If you’ve ever noticed your clients nail flashcard naming tasks and then watched nothing budge in real-life conversation, you’re in good company. Many of us were taught to assume that once a skill shows up on our therapy cards, it will “naturally” show up in life. Behavioral science has a name for that habit. It’s called “train and hope,” and it rarely gets us where our clients need to go. Generalization is not automatic. It is something we plan for at the outset, embed in treatment, and measure on purpose.

This post is a practical guide for SLPs who want their therapy to carryover. I’ll define what we mean by generalization, highlight five evidence-informed principles you can build into sessions, and teach you how to troubleshoot and support families with generalization.

What exactly are we trying to generalize?

In aphasia treatment, two big types of generalization help us think clearly about goals and outcomes:

Stimulus (or vertical) generalization: a behavior learned under one set of conditions appears under different conditions, like trained words showing up with a novel communication partner or in a new setting.
Response (or horizontal) generalization: training one behavior improves a related behavior, like gains on treated items extending to untreated items or to a related task.

When I write a plan, I ask a simple question out loud: generalization of what behavior to what behavior? Then I choose measures that match the person, the target, and the context. For example, if I’m treating spoken verbs, my acquisition goal might be ‘produce trained verbs,’ and my generalization aim could be ‘use trained verbs in conversation.’

That clarity keeps us honest. It also helps caregivers understand why we’re doing what we’re doing.

Five principles you can build into therapy

A recent AJSLP tutorial synthesized decades of mixed advice into five practical principles for maximizing carryover. Think of these as ingredients you can add to many approaches you already use.

1. Include instructional strategies that offer cognitive support

Aphasia rarely travels alone. Executive functions, awareness, and strategy use influence both treatment response and carryover. When we integrate metacognitive scaffolds, explicit strategy instruction, and error-reducing practice where appropriate, we help skills show up in the messy, unpredictable contexts of daily life.

Why this matters: Generalization requires the brain to recognize when a learned skill is relevant in a new situation and execute it without the scaffolding present during training. People with executive function challenges struggle with this. By explicitly teaching when and how to use strategies, we build the cognitive bridge between clinic performance and real-world use.

Use it tomorrow: Add a two-minute strategy check-in at the end of your session. Ask your client to name one situation this week where they might use what they practiced today. Check in during the next session to see how they did.

2. Use client-led, flexible training

Looser, person-selected training that varies stimuli, partners, settings, and supports prepares skills to travel. This aligns naturally with participation-focused work, where success is measured by life enhancement and engagement, not just test points.

Why this matters: The brain doesn't magically transfer learning from one domain to a completely unrelated domain. Generalization follows the architecture of the language system. When we activate networks through structured practice, that activation spreads to related concepts that share features with the trained items.

Use it tomorrow: Let your client choose three items from your prepared set instead of drilling your predetermined list. Vary who's in the room. Change locations mid-session if you can.

3. Treat more complex items or forms when appropriate

Training more complex structures can sometimes promote broader change, especially when sentence-level work helps speakers move beyond isolated words. Clinical judgment matters here. Complexity-heavy sentence protocols may be a poor fit for individuals with severe agrammatism or significant comprehension impairments. Match complexity to the person and the goal.

Why this matters: If you can produce a complex structure, you can also produce the simpler structures embedded within it. However, the reverse is not true. Treating simple structures doesn't automatically improve complex ones.

Use it tomorrow: If your client is working on verbs, try shifting from isolated verb naming to verbs in simple sentences. See what happens.

4. Train the underlying mechanism

Pick an approach because its active ingredients and mechanism of action match your target and your client, then state those ingredients explicitly. When we're clear about what mechanism we're targeting, we can predict what kind of generalization to expect and know what to tweak if carryover stalls.

Why this matters: Understanding the mechanism tells you what kind of generalization to expect. If you're targeting phonology, expect generalization to phonologically similar items. If you're targeting semantics, expect generalization to semantically related items.

Use it tomorrow: Write down one sentence about what mechanism you're targeting in your next session. "I'm strengthening semantic networks" or "I'm rebuilding phonological encoding." Say it out loud to your client's family member.

5. Go beyond the single-word level

If your goal is better daily conversation, plan to work at the sentence and discourse levels. The evidence base keeps growing: treating verbs and sentences is linked to stronger generalization than treating isolated words, and discourse-level work shows positive effects on word and sentence production.

When researchers looked across studies at verb and sentence treatments, they found that most people improved on treated sentences, more than half generalized to untreated sentences, and about 70% showed generalization to discourse tasks like picture description, personal narratives, and storytelling.

Why this matters: Conversation happens at the sentence and discourse level. When we train at the level where the skill will actually be used, we're building in the contextual cues and processing demands that will be present in the generalization context. Word-level training creates word-level skills. Sentence-level training creates sentence-level skills already closer to the demands of real communication.

