---
kind: concept
slug: text-simplification
title: Text simplification
summary: >-
  Rewriting dense text so a reader without specialist training can follow it. Two papers so far. In a radiology study, 15 radiologists rated 45 chatbot rewrites: mostly correct, but with errors and dropped results. A service robot restates answers in German easy language through an external API, with no measurements.
tags: [text-simplification, medical-nlp, llm, human-evaluation, human-robot-interaction]
papers: ["2212.14882", "2412.05022"]
related: [prompt-engineering, human-evaluation, llm-evaluation, text-summarization, human-in-the-loop, llm-ethics, neural-machine-translation]
status: draft
updated_at: 2026-10-25
---

## Overview

Text simplification rewrites a text so that a reader without specialist
training can follow it, while the content stays the same. Two papers with
cards apply it in different settings. In the radiology study, the content
that must survive is the medical content: a simpler text that changes a
finding has failed, however readable it is. The study checks this with
radiologist ratings of factual correctness, completeness and potential harm
[arxiv:2212.14882 §3.3]. In the service-robot paper, the rewrite restates
official information in German easy language on request. The paper does not
measure whether the restated text is understood or stays accurate
[arxiv:2412.05022 §6] [§7].

## Variants

### Prompted chat model, expert-rated

One fixed instruction, asking the model to explain the report to a child in
simple language, is sent with each report to a general-purpose chat model. Each
input is run 15 times, and every run starts a new chat so that earlier turns do
not influence the output [arxiv:2212.14882 §3.2]. The prompt was picked from
six informal trials on a different fictitious report and is not claimed to be
the best prompt [arxiv:2212.14882 §3.2] [§5.3]. Ratings: the median factual
correctness and completeness were both 2 (agree), and the median potential harm
was 4 (disagree) [arxiv:2212.14882 tab. 1]. The harm statement describes the
risk of wrong conclusions, so the low harm median means raters did not see
much risk. Card: [ChatGPT Makes Medicine Easy to Swallow](/papers/2212.14882).

### Easy-language service inside a robot dialogue

A humanoid service robot answers a customer's question from a stored knowledge
base. When the customer asks, the stored answer is sent over the internet to
an external easy-language service (SUMM AI), and the result is spoken and shown
on a tablet [arxiv:2412.05022 §6]. The paper names no prompt, model or service
settings, and the service's version is not stated [arxiv:2412.05022 §6]. Easy
language is offered for German only [arxiv:2412.05022 §6]. Translation is a
separate mode, through DeepL, and only Danish was tested [arxiv:2412.05022
§6]. The paper reports no quantitative results. Workshop feedback with
public-service staff is described as positive, without numbers
[arxiv:2412.05022 §7]. Card: [Get It Right: Improving Comprehensibility with
Adaptable Speech Expression of a Humanoid Service Robot](/papers/2412.05022).

### Other variants

None else on record. Other simplification methods exist, but no paper with a
card describes one, so this page cannot compare them.

## When to use which

The two studies cover one model in one interface, and one service on one
robot. The rows state only what their results support.

| Situation | Use | Why |
|---|---|---|
| Output goes to a patient with no clinician in the loop | Do not rely on this approach alone | Raters found misread terms, invented findings and missed results. The authors recommend physician oversight [arxiv:2212.14882 §4.2] [§5.2] |
| Output is checked by a clinician before release | Prompted chat model, with a review step | Most raters judged the rewrites correct, complete and not harmful [arxiv:2212.14882 tab. 1], and the authors recommend that a radiologist checks them [arxiv:2212.14882 §5.2] |
| Results must be reproduced exactly | Store the outputs; do not plan on re-running them | The interface had no temperature or seed control, and the authors call exact reproduction "difficult, if not almost impossible" [arxiv:2212.14882 §3.2] [§5.2] |
| Output length matters | Measure length against the source | Median word counts rose for knee MRI (222 to 414) and head MRI (117 to 150), but were similar for oncology CT (241 to 249) [arxiv:2212.14882 tab. 4] |
| Citizen-facing answers in German, from a fixed knowledge base | Easy-language service, with your own accuracy check | The authors show the flow works in a demo. The paper reports no accuracy or comprehension measure [arxiv:2412.05022 §6] [§7] |
| Target language is not German | Translation service, tested in your language | Easy language covers German only. Translation is shown for Danish only, though the service lists 27 languages [arxiv:2412.05022 §6] |
| Customer data would be sent to a third party | Check the data flow first | Customer utterances and answers pass through external services. The paper names data protection as open and describes no safeguards [arxiv:2412.05022 §6] [§7] |
| Spoken input to a robot | Chat dialogue mode over keyword spotting | The authors chose Chat because keyword spotting handles only known phrases. The gain is described without numbers [arxiv:2412.05022 §6.2] |
| A baseline simplifier is needed | Not answerable from these studies | No other system was compared [arxiv:2212.14882 §3.2] [§4]; [arxiv:2412.05022 §6] |
| A rating design is needed | Three Likert statements plus free-text highlights | This is the design the radiology study used [arxiv:2212.14882 §3.3] |

