Why Do Humans Have Linguistic Intuition?

Lay Summary

Sometimes in scientific progress, something that seems obvious and common sense turns out to be ‘upside down’. One famous example is the idea that the Sun revolves around the Earth: this initially seems like common sense but scientific discovery showed that actually the Earth revolves around the Sun. This paper argues for a similar inversion in our understanding of linguistic intuition. Some sentences, when you read or hear them, feel ‘not right’, even if the meaning seems clear. For instance, I don’t want to go to the cinema is acceptable in English but I don’t want going to the cinema is not, even though this second sentence seems to mean the same as the first. Many people assume that this sense of ‘not rightness’ occurs because we detect what is and is not grammatical. Building on the findings of several different fields, this paper reaches a different conclusion: sentences seem ‘not right’ when they violate expectations of relevance. This hypothesis is important because it undermines many common assumptions about language, assumptions that are widespread both in academic research and in the public imagination. It also explains many otherwise surprising facts about when the feeling of ‘not rightness’ occurs.

Introduction: Whence linguistic intuition?

Native speakers show a great deal of intuitive agreement over whether a given sentence is ‘acceptable’ in their language. John speaks French is acceptable in English but Speaks John French is not. These intuitions are immediate, unreflective and involuntary. They occur even if a meaning of the sentence is apparently clear, and even if it differs from acceptable sentences only superficially: I don’t want to go to the cinema is acceptable in English but I don’t want going to the cinema is not. Crucially, linguistic intuitions are accompanied by a distinctive, psychological sense that could be called ‘oddness’ or ‘wrongness’. My goals in this paper are, first, to raise the question of why humans have linguistic intuitions at all; and second, to provide an answer, based on a synthesis of theoretical principles and empirical knowledge from several fields.¹

These issues are of foundational importance because linguistic intuitions are themselves central to the scientific understanding of language and cognition in many ways. First, they are salient and naturally occurring cognitive phenomena in their own right, and as such they are an important target of explanation. Second, linguistic intuitions plainly reflect something important about individuals’ knowledge of language, and as such any theory of language should account for them. Third, field linguists habitually elicit linguistic intuitions from native speakers as part of learning and documenting languages (‘Is this ok in your language?’). Fourth, linguistic intuitions are widely used as critical empirical data for basic questions about the nature of language, and grammar in particular (Chomsky, 1965; Schütze, 1996/2016; Wasow & Arnold, 2005; Schindler et al., 2020; Gross, 2021; Scholz et al., 2022; inter alia). So linguistic intuitions are critical in many ways, and explaining why humans have them is a foundational issue for the human sciences.

Yet despite this basic importance, we have no clear explanation or understanding of why humans have linguistic intuition in the first place. Why should any sentence trigger a psychological sense of oddness? The question is not about what specific grammatical phenomena trigger linguistic intuition, nor is it about explaining particular syntactic or morphological phenomena. Those topics are plainly related and they have been the focus of much investigation, but they are not the focal issue here. Enormous literatures have used linguistic intuitions to investigate grammar itself, but there is very little if any focused attention on the question of why such intuitions exist in the first place.

The answer is not as simple as a mismatch between experience and expectation, because many aspects of everyday life mismatch with expectation without generating a psychological sense of oddness. Nor is it simply the violation of norms, because again, many everyday behaviors violate norms without generating a sense of oddness. So why do some sentences? This is an important question, without an answer.

There is, of course, a widespread assumption that humans have some cognitive capacity that functions to distinguish the grammatical from the ungrammatical. That is to say, the very existence of linguistic intuitions, and generally high levels of agreement between speakers, seem good a priori reasons to suppose that humans must have some species-universal sense of what is grammatical and what is not. Indeed this much can seem almost common sense. I call it the grammaticalness assumption. It is sometimes stated explicitly — “The only thing we can say directly is that the speaker has an ‘intuitive sense of grammaticalness’” (Chomsky, 1975, p.95) — but more often the grammaticalness assumption goes unsaid: so much so that linguistic intuitions are sometimes called grammatical intuitions. Moreover, if not all then certainly the overwhelming majority of past discussion about the nature of linguistic intuition seems to take the grammaticalness assumption as a tacit background fact. One example is debate and research about whether linguistic intuitions are affected by individuals’ own expert knowledge or prior assumptions about language (e.g. Devitt, 2006; 2010; Culbertson & Gross, 2009; Dąbrowska, 2010; Fitzgerald, 2010; Maynes & Gross, 2013; Drożdżowicz, 2018; Rey, 2020). Another example is discussion of how linguistic intuitions are affected by the parsability of sentences (e.g. Leivada & Westergaard, 2020).

Yet from an evolutionary and, in particular, adaptationist perspective, the grammaticalness assumption is hard to justify. What ecological conditions would select, not simply for a sensitivity to what is and is not grammatical, but for a cognitive capacity that delivers an intuitive association between ungrammaticality, on the one hand, and a psychological sense of oddness on the other? To what fitness enhancing task would such a capacity contribute? I am not asking how these cognitive dispositions might emerge in our species (a question about phylogeny), I am asking why our species should have such dispositions in the first place (a question about function, or adaptation). As far as I am aware, no detailed, focused or theoretically principled answer to this question has ever been developed.

Chomsky has speculated that perhaps an incidental biological mutation simply happened to generate the cognitive capacities and dispositions necessary for language (“Perhaps it was a side effect of increased brain size… or perhaps some chance mutation”: 2010, p.59), and perhaps these could include, by chance, an intuitive association between ungrammaticality and a psychological sense of oddness. If such speculation is correct then the intuitive association between ungrammaticality and a psychological sense of oddness is pure happenstance and has no theoretically principled explanation. This speculation is widely criticized on grounds of evolutionary implausibility (Pinker & Bloom, 1990; Chater et al., 2009; Planer, 2017; Hurford, 2018; Martins & Boeckx, 2019; de Boer et al., 2020) — but at the same time, it is not as if there are other possible justifications of the grammaticalness assumption with greater plausibility. The grammaticalness assumption has rather been taken for granted, by Chomskyians and non-Chomskyians alike, and put to use as a foundational assumption in the investigation of grammar itself.

Here I present and develop a cognitive and adaptationist approach to the question of why humans have linguistic intuition, grounded in theoretical and empirical knowledge from several fields, in particular cognitive linguistics, the cognitive psychology of human communication, and adaptationist approaches to the human mind. I shall argue that linguistic intuitions are byproducts of cognitive capacities and dispositions that have as their proper biological and computational function the interpretation of communicative stimuli. One important corollary of my analysis will be that the grammaticalness assumption is both unnecessary and unwarranted.

Here is the main idea. Research on the cognitive foundations of human communication has uncovered how all stimuli perceived as communicative are interpreted on the basis of an unconscious presumption of optimal relevance. I shall describe linguistic intuitions as byproduct effects of this presumption. Specifically, I shall argue that linguistic intuitions are triggered by, and only by, sentences that logically cannot have the property of optimal relevance. It is this logical impossibility that triggers the psychological sense of oddness. I call this the byproduct hypothesis of linguistic intuition.

For clarity: the byproduct hypothesis is not that ‘being irrelevant’ triggers intuitions of unacceptability. That cannot be right. Saying something utterly irrelevant in a given context can be unusual, even strange, but it does not entail sentence unacceptability. The byproduct hypothesis is rather that linguistic intuitions are caused by sentences that logically cannot have the property of optimal relevance. Put simply, the psychological sense of oddness is not caused by irrelevance, it is caused by a logical impossibility of optimal relevance. These two qualities are not the same. I will elaborate further on this contrast in §4.3.

I am aware that the byproduct hypothesis violates apparent common sense. It states, in effect, that the apparently plain observation that linguistic intuitions are intuitions about grammar, or grammaticalness, is in fact misleading. It sometimes happens that new scientific understanding upturns what previously seemed obvious and uncontroversial. I believe that is the case here.

Figure 1 summarises the content and structure of the paper. In §1 I specify two key assumptions that underpin the byproduct hypothesis. Both are longstanding, influential and supported by very large bodies of empirical and theoretical research, which I shall briefly summarise. The first assumption is that constructions — mappings between particular forms and particular communicative functions, or “meanings” — are basic building blocks for languages, and that languages can be described wholly in these terms (“It’s constructions all the way down”: Goldberg, 2003, p.223). The second assumption comes from the Relevance Theory tradition in the cognitive psychology of human communication. Specifically, I assume that humans have core cognitive capacities for communication which include, in particular, a disposition and capacity to interpret communicative stimuli on the basis of a presumption of optimal relevance. In §2 I summarize how these two assumptions together motivate the byproduct hypothesis of linguistic intuition.

Figure 1. Figure 1. Structure of the paper.

In §3 I motivate and clarify the hypothesis further, by highlighting a close parallel between linguistic intuitions and the psychological sense of oddness that accompanies so-called ‘impossible objects’: visual illusions in which stimuli appear as objects but which are logically impossible to interpret in any way consistent with the unconscious presumptions that govern the interpretation of all stimuli perceived as objects. This parallel helps to make clear how linguistic intuitions are, at bottom, not properly about grammaticalness at all: they are rather about communicativeness (more technically, ‘ostensiveness’). In §4 I describe how the byproduct hypothesis predicts linguistic intuitions. Specifically, I derive three general ways in which sentences logically cannot have the property of optimal relevance, and I describe how each triggers intuitions of unacceptability. In §5 I show how the byproduct hypothesis explains four otherwise surprising features of linguistic intuition. I conclude by placing the byproduct hypothesis in a broader cognitive and evolutionary context, as part of the deeply social nature of human minds.

A note on terminology. It is important to distinguish acceptability from grammaticality. In the literature on linguistic intuitions these two terms are used in a range of ways, some contrary to one another. Following some others (e.g. Bard et al., 1996; Hornstein, 2013; Leivada & Westergaard, 2020), I shall use (un)acceptability to refer to the spontaneous intuitions (‘linguistic intuitions’) that people have about sentences, and (un)grammaticality to refer to the relationship between a sentence and the ordinary practices (‘norms’, ‘conventions’) of a language community. These two qualities overlap to a very considerable degree — most sentences are either acceptable and grammatical, or unacceptable and ungrammatical — but they are not the same. Indeed they can sometimes dissociate: there are unacceptable-grammatical and acceptable-ungrammatical sentences. I shall give examples, and explain when and why this dissociation happens, in §5.2. Before then, in §1-§4, my target is the intuitions themselves. People have them and can report on them. But why do humans have them in the first place?

1. Two assumptions

1.1. Constructions all the way down

The first foundational assumption here is that grammars are best described and understood in terms of constructions: mappings between particular forms and particular communicative (micro-)functions, or “meanings” (e.g. Langacker, 1987; Fillmore, 1988; Fillmore et al., 1988; Michaelis & Lambrecht, 1996; Kay & Fillmore, 1999; Croft, 2001; Tomasello, 2003; Goldberg, 2003; 2019; Michaelis, 2006; Bybee, 2013;

Figure 2. Table 1. Examples of constructions, at varying levels of complexity & abstraction. Adapted from (Goldberg, 2013a). Sentences are assemblies of constructions. For example, the sentence The boy bakes his sister a cake assembles together: ditransitive construction; V-s; two objects; and the lexical constructions The, boy, bake, his, sister, a and cake.

Each construction serves its own communicative function, or ‘meaning’. For instance, ‘Kick the bucket’ has the communicative function to raise the concept of death in a mildly light-hearted way. ‘V-ed’ has the communicative function to express the past-ness of an action. ‘What’s X doing Y?’ has the function to request resolution of some perceived incongruity. The constructions in the last four rows of Table 1 all have information packaging functions i.e. they all package or arrange information for particular discourse goals, such as emphasis, questioning, and placing information into the foreground or background of discourse. Languages are vast arrays of constructions, each one with a different communicative function.

Over the past four decades this constructionist approach to describing and analysing grammars has advanced very considerably, with many empirical successes, resolving issues and advancing understanding at all levels of linguistic analysis. This is not to say that “constructions all the way down” is a universally accepted assumption, but it does have a proven track record and supports a very substantial body of knowledge (see citations above).

Two general aspects of constructionist approaches are particularly important for the byproduct hypothesis of linguistic intuition.

First, no two constructions with the same form can have the same communicative function. This is commonly known as the Principle Of No Synonymy. The idea that there are no true synonyms in languages has a long history and is certainly not unique to constructionist approaches, but it does play an especially foundational role in this literature (Bolinger, 1968; Clark, 1987; Goldberg, 1995; Leclercq & Morin, 2023; inter alia). Another way to present it is to say that every construction ‘covers’ a specific range of possible communicative functions, or meanings, and no two constructions have precisely the same coverage (Goldberg, 2019). For instance, while subway, metro and underground are not different in their denotation (they all describe subterranean railway systems) they nevertheless all have different coverage because they vary in their connotations or their sociolinguistics: each is the more common term in different parts of the world. As such, they do not have the exact same communicative functions.

Second, constructionist approaches have long aimed to explain grammatical phenomena in terms of the functions of the constructions involved. As one simple example, consider the different syntax and semantics of cardinal one and anaphoric one (Goldberg & Michaelis, 2017). Anaphora are expressions whose interpretation depends on other expressions used in the discourse: the word “one” earlier in this paragraph (“As one simple example…”) is an example, where it refers to explanations of grammatical phenomena in terms of the functions of the constructions involved. The key grammatical phenomenon here is that cardinal one and anaphoric one have different syntax and semantics. In particular, cardinal one, and not anaphoric one, receives emphasis and asserts the quantity one. Contrast “She left one behind [rather than two]” (cardinal) with “She left one behind [where one refers to something already present in the discourse]” (anaphoric). Or contrast “It was one of them [and not two of them]” (cardinal) with “It was one of them” (anaphoric). This difference occurs because these are two constructions with the same form but different functions. Specifically, while cardinal one has the function to assert quantity, anaphoric one has the function to assert existence or quality. So the grammatical phenomenon (syntactic and semantic contrasts between anaphoric one and cardinal one) is explained in terms of the functions of different constructions (asserting quantity vs asserting existence). This is just a simple example, of course. Other examples involve far more complexity. The important point here is just that over the past four decades this ‘functions of constructions’ approach has been successfully used to explain a wide and diverse array of grammatical phenomena in an especially parsimonious way (Croft, 2001; Goldberg & Suttle, 2010; Sag, 2010; Auer & Pfänder, 2011; Hilpert, 2014; Goldberg, 2016; Abeillé et al., 2020; Hoffmann, 2022; inter alia).

The analysis of linguistic intuitions that I present in this paper assumes that constructionist approaches are on the right track i.e. that grammars are indeed best described and understood in terms of constructions, and that grammatical phenomena can and are best explained in terms of the functions of constructions. At the same time, this paper also supports and reinforces the constructionist agenda to the extent that its conclusions provide a simple and distinctive explanation of an important empirical phenomenon. It is a normal part of scientific progress, that a premise (or set of premises) are supported to the extent that, by accepting those premises, we gain better explanations than before, where ‘better’ is understood in terms of ordinary scientific desiderata such as simplicity, plausibility, parsimony, consistency with a wide range of evidence and so on. For example, Newton’s laws of motion were

supported by the fact that they provided for better explanations of basic empirical phenomena such as the movements of celestial objects. In our case, the basic empirical phenomenon is the existence of linguistic intuition, including the psychological sense of oddness. In §2-§5 I will present what I believe is an especially parsimonious explanation of why humans have linguistic intuitions, at the computational level. As such, these arguments build on, enrich and support the constructionist agenda.

To do this, constructionist approaches must be paired with a key insight from the cognitive psychology of human communication (see also Scott-Phillips, 2025).

1.2. Audience presumptions of optimal relevance

The second foundational assumption here is that all stimuli perceived as communicative are interpreted on the basis of a presumption of optimal relevance.

Any behavior humans can perform, they can perform in a communicative (more technically, ‘ostensive’) manner: overtly attracting attention and providing evidence of meaning. For instance, I sometimes tilt my coffee cup simply as a byproduct of moving my arms and wrist, but I can also tilt the coffee cup in a way that will attract the attention of a waiter and simultaneously be understood as a request for another coffee. Even the most basic human activities, such as walking or eating, can be done in communicative ways and hence indicate attitudes such as irony, sadness, respect or hostility. Producing language — i.e. assembling constructions — is, of course, one especially important way of being communicative.

How are communicative behaviors, linguistic or otherwise, understood at all? Human communication is flexible and open-ended, and the interpretation of communicative behaviour, including language use, is always intersubjective and context dependent (Sperber, 1995; Carston, 2002; Recanati, 2004; Verhagen, 2005; 2015; Langacker, 2008; Zlatev et al., 2008; Ludlow, 2014; Assimakopoulos, 2017; Geeraerts, 2021; Tantucci, 2021; inter alia). To take a very simple example, consider the simple utterance “We’re on time”. First, there can be no question that the utterance must be interpreted in a context dependent way, because it uses a pronoun (“We”) and so who the utterance actually refers to depends on who the utterer is. Second, the expression “..on time” can be used in many ways, for instance to mean “..not late even

though we expected we would be”, “Don’t worry, everything is fine”, “Oh, I read my watch wrong, actually it’s 12 o’clock”, “..literally standing on a clock”, and so on. The point here is that language use is always and inherently open-ended, flexible and context dependent, which in turn raises the question of how — how just possibly? — do we ever understand one another? How can we possibly converge on a (more or less) accurate interpretation of any utterance? Without some answer to such questions, the very possibility of human communication is quite mysterious.

The past four decades of research in the cognitive psychology of human communication have uncovered a general answer. Stimuli perceived as communicative are interpreted on the basis of a presumption of optimal relevance. Here is a simple example, to illustrate the basic idea (adapted from Sperber & Wilson, 1986/1995, p.51-2). A holidaymaker in a foreign country sets off for a walk. The weather is good, with only a few light clouds in the distance. However, as they leave the hotel grounds, a local makes eye contact, points towards the clouds and shakes their head. The holidaymakers understand this as a warning: they infer that local knowledge is that the clouds that are presently distant are probably rain clouds headed this way. How and why did they hit on this understanding? How just possibly? How does a point and a shake of the head mean “Those are probably rain clouds”? The answer is that, having recognised the local’s behaviour as communicative, the holidaymakers’ cognitive systems determine its meaning by, in effect, presuming that the behaviour has the property of optimal relevance, given the local’s abilities and knowledge. With this presumption — only with this presumption — the holidaymakers’ cognitive systems are able to reverse engineer the local’s intended meaning.

These cognitive capacities and dispositions on the audience side marry with cognitive capacities and dispositions on the communicator side. Audiences interpret stimuli perceived as communicative on the basis of a presumption of optimal relevance; and communicators produce stimuli that, if they are interpreted on on the basis of a presumption of optimal relevance, are indeed likely to generate interpretations that converge on their (the communicator’s) intended meaning. This pair of insights together form the Communicative Principle of Relevance, that is central to Relevance Theory (Sperber & Wilson, 2002; Scott-Phillips, 2024).

