ARTICLE https://doi.org/10.1057/s41599-022-01440-w OPEN Determining the cognitive biases behind a viral linguistic universal: the order of multiple adjectives Evelina Leivada 1✉ When people are asked to create a phrase with the elements {blue, earrings, beautiful}, they produce ‘beautiful blue earrings’. Several theories have been proposed about the origins of 1234567890():,; this universal tendency to order multiple adjectives in a specific way: an innate universal hierarchy with designated positions for each category of adjectives, sensitivity to the defi- niteness of the adjectival denotation, availability and psychological closeness of the adjective attributes to the speaker, the encoding of subjective vs. objective properties, and the adjective’s phonological weight. Although these theories have strong descriptive power, they often focus on what happens at the phenotypic level without explaining what cognitive needs trigger this behavior. Through a timed task that measures acceptability in ‘Adjective- Adjective-Noun’ sequences that either comply with the universal order or violate it, we adduce evidence for the high acceptability of the violating orders, whose processing did not take longer than that of the compliant orders, as should have happened if the former were non-canonical. The results suggest that ordering preferences exist but are not invariable, as one would expect if a strong linguistic universal was involved. We track the origin of adjective ordering preferences to the synergistic interplay of three cognitive biases: Zipf’s Law, Intol- erance of Ambiguity, and Novel Information Bias. Last, we show that the linguistic mani- festation of these preferences is sensitive to the statistical distribution of the input data, resulting to variation even among speakers of the same language. 1 Universitat Rovira i Virgili, Tarragona, Spain. ✉email: evelina.leivada@urv.cat HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | (2022)9:436 | https://doi.org/10.1057/s41599-022-01440-w 1 ARTICLE HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | https://doi.org/10.1057/s41599-022-01440-w A Introduction djectives are ordered in a specific way across typologically (Whorf, 1945), objective (Hetzron, 1978; Scontras et al., 2017), different languages. This property of human language absolute (Sproat and Shih, 1991), or object-oriented (Stavrou, amounts to the most documented linguistic universal in 1999), (ii) the adjective’s phonological weight/length (Wulff, digital media, repeatedly going viral every year since 2014 2003; Kotowski, 2016; Scontras et al., 2017), and (iii) the noun- (Waldman, 2014; Hanson, 2015; Dowling, 2016; Horobin, 2016, specific frequency as well as the collocability and/or idiomaticity Batchelor, 2017; Nordquist, 2018; Gutoskey, 2019; Saba, 2020). of certain combinations of nominals and adjectives (Wulff, 2003; The linguistic formulation of the universal posits that different Bouchard, 2005; Svenonius, 2008; Hahn et al., 2018). classes of adjectives appear in a so-called universal order that The general tendencies are the following: Adjectives that determines the distance between an adjective and the modified denote noun-inherent, objective properties, which are object- noun (Hetzron, 1978; Dixon, 1982; Sproat and Shih, 1991; oriented, and as such less likely to cause disagreement, are placed Cinque, 1994; Scott, 2002; Laenzlinger, 2005; Teodorescu, 2006; closer to the noun. A second contributing factor is noun- Alexiadou et al., 2007; Panayidou, 2013). This universal order specificity. Certain adjectives occur more frequently with certain predicts that only (1a) is well-formed, with the rest of the nouns, and this high specificity, or else noun-specific frequency, orderings (1b–f) being ungrammatical in the absence of a impacts ordering, placing the adjectives that have a high mutual special intonation or other licensing conditions that can legit- information with the noun closer to it (Wulff, 2003; Hahn et al., imize their use (1g). 2018).2 When adjectives are relatively interchangeable in terms of order (e.g., color and shape), a third factor becomes relevant: the (1) a. A set of beautiful blue porcelain earrings. [target order] morphophonologically lengthier adjective is placed closer to the b. A set of beautiful porcelain blue earrings. noun (Wulff, 2003; Kotowski, 2016). The presence of a third c. A set of blue beautiful porcelain earrings. tendency, which is predicated on the lax application of the other d. A set of blue porcelain beautiful earrings. two, already suggests that more than one factors are at play e. A set of porcelain blue beautiful earrings. behind what is considered to be the universal, unmarked order f. A set of porcelain beautiful blue earrings. (2), such that a comprehensive answer to the question of why g. I asked for a set of BLUE beautiful porcelain earrings, adjectives are ordered the way they are across languages is unli- not white. kely to make reference to only one cause (Wulff, 2003; Kotowski Contrastive focus in (1g) legitimizes the deviation from the and Härtl, 2019). target order (1a), which shows that the adjective that denotes a The various proposals that fall under the second theory are subjective comment (‘beautiful’) is placed before the adjective that grouped together because they have a similar reasoning. They all denotes color, which in turn precedes the material adjective. (1a) propose that one or more factors (i.e., inherentness, subjectivity, is the only order that is compliant with the universal hierarchy encoding of speaker-oriented vs. object-oriented properties) are (2). The orders in (1b–f) have received various characterizations behind the attested AOPs. Recent experimental studies that in the literature, such as ungrammatical (Bever, 1970), awkward provide evidence for AOPs underscore the need for finding an (Teodorescu, 2006), semantically incorrect (Kemmerer et al., explanation as to why these factors play a role in adjective 2009), odd (Scontras et al., 2017), or marked/non-canonical ordering (Scontras et al., 2017; Fukumura, 2018). Undoubtedly, if (Smirnova et al., 2019). human language consistently deploys a strategy, some function must be served.3 The cross-linguistic preference for a specific (2) Subjective Comment > Evidential > Size > Length > Height > order over others entails that this order is likely to lead to com- Speed > Depth > Width > Temperature > Wetness > Age > municative success, possibly by facilitating referent identification Shape > Color > Nationality/Origin > Material (adapted from (Fukumura, 2018; Franke et al., 2019; Scontras et al., 2019). What Scott, 2002: p. 114). is missing from this picture is the overall underlying etiology: the This richness of views that present the structures that violate cognitive need(s) that lead speakers/signers to prefer one ordering (2) as syntactically ill-formed, semantically/pragmatically odd, over others, sculpting cross-linguistic preferences accordingly. In marked, or simply less preferred reflects the debated nature of Scott’s (2002) words, almost “all writers claim that AO[P]s can be adjective ordering preferences1 (AOPs) as pertaining to core adequately accounted for using broad “psychological” criteria, yet syntax, to the syntax-semantics interface, or to more general none of them are able to provide a convincing argument—which cognitive principles; a topic that is still under intense debate is, moreover, consistent with the data—for a psychological basis (Kotowski and Härtl, 2019). to AO[Ps]”. More recent studies have re-affirmed that a clear According to the first theory (henceforth, the syntactic origin answer to the question of what factors drive AOPs remains elu- theory, SOT), an innate, Universal Grammar-encoded syntactic sive (Trotzke and Wittenberg, 2019). Filling this knowledge gap hierarchy with designated positions for adjective classes (2) is requires (i) connecting linguistic structures observed at the phe- responsible for the grammaticality of (1a) and the ungrammati- notypic level to their cognitive underpinnings and (ii) bringing cality of (1b–f) (Cinque, 1994; Cinque, 2010; Scott, 2002; into the picture environmental triggers that may result to varia- Panayidou, 2013). This syntactic hierarchy is part of a larger spine tion within or across linguistic communities. These are the main of syntactic positions (Demonstratives > Numerals > Adjectives > objectives of the present work. More concretely, the addressed Nouns) that underlies the ordering of elements in the nominal research questions are: domain (Cinque, 2005; Alexiadou et al., 2007; Alexiadou, 2014). This larger syntactic configuration is also available in Universal (I) What cognitive needs are subserved by AOPs? Grammar and makes predictions for the behavior of different (II) Is there interspeaker variation in the attested preferences, types of adjectival modifications (Alexiadou, 2001). and if yes, how can it be reconciled with the notion of a According to the second theory (henceforth, the multifactorial