Natural language processing did not feel foreign to me when I first encountered it. Tokenisation, lemmatisation, discourse markers, these were not technical novelties, but concepts I had worked with for years in corpus linguistics and discourse analysis.
Key idea
Linguistics does not sit outside NLP. It provides the conceptual architecture that makes NLP interpretable.
From Analysis to Architecture
For two decades, my work focused on how language encodes meaning: how metaphor shapes perception, how evaluation constructs stance, how discourse reveals power and emotion. These were not abstract concerns. They were structured analytical systems.
When I began building language tools, I realised the same logic applied:
- Identify patterns in real data
- Formalise them into categories
- Define rules and constraints
- Test against variation
That is both corpus linguistics and rule-based NLP.
Building Instead of Interpreting
My transition into development was not a career pivot away from linguistics. It was a shift from interpretation to implementation.
In building Explain My Pain, I translated metaphor taxonomies into structured detection rules. Linguistic categories became YAML schemas. Interpretive insights became matching logic.
The result was not a black-box model but an interpretable pipeline.
Where I Work Now
I work at the intersection of:
- Linguistic theory
- Annotation and evaluation frameworks
- Rule-based system design
- Applied language interfaces
NLP appeals to me because it is not only about prediction. It is about representation. It is about modelling meaning in a way that can be inspected, revised, and improved.
That is where linguistics belongs: not as commentary on technology, but as its structural foundation.