Informally, ‘optimally relevant’ means ‘efficient use of cognitive resources’. More formally, the relevance of a stimulus is the trade off between the cognitive costs and the cognitive benefits created by attending to and processing the stimulus; and stimuli are optimally relevant if and only if neither costs not benefits can be improved without making the other worse off (Sperber & Wilson, 2002; Scott-Phillips, 2025). Cognitive costs are, in the most general sense, the opportunity costs of attention; and in the specific context of communication this effectively means audience processing costs. Cognitive benefits are, in the most general sense, the impact that attention has on future decision making; and in the specific context of communication this effectively means accurate enough identification of the communicator’s intended meaning. Putting all this together, the Communicative Principle of Relevance implies that when interpreting communicative stimuli, audiences presume that no alternative stimulus could suggest the same (or a very similar) meaning at lower processing cost for the audience. That is not to say that speakers are always ‘efficient’, whatever that might mean. The point is rather that communicative stimuli are (always, unconsciously) interpreted as if they have the property of optimal relevance.

A common question is, “What if communicative stimuli are not optimally relevant?”. There are two misunderstandings here. The first is that the question treats relevance as a feature of a stimulus itself. But relevance is not a feature of a stimulus, it is a feature of the relationship between a stimulus and an individual mind, at a particular moment in time. What is relevant to one mind at one given moment may not be relevant to a different mind at the same moment, or the same mind at a different moment. As such, no stimulus can ever ‘be’ optimally relevant in and of itself. The second, related misunderstanding is that, even if it were coherent to ask ‘whether’ a stimulus is optimally relevant, it would not matter. The key point is that audiences interpret communicative stimuli as if they have the property of optimal relevance (for the mind of the target audience, at that moment), regardless of what features any particular stimulus actually has.

However, there are some minimal general statements we can make about optimal relevance, or reasonably assume. For instance, we can assume that stimuli that follow conventional use entail less attentional resources than stimuli that deviate from conventional use. Hence, any stimulus that deviates from conventional use without any corresponding change in cognitive benefits cannot possibly have the property of optimal relevance. I will use this minimal assumption in §4.

The Communicative Principle of Relevance was first proposed in the 1980s and since then many research programs in experimental psychology, with both adult and infant participants, have reported results that strongly align with it (e.g. van der Henst & Sperber, 2004; Schwier et al., 2006; Gibbs Jr & Bryant, 2008; Scott-Phillips et al., 2009; Behne et al., 2012; Király et al., 2013; McEllin et al., 2018a,b; Tauzin & Gergely, 2018; Ho et al., 2021; Rubio-Fernandez et al., 2021; Vesper et al., 2021; Royka et al., 2022; Jara-Ettinger & Rubio-Fernandez, 2024). Some of these studies use especially fine-grained behavioural measures, and hence show how people follow the predictions of the Communicative Principle of Relevance in microscopically precise ways, on both the communicator and the audience side (Scott-Phillips, 2024). The principle has also been used to explain or considerably deepen our understanding of a large and diverse range of semantic, pragmatic and other expressive phenomena (e.g. Sperber et al., 1995; Blakemore, 2002; Nicolle, 1997; 1998; Gibbs & Tendahl, 2006; Wilson & Carston, 2007; Sperber, 2010; Sasamoto, 2012; 2019; Wilson & Sperber, 2012; Yus, 2016; Cave & Wilson, 2018; McCallum et al., 2020; Ifantidou et al., 2021; Scott, 2022; Wharton et al., 2022; Heintz & Scott-Phillips, 2023; Mazzarella & Pouscoulous, 2023). In this article I am, in effect, extending the reach of the Communicative Principle of Relevance into a new domain, by using it to account for the existence and nature of linguistic intuition.

2. The byproduct hypothesis of linguistic intuition

Both assumptions summarised in the previous section are longstanding and influential, but until very recently they have not been viewed in conjunction with one another. The idea that constructionist approaches to grammar should be closely connected with the cognitive psychology of human communication is not new (e.g. Tomasello, 1998a,b; Beckner et al., 2009; Croft, 2009; Enfield & Sidnell, 2014; Geeraerts, 2016; Schmidt, 2016), but only recently have constructionist approaches been specifically connected with the insight that audiences interpret communicative stimuli on the basis of a presumption of optimal relevance (Leclerq, 2023; Scott-Phillips, 2025; in press). Here I state how this conjunction in turn suggests the byproduct hypothesis of linguistic intuition.

When humans communicate they can make use of whatever possible communicative tools may be available. These tools include their own bodies; objects available in the immediate environment; and, critically, any communicative conventions they might share with the target audience. (Note: I use ‘convention’ to mean ‘commonly known solution to a recurrent coordination problem’. This use does not include notions such as ‘arbitrariness’. For discussion see Scott-Phillips, 2025, §3.1.) These communicative conventions can be non-linguistic, such as pointing or nodding, but if communicator and audience share a language then the communicative conventions they have available includes, by definition, a vast array of constructions (Scott-Phillips, 2025). To share a language is to have shared knowledge of a vast array of socio-cognitive conventions, called constructions, that are used for communicative ends. So when audiences interpret a sentence, interpretation proceeds on the basis of a presumption that whatever the intended meaning might be, no alternative assembly of constructions could suggest the same (or an extremely similar) meaning at lower cognitive cost.

But what happens when a cognitive process that has the function to interpret incoming stimuli as if they have a certain property perceives that a stimulus is assembled in such a way that it is logically impossible for the stimulus to ever have that property? As I mentioned in the Introduction, this is not a question about what happens when people say strange or irrelevant things, or use far more words than they might otherwise need to. Such behaviours may be unusual or frustrating, but they do not entail a logical impossibility of optimal relevance. The question is about what happens when there is indeed a logical impossibility of optimal relevance. The byproduct hypothesis is, in effect, an answer to this question. It states that what happens is a psychological sense of oddness.

3. Unacceptable sentences and impossible objects: A close parallel

To help motivate the byproduct hypothesis further, and to highlight what it may share with other domains of cognition, here I describe how the byproduct hypothesis is closely parallel to what happens with so-called impossible objects (Figure 2).

Figure 3. Figure 2. Impossible trident; Penrose triangle; Shepard elephant (all three images are in the public domain). In each image, individual parts are straightforwardly interpretable on their own, but they clash with other parts. This in turn makes it logically impossible to interpret the stimulus in any way consistent with the presumption of physical cohesiveness that governs the interpretation of all stimuli perceived as objects.

Whenever stimuli are perceived as objects, they are interpreted on the basis of a presumption that they have certain properties (Spelke, 1990; 2022; Baillargeon, 2004; Carey, 2009; Bai et al., accepted). These properties include, at a minimum, physical cohesiveness, subjection to gravity, bondedness and rigidity. Thus, when stimuli appear to be objects but are also assembled in ways that make it logically impossible for the stimulus to ever have these properties, we spontaneously intuit that something is not right. This is specifically what happens with impossible objects. The stimuli in Figure 2 appear to be objects and yet any interpretation of them as objects violates the presumptions that govern the interpretation of all stimuli perceived as objects. For example, it is logically impossible for the Shepard elephant to be physically cohesive, because its body and its legs are in mutual contradiction in this respect.

Both object perception and social interaction are domains of high adaptive importance for humans, and there is, I am suggesting, a tight parallel here (Table 2). In the case of impossible objects, when the different parts are assembled in mutually contradictory ways, it becomes logically impossible for the image to be interpretable in any way concordant with the presumption of physical cohesiveness that underpins the interpretation of all stimuli perceived as objects. In the case of many unacceptable sentences, when the different parts — i.e. the constructions — are assembled in mutually contradictory ways, then it becomes logically impossible to interpret the sentence in any way concordant with the presumption of optimal relevance that underpins the interpretation of all stimuli perceived as communicative. (I elaborate exactly on how sentences can have this property in §4.3, below.)

Figure 4. Table 2. Unacceptable sentences are to communication what impossible objects are to perception. They are logically uninterpretable in ways consistent with the spontaneous and unconscious assumptions that govern and guide the respective cognitive process.

This parallel helps to highlight and clarify several subtle but important aspects of the byproduct hypothesis.

First, it provides the means for a more informal and intuitive presentation of the main idea. The byproduct hypothesis states that, just as the physical features of impossible objects effectively ‘pull’ interpretation in two (or more) mutually contradictory directions with respect to physical cohesiveness, the functions of the constructions of many unacceptable sentences effectively ‘pull’ interpretation in two (or more) mutually contradictory directions with respect to optimal relevance (and I shall describe when and how this can occur in the next section). This creates, in both cases, an unresolvable tension, the consequence of which is a psychological sense of oddness.

Second, the parallel with impossible objects highlights an important distinction between what the intuitions are about and how they are generated. In the case of impossible objects, what the intuitions are about is always objectness. That is to say, the intuitions are about whether the stimulus can possibly be concordant with the presumptions that govern the interpretation of all stimuli perceived as objects. In the case of unacceptable sentences, what the intuitions are about is always communicativeness (and not grammaticalness). That is to say, the intuitions are about whether the stimulus can possibly be concordant with the presumption of optimal relevance that governs the interpretation of all stimuli perceived as communicative.

Third, the parallel with impossible objects aids specification and clarification of the relationship between processes of interpretation and the properties of perceived stimuli. When we perceive stimuli as objects, we do not ‘track’ or ‘monitor’ their bondedness, physical cohesiveness or other aspects of objectness. Rather, we interpret the stimuli as if it has those properties (or put in other words: on the basis of a presumption that it has those properties). Similarly, when we perceive stimuli as communicative, we do not ‘track’ or ‘monitor’ their relevance, let alone their optimal relevance. Indeed, research in the Relevance Theory tradition has long rejected any idea that humans mentally represent the relevance of incoming stimuli (e.g. Sperber, 2005; Allott, 2013; Sperber & Wilson, 2025). Rather, we interpret communicative stimuli as if they have the property of optimal relevance (or put in other words: on the basis of a presumption that it has the property of optimal relevance). In both domains, a psychological sense of oddness occurs when stimuli cannot actually be interpreted in this ‘as if ’ way. As such, no mental representation, either of an object’s degree of physical cohesiveness or a communicative stimulus’ degree of relevance, is necessary for the stimulus to be intuitively perceived as ‘not right’.

4. Optimal relevance as logically impossible: Three ways

Here I use the definition of relevance, as developed in Relevance Theory (Sperber & Wilson, 1986/1995; 2002; Clark, 2013; Scott-Phillips, 2024), to derive three general ways in which a sentence logically cannot have the property of optimal relevance; and I describe how each of these three ways triggers intuitions of unacceptability.

As summarised in §1.2, relevance is a trade off between two metrics, cognitive costs and cognitive benefits on the audience side, and stimuli are optimally relevant if and only if neither of these metrics can be improved without making the other worse off. Therefore, a sentence logically cannot have the property of optimal relevance if:

● The sentence appears to have no plausible cognitive benefits in the first place (i.e. no meaning can be determined), such that there is no possible trade off of costs and benefits (i.e. no relevance).

● The sentence deviates from conventional use without any plausible change in interpretation, however small or nuanced. Such sentences raise the cognitive costs of interpretation with no plausible change in benefits. (Again, I assume that stimuli that follow conventional use entail less attentional resources than stimuli that deviate from conventional use.)

● There are mutual contradictions between the functions of two (or more) constructions within a sentence, rendering the optimisation of cognitive costs and cognitive benefits impossible. (This is the case that is closely parallel to the case of impossible objects, as described in §3 above.)

In this section I elaborate on how each of these ways can trigger the psychological sense of oddness that is the signature of linguistic intuition. I give simple examples, and I relate them to established findings in the experimental literature.

4.1. No plausible cognitive benefi ts

The most straightforward way by which a sentence logically cannot have the property of optimal relevance is if the sentence is not parseable, and therefore appears to have no plausible cognitive benefits in the first place.

The clearest example of such sentences are those that overload short term memory or other cognitive processes involved in sentence processing; and the classic class of sentences that do this are those with multiple embedded clauses:

(1) The rat the cat the dog chased killed ate the malt.

The dog chased the cat. The cat killed the rat. The rat ate the malt.

(2) The patient the nurse the clinic had hired admitted met Jack.

The clinic had hired the nurse. The nurse admitted the patient. The patient met Jack.

(3) That that that Bill left Mary amused Sam is interesting is sad.

Bill left Mary. The fact that Bill left Mary amused Sam. The fact that this amused Sam is interesting. The fact that this is interesting is sad.

These sentences all follow the grammar of English. In particular, they all use embedded clauses in ordinary and conventional ways. However, they also embed multiple clauses inside one another. This can overload short term memory or other aspects of processing, making the sentences hard or impossible to parse, and hence without cognitive benefit for the audience. Prosody and other aspects of language production can help to demarcate embedded clauses or other aspects of sentence structure, and hence reduce these impacts on language processing (e.g. Fodor et al., 2017), but if these impacts are present then cognitive benefits are reduced to zero or near zero, hence making it logically impossible for the sentence to have the property of optimal relevance.

Of course, it is not controversial that limits on short term memory and other processing factors can impact on judgements of acceptability. This has long been known, and investigated in detail (e.g. Miller, 1962; Gibson & Thomas, 1999; Phillips, 2006; Karlsson, 2007; Sprouse, 2013; inter alia). The important point here is just that this fact is consistent with — in fact it is predicted by — the byproduct hypothesis of linguistic intuition. Overloading cognitive processing renders it difficult or impossible to identify and track relations within a sentence; so such sentences entail cognitive costs for no real cognitive benefit; which in turn means that the sentence logically cannot optimise the trade off between cognitive costs and cognitive benefits. This impossibility triggers the psychological sense of oddness.

4.2. Deviation from conventional use without any plausible change in interpretation

The second way in which a sentence logically cannot have the property of optimal relevance is if it deviates from conventional use without any plausible change in interpretation, and hence no plausible change in cognitive benefits. Two simple examples were given in the Introduction:

(4) She gived me a book

Rather than “She gave me a book”

(5) I don’t want going to the cinema

Rather than “I don’t want to go to the cinema”

In these cases, the sentence uses one or more constructions in ways that deviate from conventional use, and as such the sentence entails cognitive costs beyond those entailed by conventional use. (As I mentioned in §1.2, I assume, other things equal, that stimuli that follow conventional use entail less attentional resources than stimuli that deviate from conventional use.) At the same time, these deviations from conventional use appear to offer just the same cognitive benefits i.e. the same meaning, including the same pragmatics, as following conventional use would have done. Together, these two properties (additional costs; no change in benefits) mean that these sentences logically cannot optimise the trade off between cognitive costs and cognitive benefits.

As with §4.1, there is no controversy that sentences such as (4) and (5) are routinely judged as unacceptable. Deviations from the ordinary practices of a language community impacts on judgements of acceptability. But also as with §4.1, the important point here is just that this fact is consistent with, indeed it is predicted by, the byproduct hypothesis of linguistic intuition.

Of course, some sentences deviate from conventional use to some degree but, by doing so, they warrant meanings (again, including pragmatics) that conventional use would not have done. Here are three simple, imagined examples:

(4) She smiled him a thank you

(5) Can I ninja your spot in line?

(6) They meandered the conversation to a weird place

This kind of productive deviation from conventional use is plainly important for linguistic creativity and language change (Kaschak & Glenberg, 2000; Robenalt & Goldberg, 2015; Turner, 2018; Goldberg, 2019; Hoffmann, 2019; Bergs, 2025; inter alia). We can reasonably say that such sentences incur increased cognitive costs relative to more conventional use (§1.2), but they also warrant interpretations that conventional use would not have done i.e. they have distinctive cognitive benefits; and as such it is not the case that they logically cannot have the property of optimal relevance. Hence, we cannot and should not predict that these sentences will necessarily trigger intuitions of unacceptability. They might or might not, depending on the specifics of each case. (A further complicating factor here is that many individuals’ self-reported intuitions about linguistic creativity may be affected by their own ideas about how language does or should work. For instance, self-identified prescriptivists may be more likely to say that such sentences are unacceptable purely by virtue of their unconventionality.)

4.3. Mutual contradictions in the functions of constructions

The third possible way in which a sentence logically cannot have the property of optimal relevance is if there are mutual contradictions in the functions of two (or more) constructions within a sentence, rendering the optimisation of cognitive costs and cognitive benefits logically impossible.

Information packaging constructions, and in particular the contrast between foregrounding and backgrounding, provide the clearest class of examples. Many information packaging constructions have the function to place information in the background of discourse, while others have the function to place information in the foreground (Table 1); so if these two functions are applied to the same information, they will inevitably be in contradiction. Information cannot be both foregrounded and backgrounded at the same time, and as such any assembly of constructions that applies both functions to the same information will be logically impossible to interpret in terms of optimal relevance.

As a simple example, consider the sentence, “The woman who called Uber for a ride to the restaurant lost her glasses”. It has a main clause that places certain information in the foreground, highlighted here in capital letters:

(9) THE WOMAN who called Uber for a ride to the restaurant LOST HER GLASSES

Below is the same clause but now edited with the addition of a wh-question construction, used to ask about information that is foregrounded by the main clause. Bold has been added to indicate what the wh-question places into the foreground.

(10) What did THE WOMAN who called Uber for a ride to the restaurant LOSE? (A: HER GLASSES)

Now consider how the same wh-question construction might be used to ask about information that is backgrounded by the main clause. Again, bold indicates the information that the wh-question places into the foreground.

(11) Who did THE WOMAN call for a ride to the restaurant LOST HER GLASSES? (A: Uber)

Plainly, while sentence (10) is acceptable, sentence (11) is not. Ex hypothesi, this is because in (11), and not (10), two of the constructions within the sentence have mutually contradictory functions. They are ‘trying’ to achieve two contradictory things: the main clause (in CAPITALS) is trying to foreground the fact that the woman lost her glasses, and by doing so it must background who she called for a ride; while the wh-question (in bold) is trying to foreground who she called for a ride, and by doing so it must background the fact that she lost her glasses. So there is an unresolvable tension between the functions of these two constructions. The tension can be seen visually by the fact that, in (11), “call for a ride to the restaurant” is in bold but not capitals, and “LOST HER GLASSES” is in capitals but not bold.