strong linguistic universal that should not allow for cognitive origin theory, mCOT), the order is not so rigid, such variability in its phenotypic realizations? that one can talk about ordering preferences, but not about Research question (I) embeds the phenotypic behavior into the ungrammatical or unattested orders. Under the assumptions of bigger picture, asking what triggers AOPs with the aim to explain the second theory, AOPs are the outcome of one or more factors their mosaic nature. Research question (II) imports a critical and such as (i) the encoding of properties that are noun-inherent hitherto missing comparative perspective: Experiments on AOPs 2 HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | (2022)9:436 | https://doi.org/10.1057/s41599-022-01440-w HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | https://doi.org/10.1057/s41599-022-01440-w ARTICLE report insights from a single, often monolingual population, grew up as monolingual and were consistently exposed to a without offering any explicit comparisons of people with different different language only upon relocating to another country/ developmental trajectories (e.g., monolinguals, early/late bilin- linguistic community as adults. It is very likely that, unlike the guals, second language learners etc). Such a comparative per- (monolingual) participants of previous experiments on AOPs, the spective has the potential to reveal the ways in which aspects of participants of this experiment have been recently made aware of the environment may influence the phenotypic manifestations of what is the prescriptively correct way of ordering adjectives, and innate cognitive needs. this may affect their perception of a strong linguistic universal in ways that are yet to be determined. At the time of testing, all participants had resided for a Methods minimum of 4 years outside Greece (mean: 11.8 years, SD: 9.8), Task. A timed acceptability task was used to collect two types of mainly in Scandinavia, UK, or Germany. Further information responses: (i) acceptability judgments on a 3-point Likert scale about the participants’ length of residence in their L2/n (‘correct’, ‘neither correct nor wrong’, ‘wrong’) and (ii) reaction communities is given in the Supplementary Information. Table 1 times (RTs). RTs are informative about the possible existence of presents the participant demographics. an unmarked/preferred order, because a comparison across orders (i.e., unmarked/baseline vs. marked/less preferred) should reveal an extra processing cost in the latter (Erdocia et al., 2009). Results This cost occurs because the marked stimulus deviates from the The results were analyzed using jamovi, version 1.8 (the jamovi expectations that the cognitive parser forms about upcoming project 2021; R Core Team, 2021). RTs showed a skewed right input, based on its knowledge about what is the most frequently tail; the standard logarithm (RT´ = log10(RT)) was applied to encountered, unmarked option (Imamura et al., 2016). normalize them, and then the classical 3 SD filter was used to The task consists of two orders and three conditions. All test detect outliers. As a result, 15/1251 outliers have been removed items have the same syntactic structure, featuring two adjectives from the congruent order (1.19%) and 13/1251 from the incon- and a Spelke object in the object position (e.g., ‘I bought a square gruent order (1.03%). In total, the results include 2474 data points black table’). Each of the three conditions includes one of the for the on-line measure (reaction times) and 2502 for the off-line following adjective pairs: 1. size adjective-nationality adjective, 2. measure (acceptability judgments). shape adjective-color adjective, 3. subjective comment adjective- Starting from the acceptability judgments, the results show that material adjective. Each condition has two orders with three test participants accept both the congruent and the incongruent structures per order (18 test structures in total). In the congruent sentences as correct (Fig. 1). The latter is the predominant order, the size adjective precedes the nationality adjective, answer, followed by ‘neither correct, nor wrong’, and then by complying with (2). In the incongruent order, the nationality ‘wrong’, and this pattern is observed in both orders. Splitting for adjective precedes the size adjective, violating (2). The design of the condition leaves this pattern unaltered (Fig. 2), although a more task was based on Stowe and Kaan (2006). The task was pronounced difference between the sentences that comply with implemented in Ibex Farm (Drummond, 2013). The task and the the hierarchy in (2) and those that violate it can be observed. To full dataset are available at https://repositori.urv.cat/fourrepopublic/ test these differences statistically, a generalized linear model search/item/PC%3A3607. analysis was run with the categorical dependent variable being treated as multinomial. The model showed the effect of order on Participants. All participants were neurotypical adults, capable of providing informed consent. All participants provided written Table 1 Participant demographics. informed consent prior to their involvement in the study, in accordance with the Declaration of Helsinki. Regarding ethical Gender Female 74 approval, the Norwegian Center of Research Data reviewed and Male 65 approved the study protocol (approval number: 55775/3/LH). Education Secondary 18 The task was administered to n = 139 bilingual speakers of Tertiary 121 Greek and a Germanic language, mainly Norwegian, Swedish, Handedness R 121 Danish, English, or German. All bilingual participants stated they L 18 speak Greek as their native language, and at least one Germanic language, with varying degrees of proficiency ranging from good to near native. The original sample involved 167 participants, but 28 participants were excluded on the basis of the following pre- established criteria: (i) providing a series of automatic responses (i.e., RTs below 600 ms), (ii) not completing the task, (iii) non- native knowledge of Greek, (iv) reception of speech-pathology treatment, and (v) presence of neurological disorders. Criteria (iii)-(v) were assessed on the basis of self-report. All participants were recruited online, through invitations posted on social media platforms, and completed the research online, in Ibex Farm. The language of testing was Greek. In previous work (Leivada and Westergaard, 2019), this task was run to two Greek-speaking populations that are different from the bilingual population tested in this work: monolingual speakers of Standard Greek and bidialectal speakers of Standard and Cypriot Greek. The results showed the high acceptability of the structures that deviate from the unmarked order (2), while not finding evidence for an extra processing cost associated with them. In the present work, the task is administered to speakers of Greek who Fig. 1 Judgments split by order. The bars show raw scores. HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | (2022)9:436 | https://doi.org/10.1057/s41599-022-01440-w 3 ARTICLE HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | https://doi.org/10.1057/s41599-022-01440-w Fig. 2 Judgments split by order and condition. The bars show raw scores. Bringing together the two measures, acceptability judgments and RTs, Fig. 5 shows that for both orders, the judgment ‘correct’ is the one that is associated with the shortest decision times. More importantly, unlike what one would expect if the incongruent order was marked and legitimized only under special licensing conditions, Fig. 5 shows that it is the congruent order that is associated with slightly longer RTs, when the sentences are accepted as ‘correct’. According to a generalized linear model, the effect of judgment on RTs is significant (χ2 = 339, p < 0.001). Figure 6 presents a more direct comparison of acceptability judgments and their associated RTs in the two orders. Having presented the results of the dataset produced by this experiment, the next aim is to determine the effect of language group, by comparing this dataset with that of other populations that completed the same task. Taking the group of n = 140 Fig. 3 RTs split by order. The y-axis shows inverse-transformed ms. monolingual speakers of Greek presented in Leivada and Wes- tergaard (2019) as the monolingual comparison group, we com- pare the two datasets in order to examine the possible effect of acceptability judgments to be statistically significant (χ2 = 100.38, developmental trajectory. Recall that although both groups grew p < 0.001), the effect of condition to not be significant (χ2 = 2.22, up as monolinguals in Greece, the bilinguals tested in the present p = 0.694), and the interaction order*condition to be significant study relocated as adults to a linguistic community where their L1 (χ2 = 36, p < 0.001). is not societally present. Figure 7 shows a first comparison of the Within conditions, the effect of order varies: the difference two groups: n = 140 monolingual