Design Opportunities for AIED to Support Parents Learning Literacy Michael A. Madaio Amy Ogan Carnegie Mellon University Carnegie Mellon University Pittsburgh, PA, USA Pittsburgh, PA, USA mmadaio@cs.cmu.edu aeo@cs.cmu.edu ABSTRACT face-to-face family literacy learning programs have been developed, Gaps in adult literacy present major barriers to personal and eco- these face the same challenges to participation as adult education nomic development. In-person adult education has leveraged AI programs more broadly [9], AIED systems could o�er personal- systems to personalize the educational experience, but existing ized, targeted support for parents learning literacy, in much the family literacy education programs are often di�cult for parents to same way as they provide personalized learning opportunities for complete. In this paper, we reframe parents as literacy learners as informal learning and continuing education more broadly. Exist- a rich opportunity for AIED systems to support parents learning ing approaches to ML-driven adult learning platforms, however, literacy with their children. We synthesize prior literature and out- have largely not addressed the particular needs, desires, and design line a set of design considerations and design directions for AIED considerations of parents as learners. If they mention parents at systems to provide unique supports for parents as literacy learners. all, such systems or interventions treat adults with children as a barrier to learning [39], rather than as an opportunity to leverage KEYWORDS parents’ interactions with children simultaneously learning literacy to foster mutually supportive and bene�cial learning. In this view, AIED, family learning, literacy, parents, lifelong learning parents tend to be seen as being represented by their de�cits (Cf. ACM Reference Format: [33]), rather than being agents of their own learning, with their Michael A. Madaio and Amy Ogan. 2019. Design Opportunities for AIED own values, goals, and aspirations. to Support Parents Learning Literacy. In Proceedings of Supporting Lifelong We focus in this paper on parents learning literacy, as literacy is Learning Workshop (SLLL’19). at AIED 2019, Chicago, 2019. a fundamental precursor to accessing other forms of education, ac- 1 INTRODUCTION cessing jobs or economic opportunities, in addition to, as Freire and Sen argue, providing a means by which people may resist oppres- According to the National Center for Education Statistics, nearly sion and live lives they have reason to value. We contribute here 18% of adults in the United States cannot read at an age-appropriate a reframing of parents as learners, from a de�cit-based approach level [41]. Globally, systemic shocks such as school closures, civil to an asset- and opportunity-based approach to supporting their wars and political struggles, and, in many under-resourced contexts, lifelong learning through AIED. the more mundane but everpresent demands of the agricultural In this paper, we discuss how speci�c barriers for parents’ literacy harvest cycle have led to low adult literacy rates, despite rising learning might instead be reframed as opportunities to support adult literacy worldwide [41]. These gaps have consequences for their learning, and we begin to map out the design space for AIED people’s lives, livelihoods, and wellbeing, with illiteracy limiting systems that can support parents learning literacy by drawing on access to jobs [27] and, more broadly, presenting obstacles to what their strengths, resources, and assets, rather than their obstacles. educational philosopher Paolo Freire has called "tools for resisting oppression" [20] or, more simply, what the philosopher economist Amartya Sen has called the opportunity for people "to live lives 2 PRIOR WORK they have reason to value". [44]. For decades, adult education programs have been developed to 2.1 AIED for Adult Education teach fundamental skills such as literacy [5], but in-person adult AIED systems and methods have been used in adult education in a education courses face challenges to providing this education at variety of ways over the years, from more formal, degree-granting the right time, pace, and level of di�culty for individual learners, in courses, to online learning, to self-directed informal learning op- addition to other challenges [28]. Digital learning platforms, such as portunities. Much of the work on technology in adult education massive, open, online courses (MOOCs) and educational apps, sug- has focused on what is often referred to as "work and learn", where gest possible paths forward, and when driven by machine learning learners complete certi�cations or on-the-job training [13] to im- technologies, may be able to personalize the learning experience to prove or gain new skills. Some of this AIED work has developed better support learners. However, despite their equalizing promise, virtual agents as lifelong learning coaches, as in the PAL3 personal many online educational platforms are primarily used for continu- assistant [46] to support US naval o�cers’ continuing education. ing education by people with a college degree, and are underused Others have taken a more systemic approach to using ML to sup- by those who lack fundamental literacy skills [11]. port adults’ educational development, by developing job advising Adult learners are more likely to have families than traditional agents, such as the "Continuous Cognitive Career Companion" [1]. learners. Some have argued that having a family presents "situa- Given the high rates of learner attrition in adult education courses tional barriers" to pursuing lifelong education [38]. While many [39] due in part to exogenous factors in learners’ lives, some work Copyright held by the author(s). Use permitted under the CC-BY license CreativeCommons.org/licenses/by/4.0/ SLLL’19, June 2019, Chicago 2019 Madaio and Ogan has leveraged ML to identify predictors of adult learners’ dropout in this space has instead focused on teaching parents the requisite from courses, as in work with English as a second language (ESOL) declarative knowledge about how to teach literacy, suggesting par- courses in Turkey [16]. In addition to this work on predicting ticular letter-naming activities or messages to tell children about dropout from in-person courses, prior work has leveraged machine reading [55], or designing a coaching program to help parents de- learning to identify predictors of dropout from MOOC courses velop the skills and self-e�cacy to foster their children’s literacy [39, 53]. To address these dropout risks, other distance learning pro- through joint reading activities [25]. However, prior interventions grams have attempted to use personalized instruction as a means have largely not assessed parents’ literacy (or knowledge of how to improve learner retention and outcomes in adult education, with to teach literacy) either before or after the intervention, and it is one example from Hong Kong University’s lifelong learning pro- thus di�cult for those interventions to say what, if anything, par- gram using an intelligent tutoring system (ITS) called SmartTutor ents learned from teaching their children. And yet, signi�cant prior to recommend particular content or learning strategies [10]. work suggests that teaching others is likely to have learning bene- While some of these approaches for ML-driven supports for �ts for the one doing the teaching, if sca�olded e�ectively for their adult learning are in online learning environments, others are in- respective abilities [6, 40]. This prior work, however, has largely creasingly leveraging blended learning approaches to supplement focused on peers teaching other peers, or students teaching virtual online learning with in-person instruction and learning communi- agents, and has not been designed to provide the sca�olding that ties [13]. In some cases, particularly in developing contexts where low- or non-literate parents might need (and bene�t from) when local experts in a particular skill may be scarce, adult education supporting their children’s burgeoning literacy. This suggests a courses are o�ered which combine online courses (i.e. MOOCs) with need to reframe the idea that having a family is an obstacle to par- in-person meetups to facilitate learners’ growth in these courses ents’ learning - particularly for literacy - in order to see it as an [15, 32]. This suggests an analogous approach for family learning, opportunity for AIED to design data-driven sca�olds for parents’ where a technology may augment the existing in-person networks literacy learning. of support. However, many researchers cite con�icts between adults’ sched- 3 DESIGNING AIED FOR PARENT LITERACY ules and the demands of formal adult education courses as reasons In this section, we propose a set of design considerations for AIED for the high rates of dropout in adult education courses [39]. Given systems to support lifelong literacy learning for parents, and discuss this, many ESL adults in the US use everyday technologies like possible design directions for AIED based on these . Google Translate and YouTube as sites for language learning across contexts [52]. However, these are typically not explicitly designed to support longitudinal learning, particularly for parents. 3.1 Design considerations for parent literacy AIED Design to support learning-by-teaching. Substantial prior lit- 2.2 Parents as Learners erature has demonstrated the bene�ts of learning by teaching - A common thread through much of the prior work on AIED in though the majority of this work has been with peer tutoring and adult learning is that external factors in adult learners’ lives such virtual agents [6, 40]. In this body of work, students who them- as having a family may be barriers to their learning, rather than selves are not experts in a particular domain (e.g. algebra) receive resources in their lifeworld that can be leveraged in the design of some sca�olding or support, and then teach or coach their peers, AIED systems [16]. Parents, like other adult learners, have aspi- leading to improved learning than if either student were to learn rations for their own growth and development and may want to alone [19, 43]. Taking a learning-by-teaching approach would align learn, but they may face unique barriers to completing more formal with prior research on adult learning, which has argued that adult degree-granting courses due to demands on their time and atten- learners are increasingly motivated to learn when they know why tion from children - what some have referred to as the "all-hours and how to use what they are learning, and there is a speci�c need undertaking" of child-rearing [33]. or goal to learn the content [28]. This suggests that parents may be However, this all-hours undertaking presents unique opportu- able to receive "just-in-time" learning supports to foster particular nities for parents to learn while parenting. For instance, in their literacy skills just prior to teaching those to their children [7, 35]. work on parents’ involvement in children’s new media learning, However, this requires a su�ciently robust knowledge model of Barron et al. describe a variety of roles parents play in their chil- both parents’ and their children’s literacy abilities in order to pro- dren’s learning, from teacher, collaborator, to providing learning vide these just-in-time instructional prompts for the right skills or resources [4]. DiSalvo et al. extend this taxonomy of roles to also "knowledge components" that both the parents and their children include co-learner, where the parent is also learning along with - or need. even from - their child [17]. Bannerjee et al. (2018) adopt this frame Some have argued that parents’ literacy ability and self-e�cacy for their work on English language-learning (ELL) families jointly (or, belief in their own ability) may be an obstacle to their ability to engaging in computer programming, despite a lack of expertise (or teach their children literacy [26]. However, prior work on Latino- even literacy) on the part of the parents [3]. American parents working with their children to teach Spanish While decades of research has demonstrated the crucial role that literacy suggests that by emphasizing the skills and resources that parents play in supporting their children’s literacy [31, 45], signi�- parents already possess, such as their wealth of cultural knowledge, cantly less research has focused on whether and how parents learn adults may be able to overcome gaps in explicit domain knowledge while teaching their children to read. Much of the existing work [30]. Other work has found that parents’ self-e�cacy can improve Design Opportunities for AIED to Support Parents Learning Literacy SLLL’19, June 2019, Chicago 2019 when they see that their child has learned, and that their instruction instructions, as in family learning coaches (Cf. [25]) or after-school was e�ective [26]. This suggests that an AIED system could provide family literacy classes [26]. However, families from historically personalized updates to parents on their children’s (and their own) marginalized communities may face additional barriers for access- progress, to help motivate parents and bolster their self-e�cacy. ing in-school instruction, either due to prior negative experiences Design to support co-learning with children. Literacy is so- with schools or di�erences in language [17], or, in the US, the very cial and cultural in nature, drawing on cultural knowledge and real fear of deportation due to engagement with apparatuses of developed through social interactions with others [14, 20]. In in- the state. In prior work on low-literate Latino-American parents tergenerational learning, these social interactions may allow each in the US, Wong-Villacres et al. found that school liaisons may be member of the family to support the others, in mutually bene�- able to bridge between families and schools, allowing parents to ac- cial ways [21]. Larrotta and Ramirez found that when low-literate cess information they would not otherwise have been able to [52]. Latino parents were provided with resources to support their chil- AIED systems may support this family-school learning ecology dren’s literacy, their own literacy developed as a result of engaging by identifying the literacy skills parents need most support with, with the texts their children were reading [30]. In their work on and aligning those with curricular items they could engage with the information-seeking practices Latino American families, Yip et at home. Following the model of the family-school liaison, AIED al. found that bilingual children acted as "information brokers" in systems might develop virtual literacy coaches to serve a similar accessing and communicating online information to their parents role in providing educational opportunities across school and home [54]. While this work doesn’t focus on literacy learning explicitly, contexts. it suggests possibilities for mutually bene�cial co-learning between Finally, parents in other families may provide additional support children and parents. In a di�erent domain, Roque et al. studied for parents learning literacy, be that through explicitly teaching parents co-learning with their children while developing compu- reading concepts, providing socio-emotional supports to help moti- tational media using Scratch and Makey Makey, and found them vate parents to learn, or connecting low-literate parents to broader developing and using skills that neither had when they started [42]. learning networks. Some prior work in connecting parents across All of this suggests that an AIED system might provide or identify families has studied parents’ use of technology to develop and structured "teachable moments" in which parents and their chil- maintain social networks with other parents, suggesting that such dren could engage in co-learning for literacy. This might involve networks provide a social ecology wherein parents can learn from providing reading materials at an appropriate level of di�culty more knowledgeable or experienced parents [50]. While this work for both child and parent (Cf [2, 48]) or automatically generating focused on parents’ knowledge of parenting, they are able to see structured reading comprehension questions or prompts (Cf. [24]) other parents in these networks as a model for themselves, fostering based on texts that parents and children choose together (as in [30]). motivation and self-e�cacy, as well as providing learning opportu- Building o� of the idea of parents learning-by-teaching discussed nities [50]. AIED systems might thus support parents’ use of social previously, in a co-learning approach, AIED systems might suggest networks, be they extant networks such as Facebook, Twitter, or complementary knowledge components or literacy skills to learn, WhatsApp, or more dedicated networks just for parents, as in "Par- by developing a learner model for both parents and children. That entopia" [49]. Such AIED supports might include recommending is, AIED systems might identify those sets of skills that are mas- particular clusters of parents to talk to for certain literacy topics, or tered by one party (parent or child) and not the other, and design recommending certain reading materials, parenting approaches, or activities to foster the bene�ts of co-learning that may otherwise parent-child literacy lessons based on a similar user pro�le, using a occur only serendipitously. collaborative �ltering approach. Design to engage other adults, other parents, and com- munity members in the family learning ecology. Finally, we widen the lens of focus from parents teaching or co-learning with 3.2 Design directions for parent literacy AIED children to other adults in the family and adults in other families To incorporate these design considerations for parents as learners and the community with whom parents may engage in their learn- of literacy, we suggest that the AIED �eld develop methods and ing process. Prior work suggests that literacy learning - regardless advance theories in some critical ways. of age - draws on what Gonzalez, Moll, and Amanti (2005) call First, we suggest that AIED researchers explore new methods, "funds of knowledge" [23] and Yosso (2005) call "community cul- systems, and approaches for parent-child complementary learner tural wealth" [56]. These funds of knowledge may involve stories, models. Such approaches may model the literacy knowledge of traditions, family wisdom, values and dispositions towards learning both children and their parents and suggest content (e.g. particular and literacy. Each of these can be resources that AIED - and instruc- words) or methods for parents to teach their children, or opportu- tional systems more generally - can leverage to support parents’ nities for mutual support and co-learning. As a precursor, it will be literacy learning. For instance, speech recognition systems have critical to understand how parents who have mastered a particular been developed to improve children’s early reading skills [22, 36], or literacy knowledge component may be able to develop that knowl- for children’s speech-based vocabulary practice [29]. Such speech edge component in their child’s literacy practice. AIED researchers recognition systems may additionally be able to automatically tran- may thus also explore what types of sca�olds an AIED system scribe oral traditions for family stories and sayings, and provide might need to