Research | Fernando Alva-Manchego

Text Adaptation & Readability

Sat, 01 Jan 2022 00:00:00 +0000

I develop models and resources to simplify text at the sentence, paragraph, and document level, with a focus on making information accessible to language learners, low-literacy readers, and general audiences. My work includes both system development and rigorous human-centered evaluation.

💬 Sentence and document simplification with controllable outputs (e.g. , )
📊 Evaluation of simplification metrics, with insights into their human-alignment and limitations (e.g. )
🧪 Benchmark creation (e.g. , , )
🛠️ Tools: , readability predictors for multiple languages

Funded project:

Welsh Language Technology & Multilingual NLP

Sun, 01 Jan 2023 00:00:00 +0000

I develop NLP resources and tools for Welsh and other low-resource languages, addressing both foundational linguistic challenges and applied downstream tasks. This work spans text profiling, morphological analysis, language proficiency assessment, and multilingual lexical simplification.

🏴󠁧󠁢󠁷󠁬󠁳󠁿 Welsh text complexity analysis and profiling (e.g. )
📊 CEFR language proficiency assessment for Welsh (e.g. ) and multilingual benchmarks (e.g. , )
🔡 Welsh morphology:
🌍 Multilingual lexical simplification pipelines (e.g. )
📖 Readability datasets and models for , , and more

Funded project:

NLP for Education, Healthcare & Social Impact

Wed, 01 Jun 2022 00:00:00 +0000

I apply NLP to real-world challenges in education, healthcare, and social care, working in close collaboration with the NHS, Welsh Government, and academic partners. A central theme is human-AI collaboration: designing systems that support practitioners rather than replace them.

🧑‍⚕️ NLP tools for evidence use and decision-making in social care, as part of (the Centre for Social Care and Artificial Intelligence Learning)
🎓 Research on the impact of generative AI in learning and writing (e.g. )
🗣️ Multilingual survey analysis for public service delivery (e.g. )
⚙️ Co-creation of tools with practitioners and rapid-cycle development for evaluation and impact

Funded project: