pith. sign in

arxiv: 2604.19276 · v1 · submitted 2026-04-21 · 💻 cs.HC

Designing Transparent AI-Mediated Language Support for Intergenerational Family Communication

Sora Kang , Youjin Hwang , Joonhwan Lee This is my paper

Pith reviewed 2026-05-10 02:17 UTC · model grok-4.3

classification 💻 cs.HC
keywords translationintergenerationalcommunicationconversationalfamilytransparentai-mediatedblack-box
0
0 comments X

The pith

Transparent AI translation displaying both original and interpreted messages improves conversational quality, intimacy, and usability in intergenerational family chats, whereas black-box translation disrupts flow.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The work addresses language barriers between generations in families by creating an AI chat tool called GenSync. Instead of hiding the translation, the system shows users the sender's original words next to the AI's version in the receiver's language. Researchers ran a study with 16 family pairs who tried three versions of the chat: one with no translation, one where only the translated text appeared without the original, and one where both were visible. In the visible version, conversations felt smoother and more personal because people could see what the AI changed and keep the human connection intact. The hidden version often broke the rhythm of talking because users could not tell how their words were being altered. The study used a within-subjects design so each pair experienced all three conditions and compared their experiences directly. Results indicated that seeing the translation process helped maintain intimacy and made the tool easier to use in sensitive family settings. This treats the AI as an open interpreter rather than a black box that replaces direct understanding.

Core claim

The results show that translation visibility plays a critical role in shaping conversational experiences. Transparent translation supported conversational quality, intimacy, and usability, while black-box translation often disrupted conversational flow.

Load-bearing premise

That short, controlled interactions in a lab setting with 16 dyads generalize to natural, ongoing family communication without major effects from order, translation accuracy variations, or social desirability bias in self-reports.

Figures

Figures reproduced from arXiv: 2604.19276 by Joonhwan Lee, Sora Kang, Youjin Hwang.

Figure 1
Figure 1. Figure 1: GenSync interface. We evaluated three interface versions: (A) Control, with no translation support; (B) Black-box [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Overall user experience with GenSync. Error bars [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
read the original abstract

Intergenerational linguistic differences pose challenges to effective and intimate family communication. This paper presents GenSync, a chat-based interface that supports intergenerational understanding through different forms of translation visibility. We conducted a controlled within-subjects study with 16 family dyads (32 participants), comparing three conditions: no translation, black-box translation, and transparent translation that displays both original and interpreted messages. The results show that translation visibility plays a critical role in shaping conversational experiences. Transparent translation supported conversational quality, intimacy, and usability, while black-box translation often disrupted conversational flow. These findings position intergenerational language support as a form of interpretive mediation and contribute design implications for AI-mediated communication in socially sensitive contexts.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper introduces GenSync, a chat-based interface for AI-mediated language support in intergenerational family communication. It reports on a within-subjects controlled study with 16 family dyads comparing no translation, black-box translation, and transparent translation (displaying original and interpreted messages). The findings indicate that transparent translation improves conversational quality, intimacy, and usability, while black-box translation disrupts flow. The work offers design implications for transparent AI in sensitive social contexts.

Significance. Assuming the empirical results are reliable, this study contributes meaningfully to HCI research on AI-supported communication by providing evidence for the benefits of translation visibility in family settings. It highlights how transparency can enhance intimacy and conversational flow in linguistically diverse families, offering practical design guidelines that extend beyond the lab to broader applications of interpretable AI systems. The direct participant data from family dyads strengthens the relevance to real-world intergenerational challenges.

major comments (2)
  1. §4 (Study Design): The within-subjects experiment with 16 dyads provides the core evidence for the visibility claims. However, the description lacks explicit details on condition counterbalancing, exact session lengths, and controls for translation accuracy (e.g., use of specific models or post-hoc error analysis). These omissions are load-bearing as they could introduce confounds affecting whether differences are attributable to transparency rather than other variables.
  2. §6 (Discussion): The paper draws implications for ongoing family communication, but does not sufficiently discuss threats to validity such as order effects, social desirability bias in self-reports from family members, or the impact of short lab sessions versus repeated natural use. This weakens the central claim that transparent translation supports stable improvements in intimacy and flow beyond the controlled setting.
minor comments (2)
  1. Abstract: The abstract omits any mention of statistical results, effect sizes, or participant demographics, which would provide immediate context for the strength of the reported findings.
  2. Figures: Figure captions and labels could be expanded to better explain the interface elements and condition differences without requiring reference to the main text.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive and detailed feedback. We address each major comment below and have revised the manuscript to incorporate the suggested clarifications and expansions.