provide to parents to help them teach the concepts locally relevant content for parents to use to develop reading skills. they already know, while also helping them develop concepts they Additionally, school teachers and other representatives of for- have not yet mastered. This may also involve suggesting to parents mal learning can provide one method for parents to access literacy opportunities for their children to teach them certain concepts in a SLLL’19, June 2019, Chicago 2019 Madaio and Ogan collaborative activity. A su�ciently robust AIED system may use reading ability, or may incorporate a pedagogical agent as a char- the data on children’s mastery development to suggest to parents acter in the story or playing the role of a virtual literacy coach to to learn particular skills just before working with their child on support both parents’ and children’s reading. New theories and that skill, using a "just-in-time" approach. Finally, AIED systems models for such virtual learning coaches are needed, however, to that incorporate these complementary learner models may share understand how to design these systems in ways that are supportive theirs and their child’s learning progress with the parents, to help of, and not replacing, this critical parent-child joint engagement. motivate them to continue learning themselves, if they can see that their co-learning is bene�cial for their children. However, this may require su�ciently open and comprehensible learner models to 4 CONCLUSION communicate that learning progress to parents (Cf. [8]). Global gaps in adult literacy present barriers for economic and When and where should all this learning take place? To address personal development. AIED systems have been developed to sup- this, we suggest research directions for AIED to develop methods, port adult education, but this work has not yet developed theories, theories, and systems for contextually-aware family literacy methods, or systems to explicitly support parents as learners, often learning. Given the "all-hours undertaking" of parenting [33], prior viewing the family as an obstacle, rather than a unique opportunity work has developed interventions for parents to support children’s for learning. In this paper, we synthesize prior work on AIED in literacy learning in out-of-school contexts. Some have developed adult education and parents as learners, and we suggest design prompts for parents to discuss print in the environment, using foods considerations and design directions for AIED to support parents’ at the grocery store as a way for children to learn new words [37]. literacy learning. AIED systems might take a similar approach and suggest activities There remain some large open challenges for this research space or lessons to learn in a variety of contexts based on the learners’ not yet discussed. First, for AIED systems to be e�ective, they may location or inferred activities. Similarly, prior work has developed require large corpora of training data - data which may be di�cult an intervention to send parents SMS messages with tips or hints to collect from families. Families may not want tech platforms col- on how to help support their children’s literacy [55]. AIED systems lecting data on them or their children, often with good reason, as may build o� of this work by adding intelligent recommendations recent data scandals for in-home intelligent platforms like Alexa for the preferred context (e.g. time, place, activity, etc) for these reveal. Or, in the absence of such corpora, AIED designers may activities, suggesting appropriate tips or activities for di�erent explicitly knowledge engineer such systems, though this approach contexts. "Context" in this case may also involve more than just the may be prohibitively expensive, and may not be robust to changes time and place, but may involve the parents’ use of other apps. For in learners’ needs over time. Further, many parents developing liter- instance, the work of Wong-Villacres et al. suggests that everyday acy may be bilingual parents from nondominant linguistic groups, technologies could be augmented through intelligent support to and may be literate in another language other than the dominant track parents’ learning (e.g. through their use of Google Translate) language. AIED researchers developing parent literacy tools should or suggest ways to augment their learning in apps they use on a thus be sensitive to the political dimensions of language, and not regular basis [52]. This may require leveraging a parent learner unintentionally reinforce existing systems of oppression through model as described previously, to model their growth on certain their choice of language to teach (Cf. [34]). In fact, in such situations, concepts over time, or to suggest opportunities for parents to engage AIED literacy systems may be able to support interlingual families with children’s