HYPOTHESIS AND THEORY published: 10 March 2020 doi: 10.3389/fpsyg.2020.00364 Acceptable Ungrammatical Sentences, Unacceptable Grammatical Sentences, and the Role of the Cognitive Parser Evelina Leivada 1* and Marit Westergaard 2,3 1 Universitat Rovira i Virgili, Tarragona, Spain, 2 Arctic University of Norway, Tromsø, Norway, 3 Norwegian University of Science and Technology, Trondheim, Norway A search for the terms “acceptability judgment tasks” and “language” and “grammaticality judgment tasks” and “language” produces results which report findings that are based on the exact same elicitation technique. Although certain scholars have argued that acceptability and grammaticality are two separable notions that refer to different concepts, there are contexts in which the two terms are used interchangeably. The present work reaffirms that these two notions and their scales do not coincide: there are sentences that are acceptable, even though they are ungrammatical, and Edited by: sentences that are unacceptable, despite being grammatical. First, we adduce a Urtzi Etxeberria, number of examples for both cases, including grammatical illusions, violations of Identity Centre National de la Recherche Avoidance, and sentences that involve a level of processing complexity that overloads Scientifique (CNRS), France the cognitive parser and tricks it into (un)acceptability. We then discuss whether the Reviewed by: Roelien Bastiaanse, acceptability of grammatically ill-formed sentences entails that we assign a meaning University of Groningen, Netherlands to them. Last, it is shown that there are n ways of unacceptability, and two ways of Kepa Erdocia, University of the Basque Country, ungrammaticality, in the absolute and the relative sense. Since the use of the terms Spain “acceptable” and “grammatical” is often found in experiments that constitute the core of *Correspondence: the evidential base of linguistics, disentangling their various uses is likely to aid the field Evelina Leivada reach a better level of terminological clarity. evelina@biolinguistics.eu Keywords: grammaticality, grammatical illusions, syntactic islands, parser, processing Specialty section: This article was submitted to Language Sciences, INTRODUCTION a section of the journal Frontiers in Psychology Introspective linguistic judgments about the well-formedness of linguistic stimuli have long been Received: 23 August 2019 regarded as one of the most important sources of evidence in linguistics, essentially forming its Accepted: 17 February 2020 empirical base (Wexler et al., 1975; Carr, 1990; Schütze, 1996/2016; Baggio et al., 2012). Both Published: 10 March 2020 the techniques used to elicit such judgments (e.g., controlled experiments, self-introspection, or Citation: targeted questioning about whether a specific sentence sounds fine in a specific language) as Leivada E and Westergaard M well as the type of sample that is necessary for the results to have ecological validity (e.g., a (2020) Acceptable Ungrammatical pool of participants that is randomly selected from the targeted linguistic community, a non- Sentences, Unacceptable Grammatical Sentences, and the Role random sample, or self-introspection) are a matter of debate (see Phillips, 2009; Gibson and of the Cognitive Parser. Fedorenko, 2010; Sprouse and Almeida, 2013; Branigan and Pickering, 2016). On the other hand, Front. Psychol. 11:364. no controversy exists over the fact that judgments about what forms part of a person’s linguistic doi: 10.3389/fpsyg.2020.00364 repertoire constitute a rich source of information in theoretical and experimental linguistics. Frontiers in Psychology | www.frontiersin.org 1 March 2020 | Volume 11 | Article 364 Leivada and Westergaard Acceptability, Grammaticality, and the Parser Since these judgments have such a key role in the study of Answering this question is the main goal of the present language, one would expect that the question of what they tap work. The starting point of the discussion is Chomsky’s into would be one of the first questions in linguistics to provide (1965) distinction between the terms “acceptability” and an indisputable answer to. But that does not seem to be the case. “grammaticalness,” according to which these two notions If one searches PubMed or any other database for the terms and their scales might not coincide, hence his reference to “acceptability judgment tasks” and “language” on the one hand, “unacceptable grammatical sentences”: sentences that do not and “grammaticality judgment tasks” and “language” on the form part of grammar for reasons that have nothing to do other, one will quickly discover that the relevant experiments that with grammar. The second aim of the present work is to chart will show up are the same. They all report findings that are based the variation space that is created when one disentangles the on the exact same elicitation technique. Perhaps the greatest two notions: unacceptable grammatical sentences, acceptable illustration of how the terms “acceptability” and “grammaticality” ungrammatical sentences, their respective parsability, and are used, often without a clear distinction in place, comes from the process of assigning them meaning. Last, the scales of Schütze’s (1996/2016) seminal book on linguistic judgments. grammaticality and acceptability will be discussed and it will While the title of the book talks about “grammaticality judgments be shown that they do not coincide: there are n ways of and linguistic methodology,” the very first quote given in the unacceptability, but only two ways of ungrammaticality, in the 2016 edition of the book is by Bever (1970), who claims that absolute and the relative sense. it is simultaneously the greatest virtue and failing of linguistic theory that acceptability judgments are used as the basic data (Schütze, 1996/2016: v). In the preface of the first edition, it ACCEPTABLE UNGRAMMATICAL is argued that “[t]hroughout much of the history of linguistics, SENTENCES AND UNACCEPTABLE judgments of the grammaticality/acceptability of sentences (and other linguistic intuitions) have been the major source of GRAMMATICAL SENTENCES evidence in constructing grammars” (p. xi, emphasis added). Humans are surprisingly good at providing accurate and Just as linguists and other cognitive scientists have at times consistent judgments about what forms part of their linguistic used the terms “ungrammatical” and “unacceptable” roughly repertoire.2 Although informants’ opinions about their linguistic synonymously, plurality and overlapping may characterize the behavior are not always concordant with the way they actually use of symbols like ?, *, or ??, that are employed to signal some speak (Labov, 1996; Cornips and Poletto, 2005), acceptability deviant property of the linguistic stimulus (Bard et al., 1996). To judgment tasks are reliable as a tool, and the majority of linguistic define the relevant terms, the grammaticality of a sentence refers stimuli can receive unambiguous, consistent judgments. For to whether the sentence conforms to the syntactic rules of a given example, little debate would occur among native speakers of language (Fromkin and Rodman, 1998: 106), or put another way, English about the acceptability of (1) or the unacceptability of (2). “it is a characteristic of the stimulus itself ” (Bard et al., 1996: 33). The former is a grammatically well-formed sentence of English, With respect to acceptability, the focus moves from the stimulus while the latter is a word-salad that would probably be read and to a speaker’s perception; in Bard et al.’s (1996) words, it “is a parsed in a rhythm that pertains more to lists of objects than to characteristic of the stimulus as perceived by a speaker” (p. 33). connected speech. Linguistics, however, is not a science that works exclusively with visible primitives; we cannot zoom in on a linguistic (1) John said to Mary that he likes doing linguistics. stimulus until we find and tease apart an independent, self- (2) *To he likes that linguistics John Mary doing said. contained grammatical core. This means that grammaticality, Yet, even though such judgments are largely coherent with as one of the possible elements that determine acceptability, the actual shape of speakers’ internalized grammar, there are “is not directly accessible to observation or measurement” (Lau some stimuli that have the ability to trick the cognitive parser et al., 2016: 3). The question thus becomes: How do we know into unlawfully accepting or rejecting them. Chomsky’s (1965) anything about grammaticality aside of the information provided discussion of “unacceptable grammatical sentences” mentions by acceptability? Put differently, if grammaticality is defined as several performance-associated factors that explain why a “conforming to the rules of the grammar of language X” and if the linguistic stimulus that does not violate any rule of grammar grammar of language X has the shape that its speakers’ judgments would be rejected by speakers as unacceptable. Factors such as and actual performance give it, what way do we have to capture memory limitations, processing constraints, as well as discourse, grammaticality other than the one that goes through speakers’ intonational and stylistic factors may all induce such an effect. For perception of well-formedness (i.e., acceptability)?1 the most important challenges that linguists working with small or non-standard 1 An obvious answer could be that rules of grammar could be extrapolated through languages face (Leivada et al., 2019a). For these reasons, native judgments are an corpora of naturalistic speech. Although such corpora are useful, they cannot indispensable tool for most linguists. 2 substitute judgments, for two reasons. First, they are informative only about what This is important because accuracy and stability of judgments are not present is part of a language, but cannot show the actual limits of variation. It is impossible in all types of judgments that are related to some aspect of human perception. to establish what is not licit in a language only by analyzing them (Henry, 2005). For example, in the famous “The Dress” photograph, not only did judgments of Second, big corpora with rich data that include a variety of genres are the only color perception differ across people, with some seeing the dress as blue/black and ones that can provide a faithful approximation of the actual variation space of a others as blue/brown or white/gold, but also test-retest reliability revealed switches language, and these are available only for big, standard languages. This is one of in perception across testing sessions (Lafer-Sousa et al., 2015). Frontiers in Psychology | www.frontiersin.org 2 March 2020 | Volume 11 | Article 364 Leivada and Westergaard Acceptability, Grammaticality, and the Parser example, overloading memory and processing resources through has suggested that it has deeper cognitive roots, and more nested hierarchies (3) may lead the cognitive parser to not specifically, that it derives from the parser’s preference to avoid fully register or retain all the relevant information (Gibson and tokenizing multiple, adjacent occurrences of the same type Thomas, 1999), something that is necessary in order to provide because of a general bias to provide more attentional resources an acceptability judgment that faithfully represents whether the to novel information (“Novel Information Bias”; Leivada, 2017). stimuli fall inside or outside the domain of predictions of the Acceptability is thus affected by a variety of processing factors underlying, internalized grammar. In other words, precisely and cognitive biases, and so is grammaticality. For example, because of the high complexity of some stimuli, and due to the although data that flout Identity Avoidance exist [(6); see fact that the cognitive parser works on the basis of processing Leivada, 2017 for examples of syntactic violations], there are no heuristics (Kahneman, 2011), some deviations may go unnoticed. grammatically licit structures that feature five identical, adjacent One such example is (4), which looks very similar to (3) but– complementizers, and the prediction is that such structures will unlike (3)–violates a rule of grammar. never be in use, because a grammar would never consistently deploy them. Even if grammars were able to generate a sentence (3) The patient the nurse the clinic had hired admitted met like *“John said that that that that that Mary kissed him,” Jack. Frazier (1985). cognitive biases would intervene and break this sequence of (4) *The doctor the nurse the hospital had hired met complementizers, for this degree of repetition would not be John. Frazier (1985). informative, and by means of looking like noise to the parser, In linguistic terms, the fact that (4) is missing a verb and has it would make communication infelicitous. A similar situation an argument (i.e., “the doctor”) that is not assigned any thematic arises with sentence (4): it is extremely unlikely that a language role entails a violation of Chomsky’s (1981) θ-criterion, according will consistently deploy sentences with missing verbs that have to which each argument bears one and only one θ-role, and each licensed arguments. In other words, although the rules of the θ-role is assigned to one and only one argument. Despite the grammar of a language are subject to change in a way that seriousness of this deviation, the “missing verb effect” showed in may legitimize the use/acceptability of a previously ill-formed (4) has been linked to high acceptability rates, even though the sentence and/or diminish the use/acceptability of a previously sentence is most definitely ill-formed from a syntactic point of attested one, certain changes are not expected to occur, because view (Gibson and Thomas, 1999). Moreover, this effect is neither they violate either a core principle of linguistic cognition or a restricted to one language nor is it a laboratory phenomenon general cognitive bias. that arises only in acceptability judgment tasks (Häussler and Talking about a dissociation of acceptability and Bader, 2015). Sentence (4) shows that ease of parsability may grammaticality, unacceptable grammatical sentences are influence judgments, and in this specific case, low parsability one logical possibility. One may wonder whether the other leads to not spotting a violation of a core syntactic principle. At possibility is also attested, i.e., acceptable ungrammatical the same time, high parsability does not guarantee acceptability sentences. Example (8) in Table 1 provides the missing piece or grammaticality. For example, speakers of English recognize of this dissociation (see also Ross, 2018 for the interaction