CASE STUDY · RELICANA · 2026
Bilingual Reading System
Bilingual Reading System
Designing the content model, interaction system, and adaptive reading modes for a dual-language EN/ES ebook where the reader controls which language leads.
Designing the content model, interaction system, and adaptive reading modes for a dual-language EN/ES ebook where the reader controls which language leads.
The Engagement
My Role
Content Designer
Format
ePub3 / PDF Ebook
Audience
English speakers learning Spanish
Platform
Apple Books, Kobo, PDF
Domain
Content Design, Reading UX, Language Learning
Status
Concept – interactive prototype complete
The Premise
Most bilingual reading tools treat the second language as a fallback. Jalisco Salt treats both languages as the experience.
Jalisco Salt is a bilingual EN/ES novella designed for English speakers learning Spanish. Rather than hiding the Spanish behind a toggle or separating it into a parallel document, the reading experience presents both languages simultaneously on every page – with the reader controlling which language leads.
The design challenge: how do you present two languages on the same page without one overwhelming the other, without the reader's eye bouncing between two equally weighted text tracks, and without the experience feeling like a worksheet?
The Format Problem
The initial target format was PDF – static, linear, no native interaction layer. No hover states, no toggles, no responsive reflow. In a web app, a reader's focus preference is a CSS class swap. In a PDF, it's a typography decision baked into the document at export.
This constraint turned out to be generative. When the format can't carry design intent, the content structure has to carry all of it. The hierarchy between languages, the scaffolding for learners, the pacing of the reading experience – all of it had to live in the text itself.
The real target became ePub3 – supported natively by Apple Books and Kobo. The PDF work was the specification. ePub3 is the implementation.
Content Model
Three structural models were evaluated for how the two languages would coexist on the page.
Parallel columns – English left, Spanish right – fails when sentence lengths diverge. Spanish runs 15–20% longer than equivalent English text. The visual alignment that earns the layout breaks within a few paragraphs.
Stacked blocks – full English paragraph, then full Spanish paragraph – is clean but loses tight coupling. The reader holds the English in working memory while processing the Spanish. The cognitive distance is too wide.
Sentence interleave – English sentence followed immediately by its Spanish equivalent, alternating within each paragraph – keeps both languages in close proximity and maps naturally to how bilingual comprehension develops.
Sentence interleave was selected. The cognitive load is real and worth naming – it was accepted deliberately. The product is designed for readers who want immersion, not readers who want a reference document.
Reading Modes
Four reading modes draw from the same underlying content. The mode is a reading posture, not a different document. An EN/ES pill in the toolbar controls which language is dominant – the right column is always the primary read.
Bilingual
English and Spanish alternate sentence by sentence. The EN/ES pill swaps which language leads – larger, darker, upright – and which sits below as the warm brown subordinate.
Side by side
Two columns always. Right column is always dominant – full weight, primary color. Left is always reference. The right-column convention is grounded in eye-tracking research: the eye sweeps right and lands.
Spanish only
No scaffold, no English. Appropriate from chapter 8–9 onward, after the reader has built enough pattern recognition. A content design decision as much as an interface one – it marks a threshold moment.
English only
Clean reading mode. Glosses are disabled – there is no Spanish text to annotate. The EN/ES pill dims. The toolbar communicates what's relevant to the current mode without hiding controls entirely.
Column headers were removed during iteration. The EN/ES pill already communicates which language is dominant. The layout says what the labels were trying to say – more quietly.
Live prototype – paste URL in right panel
Host on Netlify, then add the URL above
Key Design Decisions
01
Right column is always dominant
In left-to-right reading languages, the eye sweeps right and lands. Right is where attention settles, not where it starts. Placing the dominant language there aligns with natural reading momentum. On a tablet, the right side is also where the dominant thumb rests – embodied comfort, not just visual convention.
02
The toolbar labels, the layout reads
Column headers were cut because the EN/ES pill already does the labeling. Every control in the toolbar has one job: communicate the current reading posture. The page surface has one job: be readable. When those two jobs overlap, noise appears.
03
Scaffolding is opt-in, not mandatory
Glosses annotate Spanish vocabulary on first use – tap to reveal, never interrupting the line. Advanced readers ignore them. Beginners get support without leaving the page. Three levels: no gloss, inline gloss, chapter word list. The reader chooses how much help they want.
04
Controls dim, not disappear
In Spanish-only and English-only modes, irrelevant controls dim rather than vanish. A disappearing toolbar teaches the reader that the system's rules are hidden. A dimmed toolbar teaches them that the system has rules they can understand.
What Got Cut
The inline tap worked. It didn't survive. Understanding why is the design decision.
One interaction model was prototyped and set aside: paragraph-level hierarchy swap via inline tap. The idea removed the focus toggle from the toolbar entirely. Tapping the subordinate brown line promoted it to primary. Each paragraph could be independently flipped. The toolbar became pure navigation.
It was genuinely compelling – the gesture was intimate with the text, the reader's decision living exactly where its consequence appeared.
Discoverability failed: nothing signaled that the subordinate line was tappable. A hint helped on first load but felt like scaffolding for the interface rather than the language – the wrong kind of scaffolding.
Paragraph-level granularity created a mixed page state: some paragraphs ES-primary, others EN-primary, no spatial convention to orient the eye. The reader had no stable habit to rely on.
Setting it aside clarified what the toolbar toggle needed to be: a reading posture decision, not a per-paragraph edit. The inline tap solved a real problem with an interaction that introduced a bigger one.
Open Questions
The reading experience has not been tested with the actual target audience. Several design decisions rest on assumptions that user testing would validate or challenge.
Does sentence interleave sustain across novella length, or does cognitive load accumulate into fatigue by chapter 6 or 7? The model was tested on short passages. It hasn't been tested on 200 pages.
Does the right-column convention hold for readers who approach the text as Spanish primary from the start – heritage speakers, advanced learners? The spatial habit was designed for a particular reader trajectory.
At what chapter does Spanish-only mode feel like a reward rather than a removal of support? The chapter 8–9 threshold is an educated guess, not a tested one.
A content design system that surfaces its own uncertainties is more trustworthy than one that doesn't.