speakers of Greek and n = 139 between congruent and incongruent orders is significant in the bilingual speakers of Greek and a Germanic language. conditions ‘size-nationality’ (χ2 = 56.6, p < 0.001) and ‘subjective To evaluate the difference in judgments between the two comment-material’ (χ2 = 99.2, p < 0.001), but not in the condition groups, a generalized linear model analysis was run with the ‘shape-color’ (χ2 = 2.39, p = 0.302). This finding agrees with the categorical dependent variable being treated as multinomial. The results of previous experiments (Adam and Schecker, 2011; Lei- effect of language group on acceptability judgments was not vada and Westergaard, 2019) and is due to the distance effect: the found to reach significance (χ2 = 0.075, p = 0.963), but the effect closest two adjective classes are in (2), the more interchangeable of the interaction language group*order is marginally significant their members are in terms of order.4 Importantly, this finding (χ2 = 6.459, p = 0.040). However, these two results are not very does not translate into evidence in favor of SOT, because the informative about the potential differences between the two distance effect is compatible also with proposals that ground groups, because half of the sample on which they are based ordering preferences on cognitive notions (Scontras et al., 2017; consists of stimuli from the congruent order, for which the pre- Leivada and Westergaard, 2019). dominant answer in both groups is the target answer ‘correct’, as Turning to the on-line measure, Fig. 3 shows the RTs split for expected. In other words, there is no room for differences in this order. The aim is to shed light on whether the incongruent, order. For this reason, a separate analysis of accuracy in the two hierarchy-violating sentences are marked or dispreferred and, as groups was conducted, targeting the incongruent order alone. such, incur an extra processing cost compared to their hierarchy- Treating accuracy as a two-level variable (accurate vs. inaccurate), compliant, unmarked counterparts. Figure 4 brings condition the acceptability judgments given to the incongruent stimuli were into the picture. Both figures show that the obtained medians are classified as accurate if they matched the target answer (which for quite similar across orders and conditions. Treating the depen- this order is ‘wrong’) and as inaccurate if otherwise. Figure 8 dent variable as continuous, a generalized linear model showed shows the behavior of the two groups. Figure 9 shows the dis- that processing times do not differ significantly when comparing tribution of judgments across conditions in the two groups. congruent and incongruent sentences (effect of order: χ2 = 2.837, A generalized linear model examined the significance of the p = 0.092). The effect of condition and the interaction order*- effect of group and condition on accuracy in the incongruent order, condition do not reach the significance threshold either as well as their interaction, treating accuracy as logistic. The effect (χ2 = 0.111, p = 0.946 and χ2 = 2.052, p = 0.358 for condition of language group on accuracy is significant (χ2 = 7.48, p = 0.006), and order*condition, respectively). the effect of condition is significant too (χ2 = 43.40, p < 0.001), 4 HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | (2022)9:436 | https://doi.org/10.1057/s41599-022-01440-w HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | https://doi.org/10.1057/s41599-022-01440-w ARTICLE Fig. 4 RTs split by order and condition. The y-axis shows inverse-transformed ms. Fig. 5 Judgments and RTs split by order. They-axis shows inverse-transformed ms. however, the participants of the present experiment accepted them as well-formed, without taking extra time to process them compared to the congruent stimuli. Discussion The research questions (RQ) behind this experiment are the following: (I) What cognitive needs are subserved by AOPs? (II) Is there interspeaker variation in the attested preferences? Given that the time component of the experiment did not adduce evidence for an extra processing cost for the incongruent orders—as should have happened if this was a marked order that Fig. 6 A direct comparison of judgments and reaction times across the parser either disprefers or legitimizes only under special orders. The y-axis shows inverse-transformed ms. licensing conditions—the conclusion is that all the orders in (1), that have been variably described as ungrammatical, odd, awk- ward, marked, or semantically incorrect, are grammatically well- while the interaction language group*condition is not significant formed and highly acceptable. At the same time, the congruent (χ2 = 4.53, p = 0.104). Post-hoc tests with Bonferroni correction orders elicited a higher degree of acceptability than the incon- confirmed the effect of language group (p = 0.014) and the effect of gruent ones. This finding, in combination with the results from condition (condition ‘shape-color’ vs. condition ‘subjective com- other experiments that found robust ordering preferences (e.g., ment-material’: p < 0.001, condition ‘size-nationality’ vs. condition Scontras et al., 2017), begs the question: If the parser consistently ‘shape-color’: p = 0.011, condition ‘size-nationality’ vs. condition likes some orders more than others, some cognitive needs must ‘subjective comment-material’: p = 0.003). be subserved by such preferences. Which are these needs? The second important finding showed in Fig. 8 is the degree to So far, this question has not been tackled in a multifactorial which the behavior of the participants deviates from what all way that goes significantly beyond observations formed at the theories describe as the ill-formed, odd, or marked order. Both surface level (but see Kotowski and Härtl, 2019 for an exception). SOT and mCOT predict that the incongruent stimuli employed in To explain this better, an important note is due with respect to this study should be rejected or treated as marked respectively; the various notions put forth by the various subproposals within HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | (2022)9:436 | https://doi.org/10.1057/s41599-022-01440-w 5 ARTICLE HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | https://doi.org/10.1057/s41599-022-01440-w Fig. 7 Comparison of judgments across orders in two populations: monolinguals and bilinguals. The bars show raw scores. attests to its idiosyncratic, pragmatically determined nature. For this reason, nationality adjectives pose a challenge for subjectivity and inherentness accounts: A black cat is black in all linguistic communities, and it is unlikely that faultless disagreement will emerge over its blackness. The same cannot be claimed for the nationality adjective in ‘Turkish coffee’. Examples (3a-d) can all receive similar analyses to the one for Turkish coffee. (3) a. Turkish delights b. Russian salad c. Bavarian cream d. Italian dress In (3a-d), the adjective expresses an idiosyncratic relation to the noun that is determined by pragmatic conditions and Fig. 8 Accuracy in rejecting the incongruent stimuli. The comparison is does not necessarily refer to origin. Turkish delights are between the bilingual participants of the present experiment and the called Greek delights in Greece, while they are not known as monolingual participants of Leivada & Westergaard (2019). Turkish delights as Turkey. The Russian salad is not known as such in Russia and was not invented by a Russian. The Bavarian cream was neither conceived in Bavaria, nor by a mCOT: Subjectivity, inherentness, noun-specific frequency, high Bavarian, while it is perfectly possible that a dress marketed collocability, and phonological weight/length are not explanations by an Italian brand is manufactured in another country and of the observed AOPs; they are observations of what happens at designed by a non-Italian designer. In all these examples, the phenotypic level. Put differently, the fact that an adjective A is there is some type of relation expressed between the statistically likely to occur given a noun N, or that less subjective nominal and the adjective, but this relation is neither adjectives tend to appear closer to the noun, are surface obser- semantically transparent, nor always the same: It is place of vations that must be explained from a cognitive point of view; production in (3a), historical origin in (3b), first recipient in they are not explanations of the origin of AOPs themselves. They (3c), and origin of brand in (3d). tell us what happens at the phenotypic level, but not why it One could counter that ‘Spanish coast’ is cross-linguistically happens and what cognitive needs trigger this behavior. There- uncontentious if one knows basic geography. Yet, even in this fore, the aim is to tackle RQ (I) by uncovering the cognitive example, the relation between the nationality adjective and the underpinnings of AOPs, focusing on the parts of the hierarchy in noun is a matter of idiosyncratic convention and cross- (2) that have