of that (5) expresses a thought that their cognitive parser can easily grammaticality and acceptability). process, but their language does not produce it in this way. Sentence (8) instantiates a linguistic illusion called “comparative illusion” (Montalbetti, 1984). These sentences are (5) *What did Peter eat ravioli and? called illusions because they trick the parser in a way that renders It seems that a dissociation is in place, because being high acceptability ratings in experiments, even though the stimuli grammatical (i.e., not violating a rule of grammar) does are ill-formed (Wellwood et al., 2018). In linguistic terms, (8) is not guarantee acceptability either. Example (6) is in fact an ill-formed because the main clause subject calls for a comparison unacceptable grammatical sentence.3 Speakers would not judge of cardinalities of sets, but in the absence of a bare plural in the it as acceptable as (1), but it is a grammatically well-formed embedded clause subject, no comparison set is made available sentence of English, in the sense that no rule of grammar is (Phillips et al., 2011; O’Connor et al., 2012; Wellwood et al., 2018). violated. Its structure is analogous to that of (7). Linguistic illusions are the outcome of a partial-match strategy that is operative during processing (Reder and Kusbit, 1991; (6) Dogs dogs dog dog dogs. Barton et al. (1987). Kamas et al., 1996; Park and Reder, 2004). (7) Cats (that) dogs chase love fish. The difficulty of (6) suggests that the types of structures that are TABLE 1 | A dissociation of grammaticality and acceptability. actually attested in language are influenced by biases of general cognition. One such bias seems to underlie the unacceptability Unacceptable Acceptable of (6): Identity Avoidance holds that elements of the same Grammatical (6) Dogs dogs dog dog dogs. (1) John said to Mary that he phonological and/or syntactic type are unlikely to occur in Barton et al. (1987). likes doing linguistics. immediately adjacent positions (van Riemsdijk, 2008). Although Ungrammatical (2) *To he likes that linguistics (8) *More people have been to this has long been treated as a linguistic ban, recent work John Mary doing said. Russia than I have. Montalbetti (1984). 3 “Dog” can be both a verb and a noun in English. The sentence means the Sentences that were already introduced above appear with their following: dogs that are followed by dogs follow themselves other dogs. original numbering. Frontiers in Psychology | www.frontiersin.org 3 March 2020 | Volume 11 | Article 364 Leivada and Westergaard Acceptability, Grammaticality, and the Parser When the parser receives a linguistic stimulus, its components, something that leads to high acceptability. Of course, one could concepts, and structure are matched to stored knowledge, so claim that such a sentence, despite being labeled “acceptable,” that an output is produced. However, the parser matches the would never be attested in one’s linguistic performance. However, stimulus to stored information only up to a point. In other ungrammatical sentences that are harder to parse are in fact words, a processing threshold is set and the stimulus is checked attested in naturalistic speech (12a), and the relevant data also up to this threshold, hence the notion of partial matching. Given include missing verbs in cases of center-embedding (12b). that (8) makes use of locally coherent templates (Townsend (12a) “And since I was not informed–as a matter of fact, since and Bever, 2001) that provide a “good-enough fit” (Ferreira and I did not know that there were excess funds until we, Patson, 2007) for the parser, its ill-formedness may go unnoticed, ourselves, in that checkup after the whole thing blew up, and this results in high acceptability. Evidently, the way the and that was, if you’ll remember, that was the incident in parser works–via the use of processing heuristics–mediates which the attorney general came to me and told me that one’s access to the internalized knowledge of grammar. Yet, he had seen a memo that indicated that there were no the ease with which a sentence is unambiguously parsed is more funds.”5 President Ronald Reagan, April 28, 1987. not a guarantee for either grammaticality or acceptability. (12b) That we scrutinize is a simple consequence of the fact Table 2 adds high/low parsability to the previous dissociation that none of the predictions that you 1 during the between grammaticality and acceptability. Once again, all logical months that you have been in office has turned out to possibilities are attested. be true. Häussler and Bader (2015: 14). Example (9) does not violate any rule of grammar, however, its acceptability is not comparable to that of (1) for semantico- Going back to the rest of the data in Table 2, we see pragmatic reasons that boil down to difficulties that arise “in that (5) and (11) suggest that certain ungrammatical sentences assigning a coherent meaning to the whole” (Adger, 2018: can be easily parsed too. Recent research has suggested 161). Unlike (2) or even (10), (9) can be easily parsed in a that not all ungrammatical sentences receive unclear and way that pertains to connected speech. Moreover, a coherent unreliable interpretations across speakers (e.g., Etxeberria et al., interpretation of it can be provided, and over the years there have 2018 on negation). Talking about ungrammatical sentences been various proposals that construe meanings for it.4 Perhaps that are acceptable and parsable, Otero (1972) reached the green ideas refer to environmental considerations. One could conclusion that wide acceptability is not a guarantee for build a metaphorical narrative where these ideas are colorless and grammaticality. Even sentences that have been described as sleeping because at present there is not enough effort to combat blatantly ungrammatical may actually be acceptable to some climate change, however, their sleep is furious, something that degree, and this degree varies across speakers of the same may suggest that some promising initiatives for change are under language that have different developmental trajectories (e.g., late way. Creating the right context can improve the acceptability bilinguals, heritage speakers, L1 attriters). For example, (5) is of (9) precisely because of its grammatical well-formedness and ungrammatical because extraction out of coordinated structures high parsability. is prohibited. A similar island effect has been described for Perhaps the most interesting sentence of Table 2 is (4): a extraction out of relative clauses (13). sentence that is both ungrammatical and hard to parse, yet still acceptable. Its low parsability hides the grammatical violation, (13) *Who do you like the poem that____wrote? 