literacy together on apps both use regularly. where children may be literate in a language the parents are not Finally, mirroring the use of family literacy coaches, we also (and vice versa). This may take inspiration from computer-assisted suggest that AIED develop designs for virtual family literacy language learning (CALL) systems [47], and from prior work on coaches. These may take the form of spoken dialogue systems, bilingual children supporting ESOL parents in Latino families in the pedagogical agents in intelligent storybooks, or virtual agents that US [54]. Finally, in this paper we discuss AIED design directions for can engage parents in learning literacy while supporting their chil- supporting parents’ literacy, but parents are aspirational beings, like dren. As others have identi�ed for family learning coaches [25] all people, with desires, dreams, and goals for other skills beyond and family-school liaisons in bilingual communities [52], these literacy. Future AIED research may explore how to support parents mediators play crucial roles in framing the learning experience, learning other skills, such as fundamental math skills [12], learning motivating learners, and suggesting topics or methods to learn. sign language to communicate with their children with hearing Prior work on pedagogical agents suggests that such agents can impairments [51], or learning about parenting more broadly [33], play motivational roles [18] and may be able to provide learning among many other skills. recommendations over time and across contexts [46]. Analogously, AIED researchers have the opportunity to develop new theo- a virtual family literacy coach may use the parent-child learning ries, methods, and systems to leverage parents’ interactions with models described earlier, perhaps coupled with context-awareness their children’s learning as a fertile site for mutually bene�cial to identify when, where, or how to sca�old learning. co-learning to take place. This paper lays out a set of design con- For instance, joint media engagement between parents and chil- siderations and design directions for AIED researchers to draw on dren has been shown to be critical to fostering children’s literacy for designing such systems. We hope that future AIED research development - speci�cally for the shared experience of reading views parents as agents of their own learning, with unique motiva- together [31, 45]. However, low-literate parents may lack su�cient tions, resources, and contexts for learning, and can contribute such mastery to read to their children independently. Intelligent story- systems to support lifelong learning for parents in ways that are books may thus adapt the reading level of the text to the parents’ appropriate and bene�cial for them. Design Opportunities for AIED to Support Parents Learning Literacy SLLL’19, June 2019, Chicago 2019 ACKNOWLEDGMENTS [24] Michael Heilman and Noah A Smith. 2010. Good question! statistical ranking for question generation. In Human Language Technologies: The 2010 Annual The research reported here was supported by the Jacobs Foundation Conference of the North American Chapter of the Association for Computational Fellowship 2015117013, and the Institute of Education Sciences, U.S. Linguistics. Association for Computational Linguistics, 609–617. [25] Anneli HERSHMAN, Juliana NAZARE, Ivan SYSOEV, Lauren FRATAMICO, Department of Education, Grant No. R305B150008. Juanita BUITRAGO, Mina SOLTANGHEIS, Sneha MAKINI, Eric CHUa, and Deb ROYa. 2018. Family Learning Coach: Engaging Families in ChildrenâĂŹs Early Literacy Learning with Computer-Supported Tools. (2018). [26] Kathleen V Hoover-Dempsey, Joan MT Walker, Howard M Sandler, Darlene REFERENCES Whetsel, Christa L Green, Andrew S Wilkins, and Kristen Closson. 2005. Why do [1] Bhavna Agrawal, Rong Liu, Ravi Kokku, Yi-Min Chee, Ashish Jagmohan, Satya parents become involved? Research �ndings and implications. The elementary Nitta, Michael Tan, and Sherry Sin. 2017. 4C: Continuous Cognitive Career school journal 106, 2 (2005), 105–130. Companions. In International Conference on Arti�cial Intelligence in Education. [27] Mamoru Ishikawa and Daniel Ryan. 2002. Schooling, basic skills and economic Springer, 623–629. outcomes. Economics of education review 21, 3 (2002), 231–243. [2] Sandra Aluisio, Lucia Specia, Caroline Gasperin, and Carolina Scarton. 2010. [28] Malcolm S Knowles et al. [n. d.]. The modern practice of adult education. ([n. Readability assessment for text simpli�cation. In Proceedings of the NAACL HLT d.]). 