been examined in this study: 1. size adjective- linguistic variation exists (cf. Costa Brava being described as a nationality adjective, 2. shape adjective-color adjective, 3. sub- ‘Spanish coast’ vs. a ‘Catalan coast’). To explain the jective comment adjective-material adjective. idiosyncrasy, in order to classify an object as red, a specific Starting with the condition ‘size-nationality’, nationality condition must be met: the presence of redness. This condition adjectives appear closest to the noun because of their special is a salient property of the described object. A Spanish coast nature as sociopragmatic conventions. Succinctly put, nation- though may lack any salient property that can be construed as ality/origin adjectives form idiosyncratic concepts that encapsu- Spanishness. Unlike ‘red eyes’, ‘red car’, and ‘red soil’, which late the pragmatic conventions of the linguistic community in share the tangible property of redness, ‘Spanish eyes’, ‘Spanish which they are uttered. To give a concrete example, what is often car’, and ‘Spanish soil’ denote some relation to Spain or to an referred to as Turkish coffee in Turkey, is called Greek coffee in individual from Spain, but this relation is idiosyncratic and can Greece, Bosnian coffee in Bosnia, and Cypriot coffee in Cyprus. be variably described as origin, place of production, or any link Essentially, these nationality adjectives do not refer to a property to the Spanish country, language, or culture. inherent to the described object (i.e., Turkish/Greek/Cypriot/ Similar is the case with animate referents (4a-b), whereby Bosnian coffee refers to the same type of coffee); they rather an idiosyncratic relation is expressed, without the adjective express a fixed relation between the noun on which they are denoting a quality inherent to the nominal. formed and the noun they modify in each of the respective languages/linguistic communities. This relation is fixed within, (4) a. Meghan Markle was a British duchess. but not across languages/linguistic communities, something that b. Giannis Antetokounmpo is an American star. 6 HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | (2022)9:436 | https://doi.org/10.1057/s41599-022-01440-w HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | https://doi.org/10.1057/s41599-022-01440-w ARTICLE Fig. 9 Comparison of judgments across orders and conditions in the two populations: monolinguals and bilinguals. The comparison is between the bilingual participants of the present experiment and the monolingual participants of Leivada & Westergaard (2019). Meghan Markle was a British duchess, but she does not ori- inherent, objective property of the nominal, these adjectives are ginate from Britain and, at least in early 2022, she does not have preferably placed closer to the noun, being part of a socio- the British citizenship. Giannis Antetokounmpo is an American pragmatic convention. star ever since he plays in the NBA, however he is not American, Turning to the condition ‘shape-color’, our results show that but Greek, born to Nigerian parents, and stateless in the first this is the condition with the smallest difference in terms of years of his life. Evidently, there are contexts that legitimize the acceptability ratings across the congruent and the incongruent use of these nationality adjectives as denoting various types of stimuli (Fig. 2). This finding illustrates the distance effect: the relations. Crucially, the expressed relations are a matter of con- closer two adjective classes are in (2), the more interchangeable vention, as there is no inherent British or American quality in any their members are in terms of order. The reason is not that (2) is of the referents in (4). As in (3), these adjectives express some an innate hierarchy that predicts a more rigid ordering among its relation between the country and the individual. Precisely because distant components. The explanation we propose is that shape this relation is largely idiosyncratic and does not refer to an adjectives and color adjectives are variably ordered because the HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | (2022)9:436 | https://doi.org/10.1057/s41599-022-01440-w 7 ARTICLE HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | https://doi.org/10.1057/s41599-022-01440-w expectation value assigned by the cognitive parser is the same for Intolerance. According to this bias, the cognitive parser tends to both these categories of adjectives. treat ambiguous situations as undesirable (Frenkel-Brunswik, It is a well-established fact that the parser, while processing the 1949; Tanaka et al., 2015). The linguistic manifestation of this linguistic message, forms expectations about incoming stimuli. bias amounts to the Gricean maxim of manner that suggests that Words with low expectation value, as in the final word in “She one’s conversational contribution must be as clear and as orderly spread the bread with socks” are known to elicit larger N400s (i.e., as possible, avoiding obscurity and ambiguity (Grice, 1957). Of a negative-going deflection that peaks around 400 milliseconds course, ambiguity is pervasive both in human life and in human post-stimulus onset) than expected words, showing that the language, but equally pervasive are the strategies we can use to cognitive parser reacts when encountering deviations from the dissolve ambiguities when necessary. Specifically for adjective expectations it has formed (Kutas and Hillyard, 1980). Going order, the proposal is that speakers can externalize shape>color back to the tested condition, it has been found that shape and or color>shape, placing first the property of the nominal they find color are perceived almost simultaneously (Viviani and Aymoz, most disambiguating in each context (Kemmerer et al., 2007). If 2001 and references therein). Since the parser registers them at the context poses no such need for disambiguation, other factors the same time, and even uses cues from the one to categorize the such as availability and visual syntax representation kick in. This other (e.g., the prominent role of color in mediating shape, which proposal predicts that the incongruent order color>shape is has been found in non-human primates too; Lafer-Sousa and perfectly acceptable both when color is the most appropriate Conway, 2013), it assigns them the same expectation value and discriminating factor among a set of qualities, but also when has no strong preference in ordering them in a specific way, hence color is just one quality among many equally discriminatory their high interchangeability. ones. Evidence for the second scenario comes from the accept- One explanation for any remaining weak preference for pla- ability judgments presented in the previous section. Specifically cing shape adjectives before color adjectives is that language for the condition ‘shape-color’, the two orders were found to be reflects vision. Under this account, syntactic preferences may be identical in terms of their acceptability in the out-of-context mirroring the syntax found in the perception of a visual object presentation of the stimuli, which was employed in the present (Pinna and Deiana, 2015). The visual object is a set of multiple experiment (Figs. 2 and 9). This absence of context is important, properties, both explicit (i.e., readily visible) and implicit. Visual because it shows that the two types of adjectives can be ordered attributes like shape, material, and color are often placed in the freely, even when there are no contextual needs that force the foreground, with respect to other properties like illumination, speaker to invert the usual order. density, or contouring (Pinna and Deiana, 2015). If shape is Turning to the third condition, ‘subjective comment-material’, granted a slightly more prominent position than color in the this is the domain where the clearest difference between con- visual object, the linguistic object may reflect this preference. At gruent and incongruent stimuli was found (Figs. 2 and 9). The the same time, these perceptually induced biases do not seem to explanation we propose for this finding is based on Scontras provide the full picture. Visual syntax may indeed have a role in et al.’s (2017) results, and more specifically on their brief obser- the attested orderings, but other cognitive biases come into play. vation that as noun phrases are built semantically outward from For instance, the production-driven availability bias posits that the noun, the less subjective content is the one that enters earlier the most available adjectives are placed first to ease production in the process (see also Scontras et al., 2019). Addressing the (Fukumura, 2018 and references therein). This claim is sup- question of what drives this behavior, the answer is that the more ported by evidence from both within and outside the literature subjective adjectives usually enter the computation last, due to a on adjective order, as it has been found that attributes that are cognitive bias called Novel Information Bias (NIB). more familiar or closer to the identity of the speaker tend to be NIB refers to the cognitive tendency to avoid tokenizing mentioned first (Smirnova et al., 2019). Consequently, language multiple, adjacent occurrences of the same type, because of a does not always externalize the syntax of the visual object in a general bias to provide more attentional resources to novel faithful or uniform fashion. In fact, linguistic syntax has devices information (Leivada, 2017). Consequently, this information is for overwriting the input of visual syntax, based on the com- often granted a more prominent position in the linguistic municative needs it faces in different contexts. Focus fronting message to facilitate clear and, to the degree possible, effortless (1g) for adding emphasis to one aspect of the linguistic message identification. More prominent in this case means first in lan- is such an example. This means that driven by the need to guages like English (i.e., Adj-Adj-Noun) or last in languages subserve different communicative needs, speakers produce like Spanish (i.e., Noun-Adj-Adj). This happens because one of shape > color or color > shape, depending on which order is the key abilities of the parser is to aptly keep track of sequence more likely to facilitate referent identification in a specific con- edges (Endress et al., 2009; Ferry et al., 2015). Therefore, novel text, while also taking into account the needs of both the speaker information with a low expectation value is typically placed in a (e.g., availability) and the addressee (e.g., discriminability; Danks saliently accessible position: at the edges of the Adj-Adj-Noun/ and Glucksberg, 1971; Haywood et al., 2003; Fukumura, 2018). Noun-Adj-Adj constellation. In other words, it seems that linguistic syntax plays a key role in To unpack the somewhat unclear notion of novelty, a concrete AOPs: It is the interface that both mediates the cognitive needs finding from corpus studies refers to how often an adjective subserved by AOPs and fine-tunes their externalization, collocates with a noun. For instance, Wulff’s (2003) results sug- deploying different strategies (e.g., fronting) to ensure that the gest that noun-specific frequency is a highly significant factor that linguistic message is saliently conveyed, satisfying context- mediates order: adjectives with high noun-specific frequency tend specific communicative needs. Therefore, AOPs, as a determin- to appear closer to the noun in a multi-adjective string. Noun- ing factor of nominal syntax, both underlie adjective serialization specific frequency is proportional to expectation value—due to and boil down to cognitive factors. the high collocability of the adjective and the noun, the parser If we attempt to synthesize the overarching connection expects to see them together—and inversely proportional to between variable adjective order and effective communication, it novelty: The higher the expectation value is, the lower becomes becomes clear that word order is mindful to both production- the novelty. Put another way, a high expectation value entails a and perception-driven tendencies. Weighing all these tendencies diminished degree of novelty that may surprise the addressee. The when deciding which order to produce seems amenable to ulti- notions of novelty and surprise should be understood in this mately being attributable to a general cognitive bias: Ambiguity context as referring to attributes that the parser does not actively 8 HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | (2022)9:436 | https://doi.org/10.1057/s41599-022-01440-w HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | https://doi.org/10.1057/s41599-022-01440-w ARTICLE Table 2 Frequency of constructions in the categories the noun will host adjectives that do not need the special, out- ‘subjective comment + toy’ and ‘material + toy’ based most position, because such adjectives are already remembered easily enough upon the occurrence of the noun (Lockhart and on COCA. Martin, 1969), probably due to the fact that they describe pre- dictable properties (Eichinger, 1991). In other words, besides Subjective comment Material adjective sociopragmatic conventions that form idiosyncratic concepts adjective (e.g., Russian salad, Spanish coast, American star) and give rise to Nice 22 Plastic 73 specific orders, higher-level ontological categories, such as sali- Ugly 0 Wooden 18 ently observable object-oriented attributes vs. speaker-oriented Good 19 Metal 12 evaluative attributes, appear to come into play. High collocability Bad 1 Rubber 12 Precious 3 Porcelain 1 may then convert an attribute and a nominal into a stereotype Pretty 9 Stuffed 94 (e.g., ‘fluffy kitten’), and this in turn further increases the noun- Silly 10 Furry 2 specific frequency (Posner, 1986). Bouchard (2005) also proposes Cheap 12 Fluffy 4 concept iconicity as a general principle of serialization of adjec- Cute 10 Plush 40 tives: If the adjective is likely to form an idiomatic concept with Expensive 24 Magnetic 2 the noun, it tends to be placed close to it (see also Kotowski, 2016 Total 110 Total 258 for a review). Relations between higher-level ontological classes come into play after sociopragmatic conventions that form idiosyncratic expect in a context, because it does not consider them default concepts: If you show somebody ‘a red Russian ball’, its color is dimensions of information that often appear with the nominal. immediately more apparent and less likely to cause any dis- For example, in the context of talking about kittens, the adjective agreements than its origin (Scott, 2002). Therefore, according to ‘fluffy’ has a higher expectation value and a lower degree of the literature that puts forth the subjectivity rule (e.g., “the major novelty compared to the adjective ‘boring’.5 rule is to place the more objective and undisputable qualifications Let’s illustrate what this means for the way subjective comment closer to the noun, and the more subjective, opinion-like ones adjectives are ordered in relation to material adjectives more farther away” Hetzron, 1978: p. 178; see also Scontras et al., 2017), generally, through comparing ‘nice toy’ (subjective comment) to we should observe the reversed version of the pattern we observe: ‘plastic toy’ (material). If ‘plastic’ collocates with ‘toy’ more often COLOR and not ORIGIN should be placed closest to the noun. than ‘nice’ does, the parser will not be surprised if it sees ‘plastic’ This does not happen because idiosyncratic conventions take as close to this nominal as possible. ‘Nice’, on the other hand, precedence over other relations between adjective classes. denotes a subjective comment that is primarily informative about The last factor that plays a key role in adjective order is pho- the speaker’s perception of the object, not the object itself. If an nological weight, also referred to as length in the literature. This adjective like ‘nice’ patterns with a large set of nouns, while an factor posits that when two or more adjectives are freely ordered, adjective like ‘fluffy’ or ‘plastic’ is compatible with a smaller set the lengthier adjectives tend to appear closer to the noun (Wulff, (i.e., specific Spelke objects), then ‘fluffy’ or ‘plastic’ have a high 2003; Kotowski, 2016). Although the results of the present expectation value in the contexts of nouns denoting these specific experiment are not directly informative about the weight factor, Spelke objects. In other words, we expect to see ‘fluffy’ being because this was controlled for in the experimental design (see mentioned in relation to kittens or ‘plastic’ in relation to toys, but Leivada & Westergaard, 2019), it is worth integrating it in the our expectations about ‘nice’ are weaker because this adjective overall discussion of the cognitive underpinnings of AOPs. The patterns with many nouns. When ‘nice’ and ‘plastic’ must be reason is that this factor stands out from the rest for it does not ordered in one construction that features ‘toy’, the parser refer to some semantic notion (e.g., subjectivity, inherentness, recognizes that ‘nice’ is more generic and less noun-specific, and absoluteness) or some observation over the distribution of the thus assigns it a lower expectation value in constructions that data (e.g., lemma frequency, noun-specific frequency, degree of feature this noun. This claim relies on two premises: first, there collocability). Briefly put, this factor, unlike any other, has been are more constructions in the ‘material + toy’ category than the presented in the literature as a purely phonological one (Wulff, ‘subjective comment + toy’ category, and second, the parser is 2003). However, it is unclear why or even how the articulatory- mindful of such differences in frequency. The second premise has motor interface can have a say that affects word order (i.e., already been established empirically (see Wulff, 2003). To shed syntax). It is equally unclear why the rule ‘place the lengthier light on whether the first premise holds, the Corpus of