Setting Up Working Memory Goals That Actually Move the Needle
Working memory goals in speech-language pathology sit in this weird gray area where they're obviously relevant to language but not always cleanly inside what most clinicians consider the core skill set. You see students who can't hold a sentence in mind long enough to repeat it back. You see kids who fall apart on multidirectional directions. You see adolescents whose reading comprehension tanks because they lose track of who did what to whom halfway through a paragraph. These are working memory problems, and they absolutely fall under your domain if you own them properly. The tricky part is that not everything labeled "working memory" is the same thing. There's phonological loop stuff—holding speech sounds for a couple seconds—then there's visuospatial sketchpad stuff, then there's the central executive component that manages attention and switching. Most SLPs are dealing with the phonological side, especially when it comes to auditory verbal working memory. That's the bottleneck for sentence repetition, following complex directions, and early reading decoding. If you're writing goals for that population, here's how to do it without turning them into vague fluff.
Common Pitfalls in Working Memory Speech Therapy Goals
Most goal templates I see floating around are either too generic or they're trying to measure something that doesn't map cleanly to real function. A typical bad example: "Student will follow two-step directions with 80% accuracy." That's not wrong per se but it tells you nothing about the conditions, the type of directions, or whether the student is just getting lucky on high-imagery content. Two-step direction is a spectrum. "Pick up the red block and put it on the blue mat" is fundamentally different from "Tell me the capital of France and name a fruit that starts with the same letter." One relies on auditory sequencing and retention. The other requires manipulation and retrieval under interference. Another common trap is setting goals that conflate working memory capacity with working memory training gains. Doing n-back tasks in isolation has been shown to improve performance on n-back tasks. It does not reliably transfer to reading comprehension, sentence repetition, or academic performance. I had a kid do digit span training for six months and his digit span went from 4 to 7 forward and 3 to 5 backward. He could not follow a three-step command any better than before. The training was real but it was locked to the trained modality. So if your goal is just to increase the raw number, you might be hitting a metric that doesn't predict functional improvement. The counterintuitive insight most people miss is that strategy instruction often outperforms pure capacity-building exercises for functional outcomes. Teaching a student to chunk, to rehearse subvocally, to use imagery, or to self-verbalize the steps as they process them tends to produce broader transfer than drilling raw span. The research base on this isn't huge but it's consistent enough that I would treat strategy-based interventions as first-line for most school-age and adolescent clients, with capacity exercises as adjunctive rather than primary.
Building Specific, Measurable Goals
Good goals in this area need three things spelled out: the stimulus type, the processing demand, and the success criterion that actually matters. Let me give you the structure I use and then some real examples. Condition: What is the student being asked to process? Nonword repetition? Sentence repetition? Multistep directions? Digit span forward versus backward? Each one taps different components. Behavior: What exactly counts as correct? Full repetition with no substitutions? Correct order for directions? Correct answer despite distractors?
Get the Full Details

Criterion: This is where most people go soft. 80% accuracy across 3 sessions is fine but only if the baseline is measured with the same stimulus type, the same materials, and the same administration conditions. If you baseline on nonword repetition and your goal criterion is measured on sentence repetition, you're not measuring progress on the same thing. Here is what a proper goal looks like for a middle school student with weak phonological working memory: Nonword repetition: Given a list of 2-syllable to 4-syllable nonwords presented auditorily at a rate of one per three seconds, Student will reproduce the phonological sequence correctly on 8 out of 10 trials across 3 consecutive therapy sessions, as measured by clinical recording.
Sentence repetition: Given semantically unpredictable sentences ranging from 6 to 10 words in length, Student will reproduce the exact word order and content words (omitting only function words) on 80% of trials across 4 sessions, with materials matched to the Woodcock-Johnson IV auditory processing subtests. Multistep directions: Given verbal directions containing 2 to 3 commands with embedded prepositional phrases, Student will execute the full sequence correctly on 75% of trials across 5 sessions, with novelty of objects controlled by rotating a standardized set of 12 items. Notice the specificity around presentation rate, length ranges, and material control. That last point about rotating items is critical. If you use the same physical objects every session, students learn the toy box rather than the listening skill. I learned that one the hard way with a seventh grader who was scoring 90 percent on three-step directions until I swapped the props mid-block and his accuracy dropped to 40 percent. He had memorized the location sequence, not processed the commands.
Materials and Admin Tricks That Matter
You don't need fancy software to run working memory assessments and interventions. What you need is control over variables that otherwise drift. Presentation rate, visual distraction, semantic predictability, and novelty of materials are the four axes that move most often without anyone noticing. For nonword repetition, I use a spreadsheet with pseudorandomized lists that control for phonotactic probability and syllable count. I generate lists in batches so each session uses a fresh set. The actual tool is basically an Excel sheet playing audio through a laptop at a fixed volume measured with a sound level app at the client's ear. Volume consistency matters more than people think. A difference of 5 decibels can shift performance by a full point or two on close trials. For sentence repetition, the Clinical Evaluation of Language Fundamentals fourth edition and the Test for Auditory Processing of Information have solid stimuli if you have access. If you're working without those, you can build your own bank and validate it by running a baseline group and checking that scores cluster reasonably. The goal is not perfect norm-referencing. The goal is that your materials behave consistently session to session.