4 Although much literature portrays such sentences as https://www.physicstomato.com/colorless-green-ideas-sleep-furiously/ universally ungrammatical (see Phillips, 2013 and references therein), not all speakers find such violations fully unacceptable. TABLE 2 | A dissociation of grammaticality, acceptability, and parsability. For instance, Lowry et al. (2019) found surprising rates of acceptability for five different types of island violations– High parsability Low parsability including relative clause islands that received a mean score of Grammatical/ (1) John said to Mary that (3) The patient the nurse the 3.6 in an 1–5 scale, where 1 stood for the sentence sounding acceptable he likes doing linguistics. clinic had hired admitted met perfectly natural–among late bilingual and heritage speakers of Jack. Frazier (1985). Spanish. Importantly, the two groups differed both in terms of Grammatical/ (9) Colorless green ideas (10) That that that Bill left Mary their judgments and in terms of their involuntary physiological unacceptable sleep furiously.1 amused Sam is interesting is (Chomsky, 1957) sad. Hornstein (2013). reactions that can be proxies for processing effort. In Lowry Ungrammatical/ (5) *What did Peter eat (2) *To he likes that linguistics et al. (2019) these were measured through a pupillometry unacceptable ravioli and? John Mary doing said. study: pupil dilation in the ungrammatical stimuli was observed Ungrammatical/ (11) *Fewer people have (4) *The doctor the nurse the only in the group of late bilinguals, while there was no effect acceptable been to Tromsø than I hospital had hired met John. of ungrammaticality in the heritage group. These results have. Frazier (1985). suggest that regardless of what a theory/grammar presents as Sentences that were already introduced above appear with their ungrammatical, speakers may successfully parse ungrammatical original numbering. 1 The grammaticality of (9) is fairly indisputable (Hill, 1961), however, not everybody agrees on the degree of its unacceptability. Some stimuli in a way analogous to their grammatical counterparts. scholars have talked about doubtful acceptability, marking the sentence with a However, it is an important question whether the parsing is “?” to indicate this (Armstrong, 2005), while others have described it as outright 5 unacceptable (Bauer, 2014). http://www.reagan.utexas.edu/archives/speeches/1987/042887e.htm Frontiers in Psychology | www.frontiersin.org 4 March 2020 | Volume 11 | Article 364 Leivada and Westergaard Acceptability, Grammaticality, and the Parser complete, in the sense that these speakers assign meaning to (Maillat and Oswald, 2009: 361). In this context, the hearer stops these ungrammatical stimuli. searching for meaning after finding one that sufficiently meets Understanding the process of assigning meaning is important her expectations of relevance in accordance with the previous in the context of disentangling the role of the parser in acceptable discourse. The illusion thus arises in the process of selecting ungrammatical sentences. To illustrate this, let’s consider the meaning within a manipulative context that takes advantage of comparative illusions in Tables 1, 2 [examples (8) and (11), (i) the parser’s limitations and (ii) the parser’s way of operating respectively], which are ungrammatical but trick the parser into through employing certain processing heuristics such as partial acceptability (Wellwood et al., 2018; Leivada et al., 2019b). matching or shallow processing. Although various experiments have shown that these sentences If relevance and previous context can bias an acceptability are assigned a high acceptability rating, one could say that this judgment through creating the necessary conditions for an does not entail that these sentences are actually parsed, in the illusion to arise, the bias will be even greater if a specific m is sense that speakers actually assign them a meaning m. A clear given to a participant point-blank in an experiment that asks “Is exposition of this point is given by Tim Hunter as a reply to there a meaning m such that (s, m) is acceptable?” As shown Hornstein (2013), who suggests that such sentences may sound in Tversky and Kahneman’s (1974) work, options in a task are good to speakers, but when you ask the people that gave them a evaluated relative to some reference point. Theoretically speaking, high rating what the uttered sentence means, they are unable to the reference point in standard acceptability judgment tasks is provide a meaning: the linguistic repertoire of the tested speaker: We often instruct speakers to disregard the formal prescriptive rules of grammar I don’t think there is any meaning m such that (“More people have been to Russia than I have,” m) is judged acceptable. What and focus on evaluating the stimuli on the basis of how they is true about these examples is that if you ask whether the string is use the language. If we add a given m to this picture, we alter acceptable without providing any intended interpretation–roughly, the reference point. This does not mean that such a task cannot if you ask a question of the form “Is there a meaning m such that (s, provide useful and informative findings, but that possibly the m) is acceptable?”–then people tend to say “yes.” This despite the fact obtained findings will not be tapping directly into a speaker’s that, as everyone points out, if you ask which meaning this is, people perception of her idiolect. Instead, it will be mediated by an are stumped. [. . .] Why they should make this kind of mistake (i.e., anchoring effect that may cause an adjustment to the speaker’s accept the sentence), I have no idea: presumably the answer might judgment on the basis of m. To understand this effect, consider be something like, they start searching for a meaning for the string, the following example by Kahneman (2011). and they get close enough to feel confident that a meaning can be found without getting all the way there, so they stop and answer (14a) Was Gandhi more or less than 144 years old when he “yes” (since no one is asking for the particular meaning). died? In contrast, we suggest that illusions like (8) and (11) are (14b) How old was Gandhi when he died? parsed in a way that does go through assigning m to s. In our Kahneman (2011: 122). work on comparative illusions (Leivada et al., 2019b) we obtained Of course nobody claimed that Gandhi was 144 years old when ample evidence that most speakers that judged (8) as acceptable, he died, but it has been found that when (14b) is presented after truly construed an interpretation for it. Among the ones more (14a), the provided high number functions as an anchor that frequently given by the speakers we tested are: (i) more people affects people’s estimate (Kahneman, 2011). To draw the analogy than just me have been to Russia, (ii) people have been to Russia with judgment tasks, let’s compare (15a) to (15b), and it will more times than I have, and (iii) many people have been to Russia become clear why “Is m acceptable?” does not ask the same thing more times than I have (see also Wellwood et al., 2018). Naturally, as “Is there a meaning m such that (s, m) is acceptable?.” this is not what the sentence says, but nevertheless, a meaning is assigned to the sentence. Also we suggest that one should not (15a) Assuming a scale from 1 to 5, how acceptable is s ignore the possibility that those speakers that seem stumped upon on the basis of an intended meaning m? being asked to provide an interpretation do not do so because (15b) Assuming a scale from 1 to 5, how do you rate s they never actually established an association (s, m), but because on the basis of your idiolect? in their attempt to articulate the latter, they spot the illusion. Crucially, this does not entail that at no point were they actually In (15a), the possibility of s getting a meaning is explicit and a able to put their finger on a possible meaning. possible meaning m is already given