Use it tomorrow: Take one word-level task on your schedule and bump it to sentence level. Instead of naming "coffee," have your client tell you one thing they do with coffee in the morning.

Why generalization happens (and why it sometimes doesn't)

Here's what decades of work in language science tell us: when generalization works, it follows the brain's wiring. When it fails, it usually means one of three things, the target network is too impaired, the task lacks variability, or cognitive supports are missing.

Language systems are networked, not isolated.

When you strengthen one part of a network, connected parts benefit too. This is why semantic training generalizes to semantically related words, why phonological training generalizes to phonologically similar words, and why sentence structure training generalizes to sentences with similar structures. The brain stores patterns, relationships, and rules. When we train those patterns, the benefits spread.

Generalization follows the architecture of what's spared.

Some people generalize more readily than others, and this reflects the integrity of their residual language networks. People with milder aphasia show stronger generalization than those with more severe impairments (Quique et al., 2019). This isn't about motivation. It's about having enough intact substrate to support the spread of learning from trained to untrained items. When someone isn't generalizing despite good treatment planning, it often means the networks you're trying to engage are too damaged to support that spread. For these individuals, focus on acquisition of highly functional, frequently needed items rather than expecting broad generalization.

Cognitive resources and variability matter.

Generalization requires recognizing when a learned skill is relevant and executing it without the supports present during training. When you practice a skill in only one context with one set of materials and one communication partner, the brain learns a context-specific skill. When you practice with variability, different exemplars, different partners, and different settings, the brain learns a more abstract, transferable skill.

When generalization stalls: troubleshooting steps

If carryover stalls, run through this checklist:

Are your targets and mechanisms aligned?

Does your target connect to the desired behavior, and are you engaging the right mechanism? If you’re training semantics but the breakdown is phonological, generalization may never show up.

Have you given it enough time?

Sometimes generalization just needs more dosage. You may need to continue treatment past the point where treated items have plateaued.

Have you varied people, places, and materials?

If you've been using the same stimuli, same setting, same communication partner, you may have created very specific learning that doesn't transfer. Introduce variability.

Does your client need more cognitive scaffolding?

If your client has executive function challenges, add metacognitive components. Teach them to recognize when a strategy is needed, select an appropriate strategy, and monitor whether it worked.

Are other factors interfering?

Significant cognitive impairments, depression, or fatigue may limit generalization capacity. Sometimes the issue isn't your treatment approach. It is possible that other factors are limiting the system's ability to reorganize.

Talking to families about generalization

Families need to understand what we're doing and why so they can support generalization at home.

Set realistic expectations.

"Research shows that when we work on sentence production in therapy, most people show improvements in their everyday storytelling and conversation. That's our goal. Not just better sentences in here, but better communication out there."

Explain the timeline.

"Generalization often takes longer than the direct learning. You might see him get really good at the words we practice first, and then over the next few weeks, you'll start noticing he's finding other words more easily too."

Address the "why not just practice what he needs" question.

"We do work on high-priority communication situations, but if we only practice specific phrases, people often can't use them when the situation changes even a little bit. When we work on strengthening the underlying system, the improvements spread more broadly and give him more flexibility."

Give them concrete things to notice.

"Here's what to watch for: Is he trying more often to communicate? Is he using gestures or other strategies when he can't find a word? Is he staying in conversations longer? Those are all signs his communication is improving."

Teach them to support it.

"Give extra time when he's trying to say something. Acknowledge the attempt, not just whether the words are right. Ask open questions. Create opportunities for communication about things he cares about."

Why this matters

Our clients are experts on their own lives. They tell us their goals are about activities, relationships, and participation. When we plan generalization from day one, we honor that. When we measure discourse and confidence, not just confrontation naming, we catch meaningful changes that standardized scores can miss. That is good science and good care.

Takeaways you can use this week

Ask and answer: generalization of what behavior to what behavior? Then measure that exact thing.
Integrate the five principles during planning, not as a last step.
Work beyond single words when conversation is the target, and consider verb-and sentence-level approaches with strong generalization signals.
Specify mechanisms and ingredients, so partners can reinforce them and you can adjust with purpose.
Balance probes with discourse or functional measures, and standardize how you capture them when possible.
Co-set goals and use strategies like motivational interviewing to align therapy with what the person values.

ABOUT THE AUTHOR

References:

Mayer, J. F., Madden, E. B., Mozeiko, J., Murray, L. L., Patterson, J. P., Purdy, M., Sandberg, C. W., & Wallace, S. E. (2024). Generalization in aphasia treatment: A tutorial for speech-language pathologists. American Journal of Speech-Language Pathology, 33(1), 57-73.

Quique, Y. M., Evans, W. S., & Dickey, M. W. (2019). Acquisition and generalization responses in aphasia naming treatment: A meta-analysis of semantic feature analysis outcomes. American Journal of Speech-Language Pathology, 28(1S), 230-246.

Want to see more? Follow us on Instagram for weekly updates on all things MedSLP!