This is just a simple example, to illustrate. Phenomena of this kind have long been known about: they are especially important for so-called ‘island effects’ in syntax, and other forms of long distance dependency. Constructionist approaches, aiming to explain grammatical phenomena in terms of the functions of constructions (§1.1) make a clear, precise and otherwise surprising empirical prediction in this area: that intuitions of unacceptability should occur when, and only when, information packaging constructions place into the background information that is otherwise in the foreground, or vice versa (Cuneo & Goldberg, 2023; inter alia). Furthermore, since foregroundness and backgroundness are graded qualities (they are matters of more-or-less rather than yes-no) then the prediction is for a correlation: measures of the degree to which backgroundedness and foregroundedness clash within a sentence should correlate with measures of the degree to which that sentence is judged unacceptable. Recent large scale, preregistered experimental studies have shown this effect decisively (ibid.). Baseline sentences of many different kinds were repackaged by information packaging constructions and participants were asked to judge the acceptability of the repackaged sentences. (The example sentences above are taken from the stimuli used in this study.) Results followed the prediction closely: the degree to which the repackaged sentences were judged unacceptable strongly correlated with the degree to which information that was foregrounded in the baseline sentence was backgrounded by the information packaging construction, or vice versa. Several other studies pursuing the same line of argument report complementary results (e.g. Ambridge & Goldberg, 2008; Goldberg, 2013b; 2016; Abeillé et al., 2020; Liu et al., 2022; Winckel et al., 2025).

What I am adding here is a cognitive explanation of why an unresolvable tension in the function of constructions should lead to a psychological sense of oddness at all. It is because unresolvable tensions make it logically impossible for a sentence to have the property of optimal relevance. In the example of sentence (11) above, if we entertain the interpretation that the communicator is asking about who was called by the woman who lost her glasses (Uber), then the main clause is unnecessarily costly in terms of cognitive resources, because it focuses attention on something different (THE WOMAN LOST HER GLASSES). Alternatively, if we entertain the interpretation that the communicator is asking about what was lost by the woman who called Uber (HER GLASSES), then the choice of wh-question is unnecessarily costly in terms of cognitive resources, because again it focuses attention on something different (Who… call[ed] for a ride to the restaurant?). In such ways, inherent tensions between the functions of constructions will make it logically impossible for a sentence to have the property of optimal relevance. This is just as is the case for impossible objects, where inherent tensions between different features of the objects make it logically impossible for them to have the property of physical cohesiveness (§3).

Note that this kind of explanation does not work for sentences that are plain irrelevant, in the everyday sense of that word. I mentioned in the Introduction that ‘irrelevant’ and ‘logically impossibility of optimal relevance’ are not the same, and it is only the latter that triggers intuitions of unacceptability. Consider a sentence of absurd length, packed full of apparently unnecessary information. Say, for instance, that sentence (9) is enriched so that the relative clause includes enormous amounts of additional information about the woman.

(12) THE WOMAN who called Uber for a ride to the restaurant and was especially looking forward to meeting her friend who had just come back from a long holiday where she learned how to pilot a small green plane and… [continue like this for much longer]… LOST HER GLASSES

Suppose, furthermore, that this is uttered for no apparent reason, perhaps even by a stranger in the street. Such behaviour is socially unusual and the sentence is irrelevant in the everyday sense of the term. However, there is nothing about the sentence that makes it logically impossible for it to have the property of optimal relevance. The apparent meaning is unnecessarily rich, maybe even absurdly rich, but it does not involve the logical impossibility of optimal relevance, and hence does not trigger the psychological sense of oddness.

5. Explaining signature features of linguistic intuition

Linguistic intuition has some distinctive and intriguing features. Four of the most salient are:

● Linguistic intuitions can be graded rather than discrete

● Acceptability can dissociate from grammaticality

● Utility in communication does not predict unacceptability

● Linguistic intuitions are independent of the social situation

Any good explanation of why humans have linguistic intuition should be able to account for these features. On the grammaticalness assumption, the first two are at least somewhat surprising, or noteworthy. They have, accordingly, been the focus of enormous amounts of investigation over the past 50 or so years. The second two have been sometimes used as arguments against social and communicative approaches to grammar.

In this section I describe how all four of these features are corollaries of the byproduct hypothesis of linguistic intuition. I say they are corollaries because to explain them I will appeal to no ad hoc hypotheses. Rather, these features of linguistic intuition are all straightforward consequences of the byproduct hypothesis. Specifically, they are all downstream of the three general ways, identified in §4, in which a sentence logically cannot have the property of optimal relevance.

5.1. Linguistic intuitions can be graded rather than discrete

Linguistic intuitions are sometimes matters of more-or-less rather than yes-or-no. This is so both at the level of individuals (‘I think this sentence is slightly more acceptable than that one’) and the level of populations (‘More people think this sentence is acceptable than that one’). Under the grammaticalness assumption, this graded property is not necessarily expected, and hence is in need of explanation. Accordingly, it has long been an important topic in the study of grammar (e.g. Bard et al., 1996; Schütze, 1996/2016; Sprouse, 2007; Lau et al., 2017; Francis, 2021).

Under the byproduct hypothesis, the graded nature of linguistic intuition is both expected and predicted. This is because the critical qualities that determine whether a sentence triggers linguistic intuition (§4) are themselves graded. For instance, some sentences exceed limits on short term memory only in a marginal or borderline way (§4.1). Some deviate from ordinary use only to a marginal or partial degree (§4.2). And some constructions, especially those with information packaging functions, have functions that are contrary to other functions in the same sentence in partial but not absolute ways (§4.3). Indeed, I mentioned above recent experimental research that directly investigates how partial but not absolute clashes of function predict gradedness in linguistic intuitions (Cuneo & Goldberg, 2023). So these critical qualities are all graded, meaning that the sense of a logical impossibility of optimal relevance should itself be graded. Hence, we should expect linguistic intuitions to be graded.

None of this is to deny that many linguistic intuitions are indeed matters of yes-or-no. Whether a sentence logically can or cannot have the property of optimal relevance is often clear and unambiguous. The point is just that unclear, ambiguous and partial cases are possible, predicted and expected under the byproduct hypothesis.

5.2. Acceptability dissociates from grammaticality

I mentioned in the Introduction that acceptability and grammaticality are not the same property. Following others, I use (un)acceptability for the spontaneous intuitions that are the main focus of this article, and (un)grammaticality for the relationship between a sentence and the ordinary practices (‘norms’, ‘conventions’) of a language community. This terminology hence clearly distinguishes between two different empirical phenomena: individual intuition and community practice.

These two phenomena overlap to a very considerable degree: sentences are typically either acceptable and grammatical, or unacceptable and ungrammatical. However, acceptability and grammaticality can and do sometimes dissociate (Chomsky, 1965; Montalbetti, 1984; Frazier, 1985; Barton et al., 1987; Schütze, 1996/2016; Phillips et al., 2011; Hornstein, 2013; Wellwood et al., 2018; Leivada, 2020; Leivada & Westergaard, 2020; Tubau et al., 2020; inter alia). As examples, sentences (1)-(3) in §4.1 are all grammatical but are routinely judged as unacceptable. The opposite is also possible: sentences can be acceptable but ungrammatical (I give examples below).

Here I describe how the byproduct hypothesis predicts, explains and otherwise approaches the fact that acceptability and grammaticality can dissociate.

First, the byproduct hypothesis predicts that, while they are different phenomena, acceptability and grammaticality should nevertheless overlap to a very considerable degree. This is because each tends to entail the other. Following the ordinary practices of a language community (i.e. being grammatical) will typically keep cognitive costs at a minimum for whatever cognitive benefit might be achieved (i.e. it will entail acceptability). And vice versa: keeping cognitive costs at a minimum for whatever cognitive benefit might be achieved will, almost by definition, involve following the ordinary practices of a language community (modulo small deviations that can be used creatively: see §4.2). This mutual entailment is why acceptability and grammaticality are easily conflated.

When precisely should this mutual entailment dissolve? The most obvious case is when a sentence follows ordinary practice but also exceeds limits of short term memory or other processing constraints: this possibility was described in §4.1. Such sentences are trivially easy to create: just add sufficient layers of embedded clauses.

In the other direction, the byproduct hypothesis predicts that ungrammatical-acceptable sentences should be rare or perhaps impossible, and certainly difficult to generate ex nihilo. Significant deviations from ordinary practice should entail an inherent impossibility of interpretation consistent with the presumption of optimal relevance. And indeed this prediction appears to hold: it is very hard to write or identify new ungrammatical-acceptable sentences.

That said, there are a few apparently good examples of ungrammatical-acceptable sentences. (13) and (14) are probably the most robust and studied examples. Both initially strike many people as acceptable, but they are not grammatical. That is, they do not follow ordinary practices of English.

(13) The doctor the nurse the hospital had hired met John

The nurse the hospital had hired met John. So what did the doctor do?

(14) More people have been to Russia than I have

More people have been to Russia than I have what?

So while acceptable-ungrammatical sentences are certainly rare, (13) and (14) are an existence proof that they are possible. How can this be, if the prediction is that they should not occur? Our best experimental studies to date seem to suggest that these sentences have some idiosyncratic features that cause people to fail to identify their structure i.e. they do not identify the constructions accurately — and crucially, once people do notice the actual structure then they then readily judge the sentences as unacceptable (Wellwood et al., 2018). In other words, acceptable-ungrammatical sentences quickly become unacceptable-ungrammatical sentences on further reflection, as the byproduct hypothesis predicts. Further focused research is needed here, including in languages other than English.

5.3. Utility in communication does not predict unacceptability

In the Introduction I gave a simple example of how sentences can be unacceptable even if their meaning is apparently clear: I don’t want to go to the cinema is acceptable but I don’t want going to the cinema is not. Such pairs of sentences are sometimes called ‘Why Nots’ (because they invite the question, why is one of the sentences not acceptable?: Rey, 2020) or ‘Fine Thoughts’ (because they highlight sentences that express a clear thought even though they are unacceptable: Chomsky, 2013). Any syntax textbook provides many more examples. These have long been highlighted as critical data speaking against communicative approaches to grammar, because they show that possible utility in communication does not delimit the acceptable from the unacceptable. Sentences that can be useful in communication may or may not be acceptable.

However, while the byproduct hypothesis of linguistic intuition is predicted on a communicative approach to grammar (§1.1), it does not at all predict utility in communication should delimit the acceptable from the unacceptable. On the contrary, the hypothesis states that what delimits the acceptable from the unacceptable is something different: it is the logical impossibility of optimal relevance (§2). Thus, since ‘possible utility in communication’ and ‘logical impossibility of optimal relevance’ are not the same concept, the fact that utility in communication does not delimit the acceptable from the unacceptable is expected.

In fact, the byproduct hypothesis states the conditions under which sentences may be useful in communication but unacceptable: it is when ‘possible utility in communication’ is different to ‘impossibility of optimal relevance’. §4.2 and §4.3 identified and described the two general sets of circumstances in which this will happen (and §4.1 describes how a sentence can have neither quality).

5.4. Linguistic intuitions are independent of the social situation

One entailment of the byproduct hypothesis is that linguistic intuitions are determined in context. This is because relevance, and hence optimal relevance, is a relationship between a stimulus and an information processing system, such as a mind, at a given moment and hence in a given cognitive context (Sperber & Wilson, 1986/1995; Assimakopoulos, 2017; Wilson, 2023; Scott-Phillips, 2024). So if linguistic intuitions are intuitions ultimately about the logical impossibility of optimal relevance (§2) then linguistic intuitions must be determined in context.

This fact may seem to be a problem for the byproduct hypothesis, because describing a social situation in which a sentence might be uttered is unlikely to change intuitions about (un)acceptability. (Perhaps providing a social situation can change linguistic intuitions at the margins, but I assume it does not change linguistic intuitions in general.) So how can this be? How can it be that linguistic intuitions are determined in context, but changing the social situation does not change the intuitions?

There are (at least) two important points to make here. First, the fact that (i) linguistic intuitions are determined in context does not at all entail that (ii) changing the social situation should change the intuition. There is just no inherent entailment or contradiction here. Second, the byproduct hypothesis actually predicts that changing the social situation should not change linguistic intuitions at all, because it states that what triggers linguistic intuitions is the perception of a logical impossibility of optimal relevance i.e. an impossibility regardless of social situation. Again, in §4 I identified the three ways in which a sentence can have this property.

6. Conclusion: Ultimate and proximate explanations for linguistic intuition

One broader goal of this article is simply to raise the question of why humans have linguistic intuitions at all. The issue is basic for cognitive science yet it has not received the focused attention it deserves (see Introduction). Cogent answers will deepen, reinforce, reshape or in other ways considerably advance understanding in several areas.

Understanding of biological capacities or dispositions can be pursued at both ultimate (evolutionary) and proximate (mechanistic) levels (Mayr, 1961; Ariew, 2003; Scott-Phillips et al., 2011; Dickins & Barton, 2013; Nettle & Scott-Phillips, 2023; Al-Shawaf, 2024; inter alia). Ultimate explanations identify the adaptive function(s) of a trait, if any, and proximate explanations describe how that functionality is achieved. Put another way, ultimate explanations describe why a given trait exists in a species, and proximate explanations describe how it operates. The two levels are separate, and good explanations at each level will be mutually supportive of one another.

Here I have presented both ultimate and proximate explanations of linguistic intuition. Ultimately, humans have linguistic intuition because linguistic intuitions are byproduct effects of core cognitive capacities and dispositions for human (‘ostensive’) communication, that are themselves biological adaptations to the deeply social nature of humans’ evolutionary ecology (Sperber & Wilson, 2002; Enfi eld & Levinson, 2006; Tomasello, 2008; Frith & Frith, 2010; Scott-Phillips, 2015; Heintz & Scott-Phillips, 2023). Proximately, linguistic intuitions are triggered by sentences that, one way or another, logically cannot have the property of optimal relevance.

These answers could be further enriched in many ways, for instance by identifying the physical manifestation of linguistic intuition in the brain. We know that language processing has several distinctive neurological markers, such as N400 and P600 effects, but we know less about specifi c markers of linguistic intuition and the phenomenological experience that something is ‘not right’ about a sentence (but N400 effects may relate to expectations of relevance in communication: Kourtis et al., 2020). Identifying neuroscientifi c markers of linguistic intuition might allow us to identify its natural character more precisely than at present, and in ways that do not directly equate it with judgments of acceptability.

The bigger picture is a growing trend in evolutionary approaches to the mind, towards understanding many cognitive capacities as products (and byproducts) of biological adaptation to an especially social evolutionary ecology. Over the past 30 or so years, detailed investigation has suggested that many of the most distinctive features of the human cognitive phenotype have primarily social and interpersonal functions. This includes not only capacities that are plainly social in nature, such cooperative and competitive dispositions, intuitions about fairness and so on; but also many other capacities that can seen, on fi rst blush, to be individual and ‘internal’ phenomena: examples include reasoning, episodic memory, belief formation and many of the emotions (Humphrey, 1976; Sperber & Baumard, 2012; Whiten & Erdal, 2012; Tomasello, 2014; Mercier & Sperber, 2017; Mercier et al., 2017; Mahr & Csibra, 2020; Soares da Silva, 2021; 2022; Williams, 2021; Scott-Phillips, 2022; Al-Shawaf & Shackelford, 2024; inter alia). The byproduct hypothesis of linguistic intuition fi ts this pattern. Linguistic intuitions are sometimes conceived of as being primarily individual phenomena, indeed they are primary data for the ‘internalist’ approach to language most associated with Noam Chomsky (for a recent overview see Rey, 2020). But they are better understood as byproduct effects of social cognition, specifi cally the distinctive social cognition that underpins all human communication.

Acknowledgments

Thank you to Yolanda García-Lorenzo, Christophe Heintz, Benoit Leclercq, Elena Marx, Cameron Morin, Daniel Nettle and Eva Wittenberg for comments on previous drafts. I also received many thoughtful comments following presentations I delivered while developing these ideas. Thank you to audiences at: Society for Philosophy & Psychology Annual Conference; Relevance Researchers Online Conference; Institute for Logic, Cognition, Language & Information; Pragmatics Reading Group, University College London; Evolution, Cognition & Culture Workshop.

Additional Information

Conflict of Interest

The author declares no conflicts of interest.

Statement of Data Availability

There is no original data associated with this manuscript.

AI Usage Statement

The author declares that no AI tool was used in the preparation of this manuscript or the underlying research.

References

Abeillé, A., Hemforth, B., Winckel, E., & Gibson, E. (2020). Extraction from subjects: Differences in acceptability depend on the discourse function of the construction. Cognition, 204, 104293.

Allott, N. (2013). Relevance theory. In: A. Capone, F. Lo Piparo, & M. Carapezza (Eds.), Perspectives on Linguistic Pragmatics (pp. 57-98). Springer.

Al-Shawaf, L. (2024). Levels of analysis and explanatory progress in psychology: Integrating frameworks from biology and cognitive science for a more comprehensive science of the mind. Psychological Review, 132(2), 404-415.

Al-Shawaf, L., & Shackelford, T. (Eds.) (2024). The Oxford Handbook of Evolution & The Emotions. OUP.

Ambridge, B., & Goldberg, A. (2008). The island status of clausal complements: Evidence in favor of an information structure explanation. Cognitive Linguistics, 19(3), 357-389.

Ariew, A. (2003). Ernst Mayr’s ultimate/proximate distinction reconsidered and reconstructed. Biology & Philosophy, 18, 553-565.

Assimakopoulos, S. (2017). Context in relevance theory. In: J. Blochowiak, C. Grisot, S. Durrleman, & C. Laenzlinger (Eds.), Formal Models In The Study Of Language: Applications In Interdisciplinary Contexts (pp. 221-242). Springer.

Auer, P., & Pfänder, S. (Eds.). (2011). Constructions: Emerging & Emergent. Walter de Gruyter.

Bai, D., Hafri, A., Izard, V., Firestone, C., & Strickland, B. (2025). “Core perception”: Re-imagining precocious reasoning as sophisticated perceiving. Behavioral & Brain Sciences.

Baillargeon, R. (2004). Infants’ reasoning about hidden objects: Evidence for event-general and event-specific expectations. Developmental Science, 7(4), 391-414.

Bard, E., Robertson, D., & Sorace, A. (1996). Magnitude estimation of linguistic acceptability. Language, 72, 32-68.

Beckner, C., Blythe, R., Bybee, J., Christiansen, M., Croft, W., Ellis, H., Holland, J., Ke, J., Larsen-Freeman, D., & Schoenemann, T. (2009). Language is a complex adaptive system: Position paper. Language Learning, 59, 1-26.

Behne, T., Liszkowski, U., Carpenter, M., & Tomasello, M. (2012). Twelve-month-olds’ comprehension and production of pointing. British Journal of Developmental Psychology, 30(3), 359-375.

Bergs, A. (2025). From constructional innovation to linguistic change. Functions of Language. Blakemore, D. (2002). Relevance & Linguistic Meaning: The Semantics & Pragmatics of Discourse Markers. CUP. Bolinger, D. (1968) Entailment and the meaning of structures. Glossa, 2(2), 119-127.

Bybee, J. (2013). Usage-based theory and exemplar representations of constructions. In: T. Hoffmann &

G. Trousdale (Eds.), The Oxford Handbook of Construction Grammar (pp. 49-69). OUP. Carey, S. (2009). The Origin Of Concepts. OUP.