to the speaker as part of the The second interesting issue with Hunter’s point has to do question that introduces the stimuli s. This can bias the rating of with the juxtaposition of two very different ways of eliciting s on the basis of the “anchor-and-adjust” heuristic. judgments through asking “Is s acceptable?” or “Is there To sum up, illusions do not necessarily entail that parsing fails a meaning m such that (s, m) is acceptable?” These two to produce a meaning, but that the parser can be tricked into questions do not tap into the same thing. Previous research on providing both a meaning and an acceptability rating that may the pragmatics of cognitive illusions has proposed that when not correspond to the actual status of the stimulus in terms of processing such sentences, the hearer searches for meaning what the speaker’s internalized grammar looks like. Importantly, within a manipulative communication, that is, within a tricky a number of factors contribute to this process of tricking the context that features a “manipulation (that) can be best defined parser: context, task and stimuli presentation, as well as structural in terms of the constraints it imposes on mental processing” complexity are only a few. Frontiers in Psychology | www.frontiersin.org 5 March 2020 | Volume 11 | Article 364 Leivada and Westergaard Acceptability, Grammaticality, and the Parser The relation between grammatical well-formedness and speech. In relation to the possible acceptability of these structures acceptability is a complex one. As mentioned in the Introduction, in experimental settings, the previous section has shown that the main goal of the present work is to discuss whether there is a dissociation between acceptability and grammaticality, acceptability is an indispensable gateway to grammaticality or such that we should expect some degree of discrepancy between whether there is a way of capturing grammaticality other than the the way speakers judge sentences in an experiment (where even one that goes through speakers’ perception of what is well-formed the way the stimuli are presented may influence judgments; see in their native linguistic repertoire (i.e., acceptability). Having examples 14–15), the way they actually speak, and the way that presented the dissociation between acceptability, grammaticality prescriptive grammar says they (should) speak. and the way the parser works, the next section deals with how The question still holds: Where does grammaticality come grammaticality is established and where it comes from. from? The tentative answer we offer is that grammaticality is often a formal, standardized snapshot of the way the official language looks like at a given point in time. Grammaticality is WHERE DOES GRAMMATICALITY COME constantly redefined through ever-changing acceptability, but it FROM? also reflects stable properties of general cognition. In this context, we do not know much about grammaticality outside acceptability Asking about the origin of the rules of grammar, Adger (2019) (recall that observation of naturalistic data cannot reveal what suggests that we learn them: They come from the way people is ungrammatical in a language) in the sense that there is no speak. Although this is true, the issue is more complex, because list of grammatical properties that are grammatical in and of different people speak in different ways even within linguistic themselves. They are all grammatical within a context that is communities that feature only one language. When one says called language X. Language X is constantly changing and what is that (5) and (8) are ungrammatical in English, this use of (un)grammatical today may not be (un)grammatical tomorrow, the term “ungrammatical” is not meant to be interpreted as a depending on whether the new speakers of X find it acceptable or faithful representation of every English speaker’s idiolect in an not and whether this acceptability is generalized and established individual way, precisely because even monolingual speakers in a as the norm or not. For example, Ancient Greek featured a monolingual community show variation.6 Rather it refers to some syntactic phenomenon called Attic syntax which permitted a established consensus about what is the norm in a specific variety number mismatch between the plural, neuter subject and the verb of English; a norm that the grammar books describe in detail. (16a). This structure is not a grammatically licit option in Modern Put differently, if some speakers of English, Spanish, or German Greek (16b), but not because there is something intrinsically accept to some degree or even produce to some degree island ungrammatical about it; it simply does not form part of the violations (Lowry et al., 2019 for Spanish), missing verbs in nested grammar anymore. Phrased differently, there is no notion of self- hierarchies (Häussler and Bader, 2015 for English and German), contained grammaticality that (16a) has and (16b) lacks; they just or comparative illusions7 , do we want to say that these structures form part of two different snapshots of a grammar’s domain of are grammatical in English, Spanish, and German? While it predictions at different points in time. certainly appears to be the case that some speakers’ grammars may occasionally give rise to such structures, we should take into (16a) Ta padia pezi. [Ancient Greek] account that, in relation to naturalistic data, production factors the child.PL play.3SG may endow the linguistic message with noise (i.e., false starts, “The children are playing.” infelicitous lemma retrievals, missing elements due to memory (16b) *Ta koritsia gela. [Modern Greek] constraints, etc.), which can account for how some of these the girl.PL laugh.3SG ungrammatical sentences come to be produced in spontaneous Intended meaning: “The girls are laughing.” 6 For example, Smith and Cormack (2002) discuss sequences of tense possibilities This claim is partially concordant with Chomsky et al.’s (2019); in English. With some speakers accepting “Did you know that Emily is ill?” and see also Chomsky (1993) view that in natural language there exists with others considering it unacceptable (i.e., accepting only “Did you know that Emily was ill?”), these authors capture the observed variation by suggesting that no independently given notion of grammatical well-formedness. this is “a situation in which intuitions are completely clear-cut, so the relevant Indeed, the grammatical well-formedness of a linguistic stimulus parameter has been fixed, but it has been fixed apparently at random, presumably does not boil down to an independently definable grammatical because of the paucity of distinguishing data” (p. 286). Another example is given in core, but is a mere historical “accident” that (i) refers to whether Levelt (1972), who showed that opinions about what is grammatical in a language are not uniform even among trained linguists who are native speakers of the the stimulus forms part of the standardized snapshot or not language in question. When he asked 24 linguists to judge whether the sentence and (ii) is subject to change such as the one shown in (16a-b). “The talking about the problem saved her” (Fraser, 1970, p. 91, with the example Nevertheless, this view is true only for one reading of the marked as ungrammatical) was marked as grammatical or ungrammatical in a specific