← Back
NeuroTrials.ai
Neurology Clinical Trial Database

ILAT

Communication makes intensive language therapy more efficient: a randomized controlled trial in chronic post-stroke aphasia

Share Share Copied! Compare

Year of Publication: 2026

Authors: Pulvermüller F, Jäger AT, Osterloh MR, ..., Mohr B

Journal: Brain Communications

Citation: Brain Commun 2026;8(4):fcag252

Link: https://doi.org/10.1093/braincomms/fcag252

Bottom Line

In 44 chronic post-stroke aphasia patients, 2 weeks of intensive Intensive Language Action Therapy (ILAT, 25 h) produced significantly greater gains on the primary language measure (mean AAT) and on depression scores (BDI) than the same amount of conventional utterance-centred speech-language therapy, arguing that the communicative function of language use during therapy — not just intensity — drives outcomes.

Major Points

  • Randomized comparison of two matched-intensity SLT methods (ILAT vs conventional naming/description) in chronic post-stroke aphasia
  • Primary language outcome (mAAT): significant Time × Therapy Type interaction F(1,42)=7.42, p=0.009, η²p=0.15; ILAT improved (mean Δ +1.54, 95% CI 0.73–2.35, p<0.001), CONV did not (p>0.17)
  • Production and naming subtests specifically favoured ILAT; AAT production interaction F(1,42)=6.19, p=0.02
  • Communication measures ANELT (p=0.004, d=1.9) and ACT (p=0.03) improved only in the ILAT arm
  • Primary depression outcome (BDI): significant interaction after outlier removal (F(1,38)=6.94, p=0.01); BDI fell by 8.67 points in ILAT (95% CI −12.79 to −4.54) but not in CONV; MADRS gave concordant results
  • Total intensity (25 h over 2 weeks) and content materials were carefully matched across arms, so any difference is attributed to the communicative context of language use
  • Small sample (n=44), single German-language centre, 10 post-randomization dropouts, and no long-term follow-up limit generalizability

Design

Study Type: Randomized Controlled Trial

Randomization: 1

Blinding: Semi-blinded (assessors and patients blind to therapy assignment; therapists delivering both therapies not blinded)

Enrollment Period: Not specified; single-centre recruitment, terminated when target n=42 reached (last group excluded due to COVID-19 lockdown, March 2020, suggesting recruitment spanned pre-March 2020)

Follow-up Duration: Immediate post-therapy assessment (within 2 working days after end of the 2-week intervention); no long-term follow-up in this trial

Centers: 1

Countries: Germany

Sample Size: 44

Analysis: Per-protocol (44 of 54 randomized; 10 post-randomization exclusions for newly identified exclusion criteria, TIA/second stroke, medication change, unconfirmed aphasia diagnosis, or COVID-19 termination). Repeated-measures ANOVA with factors Time (T1, T2) and Therapy Type (ILAT, CONV); planned within-group paired t-tests. ANCOVAs included age, time post-onset, and years of education as covariates.

Registration: Registered parallel-group RCT (per authors); ethics EA1/384/16, Charité Universitätsmedizin Berlin

Inclusion Criteria

  • Age ≥18 years
  • Native German speaker
  • Right-handed
  • Ischaemic or haemorrhagic stroke involving the left hemisphere
  • Diagnosed as aphasic based on the Token Test
  • At least 12 months post-stroke (chronic phase)
  • Sufficient attentional, perceptual and motor skills to partake in testing and card-game playing
  • Written informed consent

Exclusion Criteria

  • Severe visual or auditory perceptual deficits
  • Pronounced apraxia or agnosia
  • Second stroke or additional neurological disease
  • Withdrawal of informed consent