Carston, R. (2002). Thoughts & Utterances: The Pragmatics of Explicit Communication. Blackwell. Cave, T., & Wilson, D. (Eds.). (2018). Reading Beyond The Code: Literature & Relevance Theory. OUP.

Chater, N., Reali, F., and Christiansen, M. H. (2009). Restrictions on biological adaptation in language evolution. Proceedings of the National Academy of Sciences, 106, 1015-1020.

Christiansen, M., & Chater, N. (2022). The Language Game: How Improvisation Created Language & Changed the World. Basic Books.

Chomsky, N. (1965). Aspects of The Theory of Syntax. MIT Press. Chomsky, N. (1975). The Logical Structure of Linguistic Theory. Plenum.

Chomsky, N. (2010). Some simple evo devo theses: How true might they be for language? In: R. Larson, V. Déprez, & H. Yamakido (Eds.), The Evolution of Human Language (pp. 45-62). CUP.

Chomsky, N. (2013). Problems of projection. Lingua, 130, 33-49.

Clark, E. (1987). The principle of contrast: A constraint on language acquisition. In: B. MacWhinney (Ed.), Mechanisms of Language Acquisition (pp. 1-33). Lawrence Erlbaum Associates.

Croft, W. (2001). Radical Construction Grammar: Syntactic Theory in Typological Perspective. OUP.

Croft, W. (2009). Toward a social cognitive linguistics. In: V. Evans & S. Pourcel (Eds.), New Directions in Cognitive Linguistics (pp. 395-420). John Benjamins.

Culbertson, J., & Gross, S. (2009). Are linguists better subjects? British Journal for the Philosophy of Science, 60, 721-736.

Cuneo, N., & Goldberg, A. (2023). The discourse functions of grammatical constructions explains an enduring syntactic puzzle. Cognition, 240, 105563.

Dąbrowska, E. (2010). Naive v. expert intuitions: An empirical study of acceptability judgments. The Linguistic Review, 27(1), 1-23.

de Boer, B., Thompson, B., Ravignani, A., & Boeckx, C. (2020). Evolutionary dynamics do not motivate a single-mutant theory of human language. Scientific Reports, 10(1), 451.

Devitt, M. (2006). Intuitions in linguistics. The British Journal for the Philosophy of Science, 57, 481-513. Devitt, M. (2010). Linguistic intuitions revisited. The British Journal for the Philosophy of Science, 61(4), 833-865.

Dickins, T., & Barton, R. (2013). Reciprocal causation and the proximate-ultimate distinction. Biology & Philosophy, 28, 747-756.

Diessel, H. (2019). The Grammar Network: How Linguistic Structure is Shaped by Language Use. CUP. Drożdżowicz, A. (2018). Speakers’ intuitive judgements about meaning—the Voice of Performance view. Review of Philosophy & Psychology, 9(1), 177-195.

Enfield, N., & Levinson, S. (Eds.) (2006). Roots of Human Sociality. Berg.

Enfield, N., & Sidnell, J. (2014). Language presupposes an enchronic infrastructure for social interaction. In: D. Dor, C. Knight, & J. Lewis, The Social Origins Of Language (pp. 92-104). OUP.

Fillmore, C. (1988). The mechanisms of “construction grammar”. In: Annual Meeting of the Berkeley Linguistics Society, 14 (pp. 35-55).

Fillmore, C., Kay, P., & O’Connor, M. (1988). Regularity and idiomaticity in grammatical constructions: The case of let alone. Language, 64(3), 501-538.

Fitzgerald, G. (2010). Linguistic intuitions. The British Journal for the Philosophy of Science, 61(1), 123-160.

Fodor, J., Nickels, S., & Schott, E. (2017). Center-embedded sentences: What’s pronounceable is comprehensible. In: R. de Almeida & L. Gleitman (Eds.), On Concepts, Modules, and Language: Cognitive Science at its Core (pp. 139-168). OUP.

Francis, E. (2021). Gradient Acceptability & Linguistic Theory. OUP.

Frazier, L. (1985). Syntactic complexity. In: Natural Language Parsing: Psychological, Computational and Theoretical Perspectives. In: D. Dowty, L. Karttunen, & A. Zwicky (Eds.), (pp. 129-189). CUP.

Frith, U., & Frith, C. (2010). The social brain: Allowing humans to boldly go where no other species has been. Philosophical Transactions of the Royal Society B: Biological Sciences, 365(1537), 165-176.

Geeraerts, D. (2016). The sociosemiotic commitment. Cognitive Linguistics, 27(4), 527-542.

Geeraerts, D. (2021). Second-order empathy, pragmatic ambiguity, and irony. In: A. Soares da Silva (Ed.), Figurativity in Language — Intersubjectivity & Usage (pp. 19-40). John Benjamins.

Gibbs Jr, R., & Bryant, G. (2008). Striving for optimal relevance when answering questions. Cognition, 106(1), 345-369.

Gibbs Jr, R., & Tendahl, M. (2006). Cognitive effort and effects in metaphor comprehension: Relevance theory and psycholinguistics. Mind & Language, 21(3), 379-403.

Gibson, E., & Thomas, J. (1999). Memory limitations and structural forgetting: The perception of complex ungrammatical sentences as grammatical. Language & Cognitive Processes, 14(3), 225-248.

Goldberg, A. (1995). Constructions: A Construction Grammar Approach to Argument Structure. UCP.

Goldberg, A. (2003). Constructions: A new theoretical approach to language. Trends in Cognitive Sciences, 7(5), 219-224.

Goldberg, A. (2013a). Constructionist approaches. In: T. Hoffmann & G. Trousdale (Eds.), The Oxford Handbook of Construction Grammar (pp. 15-31). OUP.

Goldberg, A. (2013b). Backgrounded constituents cannot be ‘extracted’. In: J. Sprouse & N. Hornstein (Eds.), Experimental Syntax & Island Effects (pp. 221-238). CUP.

Goldberg, A. (2016). Subtle implicit language facts emerge from the functions of constructions. Frontiers in Psychology, 6, 2019.

Goldberg, A. (2019). Explain Me This: Creativity, Competition & The Partial Productivity of Constructions. PUP.

Goldberg, A., & Michaelis, L. (2017). One among many: Anaphoric one and its relationship with numeral one. Cognitive Science, 41, 233-258.

Goldberg, A., & Suttle, L. (2010). Construction grammar. Wiley Interdisciplinary Reviews: Cognitive Science, 1(4), 468-477.

Gross, S. (2021). Linguistic judgments as evidence. In: N. Allott, T. Lohndal, & G. Rey (Eds.), A Companion to Chomsky (pp. 544-556). Wiley.

Hilpert, M. (2014). Construction Grammar & its Application to English. EUP.

Heintz, C., & Scott-Phillips, T. (2023). Expression unleashed: The evolutionary & cognitive foundations of human communication. Behavioral & Brain Sciences, 46, e1, 1-53.

Ho, M., Cushman, F., Littman, M., & Austerweil, J. (2021). Communication in action: Planning and interpreting communicative demonstrations. Journal of Experimental Psychology: General, 150(11), 2246-2272.

Hoffmann, T. (2019). Language and creativity: A construction grammar approach to linguistic creativity. Linguistics Vanguard, 5(1), 20190019.

Hoffmann, T. (2022). Construction Grammar: The Structure of English. CUP.

Hoffmann, T., & Trousdale, G. (2013) The Oxford Handbook of Construction Grammar (Eds.). OUP. Hornstein, N. (2013). Acceptability and grammaticality. Blogpost.

Humphrey, N. (1976). The social function of intellect. In: P. Bateson & R.Hinde (Eds.), Growing Points in Ethology (pp. 303-317). CUP.

Hurford, J. (2018). Mutation, modularity, merge, communication & selection. Current Opinion in Behavioral Sciences, 21, 76-79.

Ifantidou, E., de Saussure, L., & Wharton, T. (Eds.) (2021). Beyond Meaning. John Benjamins Publishing Company.

Jara-Ettinger, J., & Rubio-Fernandez, P. (2024). Demonstratives as attention tools: Evidence of mentalistic representations within language. Proceedings of the National Academy of Sciences, 121(32), e2402068121.

Karlsson, F. (2007). Constraints on multiple center-embedding of clauses. Journal of Linguistics, 43(2), 365-392.

Kaschak, M., & Glenberg, A. (2000). Constructing meaning: The role of affordances and grammatical constructions in sentence comprehension. Journal of Memory & Language, 43(3), 508-529.

Kay, P., & Fillmore, C. (1999). Grammatical constructions and linguistic generalizations: The What’s X doing Y? construction. Language, 75(1), 1-33.

Király, I., Csibra, G., & Gergely, G. (2013). Beyond rational imitation: Learning arbitrary means actions from communicative demonstrations. Journal of Experimental Child Psychology, 116(2), 471-486.

Kourtis, D., Jacob, P., Sebanz, N., Sperber, D., & Knoblich, G. (2020). Making sense of human interaction benefits from communicative cues. Scientific Reports, 10(1), 18135.

Langacker, R. (1987). Foundations of Cognitive Grammar: Volume I: Theoretical Prerequisites. SUP.

Langacker, R. (2008). Constructing the meanings of personal pronouns. In: G. Radden, K.-M. Köpcke, T. Berg, & P. Siemund (Eds.), Aspects of Meaning Construction (pp. 171-187). John Benjamins Publishing Company.

Lau, J., Clark, A., & Lappin, S. (2017). Grammaticality, acceptability, and probability: A probabilistic view of linguistic knowledge. Cognitive Science, 41(5), 1202-1241.

Leclercq, B. (2023). Linguistic Knowledge & Language Use: Bridging Construction Grammar & Relevance Theory. CUP.

Leclercq, B., & Morin, C. (2023). No equivalence: A new principle of no synonymy. Constructions, 15(1).

Leivada, E. (2020). Language processing at its trickiest: Grammatical illusions and heuristics of judgment. Languages, 5(3), 29.

Leivada, E. & Westergaard, M. (2020). Acceptable ungrammatical sentences, unacceptable grammatical sentences, and the role of the cognitive parser. Frontiers in Psychology, 11, 364.

Ludlow, P. (2014). Living Words: Meaning Underdetermination & the Dynamic Lexicon. OUP.

Liu, Y., Winckel, E., Abeillé, A., Hemforth, B., & Gibson, E. (2022). Structural, functional, and processing perspectives on linguistic island effects. Annual Review of Linguistics, 8(1), 495-525.

Mahr, J., & Csibra, G. (2020). Witnessing, remembering, and testifying: Why the past is special for human beings. Perspectives on Psychological Science, 15(2), 428-443.

Martins, P., & Boeckx, C. (2019). Language evolution and complexity considerations: The no half-Merge fallacy. PLoS Biology, 17(11), e3000389.

Maynes, J., & Gross, S. (2013). Linguistic intuitions. Philosophy Compass, 8(8), 714-730. Mayr E. (1961). Cause and effect in biology. Science, 131, 1501-1506.

Mazzarella, D., & Pouscoulous, N. (2023). Ironic speakers, vigilant hearers. Intercultural Pragmatics, 20(2), 111-132.

McCallum, K., Mitchell, S., & Scott-Phillips, T. (2020). The art experience. Review of Philosophy & Psychology, 11, 21-35.

McEllin, L., Sebanz, N., & Knoblich, G. (2018a). Identifying others’ informative intentions from movement kinematics. Cognition, 180, 246-258.

McEllin, L., Knoblich, G., & Sebanz, N. (2018b). Distinct kinematic markers of demonstration and joint action coordination? Evidence from virtual xylophone playing. Journal of Experimental Psychology: Human Perception & Performance, 44(6), 885-897.

Mercier, H., Dezecache, G., & Scott-Phillips, T. (2017). Strategically communicating minds. Current Directions in Psychological Science, 26(5), 411-416.

Mercier, H., & Sperber, D. (2017). The Enigma of Reason. HUP.

Michaelis, L. (2006). Construction grammar. In: K. Brown (Ed.), The Encyclopedia of Language & Linguistics (Vol. 3, pp. 73–84). Elsevier.

Michaelis, L., & Lambrecht, K. (1996). Toward a construction-based theory of language function: The case of nominal extraposition. Language, 72(2), 215-247.

Miller, G. (1962). Some psychological studies of grammar. American Psychologist, 17, 748-762

Montalbetti, M. (1984). After Binding: On The Interpretation Of Pronouns (Doctoral dissertation, Massachusetts Institute of Technology).

Nettle, D., & Scott-Phillips, T. (2023). Is a non-evolutionary psychology possible? In: A. du Crest, M. Valkovic, A. Ariew, H. Desmond, P. Huneman, & T. Reydon (Eds.), Evolutionary Thinking Across Disciplines (pp. 21-42). Springer Nature.

Pinker, S., & Bloom, P. (1990). Natural language and natural selection. Behavioral & Brain Sciences, 13(4), 707-727.

Phillips, C. (2006). The real-time status of island phenomena. Language, 82(4), 795-823.

Phillips, C., Wagers, M., & Lau, E. (2011). Grammatical illusions and selective fallibility in real-time language comprehension. In: J. Runner (Ed.), Syntax & Semantics Volume 37: Experiments at the Interfaces (pp. 147-180). Emerald Press.

Planer, R. (2017). How language couldn’t have evolved: A critical examination of Berwick and Chomsky’s theory of language evolution. Biology & Philosophy, 32, 779-796.

Recanati, F. (2004). Literal Meaning. CUP.

Rey, G. (2020). Representation of Language: Philosophical Issues in a Chomskyan Linguistics. OUP.

Robenalt, C., & Goldberg, A. (2015). Judgment evidence for statistical preemption: It is relatively better to vanish than to disappear a rabbit, but a lifeguard can equally well backstroke or swim children to shore. Cognitive Linguistics, 26(3), 467-503.

Royka, A., Chen, A., Aboody, R., & Jara-Ettinger, J. (2022). People infer communicative action through an expectation for efficient communication. Nature Communications, 13, 4160.

Rubio-Fernandez, P., Mollica, F., & Jara-Ettinger, J. (2021). Speakers and listeners exploit word order for communicative efficiency: A cross-linguistic investigation. Journal of Experimental Psychology: General, 150(3), 583-594.

Sag, I. (2010). English filler-gap constructions. Language, 86(3), 486-545.

Sasamoto, R. (2012). Silence in ostensive communication. In: A. Yoshimura, A. Suga, & N. Yamamoto (Eds.), Observing Linguistic Phenomena (pp. 437-449). Eihosha.

Sasamoto, R. (2019). Onomatopoeia & Relevance: Communication of Impressions via Sound. Springer.

Soares da Silva, A. (2021), Measuring the impact of (non)figurativity in the cultural conceptualization of emotions in the two main national varieties of Portuguese. In: A. Soares da Silva (Ed.), Figurative Language: Intersubjectivity & Usage (pp. 387-438). John Benjamins.

Soares da Silva, A. (2022), Metaphor, emotion, and intralinguistic cultural variation: Metaphors of anger in European and Brazilian Portuguese. In: U. Schröder, M. Mendes de Oliveira, & A. Tenuta (Eds.), Metaphorical Conceptualizations: (Inter)Cultural Perspectives (pp. 189-222). De Gruyter.

Schindler, S., Drożdżowicz, A., & Brøcker, K. (Eds.). (2020). Linguistic Intuitions: Evidence & Method. OUP.

Schmid, H. (2016). Why cognitive linguistics must embrace the social and pragmatic dimensions of language and how it could do so more seriously. Cognitive Linguistics, 27(4), 543-557.

Schütze, C. (1996/2016). The Empirical Base of Linguistics: Grammaticality Judgments & Linguistic Methodology. Language Science Press.

Scholz, B., Pelletier, F., Pullum, G., & Nefdt, R. (2022). Philosophy of linguistics. In: E. Zalta (Ed.), The Stanford Encyclopedia of Philosophy.

Schwier, C., van Maanen, C., Carpenter, M., & Tomasello, M. (2006). Rational imitation in 12‐month‐old infants. Infancy, 10(3), 303-311.

Scott, K. (2022). Pragmatics Online. Routledge.

Scott-Phillips, T. (2015). Speaking Our Minds. Palgrave Macmillan.

Scott-Phillips, T. (2022). Biological adaptations for cultural transmission? Biology Letters, 18, 20220439. Scott-Phillips, T. (2024). The communicative principle of relevance. Current Directions in Psychological Science.

Scott-Phillips, T. (2025). Communication & grammar: A synthesis. Psychological Review.

Scott-Phillips, T. (in press). Language evolution: A view from relevance theory. In: B. Clark, K. Scott, & R. Sasamoto (Eds.), Cambridge Companion to Relevance Theory. CUP.

Scott-Phillips, T., Dickins, T. & West, S. (2011). Evolutionary theory and the ultimate/proximate distinction in the human behavioural sciences. Perspectives on Psychological Science, 6(1), 38-47.

Scott-Phillips, T., Kirby, S., & Ritchie, G. R. S. (2009). Signalling signalhood and the emergence of communication. Cognition, 113(2), 226-233.

Spelke, E. (1990). Principles of object perception. Cognitive Science, 14(1), 29-56. Spelke, E. (2022). What Babies Know. OUP.

Sperber, D. (1995). How do we communicate? In: J. Brockman & K. Matson (Eds.), How Things Are: A Science Toolkit For The Mind (pp. 191-199). Morrow.

Sperber, D. (2005). Modularity and relevance: How can a massively modular mind be flexible and context-sensitive? In: P. Carruthers, S. Laurence, & S. Stich (Eds.), The Innate Mind: Structure and Contents (pp. 53-68). OUP.

Sperber, D. (2010). The guru effect. Review of Philosophy & Psychology, 1(4), 583-592.

Sperber, D., & Baumard, N. (2012). Moral reputation: An evolutionary and cognitive perspective. Mind & Language, 27(5), 495-518.

Sperber, D., Cara, F., & Girotto, V. (1995). Relevance theory explains the selection task. Cognition, 57(1), 31-95.

Sperber, D. & Wilson, D. (1986/1995). Relevance: Communication & Cognition. Blackwell.

Sperber, D., & Wilson, D. (2002). Pragmatics, modularity & mind-reading. Mind & Language 17, 3-23. Sperber, D., & Wilson, D. (2025). Rethinking ostensive communication in an evolutionary, comparative, and developmental perspective. Psychological Review.

Sprouse, J. (2007). Continuous acceptability, categorical grammaticality, and experimental syntax. Biolinguistics, 1, 123-134.

Sprouse, J. (2013). Acceptability judgments. In: J. Sprouse (Ed.), The Oxford Handbook of Experimental Syntax (pp. 3-28). OUP.

Tantucci, V. (2021). Language & Social Minds: The Semantics & Pragmatics of Intersubjectivity. CUP.

