Published September 2025 | Version v1

Segmenting Italian Sentences for Easy Reading

  • 1. Università di Bologna
  • 2. ROR icon Universitat Pompeu Fabra

Description

Abstract


 Easy Read texts are essential for individuals with reading difficulties. These texts are developed according to institutional
guidelines that establish clear rules for writing and structuring content in an accessible way. A key feature of Easy Read texts
is the segmentation of sentences into smaller grammatical units, often presented on separate lines, to enhance readability.
While several studies have addressed content simplification in easy-to-read materials, much less attention has been paid to
the automatic segmentation of such texts. This project investigates whether this kind of segmentation can be automated
in a reliable and efficient way, even with limited resources. The main goal is to develop and evaluate automatic methods
for splitting texts into simpler, shorter units to support text simplification and improve overall readability. The methods
developed and evaluated are a decision tree classifier and a prompting-based method using a large language model (LLM).
The work is focused on Italian, and the application of these methodologies to this language represents a novel contribution.


Keywords
Text simplification, easy-to-read, automatic segmentation, ER resources, CLiC-it

Files

Segmenting Italian Sentences for Easy Reading.pdf

Files (695.7 kB)

Name Size
md5:d25ef93c01391f719780a90f3592578a
695.7 kB Preview Download