2010 Fifth Workshop on Innovative Use of NLP for Building Educational Applications. [29] Anuj Kumar, Pooja Reddy, Anuj Tewari, Rajat Agrawal, and Matthew Kam. 2012. Association for Computational Linguistics, 1–9. Improving literacy in developing countries using speech recognition-supported [3] Rahul Banerjee, Leanne Liu, Kiley Sobel, Caroline Pitt, Kung Jin Lee, Meng Wang, games on mobile devices. In Proceedings of the SIGCHI Conference on Human Sijin Chen, Lydia Davison, Jason C Yip, Andrew J Ko, et al. 2018. Empowering Factors in Computing Systems. ACM, 1149–1158. Families Facing English Literacy Challenges to Jointly Engage in Computer [30] Clarena Larrotta and Ysabel Ramirez. 2009. Literacy bene�ts for Latina/o parents Programming. In Proceedings of the 2018 CHI Conference on Human Factors in engaged in a Spanish literacy project. Journal of Adolescent & Adult Literacy 52, Computing Systems. ACM, 622. 7 (2009), 621–630. [4] Brigid Barron, Caitlin Kennedy Martin, Lori Takeuchi, and Rachel Fithian. 2009. [31] Jo-Anne LeFevre and Monique Senechal. 1999. The Relations among Home- Parents as learning partners in the development of technological �uency. (2009). Literacy Factors, Language and Early-Literacy Skills, and Reading Acquisition. [5] Hal Beder. [n. d.]. The Outcomes and Impact of Adult Literacy Education in (1999). the United States. PUB DATE 1998-05-00 NOTE 335p. PUB TYPE Collected Works [32] Michael A Madaio, Rebecca E Grinter, and Ellen W Zegura. 2016. Experiences Proceedings (021) Reports Research ([n. d.]), 25. with MOOCs in a West-African technology hub. In Proceedings of the Eighth [6] Gautam Biswas, Krittaya Leelawong, Daniel Schwartz, Nancy Vye, and The International Conference on Information and Communication Technologies and Teachable Agents Group at Vanderbilt. 2005. Learning by teaching: A new agent Development. ACM, 49. paradigm for educational software. Applied Arti�cial Intelligence 19, 3-4 (2005), [33] Catherine Marienau and Joy Segal. 2006. Parents as developing adult learners. 363–392. Child welfare 85, 5 (2006), 767. [7] Dale C Brandenburg and Andrea D Ellinger. 2003. The future: Just-in-time learn- [34] Elijah May�eld, Michael A Madaio, Shrimai Prabhumoye, David Gerritsen, Brit- ing expectations and potential implications for human resource development. tany McLaughlin, Ezekiel Dixon-Román, and Alan Black. 2019. Equity Beyond Advances in developing human resources 5, 3 (2003), 308–320. Bias in Language Technologies for Education. In 14th Workshop on Innovative [8] Susan Bull and Judy Kay. 2010. Open learner models. In Advances in intelligent Use of NLP for Building Educational Applications, at ACL. tutoring systems. Springer, 301–322. [35] Gordon McCalla. 2000. The fragmentation of culture, learning, teaching and [9] Trevor H Cairney and Lynne Munsie. 1995. Parent participation in literacy technology: implications for the arti�cial intelligence in education research learning. The Reading Teacher 48, 5 (1995), 392–403. agenda in 2010. International Journal of Arti�cial Intelligence in Education 11, 2 [10] B Cheung, L Hui, J Zhang, and Siu-Ming Yiu. 2003. SmartTutor: An intelligent (2000), 177–196. tutoring system in web-based adult education. Journal of Systems and Software [36] Jack Mostow, Greg Aist, Paul Burkhead, Albert Corbett, Andrew Cuneo, Susan 68, 1 (2003), 11–25. Eitelman, Cathy Huang, Brian Junker, Mary Beth Sklar, and Brian Tobin. 2003. [11] Gayle Christensen, Andrew Steinmetz, Brandon Alcorn, Amy Bennett, Deirdre Evaluation of an automated Reading Tutor that listens: Comparison to human Woods, and Ezekiel Emanuel. 2013. The MOOC phenomenon: who takes massive tutoring and classroom instruction. Journal of Educational Computing Research open online courses and why? Available at SSRN 2350964 (2013). 29, 1 (2003), 61–117. [12] Marta Civil. 2001. Parents as Learners and Teachers of Mathematics: Toward a [37] Michelle M Neumann, Michelle Hood, and Ruth M Ford. 2013. Using environ- Two-Way Dialogue. (2001). mental print to enhance emergent literacy and print motivation. Reading and [13] Hope Clark, Parminder K Jassal, Michelle Van Noy, and Pamela L Paek. 2018. Writing 26, 5 (2013), 771–793. A New Work-and-Learn Framework. In Digital Workplace Learning. Springer, [38] E Kobena Osam, Matt Bergman, and Denise M Cumberland. 2017. An integrative 23–41. literature review on the barriers impacting adult learnersâĂŹ return to college. [14] Jenny Cook-Gumperz. 2006. The social construction of literacy. Vol. 25. Cambridge Adult Learning 28, 2 (2017), 54–60. University Press. [39] Ji-Hye Park and Hee Jun Choi. 2009. Factors in�uencing adult learners’ decision [15] Edward Cutrell, Jacki O’Neill, Srinath Bala, B Nitish, Andrew Cross, Nakull to drop out or persist in online learning. Journal of Educational Technology & Gupta, Viraj Kumar, and William Thies. 2015. Blended learning in Indian colleges Society 12, 4 (2009), 207–217. with massively empowered classroom. In Proceedings of the Second (2015) ACM [40] Judy M Parr and Michael AR Townsend. 