## Implementation sketch

The radiology study has no training step, so the first sketch covers the
generation and rating procedure only. The robot paper gives no algorithm, so
the second sketch restates its request flow in our own words, with the
numbers it omits marked as missing. Both are written in our own words.

```pseudocode
# Generate candidate rewrites for one source text  [arxiv:2212.14882 §3.2]
# Input:  source text, instruction string, repeats = 15
# Output: list of candidate rewrites

def simplify(source, instruction, repeats=15):
    outputs = []
    for _ in range(repeats):
        # Fresh session each time, so earlier turns do not leak in  [§3.2]
        reply = chat_model(new_session=True, prompt=instruction + source)
        outputs.append(reply)
    return outputs
```

```pseudocode
# Build one rating sheet per rater  [arxiv:2212.14882 §3.3]
# Input:  source texts, candidates per source, rater
# Output: one (source, random candidate) pair per source

def build_sheet(sources, candidates, rater):
    # Each source gets one randomly chosen rewrite  [§3.3]
    return [(s, random_choice(candidates[s])) for s in sources]
```

```pseudocode
# Restate a stored answer on request  [arxiv:2412.05022 §6]
# Input:  customer question, mode ("easy" or "translate")
# Output: text to speak and show; no algorithm stated in the paper

def restate(question, mode):
    # Find the stored answer for the matched topic  [§6]
    answer = knowledge_base(match_topic(question))
    if mode == "easy":
        # Service is German only in the paper  [§6]
        return easy_language_service(answer, lang="de")
    # Translation service; the demo used Danish  [§6]
    return translation_service(answer, target="da")
```

Each pair in the radiology study is rated on three statements on a 5-point
scale, with a free-text follow-up for wrong text, missing information and
harmful conclusions [arxiv:2212.14882 §3.3]. Summaries use medians and
quartiles, because the ratings are ordinal [arxiv:2212.14882 §3.4].

## Contradictions

- **Harm maximum.** The text says raters never chose "strongly disagree" for
  harm [arxiv:2212.14882 §4.1] [§5.1.3]. The summary table gives a maximum of 5
  for harm, and the per-report table has 5 in every report [arxiv:2212.14882
  tab. 1] [tab. 2]. The tables are used here.
- **Share finding harm.** The text says about half of the raters found critical
  statements [arxiv:2212.14882 §5.1.3]. The follow-up table says 36% listed
  harmful conclusions [arxiv:2212.14882 tab. 3]. The paper does not explain the
  gap.
- **Between the two papers.** None known. They measure different things, and
  the robot paper reports no numbers that could be set against the radiology
  ratings [arxiv:2412.05022 §7].

## Papers

- [ChatGPT Makes Medicine Easy to Swallow: An Exploratory Case Study on Simplified Radiology Reports](/papers/2212.14882): an exploratory study of a chat model rewriting three radiology reports, rated by 15 radiologists for correctness, completeness and harm.
- [Get It Right: Improving Comprehensibility with Adaptable Speech Expression of a Humanoid Service Robot](/papers/2412.05022): a Pepper robot app that restates knowledge-base answers in German easy language or translates them, with a case study and no measurements.