Can Speech-Recognition Apps Improve Struggling Students' Writing?

          In this article in Language Arts, Betsy Baker (University of Missouri/Columbia) says that as a second-grade teacher, she was a big fan of the language experience approach; she took dictation of students’ stories and this motivated them to express themselves and share their personal experiences and linguistic and cultural heritage. As speech-recognition apps became more accurate, Baker wondered if they might take the place of the teacher transcribing students’ writing, saving time and motivating students.

In this article, she describes serving as a teacher’s aide in charge of the writing center in a first-grade classroom of 22 students in a diverse Title I school. From January to May, Baker experimented with using speech-recognition apps (Siri and Dragon on students’ iPads) to develop students’ writing skills, spending the most time with students who were having difficulty. She launched the initiative using the following steps:

-   Baker recited a poem with which students were familiar, Mary Had a Little Lamb, and they chimed in.

-   Baker said the poem aloud as she wrote it on chart paper.

-   She modeled launching a speech-recognition app on an iPad and reading the poem aloud, with students looking over her shoulder to see if the app correctly transcribed her words (it did).

-   Baker invited each small group of students at the writing center to analyze the poem’s syncopation (they bobbed their heads) and the fact that snow rhymed with go.

-   She then asked students to write a similar poem about a different animal and owner, and worked with them to come up with a poem about Gary who had a little toad whose skin was green with slime, and everywhere that Gary went the toad would jump behind.

-   She spoke the new poem into the speech-recognition app, and students watched as the words magically appeared on the iPad; she then copied the new poem on chart paper.

-   Baker then invited students to orally brainstorm their own versions of Mary Had a Little Lamb, and they wrote stories about Victoria and her little dog, Jada and a small turtle, Sponge Bob and a monster bike, etc. Students worked with pencil and paper, and then spoke their poems using the speech-recognition app.

-   In the months that followed, students composed variations of Five Little Monkeys; Row, Row, Row Your Boat; Brown Bear, Brown Bear, What Did You See?; It Looked Like Spilt Milk; and narratives in which students chose their own topics, settings, and characters.

Baker found that this approach was highly motivating for students, and they used words like limousine, dandelion, and unicorn. Interestingly, on days when the school’s Wi-Fi was down and the speech-recognition app wasn’t accessible, students reverted to less-ambitious words like car, flower, and horse. “Speech recognition apps,” says Baker, “appeared to allow students to use their rich personal oral lexicons and cultural funds of knowledge to express themselves.” An end-of-year inventory revealed that students had greatly increased their sight vocabularies, and half of the words they knew were not in standard first-grade vocabulary lists.

            In the course of this project, says Baker, students had to deal with inaccurate transcriptions by the app. There were also times when the app would come up with inappropriate words – for example, hell instead of hill and worse (there wasn’t a child-safety feature). Baker worried at first that this would undermine the process, but found that errors actually got students reading the transcriptions of their writing more carefully, frequently engaging in word analysis and talking with classmates and their teachers about whether words made sense and were spelled correctly.

            Another glitch was that students would think up elaborate stories, launch the speech-recognition app, and forget the story – and by the time they’d recalled it, the app had timed out. The app was also not very clever at dealing with students’ false starts, repetitions, side comments, long pauses, and laughter. It turns out that planning and dictating a story is a specialized literacy skill that students needed to master – including rehearsing stories before activating the app, speaking slowly and clearly, and doing the editing on a hard copy. Students found it difficult to handle the app features allowing users to italicize, boldface, underline, and indent, and sometimes accidentally deleted entire compositions. It was difficult for students to access a separate rhyming dictionary, and the speech-recognition app didn’t accommodate students’ illustrations or make it easy to import stock illustrations from the Internet.

            On balance, though, Baker endorses the use of speech-recognition apps as one tool for young writers because of the way the app can stretch their written vocabularies and get them composing more creatively and fluently. With the proper teacher support and instruction, the limitations and glitches of the app can actually enhance students’ sense-making and critical reading skills.

“Talk to Read and Write: Using Speech-Recognition Apps in a First-Grade Writing Center” by Betsy Baker in Language Arts, July 2019 (Vol. 96, #6, pp. 358-369), https://bit.ly/2SOiSTC; Baker can be reached at bakere@missouri.edu.

(Please Note: The summary above is reprinted with permission from issue #796 of 
The Marshall Memo, an excellent resource for educators.)