linguistics article, he found that judgments varied, and only 1/3 of the term “grammatical”: grammatical as actually forming part of the consulted linguists gave the judgment “ungrammatical,” in agreement with the grammar of a specific language. original source. However, we suggest there is also another reading of the term 7 One example of a comparative illusion in naturalistic speech, outside of an “grammatical.” To understand this other reading, one needs experimental setting, is the following tweet by Dan Rather: “I think there to factor in that change is not without limits. Not all changes are more candidates on stage who speak Spanish more fluently than our president speaks English.” [Available at https://twitter.com/danrather/status/ are possible and not all linguistic stimuli are candidates for 1144076809182408704] forming part of grammar. For example, as mentioned in the Frontiers in Psychology | www.frontiersin.org 6 March 2020 | Volume 11 | Article 364 Leivada and Westergaard Acceptability, Grammaticality, and the Parser section “Acceptable ungrammatical sentences and unacceptable those sentences that could be grammatical, but for some reason grammatical sentences” there are no grammatically licit are not in the language in question, yet they probably are in structures that feature five identical, adjacent complementizers, some other language. Absolute ungrammaticality (i) is not subject and the prediction is that such structures will never be to change, (ii) is not defined in relation to one given language, grammatical. Similarly, a sentence such as (4), which violates and (iii) concerns violations of some core principle of language the θ-criterion, is unlikely to ever form part of grammar.8 As and/or cognition, that is, structures that grammar would never discussed in the next section, certain changes are not expected consistently deploy. Therefore, absolute ungrammaticality has to to occur, because they violate either a core principle of language do with structures that cannot form part of grammar. (e.g., the θ-criterion in 4) or a general cognitive bias (e.g., the Comparing the scales of the two notions, acceptability Novel Information Bias in 6). In this sense, Chomsky et al. (2019) and grammaticality, it is meaningful to talk about partial are right in arguing that there exists no independently given acceptability (Sprouse, 2007), but not about partial or strong- notion of grammatical well-formedness, but we would like to add weak ungrammaticality. A rule of grammar (or more than one to their claim that there do exist independently given constraints rule of grammar) can be either violated or not, but it cannot to the set of entities that this notion can encompass. This is the be violated just a bit. Ungrammaticality cannot be a matter of other reading of the term: grammatical as having the potential to degree, only acceptability can. Put differently, a native speaker be a part of grammar, by means of not going against any of the can judge a structure in her language as more acceptable than relevant biases and communication/processing principles that another structure, but a structure forming part of a grammar underlie language and cognition. cannot be more grammatical than another structure that forms part of the same grammar. Although some scholars have talked about “partial N TYPES OF UNACCEPTABILITY AND ungrammaticality,” we would argue that this refers either to TWO TYPES OF UNGRAMMATICALITY partial unacceptability or to variation in a linguistic community. Consider, for instance, the discussion of partial ungrammaticality It is an uncontroversial claim that acceptability judgments are not in Attinasi (1974): “A hidden assumption of homogeneity, that categorical, but form a continuous spectrum (Sprouse, 2007 and the language competence of every speaker consists of the same references therein). The usual meaning of the word “continuous” structures, falters when the question of partial ungrammaticality is unbroken or undivided, hence it is the nature of a continuum is raised. How can some speakers totally reject, others partially to be undivided, or better, to permit repeated division without accept and still others totally accept certain sentences as limit (Bell, 2017). If one subscribes to the view that acceptability grammatical if each presumably speaks ‘English,’ or any other should be viewed as a continuum, one also subscribes to the view language?” (p. 280). In our view, this question has to do with that the acceptability continuum is infinitely divisible. Although gradient acceptability: that is what speakers have judgments acceptability judgment tasks that involve Likert scales feature about.9 As we have seen, grammaticality can be dissociated a finite number of options more often than not, there are from acceptability. Also, the observed variation does not entail experiments that ask speakers to judge a linguistic stimulus by or legitimize the notion of partial grammaticality, because, adjusting a slider on a continuum without any clearly delineated as mentioned in the section “Where does grammaticality categories such as “acceptable,” “somewhat acceptable” etc. come from?,” different people speak in different ways, but While the scale of unacceptability involves n positions, the grammaticality evokes an established norm that is part of a scale of ungrammaticality involves only two: Something can be formal snapshot. Speakers may deviate from this norm, either ungrammatical in the relative or in the absolute sense. The relative because language change has occurred and the norm does not sense pertains to the first reading of the term “grammatical” reflect this yet, or because their idiolect simply differs from the that was mentioned above: forming part of grammar. We call norm. But this should be referred to as “interspeaker variation,” it “relative” because it is defined in the context of a given not “partial grammaticality.” language. For example, (16b) is ungrammatical in relation to Modern Greek, but it is not ungrammatical per se. It is a potential candidate for forming part of grammar, it was grammatical in OUTLOOK the past (16a), and it may be again in the future. Similarly, (17) is ungrammatical in relation to Standard English, but this is an The present work has discussed the complex relation between accident, as it could potentially be grammatical (and in fact it is grammaticality, acceptability, and parsability. A number grammatical in many varieties of English, including e.g., Belfast of unacceptable grammatical sentences and acceptable English; Henry, 2005). ungrammatical sentences have been presented, including grammatical illusions, violations of Identity Avoidance, and (17) The children is here. sentences that involve a high level of processing complexity Relative ungrammaticality (i) is subject to change, (ii) is 9 Boeckx (2010) rightly calls the term “grammaticality judgment tasks” a misnomer, defined in the context of a specific language, and (iii) refers to because speakers lack intuitions about whether something is grammatical. In the absolute meaning of the term “grammatical,” having judgments about 8 Although the missing-verb effect can be occasionally attested in naturalistic grammaticality would entail having intuitions about the workings of all linguistic speech (12b), we argue that this has to do with production factors that