read point-by-point responses

  1. Referee: §4 (Study Design): The within-subjects experiment with 16 dyads provides the core evidence for the visibility claims. However, the description lacks explicit details on condition counterbalancing, exact session lengths, and controls for translation accuracy (e.g., use of specific models or post-hoc error analysis). These omissions are load-bearing as they could introduce confounds affecting whether differences are attributable to transparency rather than other variables.

    Authors: We agree that these procedural details are essential for allowing readers to evaluate potential confounds. The original manuscript summarized the within-subjects design at a high level but did not explicitly document counterbalancing procedures, precise session timings, or translation accuracy safeguards. In the revised version, we have expanded §4 with a dedicated 'Experimental Procedure' subsection that specifies the counterbalancing approach, reports the exact duration allocated to each condition, and describes the translation model employed together with the post-hoc accuracy verification steps performed. These additions directly address the concern and strengthen the attribution of effects to the transparency manipulation. revision: yes

  2. Referee: §6 (Discussion): The paper draws implications for ongoing family communication, but does not sufficiently discuss threats to validity such as order effects, social desirability bias in self-reports from family members, or the impact of short lab sessions versus repeated natural use. This weakens the central claim that transparent translation supports stable improvements in intimacy and flow beyond the controlled setting.

    Authors: We acknowledge that a more explicit treatment of validity threats would better contextualize the scope of our claims. The original discussion emphasized design implications but did not systematically address order effects, social desirability in family self-reports, or the ecological limitations of brief lab sessions. We have revised §6 by adding a new 'Limitations and Threats to Validity' subsection that discusses each of these issues, explains the mitigation steps taken where applicable, and qualifies the generalizability of the intimacy and flow findings to longer-term, in-situ use. This revision provides a more balanced and transparent account without altering the core empirical contribution. revision: yes

Circularity Check

0 steps flagged

No circularity: purely empirical user study with data-driven claims

full rationale

The paper reports results from a controlled within-subjects lab study with 16 family dyads comparing no-translation, black-box, and transparent translation conditions. All load-bearing claims about conversational quality, intimacy, and usability rest directly on participant self-reports and observations rather than any equations, fitted parameters, self-referential predictions, or derivation chains. No mathematical models, ansatzes, or uniqueness theorems appear; the study design and findings are self-contained against external benchmarks with no reduction of outputs to inputs by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on empirical user study outcomes; no free parameters, invented entities, or non-standard axioms beyond typical HCI experimental assumptions.

axioms (1)
  • domain assumption Participants provide honest self-reports and conditions are experienced comparably in a within-subjects design.
    Standard assumption in controlled HCI studies for valid comparison of conversational experiences.

pith-pipeline@v0.9.0 · 5407 in / 1299 out tokens · 61627 ms · 2026-05-10T02:17:35.999562+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

18 extracted references · 18 canonical work pages

  1. [1]

    ATLAS.ti: The Qualitative Data Analysis Research Software

    2025. ATLAS.ti: The Qualitative Data Analysis Research Software. https://atlasti. com

    work page 2025

  2. [2]

    Adigwe and E

    P. Adigwe and E. Okoro. 2016. Human communication and effective interpersonal relationships: an analysis of client counseling and emotional stability.Interna- tional Journal of Economics & Management Sciences5, 3 (2016), 3–6

    work page 2016

  3. [3]

    Barrie, John P

    Christopher K. Barrie, John P. Bartkowski, and Tony Haverda. 2019. The digi- tal divide among parents and their emerging adult children: intergenerational accounts of technologically assisted family communication.Social Sciences8, 3 (2019), 83

    work page 2019

  4. [4]

