Speakable is an online platform utilizing advanced speech recognition technology and large language models. It provides interactive speaking activities with instant automatic grading and feedback for students, helping them build verbal skills and confidence.
Speakable's latest capability allows generating predicted WIDA proficiency levels from student recordings using AI matching against official scoring rubrics. Here's how it works:
Teachers or administrators first set up a speak-and-respond style assignment within Speakable using a prompt reminiscent of ACCESS speaking tasks. Students are asked to provide verbal answers 1-3 minutes in length, allowing sufficient spoken content for analysis.
Prompt design best practices include:
These open-ended directions elicit authentic demonstrations of live academic language rather than scripts. Students construct original responses on the spot using vocabulary, proper tense formation, idioms and pronunciations.
For inspiration, you can find some example prompts in this article.
Teachers pressed for time can duplicate this existing Speakable activity containing WIDA-style prompts like:
We crafted those respond card questions after consulting the official sample items. Assign to students immediately with minimal tweaking required.
Over the following days or weeks, students log into Speakable and complete the respond card activity. They click to record 1-3 minute long spoken answers and submit those recordings for analysis.
Once a student submits their recording, Speakable leaps into action. Instead of an exhaustive manual review, the platform transcribes the recording and employs a sophisticated large language model (LLM) to estimate the student's WIDA level. This LLM is equipped with numerous data points, allowing for a nuanced assessment of the student's language proficiency each time a recording is submitted.
The process is streamlined yet detailed, analyzing aspects such as:
This method provides an estimation of the student's WIDA level, offering insights into their language proficiency. Unlike traditional methods, this AI-driven approach ensures a rapid, consistent, and unbiased analysis of each student's skills. By harnessing the power of LLM, Speakable brings a new level of precision and efficiency to language proficiency assessment, giving educators a valuable tool in preparing students for the high-stakes ACCESS evaluations.
With this technology, educators can identify areas of strength and improvement well in advance of the ACCESS test, ensuring that each learner is better prepared and more confident when the exam day arrives.
Gradebooks compile scores into a unified dashboard. Filters quickly spotlight students projected to fall short of targets or demonstrating encouraging development.
Lesson plans can be adapted based on real-time gaps revealed:
Often the act of systematic measurement itself ignites student motivation. Learners take pride exerting effort to reach new language milestones, celebrating wins big and small. Teachers reinforce growth, forging engagement through data-driven discourse.
Speakable aims to provide teachers timely indicators on how English learners are progressing. With AI as an ally, success scaffolding student speaking skills leading up to high-stakes ACCESS evaluations is now within reach.
Want to start predicting your students' WIDA levels before ACCESS testing? Speakable offers free professional accounts to trial respond cards plus all capabilities covered here.
Simply schedule a 30 minute call and we'll set you up with a pilot for your classroom. See firsthand how AI can provide clarity leading up to those critical exams. Our students are counting on us!
