We wholeheartedly agree with Christiansen & Chater (C&C) that “acquiring a language is learning to process” (sect. 5, para. 3) and that “there is no representation of grammatical structure separate from processing” (sect. 6.2, para. 6). We also agree with C&C's more general thesis that the structure of language cannot be understood without taking into account the constraints and biases of the language learners and users. Although the Now-or-Never cognitive bottleneck is an unavoidable constraint on language comprehension and production, fully understanding its consequences requires taking into account socio-demographic realities, namely who is doing the language learning.
C&C write that “Language will be shaped by the linguistic patterns learners find easiest to acquire and process” (sect. 5, para. 3), but what is easiest may importantly depend on who is doing the learning. Some languages are learned exclusively by infants and used in small, culturally homogeneous communities. For example, half of all languages have fewer than 7,000 speakers. Other languages have substantial populations of non-native speakers and are used in large, culturally and linguistically heterogeneous communities. For example, at present, about 70% of English speakers are non-native speakers (Gordon Reference Gordon2005).
The socio-demographic niche in which a language is learned and used can influence its grammar insofar as different kinds of learners have different learning biases. Languages with many adult learners may adapt to their socio-demographic niche by eschewing features difficult for adults to process. Indeed, as Lupyan and Dale (Reference Lupyan and Dale2010) have shown in an analysis of more than 2,000 languages, languages spoken in larger and more diverse communities (those that tend to have more non-native learners) have simpler morphologies and fewer obligatory markings (see also Bentz & Winter Reference Bentz and Winter2013). In contrast, languages used in a socio-demographic niche containing predominantly infant learners tend to have many more obligatory markings – for example, they are more likely to encode tense, aspect, evidentiality, and modality as part of the grammar, and to have more complex forms of agreement (Dale & Lupyan Reference Dale and Lupyan2012; see also Dahl Reference Dahl2004; McWhorter Reference McWhorter2001; Trudgill Reference Trudgill2011).
Such influences of the socio-demographic environment on language structure are important caveats to C&C's thesis because the Now-or-Never bottleneck, although present in all learners, depends on the knowledge that a learner brings to the language-learning task.
On C&C's account, successful language processing depends on recoding the input into progressively higher-level (more abstract) chunks. As an analogy, C&C give the example of how remembering strings of numbers is aided by chunking (re-representing) numbers as running times or dates (sect. 2, para. 7). But, of course, this recoding is only possible if the learner knows about reasonable running times and the format of dates. The ability to remember the numbers depends on the ability to chunk them, and the ability to chunk them depends on prior knowledge.
In the case of language learning, recoding of linguistic input is “achieved by applying the learner's current model of the language” (sect. 4.1, para. 3) and further constrained by memory and other domain-general processes. But both the learner's language model and domain-general constraints vary depending on who the learner is.
Infants come to the language-learning task with a less developed memory and ability to use pragmatic and other extralinguistic cues to figure out the meaning of an utterance. As a consequence, the Now-or-Never bottleneck is strongly in place. The language adapts through increased grammaticalization that binds units of meaning more tightly, thereby increasing redundancy. For example, grammatical gender of the sort found in many Indo-European languages increases redundancy by enforcing agreement of nouns, adjectives, and pronouns, making one more predictable from the other and – arguably – reducing the memory load required for processing the utterances.
Adults come to the language-learning task with more developed memories, and ability for pragmatic inference, but at the same time they are biased by pre-existing chunks that may interfere with chunks that most efficiently convey the meaning in the new language. The greater memory capacities and ability to use contextual and other pragmatic cues to infer meanings, may relax the Now-or-Never bottleneck, nudging grammars toward morphological simplification with its accompanying decrease in obligatory markings (i.e., decrease in redundancy) and increase in compositionality (Lupyan & Dale Reference Lupyan, Dale, de Busser and LaPolla2015).
This reasoning helps explain how the Now-or-Never bottleneck can create “obligatorification” (sect. 5, para. 8) and also why some languages have more obligatory markings than other languages.
In summary, although we agree with C&C that “multiple forces influence language change in parallel” (sect. 5, para. 9), we emphasize the force constituted by the learning community. Languages adapt to the specific (cognitive) learning constraints and communicative needs of the learners and speakers.
