Regular rhythmic primes improve sentence repetition in children with developmental language disorder

378

Originally posted at Biological sciences : nature.com / July 10, 2023

• Anna Fiveash, 
• Enikő Ladányi, 
• Julie Camici, 
• Karen Chidiac, 
• Catherine T. Bush, 
• Laure-Hélène Canette, 
• Nathalie Bedoin, 
• Reyna L. Gordon & 
• Barbara Tillmann 

npj Science of Learning volume 8, Article number: 23 (2023) Cite this article

• 377 Accesses
• 70 Altmetric
• Metricsdetails

Abstract

Recently reported links between rhythm and grammar processing have opened new perspectives for using rhythm in clinical interventions for children with developmental language disorder (DLD). Previous research using the rhythmic priming paradigm has shown improved performance on language tasks after regular rhythmic primes compared to control conditions. However, this research has been limited to effects of rhythmic priming on grammaticality judgments. The current study investigated whether regular rhythmic primes could also benefit sentence repetition, a task requiring proficiency in complex syntax—an area of difficultly for children with DLD. Regular rhythmic primes improved sentence repetition performance compared to irregular rhythmic primes in children with DLD and with typical development—an effect that did not occur with a non-linguistic control task. These findings suggest processing overlap for musical rhythm and linguistic syntax, with implications for the use of rhythmic stimulation for treatment of children with DLD in clinical research and practice.

Introduction

Developmental language disorder (DLD) affects ~3–7% of the population and involves delayed and disordered language comprehension and/or production that cannot be attributed to peripheral deficits or global impairments in other cognitive domains¹. Although the DLD phenotype is heterogeneous, symptoms primarily affect the domain of morphosyntax, including the use of morphological markers and complex syntax processing (encompassing the understanding and production of sentences with multiple clauses)^2,3,4. Limitations in language processing result in a struggle to understand peers, teachers, and parents, and to efficiently express thoughts, which can lead to lifelong consequences in individuals’ academic and social life⁵. Effective speech-language therapy is essential to mitigate these consequences; yet DLD is greatly understudied, especially with respect to its high prevalence, compared to other neurodevelopmental disorders⁶.

Rhythmic priming is a short-term rhythmic stimulation with demonstrated benefits for grammar task performance that might be of clinical relevance to children with language impairments, and can also inform the theoretical understanding of connections between rhythm and grammar processing^7,8. This line of research builds on previous work showing the relationship between rhythm and grammar processing in various populations⁹ and age ranges¹⁰. In general, priming refers to the effect of a stimulus (the prime) on the processing of a subsequent stimulus (the target). In rhythmic priming experiments, regular and irregular rhythmic (or other control) primes are presented before a set of naturally spoken sentences, and participants perform a language task on these sentences. Results show that for children with typical development (TD), with dyslexia, and with DLD, grammaticality judgments are improved after regular compared to irregular rhythms or other control conditions^{7,11,12,13,14,15}.

The primary hypothesis underlying this paradigm is that engaging the beat-based rhythm processing system impacts subsequent language processing via shared underlying mechanisms⁸. It is supported by frameworks positing that (1) endogenous neural oscillations synchronize with the steady and consistent beat of regular rhythms and persist once the rhythm stops (in line with dynamic attending theory^16,17), benefiting subsequent sentence processing, and (2) rhythm and language share numerous facets of neural and cognitive processing (see for example the Processing Rhythm in Speech and Music (PRISM) framework⁸ and¹⁸ for a neuroimaging meta-analysis). More broadly, rhythmic priming results fit into a research domain showing strong connections between music and language processing in the brain. Similarities between music and language in relation to syntax^19,20, rhythm^21,22, and auditory processing^23,24 suggest the potential capacity for transfer effects across domains, and mounting evidence has shown that music training can causally enhance various aspects of language processing, including tracking of the speech signal^25,26,27, phonological awareness^28,29,30, and reading^31,32. See refs. ^{33,34,35,36,37,38} for more general links between music and language processing and their neural correlates.

The current study focused on the rhythm-grammar link, and investigated whether rhythmic priming affects complex syntax task performance. We tested for its potential benefit on sentence repetition, a task sensitive to syntactic knowledge^39,40 in children with DLD and TD, with clinical and theoretical aims. It is crucial to investigate the rhythm-grammar link across different language tasks: (1) not requiring a conscious reflection upon grammaticality, and (2) reflecting clinical characteristics of children with DLD. Clinically, sentence repetition tasks are sensitive for DLD diagnosis, as the repetition of grammatically complex sentences is particularly challenging for children with DLD³⁹. Short-term enhancement of sentence repetition performance would suggest rhythmic priming as a valuable tool in clinical use to increase the efficacy of treatment programs, as previously suggested for syntax processing in populations with hearing loss⁴¹. Theoretically, assessing the rhythmic priming effect on sentence repetition would further show a benefit on grammatical processing outside of error detection, with implications for potentially shared underlying brain networks and the previously suggested rhythm-grammar processing link^10,42.