Tauzin, T., & Gergely, G. (2018). Communicative mind-reading in preverbal infants. Scientific Reports, 8(1), 1-9.

Tomasello, M. (Ed.) (1998a). The New Psychology of Language: Cognitive & Functional Approaches to Language Structure, Volume I. Routledge.

Tomasello, M. (Ed.) (1998b). The New Psychology of Language: Cognitive & Functional Approaches to Language Structure, Volume II. Routledge.

Tomasello, M. (2003). Constructing A Language. HUP.

Tomasello, M. (2008). Origins Of Human Communication. MIT Press.

Tomasello, M. (2014). The ultra‐social animal. European Journal of Social Psychology, 44(3), 187-194.

Tubau, S., Etxeberria, U., Déprez, V., & Espinal, M. (Eds.) (2020). What are (un)acceptability and (un)grammaticality? How do they relate to one another and to interpretation? Frontiers in Psychology, 11, 621267.

Turner, M. (2018). The role of creativity in multimodal Construction Grammar. Zeitschrift für Anglistik und Amerikanistik 66(3). 357-370.

Ungerer, T., & Hartmann, S. (2023). Constructionist Approaches: Past, Present, Future. CUP.

van der Henst, J., & Sperber, D. (2004). Testing the cognitive and communicative principles of relevance. In: I. Noveck & D. Sperber (Eds.), Experimental Pragmatics (pp. 141-171). Palgrave Macmillan.

Verhagen, A. (2005). Constructions of Intersubjectivity: Discourse, Syntax & Cognition. OUP.

Verhagen, A. (2015). Grammar and cooperative communication. In: E. Dąbrowska & D. Divjak (Eds.).

Handbook of Cognitive Linguistics (pp. 232-252). De Gruyter Mouton.

Wasow, T., & Arnold, J. (2005). Intuitions in linguistic argumentation. Lingua, 115(11), 1481-1496. Wellwood, A., Pancheva, R., Hacquard, V., & Phillips, C. (2018). The anatomy of a comparative illusion. Journal of Semantics, 35(3), 543-583.

Wharton, T., Jagoe, C., & Wilson, D. (2022). Relevance theory: New horizons. Journal of Pragmatics, 194, 1-5.

Whiten, A., & Erdal, D. (2012). The human socio-cognitive niche and its evolutionary origins.

Philosophical Transactions of the Royal Society B, 367(1599), 2119-2129.

Williams, D. (2021). Socially adaptive belief. Mind & Language, 36(3), 333-354.

Wilson, D. (2023). Relevance theory and context. In J. Romero-Trillo (Ed.), The Cambridge Handbook of Language in Context (pp. 247-267). CUP.

Wilson, D., & Carston, R., (2007). A unitary approach to lexical pragmatics: Relevance, inference and ad hoc concepts. In: N. Burton-Roberts (Ed.), Pragmatics (pp. 230-259). Palgrave.

Wilson, D., & Sperber, D. (2012). Meaning & Relevance. CUP.

Winckel, E., Abeillé, A., Hemforth, B., & Gibson, E. (2025). Discourse-based constraints on long-distance dependencies generalize across constructions in English and French. Cognition, 254, 105950.

Vesper, C., Morisseau, T., Knoblich, G., & Sperber, D. (2021). When is ostensive communication used for joint action? Cognitive Semiotics, 14(2), 101-129.

Yus, F. (2016). Humour & Relevance. John Benjamins Publishing Company.

Zlatev, J., Racine, T., Sinha, C., & Itkonen, E. (2008). Intersubjectivity: What makes us human? In: J. Zlatev et al. (Eds.), The Shared Mind: Perspectives on Intersubjectivity (pp. 1-14). John Benjamins Publishing Company.

Review

DOI: https://doi.org/10.25189/2675-4916.2025.V6.N3.ID868.R

Editorial Decision

EDITOR 1: Miguel Oliveira Jr

ORCID: https://orcid.org/0000-0002-0866-0535

AFFILIATION: Universidade Federal de Alagoas, Alagoas, Brasil.

-

EDITOR 2: René Alain Santana de Almeida

ORCID: https://orcid.org/0000-0002-9288-0740

AFFILIATION: Universidade Federal do Recôncavo da Bahia, Bahia, Brasil.

-

ASSESSMENT: This article proposes a principled account of intuitions of acceptability and unacceptability, treating them as by-product effects of general cognitive mechanisms for interpreting communicative stimuli within a construction-grammar and relevance-theoretic framework. The argument is carefully developed and draws on existing work in pragmatics, construction-based grammar and evolutionary thinking about communication, even though some aspects of its extension to other kinds of linguistic judgement remain open. The work will interest researchers concerned with grammaticality judgements, pragmatics, construction-based approaches to grammar, and cognitive or evolutionary approaches to language.

Rounds of Review

REVIEWER 1: Michael Pleyer

ORCID: https://orcid.org/0000-0001-6685-391X

AFFILIATION: Nicolaus Copernicus University, Toruń, Polônia.

-

REVIEWER 2: Nicholas Elwyn Allott

ORCID: https://orcid.org/0000-0001-9742-080X

AFFILIATION: University of Oslo, Oslo, Noruega.

-

REVIEWER 3: Lilian Vieira Ferrari

ORCID: https://orcid.org/0000-0001-7808-4425

AFFILIATION: Universidade Federal do Rio de Janeiro, Rio de Janeiro, Brasil.

-

REVIEWER 4: Augusto Soares da Silva

ORCID: https://orcid.org/0000-0001-7951-5194

AFFILIATION: Universidade Católica Portuguesa, Lisboa, Portugal.

-

ROUND 1

REVIEWER 1

2025-08-17 | 01:29 PM

This is a very interesting proposal dealing with the underlying cognitive mechanisms and the "why" of a key concept in much of linguistics, namely language users' intuitions that sentences are "un/grammatical" or "un/acceptable." Most research has assumed that such judgements are a reflection of language users' linguistic knowledge about the structure of their language and it has widely been used in linguistic theory building. The author proposes a truly novel explanation, based on Relevance Theory and Construction Grammar, for the sense of "oddness" we have when interpreting ungrammatical or unacceptable sentences, namely that is stems not from language users' internalised linguistic system tracking and processing grammaticality and acceptbility per se, but that instead it is related to the principle of communicative relevance. Communicators presume that communicative stimuli are "optimally relevant", that is that they are the most optimal and way to communicate a particular meaning and intention. "ungrammatical" and "unacceptable" sentences cannot be interpreted in a way that is consistent with the presumption of optimal relevance, that is that there are no plausible cognitive benefits to be derived from the sentence. This fundamentally shifts the view of what linguistic intuitions are, not direct reflections of our internalised grammatical system at work in parsing sentences, but instead a byproduct of the presumption of optimal relevance. This hypothesis is relevant not only for researchers in relevance theory and construction grammar, but for linguistic theorising in general that makes use of grammaticality and acceptability judgements, meaning it potentially has very wide appeal.However, I would like to see some more evidence of how to assess the main claim and more contextualisation.

However, for some aspects of the proposal, I would like to see a bit more evidence and discussions of its implications, to bolster the very strong - and controversial - claim made by the author.

This is a very interesting proposal dealing with the underlying cognitive mechanisms and the "why" of a key concept in much of linguistics, namely language users' intuitions that sentences are "un/grammatical" or "un/acceptable." Most research has assumed that such judgements are a reflection of language users' linguistic knowledge about the structure of their language and it has widely been used in linguistic theory building. The author proposes a truly novel explanation, based on Relevance Theory and Construction Grammar, for the sense of "oddness" we have when interpreting ungrammatical or unacceptable sentences, namely that is stems not from language users' internalised linguistic system tracking and processing grammaticality and acceptbility per se, but that instead it is related to the principle of communicative relevance. Communicators presume that communicative stimuli are "optimally relevant", that is that they are the most optimal and way to communicate a particular meaning and intention. "ungrammatical" and "unacceptable" sentences cannot be interpreted in a way that is consistent with the presumption of optimal relevance, that is that there are no plausible cognitive benefits to be derived from the sentence. This fundamentally shifts the view of what linguistic intuitions are, not direct reflections of our internalised grammatical system at work in parsing sentences, but instead a byproduct of the presumption of optimal relevance. This hypothesis is relevant not only for researchers in relevance theory and construction grammar, but for linguistic theorising in general that makes use of grammaticality and acceptability judgements, meaning it potentially has very wide appeal.However, for some aspects of the proposal, I would like to see a bit more evidence and discussions of its implications, to bolster the very strong - and controversial - claim made by the author. First, some more contextualisation of how grammaticality and acceptability judgements have been used in theorising and with what justification, maybe in a short paragraph would be helpful, for example to more recent developments associating grammaticality and acceptability intutions with parsability (e.g. Leveida & Westergraard 2020)Leivada, E., & Westergaard, M. (2020). Acceptable ungrammatical sentences, unacceptable grammatical sentences, and the role of the cognitive parser. Frontiers in psychology, 11, 364.I would also like to see a bit more detail on the proposed "psychological sense of oddness" that language users have. How exactly can we define oddness and is there a way to measure and quantify it in a way that does not conflate it with un/grammaticality un/acceptability rating?I would also would like to see a bit more discussion of how it can be attested that a particular sentence does not incure plausible cognitive benefits. That is, there deviations from conventional use that lead to language change. This also relates to inter-individual variation. For example, some resultative constructs such as "She sneezed the napkin off the table" might appear acceptable for some, but not others, and expressions like "same same but different" or "long time no see" are technically ungrammatical but acceptable for others based on context and language and cultural experience. I am not suggesting these examples are incompatible with the proposal but would like to see a bit more detail on how the distinction can be made between sentences that can incur plausible cognitive benefits and those that can't and how we can make this distinction. Why is "long time no see" or "all your base are belong to us" acceptable to some and infer additional cognitive benefits (e.g. as humourous expressions" but "I don't want going to the cinema" cannot give a plausible change in cognitive benefits (e.g. when looking at verb complementation, what is the relevance-related reason that *We decided going there *I hope doing it cannot be interpreted, e.g. as the speaker wanting to stress the progressive nature of the action? That is I would generally see some more discussion of the logic of how to arrive at the conclusion that a sentence cannot be interpreted under the presumption of optimal relevance, which then explains why we feel a sense of oddness when reading *I avoided to do it vs I avoided doing it. Regarding the section "Mutually contradictory constrains, rendering optimisation impossible", I would like to see some concrete examples of cases where constraints contradict each other, as of now the discussion remains relatively abstract, which also holds for the discussion of Cuneo & Goldberg 2023.Lastly, I was wondering if more could be said about the implications of the byproduct hypothesis and its predictions. For example, there is a wealth of work on the neuroscience of prediction, parsing, and unacceptability, such as N400 and P600 effects. How are they reinterpreted in this framework and are there testable predictions of this hypothesis that would allow distinguishing it from other models (e.g. different neural activation and involvement of a pragmatic component when encountering unacceptable sentences)Overall this is an intriguing and thought-provoking proposal, but in my view it should be somewhat more developed, especially with regard to how to arrive at conclusions that particular utterances do not incur plausible cognitive benefits and cannot be interpreted using the presumption of optimal relevance.Some minor comments: On page 3 the author points out that there have been no detailed investigations into why there is an intutitive feeling of oddness when encountering an ungrammatical sentence. They then state that Chomsky "has speculated that perhaps an incidental biological mutation simply happened to generate the relevant cognitive dispositions" but in the section the author then quotes Chomsky talks about the appearance of language and the ability for language acquisition in general, not linguistic intutitions specifically, so the phrasing here is a bit misleading.Page 9, "what people do always underdermines what they mean" could be explained in a bit more details with examples, of how something is "underdetermined"Page 24, it is not quite clear why "self-conscious" emotions such as pride and jealousy would not be seen as social. As one is jealous of someone it seems to need a social dimension to be understood?

-

REVIEWER 2

2025-08-29 | 10:54 AM

This paper argues for a seemingly radical answer to the question: Where do linguistic intuitions of acceptability and unacceptability come from? The core of the claim made here is that such intuitions depend on pragmatic principles – that is, principles that govern communication – and never on the grammar unmediated by pragmatics. In my review, I suggest that more needs to be done in this paper to explain how the mechanism explains key types of intuitions and to clarify its relation to rival accounts.

This paper proposes a seemingly radical answer to a familiar but interesting question: Why do we have intuitions of unacceptability and acceptability of sentences, of the type that are usually understood to be related to the grammatical structure of the sentences in question rather than some other factor e.g. to use of a socially inappropriate expression or to merely pragmatic infelicity? (I call this ‘grammatical (un)acceptability’ for short in what follows.)The answer proposed is that these intuitions are due to communicative principles: specifically a sentence is grammatically unacceptable if and only if “there appears to be an intrinsic impossibility of interpreting them in any way consistent with the unconscious presumption of optimal relevance that underpins the interpretation of all stimuli perceived as communicative.” (p. 1)It is suggested that this “implies that there may be no particular cognitive capacity that functions to distinguish the grammatical from the ungrammatical.” (p. 1) So far, so radical: many apparently grammatical facts are actually due to pragmatics, and this is part of a programme tending towards thorough-going rejection of the idea that there is a domain-specific mental grammar, and thus of generativism and the conception of linguistics due to Noam Chomsky. However this apparent radicalism is not fully borne out by the theory presented, which leans not only on relevance theoretic pragmatics but on a theory (or family of theories) of grammar, construction grammar. On this account, grammatical (un)acceptability intuitions are causally downstream of both grammar and pragmatic processing. This is common ground with generativists (although obviously they would disagree about which grammatical theory is right). Further, it becomes clear in Scott-Phillips’ discussion of centre-embedded sentences that, like generativists, he also sees parsing as a causal factor here, and also short-term memory limitations (to at least the extent that they are implicated in some parsing problems). As far as I can see, what is new here, then, is not any claim that the grammar isn’t involved in grammatical (un)acceptability intuitions. Rather it is the centrality of pragmatic principles and the kind of role that they play. Specifically, the claim, I think, is that the principle of communicative relevance always mediates the causal role of the grammar in these intuitions. There are no strings that are (un)acceptable only because the grammar does not/does assign a structure to them. Rather, strings that are assigned a structure by the grammar may be unacceptable if the parts that are combined are communicatively incompatible. At least I think this is the central claim here. It’s interesting and still radical, but in rather a different way from the impression given in the framing of this paper.I think there are several problems with the paper and the proposal as it stands:1) The roles and importance of the qualifiers ‘appears to be’ and ‘intrinsic’ in the condition proposed could do with some clarification.2) There’s quite a bit of ‘prior art’ here which is neglected. Scott-Phillips writes, “there is very little focused attention on the question of why humans have these intuitions in the first place.” It would be fair to say that there’s no generally accepted detailed account of the aetiology of grammatical (un)acceptability intuitions, but there’s plenty of research which has addressed this question, including work in philosophy by Michael Devitt and Gareth Fitzgerald (which is cited here), Georges Rey and Anna Drożdżowicz (not cited), among others, and work on the relation of parsing to the grammar inter alia by Colin Phillips, Shevaun Lewis and Dave Kush among many others. (See references below.)Perhaps Scott-Phillips’ point here is that work of these kinds doesn’t generally try to provide an explanation for the origin in our species of grammatical (un)acceptability intuitions (although Chomsky and others certainly have discussed the evolutionary origins of the language faculty). If so, it would be worth distinguishing more clearly two questions here: the one about the aetiology of these intuitions, and the one about their origin (where Scott-Phillips favours an account in terms of evolutionary natural selection), and noting that there have been many attempts to answer the first of them (see discussion in Gross 2021). As far as I can see, the primary claim made in the current paper is about the aetiology, not the origin, although the latter is in the background. Some evaluation of other accounts of the aetiology is necessary to the project of this paper, given that they are rivals to the account presented here.3) Relevance – and therefore optimal relevance – is relative to context. Thus it seems to follow from the proposal here that grammatical (un)acceptability intuitions are context-relative. But grammaticality is standardly assumed to be a property of sentence-types (not utterances or sentence tokens), and empirically this seems to be correct: mostly when a string of words strikes us as unacceptable and the explanation given is grammatical, providing a context doesn’t help. This looks bad for any pragmatic account of grammatical (un)acceptability intuitions, so needs discussion here. Perhaps the intention is that this is dealt with by making the criterion (apparent intrinsic) impossibility of an optimally relevant interpretation, where this entails unacceptability across contexts. If so that needs spelling out, showing how context shift cannot help.4) The role of construction grammar in the argument presented here is rather unclear. Is the claim the conditional one: “If construction grammar is the right account of grammar, then the explanation of grammatical (un)acceptability is pragmatic principles”? Or the more ambitious conjunction: “Construction grammar is the right account of grammar and the explanation of grammatical (un)acceptability is pragmatic principles”? Generativists could grant the conditional claim, since they take the antecedent to be false. It’s not clear why they should be moved by the argumentation presented in this paper if the conjunction is claimed, since the claim that construction grammar is the right account of grammar is hardly supported here.5) In the discussion of centre-embedded sentences, there’s no mention of the account in terms of the interface with prosody provided by Fodor et al (2017). Such discussion seems to be needed here, because that account appears to have better predictive powers than older accounts in terms of overloading short-term memory. It explains not only why some grammatical centre-embedded sentences are unacceptable, but also why others are not, e.g. “The rusty old ceiling pipes that the plumber my dad trained fixed continue to leak occasionally”. It would obviously be a merit of the proposal here if it could account for these facts in another way.6) This paper sets out three ways that intuitions of grammatical unacceptability can arise. The discussions of the first two ways are interesting. The third way strikes me as under-described here. The explanation points towards recent work in construction grammar that attempts to (re)describe “many grammatical phenomena” in this way, particularly island effects. But no examples are given. The brief discussion of conflicting information packaging is interesting but entirely abstract. For the argumentative purposes of this paper, this subsection needs to be compelling, because it seems to be how many of the key examples in the literature are to be accounted for. (See next point.)7) One target of this paper seems to be Chomsky’s category of ‘Fine Thoughts’, which Rey calls ‘WhyNots’: strings of words that strike us as unacceptable even though we would be able to understand what is meant by them if they were used (see e.g. Allott & Rey, 2017; Rey 2000: 21–35). Both take these data to be important to the generativist project, because they seem to establish that there’s some system in our mind that assigns one or more structures to some strings and none to others, independently of our communicative goals. (Chomsky’s view is that ungrammaticality is when the grammar assigns zero structures to a string: see Berwick et al., 2011.) It’s odd that they aren’t mentioned here. I think that in order to serve its argumentative goals, this paper needs to show how the proposal it gives can account for at least some types. To end this review on a positive note, I want to say that it seems to me as though the proposal sketched here could push this debate forwards. As noted above, Chomsky and Rey (and others) argue that ‘Fine Thoughts’/‘WhyNots’ are best explained by – and so are evidence for – a syntax system governed by its own domain-specific principles, since they are unacceptable even though they would serve for communication. I think Scott-Phillips’ proposal claims that the intuitions might derive from communicative principles nonetheless: the claim is not that the ‘Fine Thought’ strings couldn’t be used for communication, but that they could never be optimally relevant. That’s a new move in the debate, and it would be interesting if fleshing it out more reveals it to have explanatory power.