2002. Environments, processes, and Conference on Learning@ Scale. ACM, 47–56. mechanisms in peer learning. International journal of educational research 37, 5 [16] Mohammed R Dahman and Hasan Dağ. 2019. Machine learning model to pre- (2002), 403–423. dict an adult learnerâĂŹs decision to continue ESOL course. Education and [41] Bobby D Rampey, Robert Finnegan, Madeline Goodman, Leyla Mohadjer, Tom Information Technologies (2019), 1–24. Krenzke, Jacquie Hogan, and Stephen Provasnik. 2016. Skills of US Unemployed, [17] Betsy DiSalvo, Parisa Khanipour Roshan, and Briana Morrison. 2016. Information Young, and Older Adults in Sharper Focus: Results from the Program for the seeking practices of parents: Exploring skills, face threats and social networks. In International Assessment of Adult Competencies (PIAAC) 2012/2014. First Look. Proceedings of the 2016 CHI Conference on Human Factors in Computing Systems. NCES 2016-039. National Center for Education Statistics (2016). ACM, 623–634. [42] Ricarose Roque, Karina Lin, and Richard Liuzzi. 2016. âĂIJIâĂŹm Not Just a [18] Melissa C Du�y and Roger Azevedo. 2015. Motivation matters: Interactions MomâĂİ: Parents Developing Multiple Roles in Creative Computing. Singapore: between achievement goals and agent sca�olding for self-regulated learning International Society of the Learning Sciences. within an intelligent tutoring system. Computers in Human Behavior 52 (2015), [43] Rod D Roscoe and Michelene TH Chi. 2007. Understanding tutor learning: 338–348. Knowledge-building and knowledge-telling in peer tutorsâĂŹ explanations and [19] John W Fantuzzo, Judith A King, and Lauren R Heller. 1992. E�ects of reciprocal questions. Review of educational research 77, 4 (2007), 534–574. peer tutoring on mathematics and school adjustment: A component analysis. [44] Amartya Sen. 2001. Development as Freedom. Oxford Paperbacks. Journal of educational psychology 84, 3 (1992), 331. [45] Monique Sénéchal. 2015. Young Children’s Home Literacy. The Oxford handbook [20] Paulo Freire. 1970. Pedagogy of the oppressed. Bloomsbury publishing USA. of reading (2015), 397. [21] VL Gadsden. 2000. Intergenerational literacy within families. Handbook of reading [46] William R Swartout, Benjamin D Nye, Arno Hartholt, Adam Reilly, Arthur C research 3 (2000), 871–888. Graesser, Kurt VanLehn, Jon Wetzel, Matt Liewer, Fabrizio Morbini, Brent Morgan, [22] Matteo Gerosa, Diego Giuliani, Shrikanth Narayanan, and Alexandros Potami- et al. 2016. Designing a personal assistant for life-long learning (PAL3). In The anos. 2009. A review of ASR technologies for children’s speech. In Proceedings of Twenty-Ninth International Flairs Conference. the 2nd Workshop on Child, Computer and Interaction. ACM, 7. [47] Michael Thomas, Hayo Reinders, and Mark Warschauer. 2012. Contemporary [23] Norma González, Luis C Moll, and Cathy Amanti. 2006. Funds of knowledge: computer-assisted language learning. A&C Black. Theorizing practices in households, communities, and classrooms. Routledge. SLLL’19, June 2019, Chicago 2019 Madaio and Ogan [48] Sowmya Vajjala and Detmar Meurers. 2012. On improving the accuracy of [52] Marisol Wong-Villacres, Neha Kumar, and Betsy DiSalvo. 2019. The Parenting readability classi�cation using insights from second language acquisition. In Actor-Network of Latino Immigrants in the United States. In Proceedings of the Proceedings of the seventh workshop on building educational applications using 2019 CHI Conference on Human Factors in Computing Systems. ACM, 684. NLP. Association for Computational Linguistics, 163–173. [53] Diyi Yang, Tanmay Sinha, David Adamson, and Carolyn Penstein Rosé. 2013. [49] Susan Walker. 2017. Creating Parentopia: Design-Based Research to Develop Turn on, tune in, drop out: Anticipating student dropouts in massive open online an Interface for Parent Learning Communities and Networks. Philadelphia, PA: courses. In Proceedings of the 2013 NIPS Data-driven education workshop, Vol. 11. International Society of the Learning Sciences. 14. [50] Susan K Walker and Jessie H Rudi. 2014. Parenting across the social ecology [54] Jason C Yip, Carmen Gonzalez, and Vikki Katz. 2016. The learning experiences facilitated by information and communications technology: Implications for of youth online information brokers. Singapore: International Society of the research and educational design. Journal of Human Sciences and Extension 2, 2 Learning Sciences. (2014), 15–32. [55] Benjamin N York, Susanna Loeb, and Christopher Doss. 2018. One step at a [51] Kimberly A Weaver and Thad Starner. 2011. We need to communicate!: helping time: The e�ects of an early literacy text messaging program for parents of hearing parents of deaf children learn American Sign Language. In The proceed- preschoolers. Journal of Human Resources (2018), 0517–8756R. ings of the 13th international ACM SIGACCESS Conference on Computers and [56] Tara J Yosso*. 2005. Whose culture has capital? A critical race theory discussion Accessibility. ACM, 91–98. of community cultural wealth. Race ethnicity and education 8, 1 (2005), 69–91.