Semantic feature analysis (SFA) is a structured language therapy technique designed to help people retrieve words they struggle to find. It works by prompting a person to describe the features of a target word, such as what category it belongs to, what it’s used for, and where you’d find it, which activates the brain’s word-finding networks and makes the target word easier to access. Originally developed for people with aphasia (language loss after stroke or brain injury), SFA is also used in educational settings to help students build vocabulary and reading comprehension.
How SFA Works in the Brain
Word retrieval isn’t a simple one-step process. Your brain stores words in networks of meaning: “apple” is connected to “fruit,” “red,” “eat,” “tree,” and dozens of other associations. When someone has aphasia, the path to a word gets blocked or weakened, even though the person still knows the concept. SFA works by activating the surrounding network. When you describe what something looks like, what group it belongs to, and what you do with it, those features send activation signals toward the target word from multiple directions at once. This convergence of signals makes it far more likely the word will surface.
The effect is bidirectional. Thinking about a verb like “cut” activates knowledge about who typically does the cutting, what gets cut, and where cutting happens. That broader situational knowledge then feeds back and strengthens retrieval of the original word. Over time, repeatedly exercising these connections can reinforce the pathways, making word retrieval more automatic.
The Standard SFA Protocol
In its classic form, developed by Mary Boyle and Carl Coelho in 1995, SFA follows a simple cycle. The clinician shows a picture and asks the person to try naming it. Whether or not they succeed, the person then works through a set of feature prompts, generating descriptions for each one. After completing the features, they attempt to name the picture again, and then repeat the correct name.
The original protocol uses five feature categories:
- Category: What group does it belong to?
- Use: What do you do with it?
- Association: What does it remind you of?
- Location: Where do you find it?
- Characteristics: What does it look, feel, sound, or taste like?
In practice, clinicians adapt freely. Studies have used anywhere from 1 to 11 feature labels, with a median of five. Some ask for a single response per category, while others encourage generating multiple features. One modified version reduced the prompts to just three features and allowed written responses alongside verbal ones, specifically to accommodate patients who also have difficulty with speech motor planning. That simplified approach still produced long-term naming improvements.
SFA in Educational Settings
Outside of speech therapy, SFA takes the form of a vocabulary-building chart. Teachers identify key terms from assigned readings and list them in one column. Across the top row, they write characteristics that might apply. Students then fill in the grid, marking a “+” when a characteristic applies to a term and a “-” when it doesn’t. This visual organizer helps students analyze relationships between concepts, distinguish similar terms from one another, and build deeper understanding of academic vocabulary rather than memorizing definitions in isolation.
How Well SFA Works for Naming
The evidence for SFA is positive but nuanced. For words that are directly practiced during therapy, naming accuracy improves significantly, and those gains typically hold for at least two months after treatment ends. The harder question is whether those improvements transfer to words that weren’t practiced.
Generalization to untrained items is less consistent. Some studies find modest improvement on new words, while others find that the treatment effects don’t carry over to untrained picture naming or to connected speech tasks like retelling a story. In one case study combining SFA with brain stimulation, a participant improved naming of trained items by 60 to 75% compared to baseline and showed about a 20% improvement on untrained items, with those gains lasting up to four weeks. But these results varied across treatment phases, and the untrained gains were smaller and less reliable.
Long-term maintenance is another consideration. While gains on trained items can persist for months, the research suggests a gradual decline. After three to six months, only 10 to 65% of trained words are still named correctly, depending on the individual and the condition being treated.
SFA Compared to Phonological Approaches
SFA focuses on meaning. An alternative approach called phonological components analysis (PCA) focuses on sound: the first sound of the word, what it rhymes with, how many syllables it has. A head-to-head comparison of the two, with eight people receiving 12 sessions split evenly between both therapies, found that PCA actually produced more consistent results. Seven of eight participants improved significantly on items treated with PCA, compared to four of eight for SFA. Six of the seven PCA responders maintained their gains at follow-up, versus three of four SFA responders.
Interestingly, the participants with primarily meaning-based deficits didn’t benefit from the meaning-based therapy (SFA), while the sound-based therapy helped most participants regardless of their underlying impairment. This doesn’t mean SFA is ineffective, but it does suggest that the best approach depends on the individual’s specific language profile, and that PCA may be a more broadly useful starting point for some clinicians.
Who Benefits from SFA
SFA was originally developed for people with aphasia after stroke, and that remains its best-studied use. But recent research has extended it to people with primary progressive aphasia (PPA), a neurodegenerative condition where language abilities gradually decline rather than being suddenly lost. In both the logopenic variant (where word-finding pauses are the hallmark) and the semantic variant (where word meanings themselves erode), SFA improved naming abilities, broader language functioning, and quality of life scores. Follow-up scores did decline compared to immediately after treatment, but several measures remained higher than pre-treatment levels, suggesting meaningful lasting benefit even in a progressive condition.
The distinction matters because stroke-related aphasia is stable or improving, while PPA involves ongoing decline. The fact that SFA can still provide benefit in a degenerative context is notable, though it also means that maintenance therapy or repeated treatment cycles may be necessary.
Treatment Intensity and Scheduling
How often and how long someone does language therapy matters as much as the type. A large meta-analysis of aphasia treatment found that the biggest language gains came from 20 to 50 total hours of speech-language therapy. For scheduling, three to five sessions per week produced the strongest results across measures of overall language, functional communication, and comprehension. In terms of weekly hours, even modest amounts (two hours per week or less) produced substantial gains, with similar results at three to four hours per week.
The practical takeaway: frequent, consistent sessions matter more than marathon sessions. Four to five days per week, even in shorter blocks, tends to outperform less frequent but longer appointments. For SFA specifically, this means that daily or near-daily practice with the feature prompts, whether in a clinical session or at home with a caregiver, is likely to produce better outcomes than once-a-week therapy alone.