Con- adjective closer to the noun’ would be activated only with temporary American English (COCA) was searched. The first adjectives that are freely ordered (e.g., shape-color), and what search featured one of the following subjective comment adjec- prevents it from generalizing and applying more broadly, espe- tives: {nice, ugly, good, bad, precious, pretty, silly, cheap, cute, cially since it seems that all adjectives are freely ordered to expensive} + toy. The second search featured one of the following varying degrees, and there are no flat-out rejected, unacceptable material adjectives: {plastic, wooden, metal, rubber, porcelain, orders (cf. endnote 3 and Fig. 2). stuffed, furry, fluffy, plush, magnetic} + toy. The results, given in Addressing these issues, the first step is to propose that weight/ Table 2, grant support to the first premise. length is relevant across categories of adjectives, since the results To continue with the previous example, having established that of this experiment show that there are preferred orders, but not ‘plastic’ has a higher noun-specific frequency value in relation to unacceptable or ungrammatical ones. Second, the way this factor ‘toy’, it follows that the parser assigns to their co-occurrence a has been presented in the literature brings forward an overlooked high expectation value. As a result, when the speaker/signer must problem. In certain linguistic frameworks (e.g., the inverted-Y choose between placing ‘plastic’ or ‘nice’ next to ‘toy’ in a multi- model in Minimalism and its precursors; Chomsky and Lasnik, adjective string, ‘plastic’ wins as the adjective that has the higher 1977), it is hard to sustain the claim that phonology affects expectation value, because the parser prefers to have the novel, syntax, as the latter is taken to be “phonology-free” (Miller et al., less object-oriented, and less expected in the context, information 1997; Irurtzun, 2009). The solution to this problem lies in placed at the outmost position to facilitate easy retrieving and recognizing the cognitive needs of the parser. Unlike previous orienting attention accordingly.6 By extension, the slot closest to studies that presented this factor as a phonological one, we HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | (2022)9:436 | https://doi.org/10.1057/s41599-022-01440-w 9 ARTICLE HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | https://doi.org/10.1057/s41599-022-01440-w propose that its phonological repercussions are only an epiphe- language and learned aspects of its grammar through learning the nomenon, and the effect itself boils down to a cognitive principle rules. Dictionaries and books of grammar that are used in foreign called the Principle of Least Effort. language learning contexts offer explicit instructions on adjective It is well known that words that are used more frequently (e.g., use (e.g., see the lemma ‘Adjectives: Order’ in the Cambridge ‘and’, ‘the’) tend to be shorter (Zipf’s law of abbreviation; Zipf, Dictionary). In other words, it is very likely that bilinguals have 1932; 1949). Zipf theorized that this pattern is the result of received instructions about the prescriptively correct order of accommodating two competing needs: the pressure to take the adjectives in their L2/n, and these rules mention a version of the path that entails the least effort (i.e., short words need less effort hierarchy in (2). This explanation suggests that the multifactorial to produce) and the pressure for successful communication (i.e., origin of AOPs relies on different cognitive biases, but in practice, short words are more susceptible to noise in the transmission of the realization of different orders in language is also modulated the linguistic message). The Principle of Least Effort was pro- by statistical learning over the input. posed as an explanation of the law of abbreviation. Language Last, addressing the scope of the obtained results, a question strives for optimizing form-meaning mappings under competing that arises is whether these favor one of the two proposals about pressures, such that a ‘frequency-length-meaning’ relationship is the origin of AOPs, SOT, or mCOT, which have a syntactic and a formed: words that are used more frequently tend to be shorter cognitive orientation, respectively. This is not one of the two and tend to have the most frequent meanings (see Kanwal et al., research questions of this experiment, and the experimental 2017 for a recent overview). design cannot adduce evidence for either one of these theories. In If some adjectives are more generic in meaning, it follows that fact, it can be argued that a theory does not have to choose they are used more frequently than others and are compatible between these two theories, under the premise that syntax reflects with many nouns. For example, the evaluative adjective ‘good’ has cognitive principles. In this sense, the obtained results cannot a frequency of 1,130,305 in COCA, while the material adjective confirm or disconfirm any of the two theories, SOT or mCOT, ‘plastic’ has a frequency of 43,844. It is likely that genericity/ and the results do not seem more probable under one of them, frequency of meaning is what makes some adjectives appear first because it seems that the two theories together can explain them. in an Adj-Adj-Noun construction, while by virtue of Zipf’s law, it There is, however, one important disagreement between SOT and is frequency of use that makes them be shorter. Under our mCOT, and in this respect the obtained results seem to tentatively explanation, the weight/length factor has been mistakenly iden- favor the latter: rigidity. More specifically, the obtained results tified as an individual factor that pertains to phonology, suggest that it is more meaningful to talk about ordering pre- according to the literature; it is a by-product of the Principle of ferences than rigid hierarchies that predict that certain orders are Least Effort. Also, unlike many accounts of Zipf’s law that focus ungrammatical (Cinque, 2014).7 exclusively on the interaction between length and frequency, the As mentioned in the Introduction, many syntactic accounts explanation of AOPs put forth here suggests that meaning acts as within SOT accept that AOPs may be violated if the emphasis (on the main determinant of the attested orders, affecting both fre- any adjective) shifts, if the scope relations change (Alexiadou et al., quency and length. As Piantadosi (2014) puts it, word meaning is 2007 and references therein), or in cases of parallel modification the best causal force in shaping frequency. In his words, ““[h] (Ferris 1993), where a pause exists in between the adjectives (e.g., ‘a appy” is more frequent than “disillusioned” because the meaning blue, beautiful, woollen, expensive jumper’). In other words, many of the former occurs more commonly in topics people like to syntactic accounts accept that the ordering restrictions are not discuss” (Piantadosi, 2014: np). absolute, recognizing some freedom in the ordering, especially in the Another important finding is the effect of language group on presence of specific communicative needs (e.g., emphasis, contrastive providing the target acceptability judgment: ‘correct’ for the focus). From this perspective, the variation in the orderings is not at congruent stimuli and ‘wrong’ for the incongruent stimuli. The odds with all syntactic accounts. At the same time, the underlying effect of language group on accuracy provides an answer to RQ assumption in many SOT accounts is that, in the absence of any (II), about the existence of interspeaker variation in the attested special licensing conditions like emphasis, there is an unmarked preferences. As the previous discussion suggested, AOPs are order that predicts a rigid spine of adjectives. Although the obtained sensitive to the statistical distribution of the input data. results do not conclusively settle this matter, it seems that variation According to Yang’s (2000) Generalized Statistical Learning exists even in the default, unmarked setting (i.e., in the out-of-context Hypothesis, this is a general property of acquisition. The child presentation of the stimuli in the present experiment that does not learner can be parallelized to a generalized data processor, which call for emphasis or other special conditions that justify any devia- approximates the target language based on the statistical dis- tion from the unmarked order), casting some doubt on claims about tribution of the input data. If this is an accurate description of the rigidity in what is typically viewed as the unmarked order. process of extrapolating the target grammar, statistical learning can explain the obtained differences in accuracy across language groups (Figs. 7 and 8). The effect of language group (monolingual Outlook. A timed acceptability judgment task showed that vs. bilingual) in providing the target judgment in the incongruent deviations from what is often deemed as the universal order for order was found to be significant, with