For language acquisition and processing, the ideomotor theory predicts that the comprehension and the production of language are functionally based on their expected perceptual effects (i.e., linguistic events). This anticipative mechanism is central for action–perception behaviors in human and nonhuman animals, but a recent ideomotor recycling theory has emphasized a language account throughout an evolutionary perspective.

We agree with Christiansen & Chater (C&C) that language processing and acquisition are tightly constrained by the limits of sensory and memory systems. However, the human brain supports a range of cognitive functions that mitigate the effects of information processing bottlenecks. The language system is partly organised around these moderating factors, not just around restrictions on storage and computation.

The target article by Christiansen & Chater (C&C) offers an integrated framework for the study of language acquisition and, possibly, a novel role for internal speech in language acquisition. However, the “Now-or-Never bottleneck” raises a paradox for language evolution. It seems to imply that language complexity has been either reduced over time or has remained the same. How, then, could languages as complex as ours have evolved in prelinguistic ancestors? Linguistic Platonism could offer a solution to this paradox.

The Now-or-Never bottleneck has important consequence for understanding why languages have the structures they do. However, not addressed by C&C is that the bottleneck may interact with who is doing the learning: While some languages are mostly learned by infants, others have a large share of adult learners. We argue that such socio-demographic differences extend and qualify C&C's thesis.

Christiansen & Chater (C&C) propose that language comprehenders must immediately compress perceptual data by “chunking” them into higher-level categories. Effective language understanding, however, requires maintaining perceptual information long enough to integrate it with downstream cues. Indeed, recent results suggest comprehenders do this. Although cognitive systems are undoubtedly limited, frameworks that do not take into account the tasks that these systems evolved to solve risk missing important insights.

Attempts to explain linguistic phenomena as consequences of memory constraints require detailed specification of linguistic representations and memory architectures alike. We discuss examples of supposed locality biases in language comprehension and production, and their link to memory constraints. Findings do not generally favor Christiansen & Chater's (C&C's) approach. We discuss connections to debates that stretch back to the nineteenth century.

Christiansen & Chater (C&C) make two related and somewhat contradictory claims, namely that the ever abstract language representations built during Chunk-and-Pass processing allow for ever greater interference from extra-linguistic information, and that it is nevertheless the language system that re-codes incoming information into abstract representations. I analyse these claims and discuss evidence suggesting that Gestalt-like representations hijack Chunk-and-Pass processing.

Signed and spoken languages emerge, change, are acquired, and are processed under distinct perceptual, motor, and memory constraints. Therefore, the Now-or-Never bottleneck has different ramifications for these languages, which are highlighted in this commentary. The extent to which typological differences in linguistic structure can be traced to processing differences provides unique evidence for the claim that structure is processing.

Christiansen & Chater (C&C)'s key premise is that “if linguistic information is not processed rapidly, that information is lost for good” (sect. 1, para. 1). From this “Now-or-Never bottleneck” (NNB), C&C derive “wide-reaching and fundamental implications for language processing, acquisition and change as well as for the structure of language itself” (sect. 2, para. 10). We question both the premise and the consequentiality of its purported implications.

Christiansen & Chater's (C&C's) Now-or-Never bottleneck framework is similar to the Good-Enough Language Processing model (Ferreira et al. 2002), particularly in its emphasis on sparse representations. We discuss areas of overlap and review experimental findings that reinforce some of C&C's arguments, including evidence for underspecification and for parsing in “chunks.” In contrast to Good-Enough, however, Now-or-Never does not appear to capture misinterpretations or task effects, both of which are important aspects of comprehension performance.

Prior language input is not lost but integrated with the current input. This principle is demonstrated by “reservoir computing”: Untrained recurrent neural networks project input sequences onto a random point in high-dimensional state space. Earlier inputs can be retrieved from this projection, albeit less reliably so as more input is received. The bottleneck is therefore not “Now-or-Never” but “Sooner-is-Better.”

Researchers, motivated by the need to improve the efficiency of natural language processing tools to handle web-scale data, have recently arrived at models that remarkably match the expected features of human language processing under the Now-or-Never bottleneck framework. This provides additional support for said framework and highlights the research potential in the interaction between applied computational linguistics and cognitive science.

Christiansen & Chater's (C&C's) key goals for a language system have been realized by neural models for short-term storage of linguistic items in an Item-Order-Rank working memory, which inputs to Masking Fields that rapidly learn to categorize, or chunk, variable-length linguistic sequences, and choose the contextually most predictive list chunks while linguistic inputs are stored in the working memory.

Empirical evidence from dialogue, both corpus and experimental, highlights the importance of interaction in language use – and this raises some questions for Christiansen & Chater's (C&C's) proposals. We endorse C&C's call for an integrated framework but argue that their emphasis on local, individual production and comprehension makes it difficult to accommodate the ubiquitous, interactive, and defeasible processes of clarification and repair in conversation.

Christiansen & Chater (C&C) envision language function as a hierarchical chain of transformations, enabling rapid, continuous processing of input. Their notion of a “Now-or-Never” bottleneck may be elaborated by recognizing that timescales become longer at successive levels of the sensory processing hierarchy – that is, the window of “Now” expands. We propose that a hierarchical “process memory” is intrinsic to language processing.

Humans process language with their neurons. Memory in neurons is supported by neural firing and by short- and long-term synaptic weight change; the emergent behaviour of neurons, synchronous firing, and cell assembly dynamics is also a form of memory. As the language signal moves to later stages, it is processed with different mechanisms that are slower but more persistent.

We argue that to reflect participant interactivity in conversational dialogue, the Christiansen & Chater (C&C) perspective needs a formal grammar framework capturing word-by-word incrementality, as in Dynamic Syntax, in which syntax is the incremental building of semantic representations reflecting real-time parsing dynamics. We demonstrate that, with such formulation, syntactic, semantic, and morpho-syntactic dependencies are all analysable as grounded in their potential for interaction.

Christiansen & Chater (C&C) offer the Chunk-and-Pass strategy as a language processing approach allowing humans to make sense of incoming language in the face of cognitive and perceptual constraints. We propose that the Chunk-and-Pass strategy is not adequate to extend universally across languages (accounting for typologically diverse languages), nor is it sufficient to generalize to other auditory modalities such as music.

This commentary raises two issues: (1) Language processing is hastened not only by internal pressures but also externally by turn-taking in language use; (2) the theory requires nested levels of processing, but linguistic levels do not fully nest; further, it would seem to require multiple memory buffers, otherwise there's no obvious treatment for discontinuous structures, or for verbatim recall.

If memory constraints were the only limitation on language processing, the best possible language would be one with only one word. But to explain the rich structure of language, we need to posit a second constraint: the pressure to communicate informatively. Many aspects of linguistic structure can be accounted for by appealing to equilibria that result from these two pressures.