Customize an Activity

Let's customize a speaking activity from beginning to end. We'll work through Basic Info, Objectives, and Scenario, using a Spanish restaurant activity. We'll pause at each group of settings so you can see what it changes.

Start with the title. Give the activity a name students will recognize. We'll call this one Lunch at a Restaurant.

The description explains the task before students start. Keep it practical: tell them what they'll do in this conversation. Here, they'll order lunch and ask questions about the menu.

Target language controls the language being practiced. Open the menu to choose a language, or use its search field. We'll keep Spanish for this activity.

Scroll to Session duration. The minimum is required for feedback. The maximum ends the conversation. Use the number fields or the slider. For this example, set the range to three to six minutes.

Continue to Objectives. These goals guide the conversation and help shape student feedback. Write each one as something observable. We'll make the first goal more specific: order a main dish and a drink politely.

Use Add objective when you need another goal. We'll add asking for a recommendation. The list recommends three to five focused objectives, so keep the activity manageable.

You can change the order with the arrow controls, or drag a row. We'll move our new objective up one place. The remove control deletes a goal you no longer need.

Check with AI can review the wording of your objectives. Treat its suggestions as a starting point and check that they reflect what you actually want students to practice.

Scroll to Target proficiency level. This menu uses the ACTFL scale, from Novice through Distinguished. Choose the level that fits your learners. We'll use Intermediate Low, keeping the task suitable for shorter, familiar exchanges.

Extra context is optional. Additional proficiency context lets you explain needs that a level alone doesn't capture. For example, these students know food vocabulary but need help asking follow-up questions.

Target vocabulary and phrases is a private language bank for the AI, hidden from students. Add a short list, separated by lines. We'll include phrases for ordering, requesting a recommendation, and asking for the bill.

Help and tips is for students to read before practicing; it isn't used by the AI. Add preparation or strategies. We'll ask students to prepare two menu questions and use a polite expression when ordering.

Continue to Scenario. The scene tells the AI where the conversation happens, who the learner is, and what task to support. Our learner is having lunch in Madrid and needs to order a meal.

The character defines the conversation partner. Give it a role, personality, and useful behavior. We'll make the AI a friendly server who asks one question at a time and waits for the student's answer.

Scroll to Voice and pace. Open Voice to see the available voices and their descriptions. We'll keep Verse. After closing the menu, Preview plays a short sample in the target language, helping you choose how the partner sounds.

Speaking pace is separate from the voice. Choose Very slow, Slower, Natural, or Slightly faster. Natural is the default. We'll choose Slower to give these learners more time to process what they hear.

Cultural context adds local norms, etiquette, or register. It's optional. We'll describe a casual Madrid restaurant and encourage courteous requests. Adding cultural context also makes cultural-appropriateness feedback available in Evaluation.

Next is Scaffolding during the conversation. These controls change how the AI helps in the moment. They don't change feedback or scoring. We'll look at each choice before selecting support for this activity.

If a learner uses another language, the AI can act as though it doesn't understand, understand while replying only in the target language, or briefly explain in the learner's language when asked. We'll choose that last option, allowing a short explanation before returning to Spanish.

For mistakes, Keep the conversation moving prioritizes meaning. Recast naturally models the correct form inside a reply. Correct briefly gives a short correction before continuing. We'll choose Recast naturally to offer a model without stopping the exchange.

Choosing recasts reveals How often to recast. Rarely, Sometimes, and Often control how frequently that support appears. We'll keep Sometimes, balancing a useful language model with time for students to keep talking.

If a learner gets stuck, Simplify and rephrase uses easier language. Give a guided hint offers a clue or guiding question. Offer a phrase to use gives the learner a useful expression. We'll choose a guided hint, leaving room for their own answer.

Formality sets the character's register: Casual, Neutral, or Formal. Match this to your situation and learning goals. We'll choose Neutral for a friendly but courteous restaurant exchange.

Response length offers Brief, Natural, and Detailed. All three keep conversational turns short; Detailed adds specificity within a one-to-two-sentence limit. We'll choose Brief so the learner has more room to speak.

Scroll to Let the AI start the conversation. Off means the student begins. Turn it on when you want the character to open with a short greeting. We'll let our server speak first.

You can also give the AI a specific opening line. Check the optional box and write the greeting in the target language. We'll welcome the learner to the restaurant. Without a specific line, the AI creates its own greeting.

Scenario image adds a visual students see before they begin. Generate the scene creates the setting without the character. Generate the character includes the conversation partner in that setting. Select the source you want, then use Generate image.

Upload your own lets you choose a JPEG, PNG, or WebP from your device instead. The image is optional, so you can also leave the activity without one. We'll keep this example focused on its conversation settings.

When you're satisfied, click Save changes and wait for confirmation. Your activity now has a clear task, focused objectives, and a conversation partner tailored to your learners. Feedback and rubric settings are covered in the Evaluation tutorial.