bilinguals performing adjectives are highly acceptable. The reaction times component better than monolinguals. This finding is at odds with the long- showed that the acceptability of these deviating orders is not entertained claim that a linguistic universal is behind AOPs. If subject to long processing times, contrary to what should have AOPs boil down to an innate universal, how is it possible that happened if these sentences were marked and legitimized under different groups of informants variably accept sentences that special licensing conditions. Overall, three cognitive principles flatly violate it as well-formed? were identified as driving AOPs: the Ambiguity Intolerance, the The answer has to do with the statistical distribution of the Novel Information Bias, and the Principle of Least Effort. These data. The observed interspeaker variation among speakers of the principles explain why one order is deemed more natural than same language attests to the sensitivity to the statistical properties others, but crucially they do not predict the ungrammaticality of of the input. Although both groups were raised as monolinguals the other orders. Upon comparing two groups of speakers of the in Greece and were tested in their L1, bilinguals differ from same language, who differ in terms of their developmental tra- monolinguals in having relocated in an L2 community as adults. jectory (i.e., monolinguals vs. late bilinguals), we observed sig- This means that they were consistently exposed to a foreign nificant variation in their acceptability judgments of the deviating 10 HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | (2022)9:436 | https://doi.org/10.1057/s41599-022-01440-w HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | https://doi.org/10.1057/s41599-022-01440-w ARTICLE orders. This finding is unanticipated since the relevant literature adjectives. The third criterion yields unclear results. Adjectives import lexical has claimed that an innate universal is behind adjective ordering. information. A closed class, however, refers to a category of function words (i.e., determiners, complementizers) that does not readily accept new members. One cannot Precisely because the obtained results point to the existence of rule out the possibility that a speaker of a language that has only a few color adjectives, adjective ordering preferences, and not adjective ordering including red, blue, green and yellow, may describe an object as blue-green, thus restrictions that ban certain orders via an innate universal, we coining a new lexical item in a way that does not work for true function words (i.e., argue that these preferences are the outcome of the synergistic creating a new determiner with transparent meaning by spontaneously merging two interplay of specific cognitive biases in terms of origin, but in existent determiners). The last criterion is perhaps the most difficult to apply. SOT’s terms of their manifestation in language, they are subject to claim is that only direct modification adjectives are rigidly ordered, and in order to determine whether an adjective is a direct modification one or not, we employ a statistical processing and sensitivity to the input, giving rise to criterion that says that these adjectives are rigidly ordered. The circular reasoning interspeaker variation. behind this last criterion brings along several important challenges about what counts as an explanation in linguistics (Leivada, 2020; Larson, 2021). Data availability The full dataset is available at https://repositori.urv.cat/ References Adam S, Schecker M (2011) Position und Funktion: kognitive Aspekte der Abfolge fourrepopublic/search/item/PC%3A3607. attributiver Adjektive. In: Schmale G (ed.) Das Adjektiv im heutigen Deutsch. Syntax, Semantik, Pragmatik. Tübingen, Stauffenburg, pp. 157–172 Received: 24 February 2022; Accepted: 14 November 2022; Alexiadou A (2001) Adjective syntax and noun raising: word order asymmetries in the DP as a result of adjective distribution. Studia Linguist 55:217–248 Published online: 05 December 2022 Alexiadou A (2014) Multiple determiners and the structure of DPs. John Benja- mins, Amsterdam Alexiadou A, Haegeman L, Stavrou M (2007) Noun phrase in the generative perspective. Mouton de Gruyter, Berlin Batchelor K (2017) Adjective order: The red big barn. BookHive. https://www. Notes bookhivecorp.com/blog/research/adjective-order-red-big-barn/ 1 The term ‘adjective ordering preferences’ is used instead of ‘adjective ordering Bever TG (1970) The cognitive basis for linguistic structures. In: Hayes JR (ed.) restrictions’ because no order is flat-out rejected. Recent experiments have found that Cognition and the development of language. Wiley, New York, pp. 279–362 orders that deviate from (2) may be less preferred than the compliant order (Scontras Bouchard D (2005) Sériation des adjectifs dans le SN et formation de concepts. et al., 2017), but still, they are acceptable to varying degrees (Leivada and Westergaard, Recherches linguistiques de Vincennes 34:125–142 2019). Chomsky N, Lasnik H (1977) Filters and control. Linguist Inq 8(3):425–504 2 Although this factor has been argued to impact ordering independently of subjectivity/ Cinque G (1994) On the evidence for partial N-movement in the Romance DP. In: inherentness (Hahn et al., 2018), the two converge on giving rise to the same order. Cinque G, Koster J, Pollock J-Y, Rizzi L, Zanuttini R (eds.) Paths towards The interaction of these two factors poses a chicken and egg question: Did a high Universal Grammar. Studies in honor of Richard S. Kayne. Georgetown degree of collocability (due to noun-specificity) result to certain ordering preferences University Press, Washington, DC, pp. 85–110 or did subjectivity/inherentness-based ordering preferences result to a higher degree of Cinque G (2005) Deriving Greenberg’s Universal 20 and its exceptions. Linguist collocability between certain nouns and adjectives? Similarly, if a causative relation Inq 36:315–332 between high collocability and AOPs is assumed, the observation that some nouns and Cinque G (2010) The syntax of adjectives. A comparative study. MIT Press, adjectives have a high collocability is not informative about the direction of the effect. Cambridge, MA Is high collocability the origin of AOPs or did the frequency of collocability become Cinque G (2014) The semantic classification of adjectives. A view from syntax. Stud high thanks to independently motivated AOPs that favored certain combinations of Chin Linguist 35(1):3–32 adjectives and nouns more than others? Danks JH, Glucksberg S (1971) Psychological scaling of adjective orders. J Verbal 3 Consistency here refers to the fact that the adjective ordering in (1a) is attested cross- Learn Verbal Behav 10:63–67 linguistically. However, the strength of the preference for this order over others is not Dixon RMW (1982) Where have all the adjectives gone?: And other essays in equally consistent. Many theoretical, corpus, and experimental studies have shown that semantics and syntax. Mouton, Berlin adjective orderings that deviate from (1a) and (2) are both acceptable and productively Dowling T (2016) Order force: the old grammar rule we all obey without realising. used in spontaneous speech (e.g., ‘red gorgeous color’ in Wulff, 2003; sabates noves The Guardian. https://www.theguardian.com/commentisfree/2016/sep/13/ blanques ‘white new shoes’ in McNally and Boleda, 2004; kokino megalo vivlio ‘red big sentence-order-adjectives-rule-elements-of-eloquence-dictionary book’ in Alexiadou, 2014; großen schönen Audi ‘big beautiful Audi’ in Kotowski and Drummond A (2013) Ibex Farm. Available at http://spellout.net/ibexfarm Härtl, 2019; malino omorfo fustani ‘woolen beautiful dress’ in Leivada and Eichinger LM (1991) Woran man sich halten kann: Grammatik und Gedächtnis. Westergaard, 2019). Jahrbuch Deutsch als Fremdsprache 17:203–220 4 See section ‘Discussion’ for the cognitive origins of the distance effect. Endress AD, Nespor M, Mehler J (2009) Perceptual and memory constraints on 5 Almost all English dictionaries offer ‘cat/kitten/dog’ as the prototypical examples of language acquisition. Trend Cogn Sci 13(8):348–353 showing how ‘fluffy’ is used. Erdocia K, Laka I, Mestres-Missé A, Rodriguez-Fornells A (2009) Syntactic com- 6 This claim agrees with the predictions of the Subjective Distance Principle, according to plexity and ambiguity resolution in a free word order language: behavioral which attributes that are more familiar, or closer to the identity of the speaker are and electrophysiological evidenceS from Basque. Brain Lang 109:1–17 Ferris C (1993) The meaning of syntax. A study in the adjectives in english. mentioned first (Smirnova et al., 2019). Also, the current proposal that connects Longman, London adjective ordering with NIB is in agreement with evidence outside the domain of Ferry AL, Fló A, Brusini P, Cattarossi L, Macagno F, Nespor M, Mehler J (2015) On adjectives that also suggests that language has strategies to place the most unexpected/ the