References

Allott, N., & Rey, G. (2017). The many errors of Vyvyan Evans’ The Language Myth. The Linguistic Review, 34(3), 1–20. https://doi.org/10.1515/tlr-2017-0011Berwick, R. C., Pietroski, P., Yankama, B., & Chomsky, N. (2011). Poverty of the stimulus revisited. Cognitive Science, 35(7), 1207–1242. https://doi.org/10.1111/j.1551-6709.2011.01189.xDevitt, M. (2006). Intuitions in linguistics. The British Journal for the Philosophy of Science, 57(3), 481–513. http://bjps.oxfordjournals.org/content/57/3/481.abstractDrożdżowicz, A. (2017). Speakers’ intuitive judgements about meaning – the Voice of Performance view. Review of Philosophy and Psychology, 28. https://doi.org/10.1007/s13164-017-0349-0Fitzgerald, G. (2010). Linguistic intuitions. The British Journal for the Philosophy of Science, 61(1), 123–160. https://doi.org/10.1093/bjps/axp014Fodor, J. D., Nickels, S., & Schott, E. (2017). Center-embedded sentences: What’s pronounceable is comprehensible. In R. G. De Almeida & L. R. Gleitman (Eds.), On concepts, modules, and language: Cognitive science at its core (pp. 139–168). Oxford University PressGross, S. (2021). Linguistic judgments as evidence. In N. Allott, T. Lohndal, & G. Rey (Eds.), A Companion to Chomsky (pp. 544–556). Wiley Blackwell. Kush, D., Lidz, J., & Phillips, C. (2017). Looking forwards and backwards: The real-time processing of Strong and Weak Crossover. Glossa: a journal of general linguistics, 2(1), 129. https://doi.org/10.5334/gjgl.280Lewis, S., & Phillips, C. (2015). Aligning grammatical theories and language processing models. Journal of Psycholinguistic Research, 44(1), 27–46. https://doi.org/10.1007/s10936-014-9329-zPhillips, C. (2006). The real-time status of island phenomena. Language, 82(4), 795–823. http://www.jstor.org/stable/4490269Phillips, C. (2013). Some arguments and nonarguments for reductionist accounts of syntactic phenomena. Language and Cognitive Processes, 28(1-2), 156–187. https://doi.org/10.1080/01690965.2010.530960Rey, G. (2020). Representation of language: Philosophical issues in a Chomskyan linguistics. Oxford University Press. https://dx.doi.org/10.1093/oso/9780198855637.001.0001

-

REVIEWER 3

2025-08-11 | 05:37 PM

The paper main goal is to investigate what causes speakers of a language to have an intuitive sense that certain sentences are fully acceptable while others trigger a psychological sense of oddness. In order to explain this phenomenon, the author draws on theoretical and experimental knowledge provided by a vast body of literature in related fields—linguistics and psychology—, articulating this knowledge in a consistent and creative manner. More specifically, the work proposes that there is an unconscious assumption of optimal relevance that triggers the psychological sense of oddness whenever a sentence cannot be interpreted in a way that is consistent with this assumption. This impossibility, in turn, is explained in terms of three possibilities: (i) although grammatical, the sentence is not analyzable and therefore overloads short-term memory; (ii) the sentence deviates from conventional usage without any cognitive benefit; (iii) the constructions within a sentence mutually restrict each other, making the sentence contradictory and impossible to interpret.

The manuscript may be of interest for researchers working on Cognitive Linguistics, Functionalism, Psycholinguistics, Discourse analysis, etc), Cognitive Psychology and Evolutionary Psychology.

The paper presents a theoretical discussion regarding the established assumption that we have an intuitive sense of the acceptability of sentences (what is grammatical and what is not). Alternatively, it is proposed that linguistic intuitions of unnaceptability occur only when there appears to be an intrinsic impossibility of interpreting a sentence in any way consistent with the presumption of optimal relevance. This intrinsic impossibility triggers the psychological sense of oddness referred to as byproduct hypothesis of linguistic intuition.

Based on a consistent set of theoretical and empirical research from cognitive linguistics, cognitive psychology, and evolutionary psychology, the work relies on two assumptions which underpin the byproduct hypothesis: (i) the pairing between particular forms and specific communicative functions ("constructions all the way down", Goldberg, 2003) within the framework of cognitive construction grammar; (ii) the description of linguistic intuitions as byproduct effects of the audience presumption of optimal relevance, as proposed by Relevance Theory in cognitive psychology of human communication (Sperber & Wilson, 2002).

Departing from this connection between constructionist approaches to grammar and the cognitive psychology of human communication, already pointed out in the literature, the work’s main argument is that audiences interpret communicative stimuli on the basis of a presumption of optimal relevance. In this vein, the proposal describes three general ways in which the interpretation of constructions on the basis of a presumption of optimal relevance is intrinsically impossible:

(i) the sentence is not parseable; thus, it overloads short term memory or other aspects of processing. The example the author mentions are sentences that follow the grammar of English, but are hard to parse, as multiple embedded clauses (ex. “The rat the cat the dog chased killed ate the malt”/ The dog chased the cat. The cat killed the rat. The rat ate the malt.); (ii) the sentence deviates from conventional use without any plausible change in interpretation, and hence no plausible change in cognitive benefits (ex. She gived me a book (rather than She gave me a book). In this case, it is assumed that stimuli that follow conventional use entail less attentional resources from the audience than stimuli that deviate from conventional use; at the same time, the latter offer the same cognitive benefits (same meaning, including same pragmatics). Therefore, these sentences are intrinsically impossible to interpret in any way concordant with a presumption of optimal relevance, given that they combine additional costs and no change in benefits. (iii) a sentence can be intrinsically impossible to interpret in any way consistent with the presumption of optimal relevance if two (or more) constructions within a sentence constrain one another in mutually contradictory ways (ex. information packing constructions, and in particular the contrast between foregrounding and backgrounding).

-

REVIEWER 4

2025-08-05 | 03:40 PM

The author offers a highly innovative and thought-stimulating response to the fundamental questions of why humans have linguistic intuitions and how these can be explained in psychological, evolutionary, and linguistic terms, without invoking a supposed cognitive capacity to distinguish the grammatical from the ungrammatical. The author’s proposal to consider linguistic intuitions as by-product effects of core cognitive capacities for the interpretation of communicative stimuli is both theoretically and empirically well-grounded, convincing, and clearly demonstrates the inadequacy of the traditional (generativist) “grammaticalness assumption”. Furthermore, the paper brings together interdisciplinary evidence from cognitive linguistics, cognitive psychology, and evolutionary psychology. The paper fully meets the requirements for publication essentially in its current form.

-

ROUND 2

REVIEWER 1

2025-10-12 | 01:43 PM

I thank the author for addressing my comments and revising the manuscript. The changes have clarified the issues I raised and have made the overall position and structure of the paper. I therefore recommend accepting the paper.

-

REVIEWER 2

2025-10-06 | 12:12 PM

This paper takes an interesting new line on linguistic intuitions of acceptability and unacceptability, focussing on their function. This revised version strikes me as helpfully clearer about the contribution that the paper aims to make. The detailed discussion of examples in §5.3 is also very welcome.

I think there are still some issues to be dealt with, which could be addressed by relatively minor changes to the text.

In sum: I suggest that it’s important to acknowledge that the intuitions in question here are a subset of those that have been appealed to in generative (and other) grammar, and that the account of what generativists are committed to needs to be improved in certain specific ways. I also urge the author to say something about what the view expressed here commits him to on the causal aetiology of these intuitions. I finish with two worries about the role he attributes to optimal relevance.

I see that I am reviewer 2. I’m afraid that I have to live up to the billing now and do a reviewer 2 thing, namely to raise an issue I should have raised explicitly in my first review. I can only apologise for that. It’s not due to malice, but my failing to make enough time for careful thought first time around.

1) Here’s the issue. Generativist use of intuitions as data has always appealed to various different types of intuition: not just intuitions of unacceptability or ungrammaticality, but also intuitions of ambiguity, of what readings sentences can have, and of relatedness of sentences. (And of course there are many other intuitions that can reasonably be called ‘linguistic intuitions”: Gross 2021: 544; 546–8.).

In particular, generative work has made use since early work of intuitions of relatedness: e.g. that the declarative sentence corresponding to the interrogative in (1) is (2), not (3):

(1) Can eagles that fly eat?

(2) Eagles that fly can eat.

(3) Eagles that can fly eat.

This is connected to the criteria, also going back to early work, that i) a theory of grammar ‘strongly’ rather than ‘weakly’ generates: i.e. it generates structures, not just strings, and ii) that it explain how certain sentences are related through operations on those structures (transformations in early work, now the Move operation). (See Berwick et al., 2011, and Berwick, Chomsky & Piattelli-Palmarini, 2013; Smith & Allott 2016: 185–6.)

It’s also important to see that (un)acceptability intuitions have not only been used to tell us whether a string is/is not grammatical. For example, intuitions of the unacceptability of grammatical sentences, as in centre-embedding and garden path sentences are used as evidence for the kind of grammatical structure those sentences have.

All that said, it’s fine to focus in this paper on one kind of use of a subset of linguistic intuitions, and even a subset of the ones that have played key evidential roles in modern syntax. But that needs to be acknowledged, and the text needs to be amended because as it stands it’s inaccurate in places.

On this point I have three recommendations:

i) this restriction of focus should be noted, with some reference to generativist discussion of use of intuitions of readings and intuitions of relatedness (see above).

ii) the wording needs to be changed in a few places to reflect this limitation: e.g. this is too sweeping:

“Crucially, linguistic intuitions are accompanied by a distinctive, psychological sense that could be called “oddness” or “wrongness”.” (p. 2)

iii) it would also strengthen the paper if the author could say something about whether and, if so, very roughly how his programme could (in future work, presumably) explain ambiguity intuitions and intuitions of sentence-relatedness.

Incidentally, this issue is why I referred in my comments on the first version to the intuitions that the paper is concerned with as ‘(un)grammaticality intuitions’: not (as the author’s response suggests) to beg the question against the theory proposed here(!), but to isolate the phenomenon in question from all the other kinds of linguistic intuition that aren’t addressed here. I defined ‘(un)grammaticality intuitions’ thus:

“intuitions of unacceptability and acceptability of sentences, of the type that are usually understood to be related to the grammatical structure of the sentences in question rather than some other factor e.g. to use of a socially inappropriate expression or to merely pragmatic infelicity […] (I call this ‘grammatical (un)acceptability’ for short in what follows.)”This time I call them ‘putative (un)grammaticality intuitions’ to avoid further confusion.

The rest of my points concern issues that have come into focus for me with this revised version of the paper and Scott-Phillips’ helpful reply letter.

2) Contrary to what is suggested here, generativists are not committed to the claim that putative (un)grammaticality intuitions are due to a sense of what is grammatical and what is not as such. Rather it’s generally assumed that (un)acceptability is due to interaction, in particular with the parser but not only that: “Grammaticalness is only one of many factors that interact to determine acceptability” (Chomsky 1965: 11).

Of course, it is also assumed that many but not all of these failures to parse are due to the string in question being ungrammatical, and that unacceptability intuitions are often due to ungrammaticality. But as Jerry Fodor put it, “It’s always up for grabs what an intuition is an intuition of” (1999: 87). When generativists call something a grammaticality intuition they aren’t claiming that the phenomenology of how the intuition strikes the speaker/hearer gives it this status: rather, it’s a theoretical question which acceptability intuitions are grammaticality intuitions, and a further theoretical question how the acceptability/(un)acceptability relates to grammatical principles. This view is pretty clear in many generativist discussions of centre-embedding (e.g. Collins 2023: 6–7) and garden-path sentences (e.g. Collins 2008: 219 n. 19). On the general point see also Schütze (2019: 13–14).

Recommendations:

i) This issue needs some discussion to make accurate the account given here of the primary opposing view to the one proposed in this paper.

ii) Some wording needs to be changed to make statements about the view accurate. E.g. this sentence in particular needs some qualification or hedging: “It assumes that what people judge as (un)acceptable is revealing of what does and does not accord with their internal grammar.” (p. 3)

3) It is very helpful clarification that a central aim of the current paper is to explain the function of putative (un)grammaticality intuitions. But then I wonder why the paper doesn’t consider any rival views. In particular it would be good to hear a) what existing accounts of grammar and parsing are committed to on this point, and b) what it would be natural for them to say even if it goes a bit beyond what they are committed to.

For what it’s worth, it seems to me that anyone who thinks that there’s a dedicated parser can give a straightforward answer to the function question, drawing on work in recent decades on monitoring in speech comprehension (see Drożdżowicz 2017: 188 for references) and metacognitive feedback more generally. Roughly, when there’s parsing failure, there is often metacognitive feedback: we get a certain sensation, which inter alia serves the purpose of focussing our attention on the problem, like a warning light on the dashboard of a car. This seems a relatively theoretically costless account, at least in comparison to the account given here which seems to need to make more assumptions.

4) In his response letter, Scott-Phillips implies, I think, that he’s only committed to an answer on the functional question. But I am left wondering what his views here commit him to on the aetiology of putative (un)grammaticality intuitions. I think that this is an interesting issue and a natural question given what he does say, so I urge him to say a bit about it, even if it’s just that he thinks that his view on the function of putative (un)grammaticality intuitions commits him to nothing at all about their aetiology i.e. is completely neutral about how they are caused. That in itself would be interesting and worth saying!

I want to suggest (and only that) that a possible avenue for exploring what he may be committed to here connects to questions about abstraction in science, and whether his view implies that generativist syntax is wrongly idealising – and if so, with respect to acceptability or to grammaticality.

It seems to me that there are some commitments here on aetiology: in particular, the current proposal seems to be committed to agreeing with generativism that acceptability intuitions in general are causally downstream of a lot of different things: rules of grammar, parsing abilities, short-term memory limitations, prosody and how it affects parsing, pragmatic abilities, and probably more. That is, the view here as in Chomsky (1965) is that they are an interaction effect (see above).

But of course generativists (and actually many other linguists) claim that often enough, if we carefully abstract away from other factors, certain linguistic intuitions (see above for types that have been in focus) give us insight into the rules of grammar, very much as Galileo thought that abstracting away from friction allowed us to get insight into the laws of motion (Allott, Lohndal, & Rey 2021).

One way of reading the current proposal would be that it is committed to denying that this is possible: that putative (un)grammaticality intuitions never just match what grammatical rules on their own would do with the string, as it were ‘giving us a window into the grammar’. That would be a very interesting claim about the aetiology of putative (un)grammaticality intuitions.I also suspect that there’s a (very interesting) claim in the background about grammaticality, namely that there’s no (un)grammaticality tout court i.e. due solely to interactions between principles/rules of grammar; optimal relevance is always in the picture as well.

5) As I said above, the detail in §5.3 is welcome. It makes much clearer the relation of the current proposal to the work in construction grammar that it appeals to.

If I’ve understood it, the key argument here is that there’s a programme in construction grammar of trying to accounting for (all?) intuitions of unacceptability in terms of a clash between constructions such that “information packaging constructions place into the background information that is otherwise in the foreground, or vice versa” (p. 22) and that a major contribution of the current proposal is to explain why such clashes should lead to unacceptability intuitions.

One fairly minor point. There are citations on this general topic, but: given its crucial role in the argumentation, I think a citation is needed for this claim specifically: “Constructionist approaches, aiming to explain grammatical phenomena in terms of the functions of constructions make a clear, precise and otherwise surprising empirical prediction in this area: that intuitions of unacceptability should occur when, and only when, information packaging constructions place into the background information that is otherwise in the foreground, or vice versa.” (p. 23)

I finish with two worries about the positive proposal here:

6) A puzzle: Why couldn’t the sentences in (4) and (5) be optimally relevant in a case where the speaker wants to both a) say that (e.g.) Mary gave the speaker a book and b) imply something about the speaker’s (or someone else’s) non-native status (perhaps as a joke)? I suppose they could: after all, what would be a more economical way of doing both of those things? And yet, even in such contexts – where the utterance would be highly relevant – the string would strike us as unacceptable-because-ungrammatical: in fact this is why it can indicate that the speaker is a non-native speaker. I can’t see how the account proposed here could explain this. It seems to get the explanation the wrong way around, as it were.

7) Finally, a question about processing and ecological plausibility: Is is plausible that speaker/hearers spend effort working out and keeping track of the possibility/impossibility of optimal relevance for an utterance? There seems to be a clash here with assumptions made by work the author draws on. According to Sperber and Wilson, we don’t generally mentally represent relevance, because it would be too costly (Sperber 2005: 64; for more references and discussion see Allott 2013: 67) and we don’t need to. Wouldn’t it be much worse estimating the (im)possibility of optimal relevance? It seems that it would require considering many different possible scenarios.If the claim is that somehow we track possibility/impossibility of optimal relevance without representing it or calculating it, I think more needs to be said about how we could be doing this.

References (not including sources already listed in the paper)

Allott, N. (2013). Relevance theory. In A. Capone, F. Lo Piparo, & M. Carapezza (Eds.), Perspectives on Linguistic Pragmatics (pp. 57-98). Cham: Springer. Allott, N., Lohndal, T., & Rey, G. (2021). Chomsky’s “Galilean” explanatory style. In N. Allott, T. Lohndal, & G. Rey (Eds.), A Companion to Chomsky (pp. 517-528). Hoboken, NJ: Wiley Blackwell. Collins, J. (2008). Chomsky : A Guide for the Perplexed. London: Continuum. Collins, J. (2023). Generative linguistics: ‘Galilean style’. Language Sciences, 100, 101585. Smith, N. V., & Allott, N. (2016). Chomsky: Ideas and Ideals (3rd ed.). Cambridge: Cambridge University Press. Sperber, D. (2005). Modularity and relevance: How can a massively modular mind be flexible and context-sensitive? In P. Carruthers, S. Laurence, & S. Stich (Eds.), The Innate Mind: Structure and Contents (pp. 53-68). Oxford: Oxford University Press.

Author Reply

DOI: https://doi.org/10.25189/2675-4916.2025.V6.N3.ID868.A

-

ROUND 1

2025-09-26

Response letter for:Why do humans have linguistic intuition?