Baseline Characteristics

CharacteristicILAT (n=23)CONV (n=21)
Mean Age (y)62.4 (SE 2.31; 95% CI 57.7–67.2)59.4 (SE 2.76; 95% CI 53.7–65.2)
Sex - Male~78% (mean gender code 1.78 on 1=female, 2=male scale)~67% (mean gender code 1.67)
Months since stroke (mean)56.7 (SE 12.14; 95% CI 31.6–81.9)74.5 (SE 14.66; 95% CI 43.9–105.1)
Years of education (mean)13.3 (SE 1.29; 95% CI 10.6–16.0)15.5 (SE 0.55; 95% CI 14.4–16.7)
Aphasia severity (mean code, 1=mild/2=moderate/3=severe)1.96 (SE 0.15; 95% CI 1.65–2.26) — ~6 mild, 12 moderate, 5 severe
Aphasia subtype5 global, 3 Wernicke, 8 amnesic (anomic), 5 Broca, 2 others (mixed)6 global, 3 Wernicke, 3 amnesic (anomic), 8 Broca, 1 other
Aphasia severity (mean code)2.19 (SE 0.15; 95% CI 1.88–2.50) — ~2 mild, 12 moderate, 7 severe

Arms

FieldILATControl
InterventionIntensive Language Action Therapy — social-communicative therapy using request and planning language games in groups of 3 patients + 1 therapist, with picture cards and matched target utterances; 2.5 h/day, 5 days/week, for 2 weeks (25 h total). Constraint on non-verbal communication used in classic CIAT was NOT appliedUtterance-centred conventional speech-language therapy — naming, picture description, phonological/semantic distinctions and sentence completion using the same picture card sets and matched carrier phrases; groups of 3 patients + 1 therapist; 2.5 h/day, 5 days/week, for 2 weeks (25 h total)
Duration2 weeks2 weeks

Outcomes

OutcomeTypeControlInterventionHR / OR / RRP-value
Mean of four Aachen Aphasia Test (AAT) subtests (mAAT) — language performance; assessed within 2 working days before (T1) and after (T2) the 2-week interventionPrimaryControl (CONV): No significant change (t(20)=1.44, p>0.17) · Intervention (ILAT): Significant improvement (t(22)=3.93, p<0.001; mean Δ = +1.54, 95% CI 0.73–2.35) · Interaction Time × Therapy Type: F(1,42)=7.42, p=0.009, partial η² = 0.15 · Main effect of Time: F(1,42)=16.26, p<0.001, partial η² = 0.28 · ANCOVA covariates: Age, time post-onset, years of education — all p>0.11 (did not alter main result)
AAT production subscore (AATpro; naming + repetition)SecondaryInteraction: F(1,42)=6.19, p=0.02, η²p=0.13 · ILAT: t(22)=−3.57, p=0.002, d=1.13, 95% CI 0.47–1.79 · CONV: ns (p>0.33)
AAT comprehension (AATcom)SecondaryEffect of Time: F(1,42)=6.88, p=0.01, η²p=0.14 · Interaction: ns
AAT Naming subtestSecondaryInteraction: F(1,42)=4.67, p=0.04, η²p=0.10 · ILAT: t(22)=−2.43, p=0.03, Δ=1.35, 95% CI 0.20–2.50 · CONV: ns
AAT Token TestSecondaryInteraction: trend F(1,42)=3.52, p=0.07 · ILAT: t(22)=−2.20, p=0.04 · CONV: ns
AAT Repetition (ILAT)SecondaryILAT: t(22)=−3.53, p=0.002, Δ=0.91, 95% CI 0.38–1.45 · CONV: trend (p=0.07 two-tailed)
AAT Comprehension (ILAT)SecondaryILAT: t(22)=−2.26, p=0.03, Δ=2, 95% CI 0.17–3.38 · CONV: ns (p>0.32)
Amsterdam Nijmegen Everyday Language Test (ANELT)SecondaryEffect of Time: F(1,39)=8.84, p=0.005, η²p=0.19 · ILAT: t(19)=−3.33, p=0.004, d=1.9, 95% CI 0.71–3.09 · CONV: ns (p>0.33)
Action Communication Test (ACT)SecondaryEffect of Time: F(1,42)=8.47, p=0.006, η²p=0.17 · ILAT: t(22)=−2.32, p=0.03, d=1.67, 95% CI 0.18–3.17 · CONV: ns (p>0.08)
Communicative Activity Log (CAL) — quantity of communication (rated by relatives)SecondaryEffect of Time: F(1,40)=30.14, p<0.001, η²p=0.43 · ILAT: t(22)=−4.22, p<0.001, Δ=7.76, 95% CI 3.94–11.58 · CONV: t(18)=−3.58, p=0.002, Δ=8.84, 95% CI 3.65–14.04
Beck Depression Inventory (BDI-V) — primary depression outcome (after removing 4 outliers, 2 per arm, >2 SD from mean)SecondaryInteraction: F(1,38)=6.94, p=0.01, η²p=0.15 · Main effect of Time: F(1,42)=14.89, p<0.001, η²p=0.26 · ILAT: t(20)=4.38, p<0.001, Δ = −8.67, 95% CI −12.79 to −4.54 · CONV: ns (p>0.25)
Montgomery–Åsberg Depression Rating Scale (MADRS)SecondaryEffect of Time: F(1,42)=8.88, p=0.005 · ILAT: t(22)=−2.62, p=0.016, Δ = −2.7, 95% CI −4.83 to −0.56 · CONV: ns (p>0.13)
OverallAdverseBehavioural rehabilitation intervention with no drug exposure; no serious therapy-attributable adverse events reported
Post-randomization exclusions/dropouts (n=10)Adverse1 transient ischaemic attack during therapy; 1 second stroke; 1 newly diagnosed Parkinson’s disease; 1 medication change affecting attention/perception/motor performance; 3 previous aphasia diagnosis not confirmed by Token Test at T1; 3 in one group whose therapy was terminated by first COVID-19 lockdown (March 2020). 7 clinically related events split 3 CONV / 4 ILAT
Serious Adverse EventsAdverseNot systematically tabulated; only 1 TIA and 1 second stroke among 54 randomized participants (both counted as protocol exclusions rather than intervention adverse events)