You have
Access
We wholeheartedly agree with Christiansen & Chater (C&C) that “acquiring a language is learning to process” (sect. 5, para. 3) and that “there is no representation of grammatical structure separate from processing” (sect. 6.2, para. 6). We also agree with C&C's more general thesis that the structure of language cannot be understood without taking into account the constraints and biases of the language learners and users. Although the Now-or-Never cognitive bottleneck is an unavoidable constraint on language comprehension and production, fully understanding its consequences requires taking into account socio-demographic realities, namely who is doing the language learning.
C&C write that “Language will be shaped by the linguistic patterns learners find easiest to acquire and process” (sect. 5, para. 3), but what is easiest may importantly depend on who is doing the learning. Some languages are learned exclusively by infants and used in small, culturally homogeneous communities. For example, half of all languages have fewer than 7,000 speakers. Other languages have substantial populations of non-native speakers and are used in large, culturally and linguistically heterogeneous communities. For example, at present, about 70% of English speakers are non-native speakers (Gordon Reference Gordon2005).
The socio-demographic niche in which a language is learned and used can influence its grammar insofar as different kinds of learners have different learning biases. Languages with many adult learners may adapt to their socio-demographic niche by eschewing features difficult for adults to process. Indeed, as Lupyan and Dale (Reference Lupyan and Dale2010) have shown in an analysis of more than 2,000 languages, languages spoken in larger and more diverse communities (those that tend to have more non-native learners) have simpler morphologies and fewer obligatory markings (see also Bentz & Winter Reference Bentz and Winter2013). In contrast, languages used in a socio-demographic niche containing predominantly infant learners tend to have many more obligatory markings – for example, they are more likely to encode tense, aspect, evidentiality, and modality as part of the grammar, and to have more complex forms of agreement (Dale & Lupyan Reference Dale and Lupyan2012; see also Dahl Reference Dahl2004; McWhorter Reference McWhorter2001; Trudgill Reference Trudgill2011).
Such influences of the socio-demographic environment on language structure are important caveats to C&C's thesis because the Now-or-Never bottleneck, although present in all learners, depends on the knowledge that a learner brings to the language-learning task.
On C&C's account, successful language processing depends on recoding the input into progressively higher-level (more abstract) chunks. As an analogy, C&C give the example of how remembering strings of numbers is aided by chunking (re-representing) numbers as running times or dates (sect. 2, para. 7). But, of course, this recoding is only possible if the learner knows about reasonable running times and the format of dates. The ability to remember the numbers depends on the ability to chunk them, and the ability to chunk them depends on prior knowledge.
In the case of language learning, recoding of linguistic input is “achieved by applying the learner's current model of the language” (sect. 4.1, para. 3) and further constrained by memory and other domain-general processes. But both the learner's language model and domain-general constraints vary depending on who the learner is.
Infants come to the language-learning task with a less developed memory and ability to use pragmatic and other extralinguistic cues to figure out the meaning of an utterance. As a consequence, the Now-or-Never bottleneck is strongly in place. The language adapts through increased grammaticalization that binds units of meaning more tightly, thereby increasing redundancy. For example, grammatical gender of the sort found in many Indo-European languages increases redundancy by enforcing agreement of nouns, adjectives, and pronouns, making one more predictable from the other and – arguably – reducing the memory load required for processing the utterances.
Adults come to the language-learning task with more developed memories, and ability for pragmatic inference, but at the same time they are biased by pre-existing chunks that may interfere with chunks that most efficiently convey the meaning in the new language. The greater memory capacities and ability to use contextual and other pragmatic cues to infer meanings, may relax the Now-or-Never bottleneck, nudging grammars toward morphological simplification with its accompanying decrease in obligatory markings (i.e., decrease in redundancy) and increase in compositionality (Lupyan & Dale Reference Lupyan, Dale, de Busser and LaPolla2015).
This reasoning helps explain how the Now-or-Never bottleneck can create “obligatorification” (sect. 5, para. 8) and also why some languages have more obligatory markings than other languages.
In summary, although we agree with C&C that “multiple forces influence language change in parallel” (sect. 5, para. 9), we emphasize the force constituted by the learning community. Languages adapt to the specific (cognitive) learning constraints and communicative needs of the learners and speakers.