far the most common use in a dev team) • Commit (general English) = dedicate yourself, make a promise ("are you committed to this project?") • Commit (databases) = confirm a transaction permanently (in the ACID sense: atomicity, consistency, isolation, durability)
In standup/Slack: "commit" almost always means the Git operation.
4 / 5
A frontend developer says:
"The app stores a refresh token in localStorage and uses it to get new access tokens without requiring the user to log in again."
What does "token" mean in this context?
Token (authentication) — a credential string that proves identity or grants access to resources, without requiring a password on every request. Access token: short-lived (minutes to hours), used on every API request in the Authorization header. Refresh token: long-lived (days to weeks), stored securely — used only to obtain a new access token when the old one expires. JWT (JSON Web Token): a common token format — contains encoded claims (user ID, roles, expiry) signed with a secret key.
The same word, different contexts: • Token (auth/API) = credential string — JWT, OAuth access token, API key-style token (this exercise) • Token (NLP/AI) = a unit of text that a language model processes; "this prompt uses 1,200 tokens" — GPT charges by token, which roughly = ¾ of a word • Token (blockchain) = a digital asset on a blockchain (ERC-20 token, NFT) • Token (everyday English) = a physical coin substitute
In an auth discussion: always the credential meaning.
5 / 5
A data scientist presents results:
"We trained three models on the dataset — a random forest, a gradient boosting model, and a logistic regression baseline. The random forest outperformed the others with 94% accuracy."
What does "model" mean here?
Model (machine learning / AI) — a trained algorithm that has learned patterns from data and can make predictions, classifications, or generate output for new inputs. Training a model = fitting its parameters to minimise error on training data. Examples: decision tree, neural network, linear regression, transformer (the architecture behind GPT). Key vocabulary: train a model, evaluate a model, deploy a model, fine-tune a model, model accuracy/precision/recall, baseline model (the simplest model used for comparison).
The same word, different contexts: • Model (ML/AI) = trained algorithm (this exercise — extremely common in modern tech teams) • Model (MVC architecture) = the data / business logic layer in the Model-View-Controller pattern • Model (3D graphics) = a mesh geometry object (.obj, .gltf files) • Model (data modelling) = the schema/structure of your data (ER model, domain model) • Model (everyday English) = a person who wears clothes; also a miniature representation ("model airplane")
In a data science context: always the ML training meaning.