We additionally measured individual differences in key demographic and cognitive characteristics to investigate links with the rhythmic priming effect. Chronological age was included, as grammatical sentence processing and production was expected to increase with age. Reading age was included, as previous results in TD children have shown a correlation with increased reading age and benefit of rhythmic primes on grammaticality judgements¹³. Digit span (i.e., the number of digits that could be recalled in sequence) was measured to investigate the impact of short-term memory capacity on sentence repetition, and beat-based rhythm perception abilities (measured with the beat alignment test⁴³) were measured to investigate whether participants with greater rhythmic abilities might benefit more from the rhythmic primes. The goal of including these characteristics in the analysis was to establish profiles of children that could benefit the most from rhythmic priming in the present experimental paradigm, and then potentially within speech-language therapy.

Results

Regular rhythmic primes improve sentence repetition

French-speaking children with DLD and age-matched children with TD aged 5.4–13 years listened to regular or irregular rhythms followed by sets of six sentences. After listening to each sentence, they repeated the sentence as accurately as possible. Sentence repetitions were recorded and scored blindly offline, with possible scores of 0, 0.5, or 1, focusing on grammatical features of each reproduced utterance (see Methods for more information). A control task was run where children listened to regular or irregular rhythms before performing a visual cancellation task (cross-out as many animals as possible in a given time period).

To investigate whether regular rhythmic primes improved sentence repetition performance compared to irregular rhythmic primes, cumulative link mixed models were run. Prime (regular, irregular) and group (DLD, TD) were included as fixed effects, and participants and sentences as random effects (details in Methods). Overall, both prime and group had significant effects (in the expected directions) on sentence repetition scores (Fig. 1). Prime, χ² (1) = 6.36, p = 0.01, AIC = 1202.2, and group, χ² (1) = 29.81, p < 0.001, AIC = 1178.7, significantly improved the intercept-only model (AIC = 1206.51). The prime x group interaction, χ² (1) = 0.39, p = 0.53, AIC = 1175.73 did not improve the model with prime and group as fixed effects (AIC = 1174.12). The final base model (with fixed effects of prime and group), revealed higher performance after regular, emmean = 2.67, SE = 0.45, 95% CI [1.78, 3.55] than irregular, emmean = 2.20, SE = 0.45, 95% CI [1.33, 3.08] primes, estimate = 0.46, SE = 0.18, z-ratio = 2.551, p = 0.01, d = 0.44, 95% CI around effect size [0.10, 0.79], and higher performance for children with TD, emmean = 4.16, SE = 0.52, 95% CI [3.13, 5.18] than children with DLD, emmean = 0.71, SE = 0.50, 95% CI [−0.26, 1.68], estimate = 3.44, SE = 0.52, z-ratio = 6.66, p < 0.001, d = 2.32, 95% CI around effect size [1.64, 3.00]. These results show that regular rhythmic primes significantly improved sentence repetition performance across both participant groups, extending prior work showing priming effects for grammaticality judgment tasks^7,11,12,13.

To investigate the potential influence of specific demographic and cognitive covariates (chronological age, reading age, digit span, rhythm perception), we added each covariate (main effect and interactions with prime, group, and prime x group) into the base model separately. The same procedure of adding and removing effects and interactions for likelihood-ratio testing was done for each covariate. (1) For chronological age, adding the fixed effect significantly improved the base model, χ² (1) = 27.54, p < 0.001, AIC = 1148.6, but interactions with other factors did not, reflecting increased performance with age, trend = 1.04, SE = 0.16, p < 0.001, z-ratio = 6.34, r = 0.74, 95% CI around effect size [0.61, 0.82], see Fig. 2A. (2) For reading age, adding the fixed effect and interactions showed a significant prime x group x reading age interaction, χ² (1) = 5.71, p = 0.017, AIC = 1042.1. This interaction revealed a significant trend for increased performance with reading age after regular primes in children with TD, trend = 1.26, SE = 0.40, p = 0.002, z-ratio = 3.13, r = 0.48, 95% CI around effect size [0.20, 0.66], but not after irregular primes in children with TD (p = 0.28) and not in children with DLD after regular (p = 0.35) or irregular (p = 0.10) primes. Note that a reduced base model was made for this analysis, since three children (two with DLD and one with TD) had missing reading age data as they were not yet able to read. Finally, the model did not significantly improve when adding (3) age-adjusted digit span scores, χ² (4) = 5.01, p = 0.29, AIC = 1177.1, or (4) age-adjusted rhythm perception scores, χ² (4) = 1.51, p = 0.83, AIC = 1180.6.

These results show that general performance on the task increased with age (see Fig. 2A), but was not influenced by digit span or beat-based rhythm perception scores. Importantly, the rhythmic priming effect (i.e., the benefit of the regular compared to irregular rhythmic primes) was not influenced by age or individual differences in cognitive abilities in children with DLD, suggesting benefits across ages and cognitive abilities. However, higher reading age was associated with an increased priming effect in children with TD, supporting previous results¹³. The effect sizes of group and age were very large, showing that these predictors strongly influenced sentence repetition, as might be expected. The effect size of prime in comparison was small (though approaching a medium effect size); however, it emerged across the full sample regardless of age, suggesting potential for clinical applications.

Read the FULL RESEARCH PAPER HERE