Language as a central axis for studying Alzheimer's disease: From pathogenesis and clinical practice to scientific theory

Language occupies a central position in human cognition. It is not only one of its most natural and complex manifestations, but also one of the most distinctive and valuable human capacities. As such, language is particularly vulnerable to diseases or impairments that affect its primary neural substrate, namely the brain. Consequently, linguistic performance has come to be regarded as a key biomarker for many, if not all, brain-related disorders, including dementias. In this context, Alzheimer's disease (AD) is currently understood as a neurodegenerative condition characterized by prominent linguistic symptomatology,^1,2 which coexists from the earliest stages with impairments in episodic memory and, crucially, may already be evident during the preclinical phase, ³ showing particular sensitivity to the presence of AD-related biomarkers (e.g., amyloid-β). ⁴ It is therefore unsurprising that difficulties in episodic retrieval (memory) and in lexicalization or naming (language) are often conceptualized as ‘two sides of the same coin’; that is, interrelated and interconnected changes. The inability to retrieve an episode necessarily entails a failure to engage in language-mediated conscientization—that is, to access the lexical representations of events. ⁵

On this basis, the extent to which language impairment in AD can be considered independent of deficits in other cognitive functions remains a matter of ongoing debate. Is it plausible to determine whether language disorders in AD arise from impairment of language as a mediating system—that is, as a tool for modulating and supporting human cognition across multiple domains (e.g., thinking, perception, interaction, or memory), ⁶ or, alternatively, whether language itself constitutes the primary function affected, albeit not entirely independently of other cognitive processes? Teruo Yokoi's position paper ⁵ supports the view that AD primarily impairs language-mediated conscientization, resulting not only in word-finding difficulties (anomia) but also in widespread cognitive deficits across multiple domains. Given that cognitive capabilities are mediated through language, one plausible interpretation is that linguistic impairment in AD primarily reflects early-stage cognitive dysfunction. ⁵ At first glance, this relationship seems theoretically reasonable and likely. Language comprehension and production, as cognitive functions, rely on general cognitive abilities, ⁷ and studies have also shown that changes in language tend to correspond with the level of cognitive and functional decline in AD: greater cognitive deterioration is associated with more noticeable language impairments.^8,9 This would confirm that the relationship between cognition and language, at least in the context of impairments, is implicational in nature: language functions as a mediator of cognition.

This view is sufficiently well established. For instance, a growing body of research^10–12 advocates the use of assessment tools with increased cognitive load to detect language disorders for the early identification of AD. Moreover, higher-order human cognitive functions, such as memory, thought, and judgment, are largely mediated through language, ⁵ making it difficult to conceive of remembering an event, referring to a concept, or engaging in reasoning without linguistic support. ¹³ From this perspective, many higher cognitive functions, including memory, orientation, calculation, attention, emotion, and social cognition, are channeled through language, rendering it reasonable to interpret linguistic disruption as a consequence of broader cognitive decline.

However, the extent to which language, and, consequently, its impairments, can be approached independently of other cognitive abilities remains an open question. To what degree is it implausible that difficulties in lexical access, the construction of syntactically complex sentences, or the production of pragmatically appropriate discourse reflect primary linguistic impairments rather than deficits in memory, executive functioning, or reasoning? Although it is neither desirable nor possible to fully dissociate linguistic decline from broader cognitive impairment, despite evidence that patients with AD are often more aware of their language difficulties than of other cognitive dysfunctions, ¹⁴ it remains reasonable to propose that language, as a function in its own right, may undergo a relatively autonomous decline and constitute an independent symptomatological pathway. As a cognitive function, language relies on its own neuroanatomical network. Although this network is partially shared with other cognitive functions, and depends on the proper functioning of their respective neural systems, it may nonetheless exhibit impairments that are specific to language itself. It is therefore reasonable to expect that the autonomy of language, even if only partial or relative, is not merely neurofunctional but also symptomatological. In other words, language disturbances may emerge in an isolated manner under conditions involving deterioration of the neurophysiological networks that support linguistic processing.

Evidence for a dissociation between language and cognition has been documented in adult speakers with acquired brain damage: in one case, the speaker exhibited impairments in lexical retrieval and irregular inflection but preserved declarative memory, whereas in another case, the speaker showed impaired declarative memory without deficits in inflection. ¹⁵ Moreover, primarily linguistic neurodegenerative trajectories, such as different clinical variants of PPA, not only demonstrate dissociations between language and other cognitive domains,^16,17 but also between specific linguistic functions themselves.^18–20 Similar patterns of dissociation have also been reported in speakers with AD, who, for example, may exhibit differential impairment of structural versus semantic processing of objects ²¹ (language–cognition dissociation) and show greater deficits in certain language functions (e.g., inferencing) compared with others (e.g., embedding processing) ²² (language–language dissociation).