Subgroup Analysis

No prespecified drug-style subgroup analyses. ANCOVA adjusting for age, time post-onset and years of education did not change the primary result (all covariate p>0.11). Baseline aphasia severity, aphasia subtype, and lesion topography were comparable across arms.

Criticisms

  • Small sample (n=44 analysed of 54 randomized) limits statistical power for subgroup analyses and generalizability
  • Single-centre study in native German speakers at one Berlin laboratory — external validity to other languages, healthcare systems, and rehabilitation settings is untested
  • Semi-blinded design: assessors and patients were blinded to therapy assignment, but therapists delivering both therapies were not — possible performance bias
  • No long-term follow-up in this trial; authors rely on earlier work to argue for durability of ILAT gains (1–2 years)
  • BDI interaction only became clearly significant after post-hoc removal of 4 outliers (2 per arm), and the pre-outlier interaction was near-significant (p=0.05)
  • Communicative Activity Log ratings by relatives, who accompanied patients daily to therapy, may be affected by expectation bias, potentially inflating between-timepoint improvements in both arms
  • 10 post-randomization exclusions (~19%) analysed per-protocol rather than by ITT could bias effect estimates
  • Chronic PSA only (≥12 months post-stroke); results should not be extrapolated to acute or subacute aphasia, where prior data show no ILAT/CIAT advantage
  • Two of three communication measures (ANELT, ACT) showed within-group improvement with ILAT only but interaction terms with therapy type were not significant, so between-arm superiority on daily communication is inconclusive

Funding

Not explicitly stated in the extracted full-text sections; ethics approval EA1/384/16 (Charité Universitätsmedizin Berlin, Campus Mitte). Investigator-initiated academic trial, Brain Language Laboratory, Freie Universität Berlin.

Based on: ILAT (Brain Communications, 2026)

Authors: Pulvermüller F, Jäger AT, Osterloh MR, ..., Mohr B

Citation: Brain Commun 2026;8(4):fcag252

Content summarized and formatted by NeuroTrials.ai.