introduce and cognitive factors that determine the limits of grammar, and no speaker (or noise to the linguistic message. linguist for that matter) has that. Frontiers in Psychology | www.frontiersin.org 7 March 2020 | Volume 11 | Article 364 Leivada and Westergaard Acceptability, Grammaticality, and the Parser that overloads the cognitive parser. Focusing on acceptable language-specific factors? Or as Ott (2014): 290) asks is “*What ungrammatical sentences, we have argued that in many cases does John like and oranges?” ungrammatical (in the absolute their acceptability entails that a meaning has been assigned sense that it cannot be generated by the grammar), given that to them. Also, two notions of ungrammaticality have been speakers can easily assign it a transparent interpretation (e.g., introduced: (Un)grammaticality in the relative sense refers to which x: John likes x and oranges)? The answer is currently the whether the stimulus falls within the domain of predictions unclear to us, and it probably needs novel experimental work of a given grammar or not. (Un)grammaticality in the absolute to be properly discussed. Recognizing this uncertainty does not sense refers to whether the stimulus has the potential to mean undermining the proposed distinction between absolute be a part of grammar or not. Relative (un)grammaticality is and relative ungrammaticality. It rather suggests that progress is an ever-changing property of the stimulus, whereas absolute underway, or as (Feynman, 1998: 27) puts it, “[b]ecause we have (un)grammaticality is stable. In both readings of the term, the doubt, we then propose looking in new directions for new grammaticality is defined by something that is external to the ideas. The rate of the development of science is not the rate at stimulus (be it the grammar of a specific language or principles which you make observations alone but, much more important, of general/linguistic cognition), and it is not an inextricable the rate at which you create new things to test.” property of the stimulus itself. Put differently, there is no list of properties that are (relatively/absolutely) grammatical in and of themselves, or as Chomsky et al. (2019) phrase it, there is no AUTHOR CONTRIBUTIONS independently given notion of grammatical well-formedness in natural language. EL and MW conducted the research behind this work. EL drafted Through disentangling the various uses of the terms a first manuscript, which MW revised. Both authors contributed “acceptable” and “grammatical,” the overarching aim of this equally to the final editing of this work. work has been to aid the field in reaching a more adequate level of terminological clarity for notions that pertain to the evidential base of linguistics. Many details of the distinction FUNDING between relative and absolute (un)grammaticality are left to be worked out, and this will likely be the topic of future This work was supported by the European Union’s Horizon 2020 work. To give just one example, when we deal with island Research and Innovation Program under the Marie Skłodowska- effects of the sort discussed above, do we deal with absolute Curie Grant Agreement No. 746652. The publication charges for ungrammaticality that is universal and derives from processing this manuscript have been funded by a grant from the publication or other principles of language/cognition, or with relative fund of UiT The Arctic University of Norway. The funders had ungrammaticality that is manifested in different ways across no role in study design, data collection and analysis, decision to different languages, precisely because it is defined on the basis of publish, or preparation of the manuscript. REFERENCES Boeckx, C. (2010). Language in Cognition. Uncovering Mental Structures and the Rules Behind Them. Maden: Wiley-Blackwell. Adger, D. (2018). “The autonomy of syntax,” in Syntactic Structures After 60 Years: Branigan, H. P., and Pickering, M. J. (2016). An experimental approach to linguistic The Impact of the Chomskyan Revolution in Linguistics, eds N. Hornstein, H. representation. Behav. Brain Sci. 40:e282. doi: 10.1017/s0140525x16002028 Lasnik, P. Patel-Grosz, and C. Yang, (Berlin: Mouton de Gruyter), 153–176. Carr, P. (1990). Linguistic Realities: An Autonomist Metatheory for the Generative Adger, D. (2019). Where do the rules of grammar come from? Psychol. Today. Enterprise. Cambridge: Cambridge University Press. Available online at: https://www.psychologytoday.com/us/blog/language- Chomsky, N. (1957). Syntactic Structures. The Hague: Mouton. unlimited/201908/where-do-the-rules-grammar-come (accessed February 27, Chomsky, N. (1965). Aspects of the Theory of Syntax. Cambridge, MA: MIT Press. 2020). Chomsky, N. (1981). Lectures on Government and Binding. Dordrecht: Foris Armstrong, N. (2005). Translation, Linguistics, Culture. Clevedon: Multilingual Publications. Matters. Chomsky, N. (1993). “Aminimalist program for linguistic theory,” in The View from Attinasi, J. T. (1974). The sociolinguistics of William Labov. Biling. Rev. 1, 279–304. BUILDING 20: Essays in Linguistics in Honor of Sylvain Bromberger, eds K. Hale, Baggio, G., van Lambalgen, M., and Hagoort, P. (2012). “Language, linguistics and and S. J. Keyser, (Cambridge, MA: MIT Press), 1–52. cognition,” in Handbook of the Philosophy of Science: Philosophy of Linguistics, Chomsky, N., Gallego, A. J., and Ott, D. (2019). Generative grammar and the facu- Vol. 14, eds R. Kempson, T. Fernando, and N. Asher, (Amsterdam: Elsevier), lty of language: insights, questions, and challenges. Catalan J. Linguist. Available 325–355. doi: 10.1016/b978-0-444-51747-0.50010-x online at: https://ling.auf.net/lingbuzz/003507 (accessed February 27, 2020). Bard, E. G., Robertson, D., and Sorace, A. (1996). Magnitude estimation of Cornips, L., and Poletto, C. (2005). On standardising syntactic elicitation linguistic acceptability. Language 72, 32–68. techniques (part 1). Lingua 115, 939–957. doi: 10.1016/j.lingua.2003.11.004 Barton, G. E., Berwick, R. C., and Ristad, E. S. (1987). Computational Complexity Etxeberria, U., Tubau, S., Deprez, V., Borràs-Comes, J., and Espinal, M. T. (2018). and Natural Language. Cambridge, MA: MIT Press. Relating (Un)acceptability to Interpretation. Experimental investigations on Bauer, L. (2014). Grammaticality, acceptability, possible words and large corpora. negation. Front. Psychol. 8:2370. doi: 10.3389/fpsyg.2017.02370 Morphology 24, 83–103. doi: 10.1007/s11525-014-9234-z Ferreira, F., and Patson, N. D. (2007). The ‘good enough’ approach to language Bell, J. L. (2017). “Continuity and Infinitesimals,” in The Stanford Encyclopedia of comprehension. Lang. Linguist. Compass 1, 71–83. doi: 10.1111/j.1749-818x. Philosophy, Summer 2017 Edition, ed. E. N. Zalta. Available online at: https: 2007.00007.x //plato.stanford.edu/archives/sum2017/entries/continuity/ (accessed February Feynman, R. P. (1998). The Meaning of It All. Thoughts of a Citizen Scientist. 