Different architectures for different problems.
Oxford answers: “What does X mean?” and “What’s similar to X?”
Linguabase answers those too, plus: “How hard is this word?” “What hints work without sharing morphology?” “Which sense of X connects to which sense of Y?”
Traditional dictionaries and thesauruses are designed for word substitution—helping writers replace one word with another. Linguabase is designed for word games—validated vocabulary, difficulty rankings, readable definitions, content filters, and semantic connections that power hints, clues, and exploration mechanics.
Deep Dive
Words with Spaces — an interactive essay on the half-million compound phrases missing from dictionaries.Traditional dictionaries exclude most compound phrases—not because they aren’t words, but because they contain spaces. “Boiling water.” “Saturday night.” “Best friend.” These function as single semantic units—they name things—but dictionaries leave them out. Oxford includes opaque idioms like “red tape” and “cold feet” (where meaning isn’t predictable from parts), but skips transparent compounds that are just as real.
The result: roughly half a million English expressions that function as words aren’t in any dictionary.
| Oxford | Linguabase | |
|---|---|---|
| “hot dog” (opaque) | ✓ | ✓ |
| “red tape” (opaque) | ✓ | ✓ |
| “boiling water” (transparent) | ✗ | ✓ |
| “front door” (transparent) | ✗ | ✓ |
| “best friend” (semi-opaque) | ✗ | ✓ |
For word games, this matters. Players think in concepts, not orthographic conventions. If your game accepts “discombobulate,” it should probably accept “paper towel.”
Linguabase includes ~200K multi-word expressions—compounds, phrasal verbs, and collocations that traditional dictionaries systematically exclude.
| Dimension | Oxford SELD | Linguabase |
|---|---|---|
| Architecture | Dictionary + thesaurus cross-refs | Complete vocabulary stack |
| Relationship count | 600K synonyms | 40M connections (400K vocabulary) |
| Relationship types | 2 (synonym, antonym) | Weighted by strength |
| Graph operations | None (flat lookup) | Pathfinding, distance, sense annotation |
| Sense handling | IDs for disambiguation | Balanced representation across senses |
| Directionality | None | Directional weights |
| Validation | Human lexicographers | Human + LLM + confirmation filtering |
| False cognate handling | Unknown | 291K removed via LLM audit |
| Gestalt/experiential | None | Visual, sensory, cultural, emotional |
| Usage examples | Example sentences | 1.46M from literature, journalism, scholarly sources |
| Definitions | Fragmented numbered senses | Readable flowing paragraphs |
| Languages | 50+ | English only |
| Audio | Yes | No |
| Brand | 150+ years, Oxford name | New (IDEA.org) |
Oxford’s thesaurus is excellent at what it does — finding substitute words. But that’s a different problem than finding associated concepts.
Oxford helps writers find substitute words. Linguabase helps applications find where users’ minds might go—including senses (card games, guitar parts, dental work) that thesauruses don’t cover at all. See how we build sense-aware associations →
The clearest difference is in how definitions are structured.
Oxford: 5 verb senses, 5 noun senses, sub-senses, phrases, derivatives, etymology — fragmented across 50+ lines. Designed for scholarly completeness.
Linguabase: One readable paragraph covering all senses naturally. Designed for display and AI consumption.
Oxford has 150 years of brand equity. We have over a decade of building vocabulary data for word games.
We’re not offering a cheaper Oxford. We’re offering something Oxford doesn’t provide: a complete vocabulary stack designed for game mechanics—difficulty rankings, content filters, weighted associations, and hints that don’t give away the answer.
If you need to embed dictionary lookups in an e-reader, use Oxford. If you need the data layer for word games—that’s what we built. See licensing options →