While basic reading strategies, such as the Z- and F-patterns are consistent around the world, readers face significant challenges when documents are poorly adapted to their script and cultural norms. The document text itself may have undergone a water-tight process of forward translation and review, but localisation is more than that. It is about aligning the visual structure of the document, the reading flow and formatting expectations to the target audience’s linguistic and cognitive habits.

The Reader Is Thrown Off

This is otherwise known as ‘disrupted visual anchoring’. Left-to-right (LTR) readers are trained to anchor their attention to the left margin, where they expect to find key headings, bullet points and calls-to-action. Right-to-left (RTL) readers, such as those reading Arabic or Hebrew, instinctively anchor to the right margin.

When a text in an RTL script is translated but remains in an LTR layout, (believe me, this happens…) the reader experiences “visual dissonance”. This is where their eyes are drawn to right-hand side margins, but critical information may still be aligned left. This forces inefficient scanning patterns, more backward eye jumps and more effort to find the information that they need.

To illustrate this, a shared-stem question in a Clinical Outcome Assessment imposes a specific cognitive burden even in the source language. A visualisation of the eye-tracking involved in a normal LTR English-language item could be as follows (example provided by AI):

Over the past 7 days, how often did you experience the following symptoms?

What happens when a native English person and LTR reader processes this?

-            The reader will fixate on the shared stem first, to understand the timeframe and context

-            Then, the reader will move down to the first symptom (headaches) and scan across the response options horizontally.

-            After the first two items, they may look back at the stem to remind themselves of the timeframe (the past 7 days), especially if the stem is long or complex.

-            Then, once familiar with the structure, they continue scanning down the list in a more linear fashion.

The result is a different reading strategy, as a mocked-up eye-tracking model shows:

If the RTL script version, such as Arabic, of this same item is visually localised to meet the expectations of RTL readers, the processes involved should be a mirror image of the LTR version. However, if it is translated and still only maintains a LTR appearance it would look as follows:

خلال الأيام السبعة الماضية، كم مرة شعرت بالأعراض التالية؟

(again, believe it or not, this is not uncommon, although this translation was created by AI to deliberately keep the LTR directionality).

This disrupts reading, causing the reader to go through the following processes:

-             Right-to-left search first, then correction to left margin.

-            Multiple exploratory fixations across the page.

-            Inefficient left-to-right saccades (the opposite of the reader's habit).

-             Frequent regressions to labels/items.

-            Overall slower reading flow and more effort.

In this instance, the eye-tracking may look as follows:

Inefficient Use of COA Real Estate

In Arabic-speaking countries, important information that remains on the left side of a COA or an information leaflet may be overlooked because of the reader’s right-margin scanning bias. Even when the translation is correct, the positioning will make the document seem foreign and out-of-place to the reader. In RTL scripts, the prime visual real estate shifts with the reading direction. Therefore, layouts need to be mirrored to preserve the natural reading flow.

Otherwise, readers may skip or miss instructions, misunderstand key information. In a health context, this can undermine safety, informed consent and patient understanding.

The eye-tracking pattern of left-aligned Arabic text could be visualised as follows:

When COAs are translated without adapting the layout to match a reader’s natural scanning behaviour, such as leaving the Arabic translation in a left-aligned, LTR structure, the result is awkward design. However, it is more than that: it is a potential threat to data integrity and patient safety. It causes visual dissonance and disrupted reading that slows down the processing of the document. It increases the risk of skipped content and diminishes trust in the assessment. For clinical outcomes research, where comprehension is critical, respecting the reader’s directional habits is not optional – it is essential.

Thank you for reading,

Mark Gibson

Leeds, United Kingdom, March 2025