27, 2020). Reading: Perseus Books Group. Bever, T. G. (1970). “The influence of speech performance on linguistic structure,” Fraser, B. C. (1970). “Some remarks on the action nominalization in English,” in in Advances in Psycholinguistics, eds G. B. Flores d’Arcais, and W. J. M. Levelt, Readings in English Transformational Grammar, eds R. A. Jacobs, and P. S. (Amsterdam: North-Holland Publishing Co.), 4–30. Rosenbaum, (Waltham, MA: Ginn), 83–98. Frontiers in Psychology | www.frontiersin.org 8 March 2020 | Volume 11 | Article 364 Leivada and Westergaard Acceptability, Grammaticality, and the Parser Frazier, L. (1985). “Syntactic complexity,” in Natural Language Parsing. O’Connor, E., Pancheva, R., and Kaiser, E. (2012). “Evidence for online repair of Psychological, Computational and Theoretical Perspectives, eds D. R. Dowty, Escher sentences,” in Proceedings of Sinn und Bedeutung, Vol. 17, eds E. Chemla, L. Karttunen, and A. Zwicky, (Cambridge: Cambridge University Press), V. Homer, and G. Winterstein, (Paris: École Normale Supérieure), 363–380. 129–189. Otero, C. (1972). Acceptable ungrammatical sentences in Spanish. Linguist. Inq. 3, Fromkin, V., and Rodman, R. (1998). An Introduction to Language, 6th Edn. 233–242. Orlando, FL: Holt, Rinehart and Winston. Ott, D. (2014). Syntactic islands by Cedric Boeckx (review). Language 90, 287–291. Gibson, E., and Fedorenko, E. (2010). Weak quantitative standards in doi: 10.1353/lan.2014.0008 linguistics research. Trends Cogn. Sci. 14, 233–234. doi: 10.1016/j.tics.2010. Park, H., and Reder, L. M. (2004). “Moses illusion: implication for human 03.005 cognition,” in Cognitive Illusions: A Handbook on Fallacies and Biases in Gibson, E., and Thomas, J. (1999). Memory limitations and structural forgetting: Thinking, Judgment, and Memory, ed. R. F. Pohl, (Hove: Psychology Press), the perception of complex ungrammatical sentences as grammatical. Lang. 275–292. Cognit. Process. 14, 225–248. doi: 10.1080/016909699386293 Phillips, C. (2009). “Should we impeach armchair linguists?,” in Japanese/Korean Häussler, J., and Bader, M. (2015). An interference account of the missing-VP effect. Linguistics, Vol. 17, eds S. Iwasaki, H. Hoji, P. M. Clancy, and S.-O. Sohn, Front. Psychol. 6:766. doi: 10.3389/fpsyg.2015.00766 (Stanford, CA: CSLI Publications), 49–64. Henry, A. (2005). Non-standard dialects and linguistic data. Lingua 115, 1599– Phillips, C. (2013). “On the nature of island constraints I: language processing and 1617. doi: 10.1016/j.lingua.2004.07.006 reductionist accounts,” in Experimental Syntax and Island Effects, 64-108, eds J. Hill, A. A. (1961). Grammaticality. Word 17, 1–10. doi: 10.1080/00437956.1961. Sprouse, and N. Hornstein, (Cambridge: Cambridge University Press). 11659742 Phillips, C., Wagers, M. W., and Lau, E. F. (2011). “Grammatical illusions and Hornstein, N. (2013). Acceptability and Grammaticality. Available online at: selective fallibility in real-time language comprehension,” in Experiments at the http://facultyoflanguage.blogspot.com\penalty-\@M/2013/02/acceptability- Interfaces, Vol. 37, ed. J. Runner (Bingley: Emerald Publications), 147–180. and-grammaticality.html (accessed February 22, 2013). Reder, L. M., and Kusbit, G. W. (1991). Locus of the Moses illusion: imperfect Kahneman, D. (2011). Thinking, Fast and Slow. New York, NY: Farrar, Straus and encoding, retrieval, or match? J. Mem. Lang. 30, 385–406. doi: 10.1016/0749- Giroux. 596x(91)90013-a Kamas, E. N., Reder, L. M., and Ayers, M. S. (1996). Partial matching in the Ross, D. (2018). Conventionalization of grammatical anomalies through Moses illusion: response bias not sensitivity. Mem. Cognit. 24, 687–699. doi: linearization. Stud. Linguist. Sci. 42, 1–28. 10.3758/bf03201094 Schütze, C. T. (1996/2016). The Empirical Base of Linguistics: Grammaticality Labov, W. (1996). “When intuitions fail,” in Papers from the Parasession on Theory Judgments and Linguistic Methodology. Berlin: Language Science Press. and Data in Linguistics Chicago Linguistic Society, Vol. 32, eds L. McNair, K. Smith, N., and Cormack, A. (2002). Parametric poverty. Glot Int. 6, 285–287. Singer, L. Dolbrin, and M. Aucon, (Chicago, IL: Chicago Linguistic Society), Sprouse, J. (2007). Continuous acceptability, categorical grammaticality, and 77–106. experimental syntax. Biolinguistics 1, 123–134. Lafer-Sousa, R., Hermann, K. L., and Conway, B. R. (2015). Striking individual Sprouse, J., and Almeida, D. (2013). The empirical status of data in syntax: a differences in color perception uncovered by “the dress” photograph. Curr. Biol. reply to Gibson and Fedorenko. Lang. Cogn. Process. 28, 222–228. doi: 10.1080/ 25, R545–R546. doi: 10.1016/j.cub.2015.04.053 01690965.2012.703782 Lau, J. H., Clark, A., and Lappin, S. (2016). Grammaticality, acceptability, and Townsend, D., and Bever, T. G. (2001). Sentence Comprehension: Integration of probability: a probabilistic view of linguistic knowledge. Cogn. Sci. 41, 1202– Habits and Rules. Cambridge, MA: MIT Press. 1241. doi: 10.1111/cogs.12414 Tversky, A., and Kahneman, D. (1974). Judgment under uncertainty: heuristics and Leivada, E. (2017). What’s in (a) label? Neural origins and behavioral biases. Science 185, 1124–1131. manifestations of identity avoidance in language and cognition. Biolinguistics van Riemsdijk, H. (2008). “Identity avoidance: OCP effects in Swiss relatives,” 11, 221–250. in Foundational Issues in Linguistic Theory: Essays in Honor of Jean-Roger Leivada, E., D’Alessandro, R., and Grohmann, K. K. (2019a). Eliciting big data from Vergnaud, eds R. Freidin, C. P. Otero, and M. L. Zubizarreta, (Cambridge, MA: small, young, or non-standard languages: 10 experimental challenges. Front. MIT Press), 227–250. Psychol 10:313. doi: 10.3389/fpsyg.2019.00313 Wellwood, A., Pancheva, R., Hacquard, V., and Phillips, C. (2018). The anatomy of Leivada, E., Mitrofanova, N., and Westergaard, M. (2019b). “The impact of a comparative illusion. J. of Semant. 35, 543–583. doi: 10.1093/jos/ffy014 bilingualism in processing cognitive illusions,” in Talk at the Capturing and Wexler, K., Culicover, P., and Hamburger, H. (1975). Learning-theoretic Quantifying Individual Differences in Bilingualism workshop, (Tromsø: UiT-The foundations of linguistic universals. Theor. Linguist. 2, 215–224. doi: 10.1515/ Arctic University of Norway). thli.1975.2.1-3.215 Levelt, W. J. M. (1972). Some psychological aspects of linguistic data. Linguist. Ber. 17, 18–30. Conflict of Interest: The authors declare that the research was conducted in the Lowry, C., Madsen, C. N. II, Phillips, I., Martohardjono, G., and Schwartz, R. G. absence of any commercial or financial relationships that could be construed as a (2019). “Gradience in Spanish island violations: a psychophysiological study potential conflict of interest. of two bilingual groups,” in Proceedings of the Experimental Psycholinguistics Conference, Palma. Copyright © 2020 Leivada and Westergaard. This is an open-access article Maillat, D., and Oswald, S. (2009). Defining manipulative discourse: the distributed under the terms of the Creative Commons Attribution License (CC BY). pragmatics of cognitive illusions. Int. Rev. Pragmat. 1, 348–370. doi: 10.1163/ The use, distribution or reproduction in other forums is permitted, provided the 187730909x12535267111651 original author(s) and the copyright owner(s) are credited and that the original Montalbetti, M. (1984). After Binding:On the Interpretation of Pronouns. Doctoral publication in this journal is cited, in accordance with accepted academic practice. No dissertation, Massachusetts Institute of Technology, Cambridge, MA. use, distribution or reproduction is permitted which does not comply with these terms. Frontiers in Psychology | www.frontiersin.org 9 March 2020 | Volume 11 | Article 364