One practical trick for digit span: don't read digits at a conversational pace. Read them at roughly one per second. Faster and you're giving them articulatory rehearsal time that inflates scores. Slower and you're giving them extra processing window that deflates them. Use a metronome app if you have to. I started doing this after realizing my baseline numbers were all over the place depending on whether I was having a rushed session or a slow one.
What Actually Works for Intervention
There are three evidence-backed approaches I rely on, listed in rough order of effect size for functional outcomes. Strategy instruction with gradual release: This is teaching the student to chunk, to use imagery, or to self-present the material. For sentence repetition, the most effective strategy I've seen is taught explicitly: have the student listen to the whole sentence first, then silently rehearse it once before repeating. Not aloud. Silently. Then repeat. Most kids who struggle with WM immediately start vocalizing as they hear, which floods the phonological store and creates interference. The silent rehearsal step is counterintuitive to them but it consistently improves accuracy within the first three sessions. Adaptive span training with transfer tasks: If you do span training at all, make it adaptive so it tracks just above current capability, and always include a transfer task in the same session. After the span drill, do sentence repetition or direction following using the same length range that was just trained. The transfer task is where the learning shows up. Without it, you're just making kids better at the drill.
Reduced cognitive load through environmental modification: This is the one everyone ignores because it doesn't look like therapy. Eliminating visual clutter during auditory WM tasks, reducing background noise, giving written supports when appropriate, and breaking complex instructions into smaller chunks before asking for independent recall. For a student with an IQ in the average range but a phonological WM deficit, these modifications often produce immediate performance gains that are larger than most targeted interventions. They're also not permanent. You fade them systematically as the student builds capacity. A real case from my practice: a 10-year-old with a history of recurrent otitis media, average nonverbal reasoning, and a clinical presentation of poor oral directions following and labile reading fluency. Baseline digit span forward was 4, backward was 3. Nonword repetition accuracy on 3-syllable items was 35 percent. We spent eight weeks on strategy instruction for sentence repetition—silent rehearsal, self-chunking, and error monitoring—while running adaptive digit span at the upper limit of his capacity three times per week. By week four, his sentence repetition accuracy on 7-word semantically unpredictable sentences had jumped from 20 percent to 65 percent. By week eight, he was at 82 percent and his digit span forward had stabilized at 6. Reading fluency on grade-level passages improved by roughly one correct word per minute, which is small but meaningful over a school year. The thing nobody tells you about working memory work is that progress is noisy. You will have sessions where a kid who was doing 70 percent drops to 40 percent because the teacher changed the room layout or he didn't sleep well or you used a new set of materials he hasn't seen. That doesn't mean the intervention failed. It means working memory is fragile and context-sensitive. Plan for that in your data collection. Average across at least three sessions before changing your approach, and track environmental variables alongside performance numbers.

There are also populations where working memory goals hit a hard ceiling. Students with significant intellectual disability, severe auditory processing disorder without compensatory cognitive strategy ability, or active neurological regression will not benefit from standard phonological WM interventions in any meaningful way. In those cases, the goal shifts to environmental accommodation and alternative communication modes. Saying that out loud matters because there is a lot of pressure in school settings to keep pushing remediation when accommodation is the more honest clinical move.
Documentation and Measurement Practicalities
Keep your baseline data tight. I recommend running two separate baseline probes for each goal area before writing the IEP. One with high-imagery content, one with low-imagery or semantically unpredictable content. The gap between them tells you whether the student's weakness is purely auditory-phonological or whether semantic processing is also a factor. That distinction changes your entire intervention plan. When you graph progress, plot individual session data points rather than rolling averages. Rolling averages smooth out the variability that is actually diagnostic. If you see a sudden drop after a material switch, that's information. If you see a gradual downward trend across sessions, that's also information. Both get lost in averaged scores. Report the stimulus properties alongside accuracy. Saying "80 percent on sentence repetition" is not enough for anyone reviewing your data to know what that actually means. Include the mean sentence length, the semantic predictability rating, and the presentation rate. Three lines of metadata turn a vague number into something defensible during an IEP review or a third-party audit.
I also stopped measuring working memory goals in isolation from language outcomes whenever possible. A student's sentence repetition accuracy and their direction-following accuracy should be reported in the same summary, not in separate binder sections. The reason is simple: they're the same underlying capacity expressed through different modalities. When you present them together, patterns emerge that justify continued SLP involvement versus referral to psychology for a more comprehensive cognitive evaluation. You want to be able to say whether the working memory deficit is narrow and phonological or broad and affecting multiple cognitive domains.
