The ability to remember relationships between non-adjacent elements is crucial for cognitive development, particularly language acquisition. This capacity, termed non-adjacent dependency (NAD) learning, is not unique to humans; it's also observed in other species, potentially rooted in sensory-motor integration and motor control. In humans, memory for temporally organized, rule-based auditory patterns is essential for grammar learning. Previous research has shown that even 4-month-olds can encode NADs in spoken sentences, and 7-month-olds demonstrate behavioral evidence of tracking NADs. However, memory retention of NADs in infants during their first year, and the role of sleep in this process, remained largely unexplored. Sleep plays a vital role in memory consolidation, particularly in early lexical-semantic learning. Sleep spindles, associated with increased synaptic plasticity, are implicated in sleep-dependent memory consolidation. Studies show that sleep spindles in different brain regions contribute to different aspects of memory consolidation, with frontal spindles linked to detailed episodic memory. This study aimed to investigate whether 6- to 8-month-old infants retain NADs in memory, whether sleep after learning influences this retention, and if the consolidation process differs from that of lexical-semantic memory.

Existing research demonstrates NAD learning in various species, suggesting evolutionary roots in sensory-motor processes. In humans, the ability to encode and process NADs is a precursor to grammar learning. Studies with 4-month-olds show they can encode NADs in spoken sentences, while 7-month-olds exhibit behavioral evidence of tracking NADs in artificial grammars. However, studies on memory retention of NADs in infants and the influence of sleep on this process have been lacking. Research on sleep and memory in infants has established the importance of sleep for lexical-semantic learning, with sleep spindles identified as key players in memory consolidation. Studies on sleep spindles and lexical-semantic memory have shown regional differences in their contributions, with frontal spindles associated with detailed memory of specific object-word pairings. This study builds upon this foundation by examining the role of sleep and sleep spindles in the consolidation of NAD memory in infants, comparing it to the consolidation of lexical-semantic memory.

Eighty-five monolingual German-speaking infants (mean age 7 months) participated. Infants were divided into two groups: a nap group and a wake group. In the encoding phase, infants passively listened to sentences from an Italian miniature language designed to test NADs. These sentences contained rule-based morphosyntactic dependencies between nonadjacent elements (auxiliary/modal verb and suffix). The encoding phase involved familiarization with these dependencies. Following a retention period (nap for the nap group, awake for the wake group), infants underwent a memory test phase. In this phase, infants heard new sentences: some with the familiar NADs (regular) and some with violated NADs (irregular). Half the sentences used old verb stems from the encoding phase; the other half used new ones. Event-related potentials (ERPs) were recorded during both phases to measure brain activity. Sleep was recorded polysomnographically for the nap group, measuring total sleep time, sleep stage duration, and sleep spindle activity (amplitude, number, density) in frontal, central, and parietal regions. Data analysis involved ANOVAs to examine ERP differences related to familiarity (encoding), regularity (memory test), and sleep group. Correlations were analyzed to explore relationships between sleep parameters (spindle activity) and memory-related ERP components.

The study revealed that 6- to 8-month-old infants retained familiarity with NADs and generalized this knowledge to new verb stems, regardless of whether they napped or stayed awake. Three familiarity effects were observed during encoding, reflecting early, mid, and late latency ERP responses. In the memory test phase, early and late memory effects mirrored the encoding familiarity effects, demonstrating retention of the NAD structure. Notably, a mid-latency memory effect differentiated the nap and wake groups. The nap group showed a polarity-reversed effect, suggesting a qualitative shift in NAD processing during sleep. This sleep-dependent change suggests the emergence of a new type of memory representation during the nap. Furthermore, frontal spindle activity correlated with specific memory for old speech phrases. Infants with high frontal spindle activity showed distinct ERP responses to old versus new verb stems, indicating the consolidation of specific memories during sleep. This contrasts with previous findings on lexical-semantic learning, where central-parietal sleep spindles were more relevant.

The findings demonstrate that infants as young as 6 months possess the capacity to retain and generalize NADs in speech, a fundamental aspect of grammar learning. Contrary to prior research on lexical-semantic learning in infants, sleep does not seem essential for the retention of morphosyntactic regularities. However, sleep significantly affects the processing of these regularities, suggesting that sleep contributes to a refinement or evolution of memory representations. The correlation between frontal sleep spindles and specific memory for old speech phrases suggests that sleep-dependent consolidation can occur alongside changes in generalized memory. The observed N400-like memory effect in the nap group resembles that found in adults learning NADs, possibly indicating an early contribution of the prefrontal cortex to memory processes, even though it is immature at this age. The results highlight both generalized and specific memory mechanisms involved in early language acquisition.

This study demonstrates that 6-month-old infants possess memory mechanisms crucial for grammar learning, specifically the ability to retain and generalize NADs in speech. While sleep is not necessary for initial retention, it significantly impacts the processing and potential refinement of these memories. The correlation between frontal sleep spindles and specific memory highlights the complex interplay between sleep, memory consolidation, and brain development in early language acquisition. Future studies should explore the specific neural mechanisms involved in these processes, considering factors like hippocampal involvement and individual differences in cognitive abilities.

The study's limitations include the inability to predict ERP components a priori due to the lack of prior research on this specific age group. The study also cannot definitively rule out the possibility that individual differences or developmental factors contribute to the relationship between frontal sleep spindles and specific memory. Further research is needed to clarify the direct role of sleep spindles in consolidation versus the possibility that higher spindle activity reflects pre-existing cognitive abilities. Finally, the sample size, while based on prior research, could be further expanded to enhance generalizability.