Verified by 
Prediction has been touted as a “canonical cortical computation” [i], and essential for basic linguistic processes [ii]. Naturally, the efficiency with which the brain can implement its various forms of predictive processing will change as a person ages, but it has previously been unclear how this more specifically impacts language.
A study this week sought to address this issue [iii]. A range of work in psycholinguistics seems to point to certain differences in how older and younger adults use context-based linguistic predictions to understand language. For instance, if words in a narrative are semantically related in some way, this can be used to anticipate upcoming material. If someone walks into an office meeting, abruptly shouts "Ainsley Harriott", and then departs, our understanding of appropriate context for language use informs us that some kind of violation has occurred. Likewise, certain grammatical tendencies of a language might prime a particular interpretation of a sentence over another, as when “The old man the boats” is deemed ungrammatical if we initially assume “The old man” to be referring to a person, rather than the action (i.e. manning boats) of a group of people (i.e. old people).
Previous work has shown that, in younger adults, N400 responses showed a graded facilitation for words as a factor of their semantic relationship with some anticipated target word [iv]. “N400” here refers to a negative-going waveform detected via electroencephalographic brain recordings which peaks approximately 400ms post-stimulus, which in the context of language has often been associated with semantic memory and integration processes.
For instance, in the sentence “I take my coffee with cream and …”, for younger adults a reduction in N400 responses has been found when the next word was expected (“sugar”) relative to unexpected (“socks”), suggesting that the N400 partly reflects predictive activations of semantic features. In other words, the N400 indexes that the brain is somehow generating a set of appropriate representations to aid future inferences. In older adults, there was found to be a weaker facilitation effect. There is also generally a reduction in the recruitment of predictive processing in language in older adults [v].
Other work has highlighted how brain regions that have often been touted as “the seat of syntax” – left inferior frontal regions – are, when damaged, not detrimental to basic linguistic phrase constriction, and that these regions are rather involved in forms of top-down structural predictions [vi]. The widespread prevalence of prediction in language, even in these classical “language areas”, should not be underestimated.
In this recent study by Michael Broderick and colleagues from Trinity College Dublin, a section of an audiobook version of the novel The Old Man and the Sea by Ernest Hemingway was played to two groups: a younger (19-38 years) and older group (55-77 years) [iii]. The authors predicted that during the processing of the narrative, older participants would exhibit a reduced ability (via the strength of their N400 deflections) to pre-activate semantic features of anticipated words. This effect was also hypothesized to correlate with their verbal fluency, or the number of words the participants could voluntarily elicit as belonging to a given semantic category over a fixed period of time.
Modeling the degree of lexical surprisal and semantic dissimilarity of words in the narrative, the authors showed lexical surprisal did indeed modulate the N400 in older participants, but that this modulation was delayed relative to younger participants. On the other hand, semantic dissimilarity was utilized much less by older participants, which also correlated with verbal fluency measures. This suggests that when older participants do predict upcoming material, semantic dissimilarity is not their primary tool.
More broadly, these results indicate that the N400 component reflects a complex summation of independent language-related predictive processes – a conclusion which is not necessarily in itself surprising, but which certainly helps to motivate future work to independently investigate the progressive changes in how adults recruit distinct predictive processes to solve different aspects of language processing. For example, how do these and other predictive processes go to work across different age groups during the processing of questions, or the relations between pronouns and referents, or in different types of grammatical constructions?
Similar questions arise here: By leaving behind semantic dissimilarity measures during language processing, does the aging brain correspondingly recruit new forms of predictions, or are there other, non-linguistic compensation mechanisms at play, too? How do other systems of cognition aid the degenerating language system over time?
Other recent work has used magnetoencephalography (MEG) to observe the dynamics of prediction processes in language comprehension, comparing predictable sentence endings with unpredictable sentence endings [vii]. With sentences providing enough information for listeners to easily predict upcoming words, the authors found coupling between posterior alpha and frontal high-frequency gamma oscillations (or repetitive patterns of neural activity); a form of neural interactivity that has been proposed to be involved in early stages of language processing when we begin to pair different words together into larger units [viii]. The authors conclude their study by claiming that forms of prediction and language processing "might be supported by the coupling between the alpha and gamma activities” – a very real possibility, and a direction for research into the aging brain that complements work on the N400 and related components.
References
Keller, G.B., & Mrsic-Flogel, T.D. (2018). Predictive processing: a canonical cortical computation. Neuron 100(2): 424–435.
Friston, K.J., & Frith, C.D. (2015). Active inference, communication and hermeneutics. Cortex 68: 129–143.
Broderick, M.P., Di Liberto, G.M., Anderson, A.J., Rofes. A., & Lalor, E.C. (2021). Dissociable electrophysiological measures of natural language processing reveal differences in speech comprehension strategy in healthy ageing. Scientific Reports 11: 4963.
Federmeier, K.D., & Kutas, M. (1999). A rose by any other name: Long-term memory structure and sentence processing. Journal of Memory and Language 41: 469–495.
Wlotko, E.W., Lee, C.-L., & Federmeier, K.D. (2010). Language of the aging brain: Event-related potential studies of comprehension in older adults. Language and Linguistics Compass 4: 623–638.
Matchin, W., Hammerly, C., & Lau, E. (2017). The role of the IFG and pSTS in syntactic prediction: Evidence from a parametric study of hierarchical structure in fMRI. Cortex 88: 106–123.
Wang, L., Hagoort, P., & Jensen, O. (2018). Language prediction is reflected by coupling between frontal gamma and posterior alpha oscillations. Journal of Cognitive Neuroscience 30(3): 432–447.
Murphy, E. (2020). The Oscillatory Nature of Language. Cambridge: Cambridge University Press.