This is a very interesting proposal dealing with the underlying cognitive mechanisms and the"why" of a key concept inmuch of linguistics, namely language users' intuitions that sentences are "un/grammatical" or"un/acceptable." Mostresearch has assumed thatsuch judgements are a reflection of language users' linguistic knowledge about thestructureof their language and it has widely been used in linguistic theory building. The author proposes a truly novelexplanation, based on Relevance Theory and Construction Grammar, for the sense of "oddness" we have wheninterpreting ungrammatical or unacceptable sentences, namely that is stems not from language users' internalisedlinguistic system tracking and processing grammaticality and acceptbility per se, but that instead it is related to theprinciple of communicative relevance. Communicators presume that communicativestimuli are"optimally relevant",that is that they arethe most optimal and way to communicate a particular meaning and intention. "ungrammatical"and "unacceptable" sentences cannot be interpreted in a way that is consistent with the presumption of optimalrelevance, that is that there are no plausible cognitive benefits to be derived from the sentence. This fundamentallyshiftsthe view of what linguisticintuitions are, not directreflections of ourinternalised grammaticalsystem at work inparsing sentences, but instead a byproduct of the presumption of optimalrelevance. This hypothesisisrelevant not onlyfor researchersin relevancetheory and construction grammar, but forlinguistictheorising in general that makes use ofgrammaticality and acceptability judgements, meaning it potentially has very wide appeal.

I am obviously very gratified to read this positive summary and assessment.

However, for some aspects of the proposal, I would liketo see a bit moreevidence and discussions of itsimplications, tobolsterthe very strong - and controversial -claim made by the author.

I agree that the implications of my analysis are both significant and controversial. I am thereforehappy to add additional clarity, as detailed below.

First, some more contextualisation of how grammaticality and acceptability judgements have been used in theorisingand with what justification, maybe in a short paragraph would be helpful, for example to more recent developmentsassociating grammaticality and acceptability intutions with parsability (e.g. Leveida & Westergraard 2020)

I am torn two ways on this comment. I understand the recommendation and I agree with it up toa point. However, I very much do not want to risk any confusion or conflation between my question inthis paper (Why do humans have linguistic intuition?) and issues that are related but nevertheless different. To balance these two factors I have added a few sentences in the Introduction, in a couple ofdifferent places, to provide this extra context without risking any distraction from the focal question. Forexample, one of these small additions is this passage: “..So much so, in fact, that linguistic intuitions areoften called grammatical intuitions, and used to investigate grammar. The underlying assumption thatwhat people judge as (un)acceptable is revealing of what does and does not accord with their internalgrammar.”

I would also like to see a bit more detail on the proposed "psychological sense of oddness" that language users have.How exactly can we define oddness and isthere a way to measure and quantify it in a way that does notconflateit withun/grammaticality un/acceptability rating?

This paper takes the psychological sense of oddness to be a perceivable empirical phenomenon,and as such as something in need of explanation. As such the challenge (as I see it) is not so much to“define” the sense of oddness, but rather to describe it in ways that capture its natural character. And Ithink that the best way to do this is likely to be through neuroscientific effects, as this Reviewer suggestsin another comment below. Thus, to address both comments, I have added the following passage to theConclusion section of the paper:

“..This proximate answer could be enriched by connections with neuroscience, in particularN400 or P600 effects. We know these effects occur in language processing, but exactly what they revealremains an area of active research and debate. (Such connections might also allow us to identify thenatural character of the psychological sense of oddness more precisely than at present, and in ways thatdo not conflate it with judgements of acceptability.)”

I would also would like to see a bit more discussion of how itcan be attested that a particularsentence does not incureplausible cognitive benefits. That is, there deviations from conventional use that lead to language change. This alsorelates to inter-individual variation. For example, some resultative constructs such as "Shesneezed the napkin off thetable" might appear acceptableforsome, but not others, and expressionslike"samesame but different" or"long time nosee" are technically ungrammatical but acceptablefor others based on context and language and culturalexperience.Iam not suggesting theseexamples areincompatible with the proposal but would liketo see a bit more detail on how thedistinction can be made between sentences that can incur plausiblecognitive benefits and thosethatcan't and how we can make this distinction. Why is "long time no see" or "all your base are belong to us" acceptable to some and inferadditional cognitive benefits (e.g. as humourous expressions" but "I don't want going to the cinema" cannot give aplausible change in cognitive benefits (e.g. when looking at verb complementation, what is the relevance-relatedreason that *We decided going there *I hope doing it cannot be interpreted, e.g. as the speaker wanting to stress theprogressive nature of the action? That is I would generally see some more discussion of thelogic of how to arrive at theconclusion that a sentencecannot beinterpreted underthe presumption of optimalrelevance, which then explains whywefeel a sense of oddness when reading *I avoided to do it vsI avoided doing it.

I have added a paragraph to the end of §5.2 to address this. It reads:

“This kind of productive deviation from conventional use is plainly a critical aspect of linguisticcreativity and language change (Kaschak & Glenberg, 2000; Robenalt & Goldberg, 2015; Turner, 2018;Goldberg, 2019; Hoffmann, 2019; Bergs, 2025; inter alia). We can reasonably say that such sentencesincur increased cognitive costs relative to conventional use, but they also warrant meanings thatconventional use would not have done, and as such we cannot say that people will in general perceivethem as intrinsically impossible to interpret on the basis of a presumption of optimal relevance. Hence,we cannot and should not predict that these sentences will trigger intuitions of unacceptability. On thecontrary, if we should predict anything about such cases, we should predict significant amounts ofindividual variation. (This point is further reinforced by the fact that many individuals’ assessments ofsentences such as (6) to (8) will be affected by their own ideas about how language does or should work.For instance, self-identified prescriptivists may be more likely to say that such sentences areunacceptable.)”

Regarding the section "Mutually contradictory constrains,rendering optimisation impossible",I would liketo seesomeconcreteexamples ofcases whereconstraintscontradicteach other, as of now the discussion remainsrelatively abstract,which also holdsforthe discussion of Cuneo & Goldberg 2023.

Reviewer 2 also asks for a concrete example here. I have now rewritten §5.3 to include them. Thespecific example sentences are sentences (9) to (12). This new version of §5.3 also makes much clearerwhat I mean by “..intrinsically impossible to interpret…”.

Lastly, I was wondering if more could be said about the implications of the byproduct hypothesis and its predictions.For example, there is a wealth of work on the neuroscience of prediction, parsing, and unacceptability, such asN400and P600 effects. How are they reinterpreted in this framework and are there testable predictions of this hypothesisthat would allow distinguishing it from other models(e.g. different neural activation and involvement of a pragmaticcomponent when encountering unacceptablesentences)

This is a good question. A proper answer would need detailed review and analysis of the relevantneuroscience. While that is obviously beyond the scope of this paper, I agree with the reviewer that this isa direction worth highlighting. As mentioned above, I have addressed this through a comment in theConclusion, that also addresses the question of how to best characterise the psychological sense ofoddness.

Overall this is an intriguing and thought-provoking proposal, but in my view itshould besomewhat more developed,especially with regard to how to arrive at conclusions that particular utterances do not incur plausible cognitivebenefits and cannot beinterpreted using the presumption of optimalrelevance.

Again, I’m gratified by this positive assessment. I am also grateful for the reviewer’s detailedreading. I am hopeful that the enrichment of §5.3 provides the development that is recommended here.

Some minorcomments:

On page 3 the author points out that there have been no detailed investigationsinto why thereis an intutitivefeeling ofoddness when encountering an ungrammatical sentence. They then state that Chomsky "hasspeculated that perhapsan incidental biological mutation simply happened to generate the relevant cognitive dispositions" but in the sectionthe author then quotes Chomsky talks about the appearance of language and the ability for language acquisition ingeneral, not linguisticintutitionsspecifically,so the phrasing hereis a bit misleading.

That’s true, but Chomsky’s notion of the language faculty includes the grammaticalnessassumption, so the connection seemed justified to me. Nevertheless, it’s fair to say that this was slightlyunclear in the previous version. I’ve made an edit here to remove any possible misleading. The passagenow reads, “Chomsky has speculated that perhaps an incidental biological mutation simply happened togenerate the cognitive capacities and dispositions necessary for language (“Chomsky has speculated thatperhaps an incidental biological mutation simply happened to generate the cognitive capacities and dispositions necessary for language (“Perhaps it was a side effect of increased brain size… or perhapssome chance mutation”: 2010, p.59), and perhaps these dispositions could include an intuitiveassociation between ungrammaticality and a psychological sense of oddness. Such speculation hasbeen…”.

Page 9, "what people do always underdermines what they mean" could be explained in a bit more details withexamples, of how something is"underdetermined"

I’ve now enriched and elaborated this paragraph. I removed the term “underrdeterminacy” andmade the point more general. I also connected it to points made by Reviewer 3 about intersubjectivity inlanguage use. The paragraph now reads:

“How are communicative behaviors, linguistic or otherwise, understood at all? Humancommunication is flexible and open-ended, and the interpretation of communicative behaviour,including language use, is always intersubjective and context dependent (Sperber, 1995; Carston, 2002;Recanati, 2004; Verhagen, 2005; 2015; Langacker, 2008; Zlatev et al., 2008; Ludlow, 2014;Assimakopoulos, 2017; Geeraerts, 2021; Tantucci, 2021; inter alia). To take a very simple example, considerthe simple utterance “We’re on time”. First, this must be interpreted in context by the very fact that ituses a pronoun (“We”) and so who the utterance refers to depends on who the utterer is. Second, thismust be interpreted in context because the expression “..on time” can be used in many ways, for instanceto mean “..not late even though we expected we would be”, “Don’t worry, everything is fine”, “Oh, I readmy watch wrong, actually it’s 12 o’clock”, “..literally standing on a clock”, and so on. The point here is thatlanguage use is inherently open-ended and flexible, and that point in turn raises the question of how —how just possibly? — do we ever understand one another? How can we possibly converge on a (more or less)accurate interpretation of any utterance? Without some answer to this question, human communicationis quite mysterious.”

Page 24, it is not quite clear why "self-conscious" emotions such as pride and jealousy would not be seen associal. Asoneisjealous ofsomeoneitseemsto need a social dimension to be understood?

I’ve edited this sentence to be clearer and simpler. It now reads, “..individual and ‘internal’phenomena: examples include reasoning, episodic memory, belief formation and many of theemotions…”.

Reviewer 2 (Nicholas Elwyn Allott)

This paper proposes a seemingly radical answer to a familiar but interesting question: Why do we have intuitions ofunacceptability and acceptability of sentences, of thetypethat are usually understood to berelated to the grammaticalstructure of the sentences in question rather than some otherfactore.g. to use of a socially inappropriateexpression orto merely pragmaticinfelicity?(Icall this‘grammatical (un)acceptability’ forshort in what follows.)

This summary of the paper, while generally accurate, has some choices of terminology thatembed some assumptions that I do not accept. I state them here for clarity’s sake.

First, the alternative to the grammaticalness assumption is not “socially inappropriateexpression” nor “merely pragmatic infelicity”. The alternative is an impossibility of interpretationconsistent with the presumption of optimal relevance. I stated this explicitly before: “..the psychologicalsense of oddness is not caused by irrelevance, it is caused by an intrinsic impossibility of optimalrelevance. These two qualities are not the same.” The rewritten §5.3 now elaborates further and gives aspecific example.

Second, the paper consistently describes the empirical phenomenon in question as “linguisticintuitions” and not “grammatical (un)acceptability”, which is the term the reviewer prefers here. I do thisbecause the expression “grammatical (un)acceptability” tacitly embeds the very assumption that thispaper is questioning. I have now made this explicit in the Introduction: “..more often thegrammaticalness assumption goes unsaid. So much so, in fact, that linguistic intuitions are often calledgrammatical intuitions, and used to investigate grammar…”.

The answer proposed is that these intuitions are due to communicative principles: specifically a sentence isgrammatically unacceptable if and only if “there appears to be an intrinsic impossibility of interpreting them in anyway consistent with the unconscious presumption of optimalrelevancethat underpinstheinterpretation of allstimuliperceived as communicative.” (p.1)It issuggested that this“impliesthat there may be no particularcognitivecapacitythat functions to distinguish the grammatical from the ungrammatical.” (p. 1) So far, so radical: many apparently grammatical facts are actually due to pragmatics, and this is part of a programme tending towards thorough-goingrejection of the idea that there is a domain-specific mental grammar, and thus of generativism and the conception oflinguistics duetoNoam Chomsky.

However this apparent radicalism is not fully borne out by the theory presented, which leans not only onrelevance theoretic pragmatics but on a theory (or family of theories) of grammar, construction grammar. On thisaccount, grammatical (un)acceptability intuitions are causally downstream of both grammar and pragmaticprocessing. This is common ground with generativists (although obviously they would disagree about whichgrammatical theory is right). Further, it becomesclearin Scott-Phillips’ discussion ofcentre-embedded sentencesthat,like generativists, he also sees parsing as a causal factor here, and also short-term memory limitations(to at least theextent that they areimplicated in some parsing problems).

As far as I can see, what is new here, then, is not any claim that the grammarisn’t involved in grammatical(un)acceptability intuitions. Rather it is the centrality of pragmatic principles and the kind of role that they play.Specifically, the claim, I think, is that the principle of communicativerelevance always mediatesthecausalrole of thegrammar in these intuitions. There are no strings that are (un)acceptable only because the grammar does not/doesassign a structure to them. Rather, strings that are assigned a structure by the grammar may be unacceptable if theparts that are combined are communicatively incompatible. At least I think this is the central claim here. It’sinteresting and stillradical, but in rather a different way from theimpression given in theframing of this paper

As above, this summary is more-or-less correct but the terminology is misleading. I must pushback on two points in particular.

First, I would not say that the communicative principle of relevance “mediates” the role ofgrammar, because that terminology gently implies that pragmatics serves grammar. The argument inthis paper is based on the opposite picture: grammar serves pragmatics. This picture is presented in §2(and citations therein). So if we must talk of factors mediating one another, then this paper is based onthe idea that grammar mediates pragmatics, not the other way around.

Second, nowhere does the paper mention strings or structures. The reviewer has restated themain claim of the paper as “There are no strings that are (un)acceptable only because the grammar doesnot/does assign a structure to them”. I can neither endorse nor reject this summary, because I don’tbelieve that talking of grammars assigning structures to strings is the best way to describe the relevantempirical phenomena in the first place (for reasons summarised in §2.1).

Ithink there areseveral problems with the paper and the proposal asitstands:

1) The roles and importance of the qualifiers‘appearsto be’ and ‘intrinsic’ in thecondition proposed could do with someclarification.

“Appears to be” was not a qualifier of the scientific claim, it was a description of the key empiricalphenomenon. I could have simply written, “Sentences are unacceptable when, and only when, thesentence is intrinsically impossible to interpret in any way consistent with the unconscious presumptionof optimal relevance…” rather than “Sentences are unacceptable when, and only when, there appears tobe an intrinsic impossibility of interpreting the sentence in any way consistent with the unconsciouspresumption of optimal relevance”. However, while the first of these phrasings is simpler and easier tofollow, it is also misleading, because it implies that the critical empirical phenomenon is the sentence,when in fact the critical empirical phenomenon is an audience’s attempted interpretation of thesentence. Hence, the second phrasing is more precise.

Nevertheless, I’m grateful to the reviewer for raising this issue, because this is a subtle point butcritical to my article. This is why I have changed my description of the key trigger of linguistic intuitionsthroughout the paper, from “appears to be an intrinsic impossibility” (and other such phrases), tophrases that talk of the logical impossibility of having the property of optimal relevance.

2) There’s quite a bit of ‘prior art’ here which is neglected. Scott-Phillips writes,“thereis very littlefocused attention onthe question of why humans have these intuitions in the first place.” It would be fair to say that there’s no generallyaccepted detailed account of the aetiology of grammatical (un)acceptability intuitions, but there’s plenty of researchwhich has addressed this question, including work in philosophy by Michael Devitt and Gareth Fitzgerald (which iscited here), Georges Rey and Anna Drożdżowicz (not cited), among others, and work on the relation of parsing to thegrammarinter alia by Colin Phillips, Shevaun Lewis andDave Kush among many others. (Seereferences below.)

Perhaps Scott-Phillips’ point here is that work of these kinds doesn’t generally try to provide an explanationfor the origin in our species of grammatical (un)acceptability intuitions(although Chomsky and otherscertainly havediscussed the evolutionary origins of the language faculty). If so, it would be worth distinguishing more clearly twoquestions here: the one about the aetiology of these intuitions, and the one about their origin (where Scott-Phillipsfavours an account in terms of evolutionary natural selection), and noting that there have been many attempts to answer the first of them (see discussion in Gross 2021). Asfar asIcan see, the primary claim madein thecurrent paperis about the aetiology, not the origin, although thelatterisin the background. Someevaluation of other accounts of theaetiology is necessary to the project of this paper, given that they arerivalsto the account presented here.

The reviewer anticipates my response: this prior work is not about the same topic as this paper. Itis about a nearby but nevertheless different topic. To be precise, the question of my paper is neither of thetwo possibilities raised by the reviewer. The precise question is the one in the title: Why do humans havelinguistic intuition? Not, “How did these intuitions emerge in our species?”, but “Why does our specieshave these intuitions at all?”. (In evolutionary theory, these are often characterised as the phylogenyquestion and the functional question, respectively.)

That said, I see now that my use of the word “aetiology” was misleading, and I’m grateful to havethis brought to my attention. I’ve now removed it, made some other small but important edits to theIntroduction, and also added some new passages to §3, all so the exact scope of the issue is clear. Indoing so I have added citations to the most relevant prior work named by the reviewer here.

3) Relevance – and therefore optimal relevance – is relative to context. Thus it seems to follow from the proposal herethat grammatical (un)acceptability intuitions are context-relative. But grammaticality isstandardly assumed to be aproperty of sentence-types (not utterances or sentence tokens), and empirically this seemsto becorrect: mostly when astring of words strikes us as unacceptable and theexplanation given is grammatical, providing a context doesn’t help.This looks bad for any pragmatic account of grammatical (un)acceptability intuitions, so needs discussion here.Perhaps the intention is that this is dealt with by making the criterion (apparent intrinsic) impossibility of anoptimally relevant interpretation, where this entails unacceptability across contexts. If so that needs spelling out,showing how contextshiftcannot help.

Yes, it follows from my arguments that linguistic intuitions are context sensitive. The reviewer isalso correct that providing a context tends to not much change linguistic intuitions. This is a challenge tomy arguments, and so I have added a new subsection, §6.4, to address it. The key new paragraph is asfollows:

“There are (at least) two important points to make here. First, the fact (i) that linguistic intuitionsare determined in context does not at all entail (ii) that changing the social situation should change theintuition. There is just no inherent contradiction between (i) and (ii). Second, the byproduct hypothesis actually predicts that changing the social situation should not change linguistic intuitions at all, because itstates that what triggers linguistic intuitions is the perception of a logical impossibility of optimalrelevance i.e. an impossibility regardless of social situation. In §5 I identified the three ways in which asentence can have this property.”