While language serves as a primary medium for the expression of ideas, reasoning, and judgment, it also constitutes a distinct and considerably autonomous cognitive function, even if one assumes that “the fundamental function of language, which supports thinking and communication, is to create existence”. ⁵ Accordingly, even if language acts as a conduit for alterations in other cognitive domains, such as memory or judgment, insofar as “cognitive capabilities function via language”, it is reasonable to posit that language itself may exhibit significant impairments that are not entirely reducible to, or representative of, other cognitive disorders. Its impairment, therefore, should not be understood solely as a secondary or consequential outcome of decline in other cognitive functions. On the contrary, precisely because of its centrality and its high cognitive and neural demands, language may be particularly vulnerable to pathological processes in its own right, whether rare or as prevalent as those observed in AD. In fact, evidence from neuroimaging studies support this idea, demonstrating the neurocognitive dissociation of language from other cognitive functions, such as memory, ²³ thereby supporting our position that language possesses a degree of neurofunctional, cognitive, and symptomatological autonomy.

Recent findings further support this view by suggesting that AD can be sensitively identified through linguistic markers rather than through broader cognitive traits.^10,24,25 A substantial body of recent interdisciplinary research has demonstrated that changes in speech and language may emerge during the preclinical stages of AD, manifesting, for example, in the acoustic and temporal properties of speech,^10,26,27 as well as in higher-level linguistic domains, including lexical access, semantic richness, and overall verbal proficiency.^28–30 In addition, evidence indicating that differentiated neurocognitive substrates may tip the diagnostic balance toward predominantly cognitive, linguistic, or mixed symptom profiles reinforces the notion of partial functional independence. ³¹ To strengthen this position, it is necessary to investigate whether specific language measures can selectively enhance particular language abilities in speakers with AD.

Thus, while fully endorsing Yokoi's position ⁵ —that cognitive capacities operate through language, so that linguistic disturbances in AD both reflect and contribute to impairments in language-mediated cognitive functions—I also advocate for a broader cognitive-functional perspective. My position does not depart from Yokoi's proposal, but rather extends it by suggesting that linguistic impairments may be understood not only as indicators of the cognitive decline in AD, but also as possible manifestations of a primary disruption of language itself. In this sense, some language disturbances may indeed be viewed as closely intertwined with broader cognitive impairments (the “two sides of the same coin” position), reflecting the strong relationship between global cognition and language. At the same time, however, other disturbances, such as those discussed above, may be more appropriately interpreted as evidence of disruption within the linguistic system itself. These perspectives are not mutually exclusive but rather complementary, offering both an interrelational and an autonomous view of the linguistic symptomatology of AD. When a neurodegenerative process is underway, it is reasonable to expect it to affect the most central, and therefore most demanding, neurocognitive processes. In humans, this role is fulfilled by language. Although language has evolutionarily functioned as a central mediating mechanism for consciousness and conceptualization, it represents the most central cognitive function in humans and, at the same time, one of the most vulnerable to brain-related pathology. Since Emery's seminal work, ³² research in cognitive psychology, neuroscience, and linguistics has converged on the view that linguistic components deteriorate hierarchically with the progression of AD, with more complex language phenomena being disrupted earlier. Language should therefore be understood, within the broader constellation of cognitive functions, as occupying a privileged position, one that renders it a particularly sensitive and informative marker for the detection of a wide range of neurocognitive conditions across the lifespan.

In essence, this commentary aims to highlight and reinforce scientific attention on key research avenues that should take center stage in the study of AD. A deeper understanding of how and why language changes in AD can benefit from novel assessment models, which may include the development of linguistic tests that are more firmly grounded in linguistic theory and that incorporate contexts allowing language to be isolated from cognition—for example, through tests focusing on acoustic properties of speech ¹⁰ or employing semantically controlled tasks. ³³ Complementing these studies, which are crucial given the predictive and diagnostic value of linguistic features in AD, ³⁴ with research from a neurolinguistic perspective would further enable the identification of patterns of cognitive and linguistic decline in AD with greater precision.