Learning a Language with AI: Don't Forget Pragmatics

Speaking a language isn't just about stringing together correct sentences. It's about knowing what to say, to whom, how and when. Linguists call this skill pragmatics, and it develops mostly through contact with other people.

AI platforms like sapere.ai are transforming language learning: personalized exercises, instant feedback, practice at any hour. But research in language teaching is clear on one point: to become truly comfortable in a language, learners need authentic interactions. That's why our hybrid approach, which combines AI and human conversation, delivers the best results.

What Is Pragmatic Competence?

The Office québécois de la langue française (OQLF) defines pragmatics in its Vitrine linguistique as the "part or aspect of semiotics concerned with the theoretical study of the psychological and sociological components involved in the use of linguistic signs." In short, it's the ability to use language in a way that fits the context. The Common European Framework of Reference for Languages (CEFR) places it alongside two other dimensions:

• Linguistic: vocabulary, grammar, pronunciation.

• Sociolinguistic: register, cultural references, social conventions.

• Pragmatic: performing speech acts in real situations (asking, declining, thanking, apologizing) and managing the flow of a conversation.

Here's an example. "Pass me the salt" and "Could you please pass me the salt?" are both grammatically correct. Choosing the right one for the person you're talking to is pragmatics. As French linguist and researcher Francine Cicurel points out, we speak first and foremost to get something done: to reassure, promise, protest or order a coffee.

What the Research Says

1. Pragmatics: The Poor Relation of Language Learning

In their 2020 review published in Le Langage et l'Homme, Fabienne Baider, Cristelle Cavalla and Georgeta Cislaru make three observations:

• It's taught less than grammar. According to researcher Muriel Warga, pragmatic competence has long received far less attention than grammar acquisition. Teaching focuses on plural agreement rather than on levels of formality.

• Its mistakes cost more. Warga also notes that native speakers tend to judge pragmatic errors more harshly than grammatical ones. An agreement error comes across as a learner's mistake. A request that's too blunt can come across as rude.

• It remains a hurdle even for advanced learners. Since the earliest studies (Blum-Kulka, 1982; Eisenstein and Bodman, 1986), researchers have observed that this competence can hold learners back right up to the B2 and C1 levels and might contribute to people getting stuck at the intermediate plateau. A bigger vocabulary doesn't help much if you don't know which contexts to use it in.

The authors also point out that standardized tests measure this competence poorly, because language use is always rooted in a specific social context. The research they review highlights study abroad, immersion, theatre, empathy and attention to emotions such as anxiety, all of them deeply human dimensions.

2. Pragmatics by Text Message? The WhatsApp Experiment

Brenda Youonang Fenko's master's thesis (UQAM, 2025) starts from a familiar observation in Quebec: many adult immigrants in French-language programs have little contact with French speakers outside the classroom. As a result, they have few opportunities to converse in French or to practise pragmatic competence.

How the study worked:

1. For three weeks, three upper-intermediate learners exchanged messages on WhatsApp with French-speaking conversation partners.

2. Each week, the class read the exchanges, discussed them and gave guided feedback with the teacher.

What the study observed:

• Learners diversified their speech acts (initiating, responding, clarifying, negotiating) to suit their needs.

• They adjusted their communication week after week. One learner asked for more clarification, while another moved from initiating conversations to responding more readily. One partner noticed that a learner became more open and engaged as the weeks went by.

• The context of the exchange, cultural differences and social relationships were the most decisive factors.

• Reflective feedback in class encouraged learners to think about politeness and speech acts.

The author's conclusion: messaging exchanges, when paired with thoughtful teaching support, can be a practical lever for developing pragmatic competence. As a bonus, the project built meaningful intergenerational bridges for both groups.

An important caveat: this is a qualitative study of three cases, and the author interprets it cautiously. It doesn't establish a general rule, but it clearly illustrates the role a real human conversation partner plays, even by text message.

What AI Does Very Well… and What It Misses

AI excels wherever learning calls for volume, repetition and personalization. That's what makes Sapere such a powerful learning tool, with fully tailored content. But pragmatics is built through the relationship with a real conversation partner, with their reactions, expectations and culture.

In other words, AI is an excellent training ground, but the real game is played in human conversation. That's where learners find out whether their message landed the way they intended.

The Hybrid Approach: The Best of Both Worlds

The UQAM study makes it clear: progress doesn't come from a digital tool alone, but from combining an accessible tool, real conversation partners and reflective guidance. That's exactly the Sapere model: an AI platform designed to be used between online classes with a Global Lingua teacher. At Sapere, the teacher remains "irreplaceable for human support, nuance and motivation." Learning happens in three stages:

Here's how to make the most of this cycle to work specifically on pragmatic competence:

1. Rehearse the situation through role-play. With a conversational agent, prepare a specific scenario: making a request to a landlord, politely declining an invitation, calling a clinic. Then ask for exercises generated from that conversation.

2. Target your professional vocabulary. The personalized vocabulary feature lets you practice the expressions specific to your field: medical, legal, technical, hospitality, finance.

3. Build on these exchanges in your online classes. Discuss tone, register and cultural codes with your teacher.

4. Try it out in real life. With a colleague, a neighbour or a conversation group, notice whether your message was received the way you intended.

5. Measure and repeat. Pragmatic competence is built through successive adjustments, week after week, through experience and your teacher's feedback.

This cycle turns every session on the platform into preparation for real life, and every conversation into a learning opportunity. That's why we recommend hybrid learning (the duo formula that combines AI + teacher) for anyone aiming for professional proficiency.

Learning a language means learning to connect with others. AI makes learning more accessible, flexible and personalized, while human conversation gives it meaning. At Sapere, we believe technology should open the door to human connection, not replace it.

References

• Baider, F., Cavalla, C., & Cislaru, G. (2020). Apprentissage des langues : compétence pragmatique, interculturalité. Le Langage et l'Homme, 55(1), 7–20.

• Youonang Fenko, B. (2025). Pratique de la compétence pragmatique en français chez les adultes issus de l'immigration par des interactions à l'écrit sur WhatsApp avec des partenaires de conversation aînés [Master's thesis in language didactics]. Université du Québec à Montréal.

• Council of Europe (2001, 2018). Common European Framework of Reference for Languages and its Companion Volume.