4) The role of construction grammarin the argument presented hereisrather unclear.Istheclaim theconditional one:“If construction grammar is the right account of grammar, then the explanation of grammatical (un)acceptability ispragmatic principles”? Or the more ambitious conjunction: “Construction grammar is the right account of grammarand the explanation of grammatical (un)acceptability is pragmatic principles”? Generativists could grant theconditional claim, since they take the antecedent to be false. It’s not clear why they should be moved by theargumentation presented in this paper if the conjunction is claimed,sincetheclaim thatconstruction grammaristheright account of grammaris hardly supported here.

It is both.

In the first instance, the primary claim is what the reviewer calls here the “conditional claim”.That is, I accept as a premise to the arguments here that constructionist approaches are on the righttrack. This is already explicit in the manuscript, for instance in the title of §2, “Two assumptions” (thefirst assumption is about constructionist approaches, the second is about the communicative principleof relevance).

But secondarily, the paper here supports the constructionist agenda to the extent that the paper’sconclusions provide a simple and distinctive explanation of the empirical phenomenon. It is a normalpart of scientific progress, that a premise (or set of premises) are supported to the extent that theconsequences of accepting those premises give better explanations than before, where “better” isunderstood in terms of ordinary scientific desiderata such as simplicity, plausibility, parsimony,consistency with a wide range of evidence, and so on.

This point was only tacit in the previous version of the manuscript. I have added a passage to §2.1to make it explicit. The key paragraph reads:

“The analysis of linguistic intuitions that I present in this paper assumes that constructionistapproaches are on the right track i.e. that grammars are indeed best described and understood in termsof constructions. At the same time, this paper also supports the constructionist agenda to the extent that its conclusions provide a simple and distinctive explanation of why humans have linguistic intuition. It isa normal part of scientific progress, that a premise (or set of premises) are supported to the extent thatthe consequences of accepting those premises give better explanations than before, where ‘better’ isunderstood in terms of ordinary scientific desiderata such as simplicity, plausibility, parsimony,consistency with a wide range of evidence, and so on. For example, Newton’s laws of motion weresupported by the fact that they provided for better explanations of basic empirical phenomena, such asthe movements of celestial objects. In this case, the basic empirical phenomenon is the existence oflinguistic intuition, including the psychological sense of oddness. In §3-§7 I will present (what I believeis) an especially parsimonious explanation of how these intuitions work and why humans have them.”

5) In the discussion of centre-embedded sentences, there’s no mention of the account in terms of the interface withprosody provided by Fodor et al (2017). Such discussion seemsto be needed here, becausethat account appearsto havebetter predictive powersthan older accountsin terms of overloading short-term memory.Itexplains not only why somegrammatical centre-embedded sentences are unacceptable, but also why others are not,e.g.“Therusty old ceiling pipesthat the plumber my dad trained fixed continue to leak occasionally”. It would obviously be a merit of the proposalhereif itcould account forthesefactsin another way.

I did not know of Fodor et al., 2017: I’m grateful to have it pointed out to me.

The effects described in that paper are straightforwardly accounted for by my proposals, for thereasons given by Fodor et al. themselves, namely that prosody can help to demarcate morphological andsyntactic phenomena, such as embedded clauses, and hence aid processing. Of course, what “aidprocessing” means depends on your analysis of grammar — generativists and constructionists tend toconceive of language processing in very different ways — but the basic point that prosody can aidlanguage processing is straightforwardly compatible. I have made this point explicit in the relevantsection of the paper (§4.1).

6) This paper sets out three ways that intuitions of grammatical unacceptability can arise. The discussions of the firsttwo ways are interesting. The third way strikes me as under-described here. The explanation points towards recentwork in construction grammar that attempts to (re)describe “many grammatical phenomena” in this way,particularly island effects. But no examples are given. The brief discussion of conflicting information packaging is interesting but entirely abstract. For the argumentative purposes of this paper, thissubsection needsto becompelling,becauseitseemsto be how many of the key examplesin theliterature areto be accounted for. (See next point.)

Reviewer 1 also suggests that §5.3 is under-described. I have therefore rewritten and lengthenedthis section, including examples

7) One target of this paper seems to be Chomsky’s category of ‘Fine Thoughts’, which Rey calls ‘WhyNots’: strings ofwords that strike us as unacceptable even though we would be able to understand what is meant by them if they wereused (see e.g. Allott & Rey, 2017; Rey 2000: 21–35). Both take these data to be important to the generativist project,because they seem to establish that there’ssomesystem in our mind that assigns one or morestructuresto somestringsand noneto others, independently of ourcommunicative goals. (Chomsky’s view isthat ungrammaticality is when thegrammar assigns zero structures to a string: see Berwick et al., 2011.)It’s odd that they aren’t mentioned here.Ithinkthat in orderto serveits argumentative goals, this paper needsto show how the proposal it givescan account for at leastsometypes.

To end this review on a positive note, I want to say that it seems to me as though the proposalsketched herecould push this debateforwards. As noted above, Chomsky and Rey (and others) arguethat ‘Fine Thoughts’/‘WhyNots’are best explained by – and so are evidencefor – a syntax system governed by its own domain-specific principles,sincethey are unacceptable even though they would serve for communication. I think Scott-Phillips’ proposal claims thatthe intuitions might derive from communicative principles nonetheless: theclaim is not that the‘Fine Thought’stringscouldn’t be used for communication, but that they could never be optimally relevant. That’s a new movein the debate,and it would beinteresting if fleshing it out morerevealsit to haveexplanatory power.

“Fine thoughts” / “Why nots” are indeed empirical phenomena relevant to this paper. (They arenot, however, a “target” of the paper. If they were, I would have mentioned them explicitly!)

At the most plain and basic level, what “Fine thoughts” / “Why nots” show is that possible utilityin communication is not what delimits the acceptable from the unacceptable. The reviewer states thatthis basic empirical observation “seem[s] to establish that there’s some system in our mind that assignsone or more structures to some strings and none to others, independently of our communicative goals”. Idon’t accept this inference. It is a further inference that need not follow from the basic empiricalobservation.

Nevertheless, the challenge is fair enough and should be addressed. I have therefore added a newsubsection, §6.3, titled “Utility in communication does not predict unacceptability” that addresses itexplicitly. The key point is that, on my analysis, “Fine thoughts” and “Why nots” are predicted and to beexpected exactly when (and to the extent that) “possible utility in communication” is different to“impossibility of optimal relevance”.

Reviewer 3 (Lilian Vieira Ferrari)

[Reviewer 3 begins their review with a lengthy summary of my article. It is accurate and I’m grateful forthe detailed reading. I’ve not included it in this response letter just to save space.]

The proposal isconsistent and deservesto beexplored in futureresearch.

As with the other reviews, I am very happy to read this positive assessment.

As pointed out in the paper’s last section (Future Directions), particularly relevant issues are the role ofintersubjectivity, common ground and commitment/ accountability/ argumentativity. Regarding these issues, thefollowing publications may berelevant:

● Verhagen, Arie (2005). Constructions ofIntersubjectivity.Discourse, Syntax, and Cognition. Oxford: OxfordUniversity Press.

● Verhagen, Arie. (2015). Grammar and cooperative communication. In: Ewa Dąbrowska & Dagmar Divjak(eds.). Handbook of Cognitive Linguistics (Handbooks of Linguistics and Communication Science 39),232–252. Berlin & Boston:De Gruyter Mouton.

● Verhagen, Arie. (2025). Intersubjectivity. In Wen & Sinha (eds.). Cambridge Encyclopedia of CognitiveLinguistics.

In these publications, the author seeks to combine Langacker's proposal within the scope of Cognitive Grammar,Fauconnier's Mental Spaces Theory, and Ducrot's theory of argumentativity, proposing that certain grammaticalitems operate in the dimension of coordination between a speaker and a listener (real or virtual) in triadic,intersubjective communication, modifying the strength or direction of the inferences associated with a particularutterance

This research agenda is indeed relevant. I have now cited it in the passage on how language use isalways intersubjective and context dependent (quoted above in response to a comment of Reviewer 1).

Other publicationsthatcan be highlighted are:

● Geeraerts, Dirk. Second-order empathy , pragmatic ambiguity, and irony, In Augusto Soares da Silva (ed),Figurativity in language – intersubjectivity and usage. Amsterdam/Philadelphia: John Benjamins, 2021.

● Tantucci, Vittorio (2021). Language and Social Minds. The Semantics and Pragmatics of Intersubjectivity.Cambridge: Cambridge University Press.

● Zlatev, Jordan, Timothy Racine, Chris Sinha, Esa Itkonen (2008).Intersubjectivity: What makes us human?In: Zlatev et al. (eds.), The Shared Mind: Perspectives on Intersubjectivity. Amsterdam/Philadelphia: JohnBenjamins Publishing Company,1-14These are also useful. I have added them to the same passage.

Reviewer 4 (Augusto Soares da Silva)

The author offers a highly innovative and thought-stimulating responseto thefundamental questions of why humanshave linguistic intuitions and how thesecan beexplained in psychological,evolutionary, and linguisticterms, withoutinvoking a supposed cognitive capacity to distinguish the grammatical from the ungrammatical. The author’sproposal to consider linguistic intuitions as by-product effects of core cognitive capacities for the interpretation ofcommunicative stimuli is both theoretically and empirically well-grounded,convincing, and clearly demonstratestheinadequacy of the traditional (generativist) “grammaticalness assumption”. Furthermore, the paper brings togetherinterdisciplinary evidence from cognitive linguistics, cognitive psychology, and evolutionary psychology. The paperfully meetstherequirementsfor publication essentially in itscurrent form.

Naturally, I’m very happy to read this very positive assessment.

Recommendationsforthe authors

1. Constructions and Construction Grammar: On page 6, please add a reference to Diessel, H. (2019), The GrammarNetwork: How Linguistic Structure is Shaped by Language Use, Cambridge University Press. This is an importantsource.

This has been added.

2. The expression “communicative function,” which is presented as definitional of the concept of construction and asessential to the Principle of Non-Synonymy, is somewhat vague. Why not simply use “meaning”? What does“communicativefunction” add to orsubtract from “meaning”?

The reviewer is right that “meaning” will be more straightforward to some readers, especiallythose coming from cognitive linguistics. But other audiences, for instance those coming from thecognitive psychology of human communication, will find “communicative function” more natural. I’venow edited the passage so both terms are available.

Moreover, the author seems to adhere to the traditional distinction between semantics and pragmatics, which is notwidely accepted – or at least notconsensual –within cognitivelinguistics.

I’m not sure why the reviewer believes that the paper adheres to any traditional distinctionbetween semantics and pragmatics. Indeed, I don’t believe I do follow any such adherence. If someclarification can be provided I will happily consider edits to make this issue clear.

3. Principle of Non-Synonymy: Formally distinctconstructions are generally (and probably always)semantically andfunctionally different. Still, especially in the case of constructional meaning, it is advisable to distinguish betweenpropositional or referential meaning and construal (Langacker 2008): differences between alternating constructionsoften occur at the level of construal, not at the level of propositional meaning. Although it is very difficult (or perhapsimpossible) to counter the Principle of Non-Synonymy at the grammatical level – i.e. in onomasiological variationbetween constructions – this is not necessarily the case in the lexicon. For example, the onomasiological variationbetween underground and subway, both referring to subterranean railway systems, illustrates a case of denotationalsynonymy –that is, there is no difference in denotational meaning, only in connotational or sociolinguistic meaning(underground is typically British English, whereas subway is typically American English). How can the Principle ofNon-Synonymy be maintained in face of such lexically-based constructions that are denotationally synonymous (seeTable1)?

I have added a sentence to the end of the relevant paragraph to address this question. It reads,“For instance, while ‘subway’, ‘metro’ and ‘underground’ are not different in their denotation (they alldescribe subterranean railway systems) they nevertheless all have different coverage because their vary in their connotations or sociolinguistics (each is the more common term in different parts of the world).” Ihave also cited Langacker, 2008 at a relevant point in the manuscript.

4. Emotions and the social nature of the human mind (p. 24): It is not only so-called “self-conscious” emotions likepride and jealousy that are social in nature;so-called “basic”emotionssuch as anger are also socially embedded. Bothanger and pride aresensitiveto social variation and cultural influences,even within thesamelanguage. See:

● Soares da Silva, A. (2021), Measuring the impact of (non)figurativity in the cultural conceptualization ofemotions in the two main national varieties of Portuguese.In A. Soares da Silva (Ed.), Figurative Language:Intersubjectivity and Usage(pp. 387–438). John Benjamins[a study on anger and pride]

● Soares da Silva, A. (2022), Metaphor,emotion, and intralinguisticcultural variation: Metaphors of angerinEuropean and Brazilian Portuguese”. In: U. Schröder, M. Mendes de Oliveira, and A. M. Tenuta (Eds.),Metaphorical Conceptualizations. (Inter)Cultural Perspectives(pp.189–222).De Gruyter.

This is a good point. I have added citations to the relevant paragraph (the final one of themanuscript).-

ROUND 2

2025-11-03

Answers to the Editors:"Please add a short, explicit statement clarifying that the paper focuses on a subset of linguistic intuitions (acceptability/unacceptability) and does not directly address other types (e.g., ambiguity judgments, relatedness intuitions). It would also strengthen the manuscript if you could briefly indicate whether, and how, the proposed framework could be extended to account for those other types in future work."Reviewer 2’s comments on this point were useful because they revealed to me that we are using the expression “linguistic intuition” in two different ways. I think my use is reasonable. Indeed, three other reviewers of this manuscript understood it as I intended, as did many audiences when I have presented this work in person (see Acknowledgements). I have therefore added a footnote to make especially explicit how I am using the term:

“For the avoidance of possible misunderstanding: I am using the expression ‘linguistic intuition’ to refer to the naturally occurring psychological sense that a sentence is ‘not right’ in some way. This sense is a cognitive phenomenon and ‘intuition’ is, in my view, a suitable term to describe it. I state this explicitly because in some literatures, or in some schools of thought, the notion of ‘linguistic intuition’ is used more broadly, to refer to a broad array of possible judgements that can be made about sentences — matters such as ambiguity, pronounceability, social appropriateness, entailment, frequency and so on — which are then used to construct theories of language (Gross, 2021). This broader class of judgments are not accompanied by a psychological sense of oddness and as such they are not my target here.”"Some formulations still slightly oversimplify the generativist position — particularly the idea that generativists equate acceptability with grammaticality. A brief clarification that generative approaches view acceptability as the result of multiple interacting factors (including but not limited to grammar) would bring the framing fully in line with the literature."I do not agree with this reading of my paper, and therefore not with the suggestion. Contrary to this reading, my manuscript does not present a generativist position on the issue at hand, and therefore it cannot oversimplify any such position! (This is related to the point above, that I am using the term ‘linguistic intuition’ differently to the reviewer.) Moreover, nowhere does my manuscript state that “generativists equate acceptability with grammaticality”. What the manuscript does say is that there is a widespread assumption (not specific to generativists) that “humans must have some species-universal sense of what is grammatical and what is not” (see Introduction). The manuscript also quotes an explicit statement to this effect from Chomsky (“The only thing we can say directly is that the speaker has an ‘intuitive sense of grammaticalness”). These passages do not amount to any claim that generativists equate acceptability with grammaticality. Indeed the passage is carefully worded to avoid any suggestion of such a claim. As such it is already in line with the existing literature.Reviewer 2 does point to one specific sentence that they felt needing hedging or qualification. I have removed that sentence and made the passage around it even clearer."A reviewer suggests that readers will expect some comment on the causal origin of linguistic intuitions. Even a short paragraph clarifying whether your account is agnostic about their aetiology — or whether it entails specific commitments — would address this point."I suspect there must be some deeper misunderstanding between Reviewer 2 and myself here. In the original submission I used the word “aetiology”, but this word can be used in a number of ways and it hence led to a misunderstanding. I therefore removed it for the new submission. However, there is and has always been, throughout the manuscript, an argument that linguistic intuitions are caused by cognitive capacities for the interpretation of communicative stimuli. Indeed this is a central point and it has always been stated clearly in the Abstract and Introduction. It seems to me that these points speak directly to Reviewer 2’s concerns. So as I say, I suspect there is some misunderstanding here. If necessary I will be happy to respond to any clarification."A short paragraph contrasting your explanation of the function of intuitions with competing accounts (e.g., parser-based or metacognitive feedback models) would help situate the contribution and highlight its distinctive explanatory value."I think such a paragraph would be potentially confusing. But I have made other edits to address the underlying issue here.Reviewer 2 writes that “It is very helpful clarification that a central aim of the current paper is to explain the function of putative (un)grammaticality intuitions”. But this summary is not right. Contrast:

(a) to explain the function of the cognitive capacities that trigger putative (un)grammaticality intuitions, with (b) to explain the function of putative (un)grammaticality intuitions. The aim of the paper is (a) and not (b). The Reviewer has read my previous response as asserting (b), and their recommendation would pertain if that was the aim. But the aim is actually (a). As such, adding a paragraph predicated on the assumption that the aim is (b) has a high chance of confusing readers.Nevertheless, re-reading again I see that while the contrast between (a) and (b) was strongly implied, it could be made more explicit still. I have therefore edited the key relevant passage in the Introduction to be fully explicit. I have also reviewed the whole manuscript with this point in mind and made some minor edits to avoid any ambiguity. The key relevant passage in the Introduction now reads as follows:

“Yet from an evolutionary and, in particular, adaptationist perspective, the grammaticalness assumption is hard to justify. What ecological conditions would select, not simply for a sensitivity to what is and is not grammatical, but for a cognitive capacity that delivers an intuitive association between ungrammaticality and a psychological sense of oddness? To what fitness enhancing task would such a capacity contribute? I am not asking how such a capacity might emerge in our species (a question about phylogeny), I am asking why our species should have such a capacity in the first place (a question about function, or adaptation). As far as I am aware, no detailed, focused or theoretically principled answer to this question has ever been developed.

“Chomsky has speculated that perhaps an incidental biological mutation simply happened to generate the cognitive capacities and dispositions necessary for language (“Perhaps it was a side effect of increased brain size... or perhaps some chance mutation”: 2010, p.59), and perhaps these dispositions could, by chance, include an intuitive association between ungrammaticality and a psychological sense of oddness. Such speculation has been widely criticized on grounds of evolutionary implausibility (Pinker & Bloom, 1990; Chater et al., 2009; Planer, 2017; Hurford, 2018; Martins & Boeckx, 2019; de Boer et al., 2020). But still, it is not as if there are other possible justifications of the grammaticalness assumption with greater plausibility. The grammaticalness assumption has rather been taken for granted, by Chomskyians and non-Chomskyians alike, and put to use as a foundational assumption in the investigation of grammar itself.”