Notes on a SIGCSE TS 2026 poster: what two hours at a poster board actually demands

I am one of nine listed authors, seventh in author order, on the poster Exploring Bilingual Coding for Inclusive Computer Science Learning at the ACM SIGCSE Technical Symposium 2026 in St. Louis, Missouri. It appeared in Posters Session #3 and is published in the proceedings at pages 1559–1560, DOI 10.1145/3770761.3777339. It is archived at SJSU ScholarWorks. It is a poster, not a paper, and the distinction matters more than it looks — it changes what the format can carry.

This is a note about the format itself: what a two-page abstract and a poster board can and cannot do, and what I would tell another undergraduate preparing one. The research belongs to the whole author team and to Dr. Ethel Tshukudu's CS Education Research Lab at San José State University, where I have worked as a student researcher since August 2024.

The format is a constraint, not a lesser paper

A full paper gets a reader who has already decided to spend twenty minutes. A poster gets a person walking past a board in a crowded hall who has decided nothing. The two artifacts are not the same thing at different lengths; they have different failure modes.

A paper fails by being unconvincing. A poster fails by being unreadable from a metre and a half away. That single physical fact drives most of the design decisions, and it is the one most first-time poster authors discover too late — usually while standing in front of the printed thing.

What the study was, in the space a poster gets

The work asks whether the dominance of English in programming notation creates a barrier for students who are not native English speakers. Sixty bilingual participants at San José State University — 40 with little or no prior programming experience, 20 experienced programmers — completed a two-hour workshop on the multilingual Hedy platform. Pre- and post-surveys measured changes in confidence, enjoyment, motivation, identity, and perceived usefulness, and open-ended questions captured the reasoning behind those shifts.

Students improved across most measured constructs. Inexperienced students gained more in confidence and enjoyment than experienced ones; experienced programmers showed smaller gains and generally preferred coding in English. In the qualitative responses, most students said English supported clarity, while bilingual students more often reported that working across both languages helped conceptual understanding, despite translation friction.

That is close to the entire finding, and it fits on a poster. Notice what it does not contain: no single headline percentage. That absence is deliberate, and it is the part I want to dwell on.

The number you want and the number you have

There is enormous pressure, on a poster more than anywhere, to produce one big number. A board with a 60-point figure in the middle attracts more foot traffic than a board with a paragraph. Every incentive in the format pushes toward a headline statistic.

The problem is that the honest version of most education findings is directional and conditional: this group improved more than that group, on these constructs, in this setting, over two hours. Compressing that into a single percentage requires choosing a comparison, and the choice is rarely as obvious after the fact as it looks on the board.

I have since had to correct my own public description of this work more than once, and both corrections ran in the same direction: I had written down something rounder and more quotable than the record supported. The fix is not a better memory. The fix is a rule — never write a number about your own research that you cannot trace to a line in the published abstract while you are writing it.

Structure that survives ninety seconds

The sequence that works at a board is not the sequence of a paper. A paper builds. A poster has to be enterable at any point, because people arrive mid-thought.

1. The question, in one sentence, at the top. Not the contribution — the question. A passer-by decides whether to stop based on whether they recognise the problem.
2. The setting, in numbers. Who, how many, how long, what tool. This is where 60 participants, two hours, and Hedy belong. It lets someone calibrate immediately.
3. One figure that carries the finding. Not three. If two figures are competing for the centre of the board, the poster has two arguments and will land neither.
4. The finding as a sentence, not a caption. Written out, in words, large enough to read from a distance.
5. Limitations, visible and unhidden. Counterintuitively this attracts the good conversations. Researchers who see an honest limitation stop to talk; researchers who see a suspiciously clean result walk on.

Preparing three explanations, not one

The single most useful preparation is writing out three versions of the work at different lengths, and rehearsing them until they are separate objects rather than truncations of each other.

• Fifteen seconds. One sentence, no methods, no caveats. It exists to let a person decide whether to stay. "We ran a bilingual coding workshop and looked at whether working in two languages changed how beginners felt about programming."
• Ninety seconds. Question, setting, finding, one limitation. This is the version you will give most often.
• Five minutes. Methods, the measures, what the qualitative data added, and where the design is weak. This is the version another researcher in the same subfield will pull out of you, and it is the one that produces useful criticism.

The rehearsal matters because the ninety-second version is not the five-minute version spoken quickly. It has a different shape. If you have only prepared the long one, you will start it, get interrupted by someone's polite exit, and never reach the finding.

What I would tell an undergraduate preparing one

Get the abstract's claims and your board's claims into exact agreement, word by word, before printing. It is easy for a poster draft to drift a little more confident than the text it is based on, because a board rewards confidence and a paragraph does not.

Know the parts of the method you did not personally run. On a nine-author poster, no single author executed everything, and the useful answer to "how did you handle X?" is often "that was another author's part of the design, and here is what I understand about it" — which is both true and more credible than an improvised answer.

Write down the questions you could not answer. Two hours at a board is an unusually dense source of external criticism, delivered free by people with no stake in flattering you. The questions that stopped you are the most valuable artifact you leave with, and they are gone by the next morning if you do not record them.

Why this line of work is worth doing

The underlying observation is not exotic. A student reading for, while, return, and else is reading English, and if English is a second language then part of their working memory is being spent on translation that a native speaker spends on the algorithm. That cost is invisible in a grade book, which records only the outcome.

I see the same pattern from the other side as a CS tutor at San José State, where a student who cannot get a loop working can often describe exactly what the loop should do, in Arabic or in English, when asked to say it aloud instead of type it. I have written separately about why syntax rather than logic is frequently the barrier and about what language-aware instruction looks like in practice.

Confidence is not a soft outcome in an introductory course. It is close to a proxy for whether a student enrols in the next one. That is the reason a two-hour intervention with a modest, conditional effect is still worth measuring carefully — and worth describing accurately, at the exact size the evidence supports.

For the full citation, the author list, and my 2025 SJSU College of Science Student Research Day presentation, see the research summary.