SUPER & Micro-Poetry: Towards An AI-Based Tool For Helping People With Cognitive Impairments To Understand And Enjoy Literature
Daily tasks such as booking a flight on-line present difficulties for people with cognitive disabilities. Such difficulty becomes a challenge when it comes to interacting with a creative text, such as poetry. In such a context, there is a need to enhance the access to creative literature by people with cognitive disabilities. An Artificial Intelligence (AI)-based tool can help to automatically adapt poetic texts so they can be more easily understood. Besides the benefits for people with cognitive disabilities, this tool could also become a useful educational tool in a foreign language classroom, where students with a limited knowledge of a target language could rely on it to adapt literary works.
Last summer, Dr. Paula Pérez-Sobrino and I started “In code blood: Enhancing the accessibility of micro-literature through a semi-automatic computer tool”, a project designed to help people with cognitive disabilities in their leisure time, and more specifically to aid them in enjoying literary texts. In this project, funded by Creative Multilingualism, we explored and mitigated the difficulties in the comprehension of poetic texts. In particular, we were interested in the so-called micro-poetry, a genre of poetic verse which is characterized by its extreme brevity and is often found on-line, mostly on Twitter.
Paula Pérez Sobrino was due to present on this work at the New Perspectives on Modern Languages Research conference in March 2020, which was postponed indefinitely due to the COVID-19 pandemic. Paula and Mari Carmen have recorded a presentation which you can view below and read in conjunction with this blog post.
Context
People with cognitive disabilities have some problems related to reading comprehension, communication, and their ability to respond to routine situations as well as to challenging scenarios. These situations are not just limited to everyday interactions such as obtaining a train ticket from a machine, but also to activities related to process the content and meaning while reading literary works such as poems and novels or while watching films. These last activities enhance the creative thinking skills of people.
The Easy-to-Read (E2R) methodology was formulated to assist people with cognitive abilities overcome the aforementioned barriers and to improve their daily life ([1], [2]). The E2R methodology provides guidelines for producing/consuming materials, content, language, illustrations, as well as graphic layout in documents. In adopting this methodology, content creators can develop material that is clear and easy to understand for different sectors of the population, but especially those with disabilities and those with limited language or reading proficiency.
Puzzled Elements In Micropoetry For People With Cognitive Disabilities
We wanted to understand which of the elements that commonly appears in micro-poetry posed the greater difficulties to readers with cognitive disabilities.
We began by analyzing collections of micro-poems published in English and Spanish on social media—on Twitter in particular, material posted or re-posted by the Micropoetry Society account (@pssms) or by the account moderated by Hernán Salcedo (@MicroPoesia) as well as posts tagged with hashtags such as #micropoetry, #microcuento, #micropoesia, #escribirparaincluir, #microverso, and #micropoema— and the Poetry4Kids website.
Micro-poems tend to use metaphorical constructions. And, while figurative language, such as metaphors, metonymies, simile, irony, and idioms, is widely used in daily conversation as well as in literature, there is evidence suggesting that people with cognitive disabilities encounter difficulties comprehending different kinds of metaphors.
We also discovered that in many cases micro-poems include:
- special characters (such as “@” for gender issues),
- abbreviations (like “tq” or “xq” in Spanish and “u”, “di”, or “lovi” in English),
- difficult words (for example, “serendipa” in Spanish or “unequivocall” in English), and
- contractions, (such as “I’m” or “don’t” in English).
These linguistic elements are not recommended according to the E2R Methodology ([1], [2]), which suggests trying to avoid using (a) special characters (\, &, <, §, #, @) where possible, (b) abbreviations (like “e.g.” or “etc”), (c) contractions (doesn’t, wasn’t, couldn’t), and (d) complex words.
Enhancing How People With Learning Disabilities Comprehend Micropoems
Technology can help people with cognitive disabilities to better understand and enjoy literature. To this end, Anabel Martín Peñalver, a graduate student at UPM working under the supervision and coordination of Dr. Mari Carmen Suárez-Figueroa, has developed a proof of concept called Suggesting micro-Poems in E2R (SUPER). SUPER draws on several Artificial Intelligent (AI) techniques, such as semantic representation of knowledge (based on ontologies), pattern matching, rules, and natural language processing.
SUPER suggests an adaptation of the original micro-poem, written in English or Spanish, without interfering, as much as possible, with the initial message. To do so, SUPER analyzes a micro-poem according to a subset of guidelines developed from the E2R Methodology (see Figure 1). The analysis determines whether the text adheres to the E2R recommendations, discussed above.
If the poem contains linguistic elements E2R methodology recommends avoiding, SUPER modifies the text by substituting those elements with more appropriate linguistic units that follow the E2R guidelines, resulting in an E2R-adapted micro-poem (see Figure 2). SUPER also calculates the readability index for both the original micro-poems and the adapted ones.
Conclusions
This project has allowed us to explore ways of assisting people with cognitive impairments to understand and enjoy micro-poems, and thus improve their leisure time. However, we are conscious that more work is still needed. Current and future research will focus on
- identifying conventional and explicit metaphors in micro-poems in an automatic fashion and
- performing experiments that involve people with learning disabilities reading and analyzing the different results obtained via SUPER.
Finally, while we anticipate SUPER having a great impact on the number of the adapted poetic texts for people with cognitive disabilities, we also recognize that it could prove useful for second language learners, who are gaining proficiency in their second language and would benefit from reading simplified adaptations of short poems.
Read more about the 'In Code Blood' project here>>
About the authors
Dr. Mari Carmen Suárez-Figueroa (@MCarmenSuarezF) is a lecturer at Universidad Politécnica de Madrid (UPM)and a researcher at Ontology Engineering Group (OEG).
Dr. Paula Pérez-Sobrino joined the Universidad de La Rioja as a lecturer in September 2019. Previously, she was a lecturer at Universidad Politécnica de Madrid and a researcher at OEG.
References
[1] Inclusion Europe: Information for All. European standards for making information easy to read and understand (2009).
[2] Norma UNE 153101:2018 EX Lectura Fácil. Pautas y recomendaciones para la elaboración de documentos (2018).