    Anol Bhattacherjee. 2001. Understanding information systems continuance: An expectation-confirmation model.MIS Quarterly25, 3 (2001), 351–370

    work page 2001

  5. [5]

    Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D. Kaplan, Pra- fulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, Sandhini Agarwal, Ariel Herbert-Voss, Gretchen Krueger, Tom Henighan, Rewon Child, Aditya Ramesh, Daniel Ziegler, Jeffrey Wu, Clemens Winter, Christopher Hesse, Mark Chen, Eric Sigler, Mateusz Litwin,...

    work page 2020

  6. [6]

    Lynn Schofield Clark. 2009. Digital media and the generation gap: qualitative research on US teens and their parents.Information, Communication & Society 12, 3 (2009), 388–407

    work page 2009

  7. [7]

    Fred D. Davis. 1989. Perceived usefulness, perceived ease of use, and user accep- tance of information technology.MIS Quarterly13, 3 (1989), 319–340. Designing Transparent AI-Mediated Language Support for Intergenerational Family CommunicationDIS Companion ’26, June 13–17, 2026, Singapore, Singapore

    work page 1989

  8. [8]

    2017.Children and Fami- lies in the Digital Age: Learning Together in a Media-Saturated Culture

    Elisabeth Gee, Lori Takeuchi, and Ellen Wartella (Eds.). 2017.Children and Fami- lies in the Digital Age: Learning Together in a Media-Saturated Culture. Routledge

    work page 2017

  9. [9]

    Hancock, Mor Naaman, and Karen Levy

    Jeffrey T. Hancock, Mor Naaman, and Karen Levy. 2020. AI-mediated commu- nication: Definition, research agenda, and ethical considerations.Journal of Computer-Mediated Communication25, 1 (2020), 89–100

    work page 2020

  10. [10]

    Ifeoma Onyeator and Nkiru Okpara. 2019. Human communication in a digital age: perspectives on interpersonal communication in the family.New Media and Mass Communication78, 1 (2019), 35–45

    work page 2019

  11. [11]

    2019.Trust in automated decision making: How user’s trust and perceived understanding is influenced by the quality of automatically generated explanations

    Andrea Papenmeier. 2019.Trust in automated decision making: How user’s trust and perceived understanding is influenced by the quality of automatically generated explanations. Master’s Thesis. University of Twente

    work page 2019

  12. [12]

    Weiland, K

    Dheeraj Rai, R. Weiland, K. Herrera, T. Shaw, and Z. Yao. 2024. Understanding the Effect of Algorithm Transparency of Model Explanations in Text-to-SQL Semantic Parsing. Preprint or In Preparation

    work page 2024

  13. [13]

    Romer and Bernard J

    Anna L. Romer and Bernard J. Hammes. 2004. Communication, trust, and making choices: advance care planning four years on.Journal of Palliative Medicine7, 2 (2004), 335–340

    work page 2004

  14. [14]

    Philipp Schmidt, Felix Biessmann, and Timm Teubner. 2020. Transparency and trust in artificial intelligence systems.Journal of Decision Systems29, 4 (2020), 260–278. doi:10.1080/12460125.2020.1819094

    work page doi:10.1080/12460125.2020.1819094 2020

  15. [15]

    2018.Family Communication

    Chris Segrin and Jeanne Flora. 2018.Family Communication. Routledge

    work page 2018

  16. [16]

    Shabaniminaabad

    M. Shabaniminaabad. 2020. Comparison of the effect of generational gaps on language use, verbal and nonverbal communication between fathers and their sons from a linguistic perspective.Journal of Research in Educational Science14, Special Issue (2020), 685–701

    work page 2020

  17. [17]

    Stiff, James P

    James B. Stiff, James P. Dillard, Lawrence Somera, Hee Sun Kim, and Christine Sleight. 1988. Empathy, communication, and prosocial behavior.Communication Monographs55, 2 (1988), 198–213

    work page 1988

  18. [18]

    Man Ping Wang, Jasmine T. Chu, K. Viswanath, Alice Wan, Tai Hing Lam, and Sophia S. Chan. 2015. Using information and communication technologies for family communication and its association with family well-being in Hong Kong: FAMILY project.Journal of Medical Internet Research17, 8 (2015), e4722

    work page 2015