edge of language acquisition: inherent constraints on encoding multi- novel material first. Focus fronting (1g) is one example. syllabic sequences in the neonate brain. Dev Sci 19:488–503. https://doi.org/ 7 One possible counterargument is that the rigid order pertains only to direct 10.1111/desc.12323 modification adjectives (Cinque, 2014), such that the attested flexibility in the ordering Franke M, Scontras G, Simonič M (2019) Subjectivity-based adjective ordering preferences is not really at odds with SOT. To evaluate this counterargument, let us maximizes communicative success. In: Goel A, Seifert C, Freksa C (eds.) examine the criteria upon which an adjective can be classified as a direct modification Proceedings of the 41st Annual Conference of the Cognitive Science Society. adjective (Cinque, 2014): direct modification adjectives are (i) non-predicative, though Cognitive Science Society, Montreal, QB, pp. 344–350 they may also have predicative uses; (ii) closer to the noun; (iii) a closed functional Frenkel-Brunswik E (1949) Intolerance of ambiguity as emotional and perceptual class; and (iv) rigidly ordered. The first criterion is somewhat ambiguous in its personality variable. J Pers 18:108–143 definition. The adjectives employed in the present experiment have predicative uses. Fukumura K (2018) Ordering adjectives in referential communication. J Mem Lang By this criterion alone, it is not clear whether they should be classified as direct 101:37–50 modification adjectives or not. The predictions of the second criterion are challenged Grice HP (1957) Meaning. Philos Rev 62:377–388 both by corpus studies and elicitation studies that have provided ample evidence (see Gutoskey H (2019) The very particular grammar rule you probably never knew— endnote 3) that the position closer to the noun can be occupied interchangeably by but use every day. Mental Floss. https://www.mentalfloss.com/article/597981/ different categories of adjectives. Although Cinque (2014) recognizes that both ‘big adjective-order-grammar-rule-you-probably-never-knew-about brown dog’ and ‘brown big dog’ are possible in English, he also argues that the former Hahn M, Degen J, Goodman N, Jurafsky D, Futrell R (2018) An information- corresponds to the natural order. He further argues that this reversal should not be theoretic explanation of adjective ordering preference. Proceedings of the taken as showing that no rigid order exists in English among direct modification 40th Annual Meeting of the Cognitive Science Society (CogSci), pp. 1–6 HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | (2022)9:436 | https://doi.org/10.1057/s41599-022-01440-w 11 ARTICLE HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | https://doi.org/10.1057/s41599-022-01440-w Hanson D (2015) Ordering adjectives… Who knew? Crafting Connections https:// Smirnova A, Romero Sanchez R, Lenarsky A (2019) Contextual determinants of www.crafting-connections.com/2015/09/ordering-adjectives-who-knew.html adjective order: beyond itsy bitsy teeny weeny yellow polka dot bikini. In: Haywood SL, Pickering MJ, Branigan HP (2003) Co-operation and co-ordination Goel AK, Seifert CM, Freksa C (eds.) Proceedings of the 41st Annual Con- in the production of noun phrases. In: Alterman R, Kirsh D (eds.) Pro- ference of the Cognitive Science Society. Cognitive Science Society, Montreal, ceedings of the 25th Annual Conference of the Cognitive Science Society. QB, pp. 2825–2831 Lawrence Erlbaum Associates, Mahwah, pp. 533–538 Sproat R, Shih C (1991) The cross-linguistic distribution of adjective ordering Hetzron R (1978) On the relative order of adjectives. In: Seiler H (ed.) Language restrictions. In: Georgopoulos C, Ishihara R (eds.) Interdisciplinary universals. Papers from the conference held at Gummersbach/Cologne, approaches to language. Kluwer Academic Publishers, Dordrecht, Germany, October 3–8, 1976. Narr, Tübingen, pp. 165–184 pp. 565–593 Horobin S (2016) Beware the bad big wolf: why you need to put your adjectives in Stavrou M (1999) The position and serialization of APs in the DP. In: Alexiadou A, the right order. The Conversation. https://theconversation.com/beware-the- Horrocks G, Stavrou M (eds.) Studies in Greek Syntax. Kluwer Academic bad-big-wolf-why-you-need-to-put-your-adjectives-in-the-right-order-64982 Publishers, Dordrecht, pp. 201–226 Imamura S, Sato Y, Koizumi M (2016) The processing cost of scrambling and Stowe LA, Kaan E (2006) Developing an experiment. Techniques and design. topicalization in Japanese. Front Psychol 7:536 Unpublished MS Irurtzun A (2009) Why Y: on the centrality of syntax in the architecture of Svenonius P (2008) The position of adjectives and other phrasal modifiers in the grammar. Catalan J Linguist 8:141–160 decomposition of DP. In: Kennedy C, McNally L (eds.) Adjectives and Kanwal J, Smith K, Culbertson J, Kirby S (2017) Zipf’s law of abbreviation and the adverbs: syntax, semantics, and discourse. Oxford University Press, New principle of least effort: language users optimise a miniature lexicon for York, pp. 16–42 efficient communication. Cognition 165:45–52 Tanaka Y, Fujino J, Ideno T, Okubo S, Takemura K, Miyata J, Kawada R, Fujimoto Kemmerer D, Weber-Fox C, Price K, Zdanczyk C, Way H (2007) Big brown dog or S, Kubota M, Sasamoto A, Hirose K, Takeuchi H, Fukuyama H, Murai T, brown big dog? An electrophysiological study of semantic constraints on Takahashi H (2015) Are ambiguity aversion and ambiguity intolerance prenominal adjective order. Brain Lang 100(3):238–256. https://doi.org/10. identical? A neuroeconomics investigation. Front Psychol 5:1550 1016/j.bandl.2005.12.002 Teodorescu A (2006) Adjective ordering restrictions revisited. In: Baumer D, Kemmerer D, Tranel D, Zdanczyk C (2009) Knowledge of the semantic constraints Montero D, Scanlon M (eds.) Proceedings of the 25th WCCFL. Cascadilla, on adjective order can be selectively impaired. J Neurolinguistics 22:91–108. Somerville, MA, pp. 399–407 https://doi.org/10.1016/j.jneuroling.2008.07.001 The Jamovi Project (2021) jamovi. (Version 1.8). Retrieved from https://www. Kotowski S, Härtl H (2019) How real are adjective order constraints? Multiple jamovi.org prenominal adjectives at the grammatical interfaces. Linguistics 57:395–427. Trotzke A, Wittenberg E (2019) Long-standing issues in adjective order and corpus https://doi.org/10.1515/ling-2019-0005 evidence for a multifactorial approach. Linguistics 57(2-S):273–282 Kotowski S (2016) Adjectival modification and order restrictions. The influence of Viviani P, Aymoz C (2001) Colour, form, and movement are not perceived temporariness on prenominal word order. De Gruyter, Berlin simultaneously. Vision Res 41(22):2909–2918 Kutas M, Hillyard SA (1980) Reading senseless sentences: Brain potentials reflect Waldman K (2014) The secret rules of adjective order. A long fascinating article— semantic incongruity. Science 207:203–205 or is it a fascinating long article? Slate Magazine https://slate.com/culture/ Laenzlinger C (2005) French adjective ordering: perspectives on DP-internal 2014/08/the-study-of-adjective-order-and-gsssacpm.html movement types. Lingua 115:645–689 Whorf BL (1945) Grammatical categories. Language 21(1):1–11 Lafer-Sousa R, Conway BR (2013) Parallel, multi-stage processing of colors, faces and Wulff S (2003) A multifactorial corpus analysis of adjective order in English. Int J shapes in macaque inferior temporal cortex. Nat Neurosci 16(12):1870–1878 Corpus Linguist 8:245–282 Larson RK (2021) Rethinking cartography. Language 97(2):245–268 Yang C (2000) Knowing and learning in natural language. MIT, Doctoral Leivada E (2017) What’s in (a) Label? Neural origins and behavioral manifestations of dissertation Identity Avoidance in language and cognition. Biolinguistics 11(SI):221–250 Zipf GK (1932) Selected studies of the principle of relative frequency in language. Leivada E (2020) Mid-level generalizations of generative linguistics: Experimental Harvard University Press, Cambridge robustness, cognitive underpinnings and the interdisciplinarity paradox. Zipf GK (1949) Human behavior and the principle of least effort: an introduction Zeitschrift für Sprachwissenschaft 39(3):357–374. https://doi.org/10.1515/ to human ecology. Addison-Wesley Press, Cambridge zfs-2020-2021 Leivada E, Westergaard M (2019) Universal linguistic hierarchies are not innately wired. Evidence from multiple adjectives. PeerJ 7:e7438. https://doi.org/10. Acknowledgements 7717/peerj.7438 I thank the participants of this experiment for their willingness to devote their time to Lockhart RS, Martin JE (1969) Adjective order and the recall of adjective-noun science. I am also grateful to Antonio Fábregas for his insights on the syntax of triples. J Verbal Learn Verbal Behav 8:272–275 adjectives in Spanish. This work received support from the European Union’s Horizon McNally L, Boleda G (2004) Relational adjectives as properties of kinds. In: O. 2020 research and innovation program under the Marie Skłodowska-Curie grant Bonami O, Cabredo Hofherr P (eds.). Empirical issues in formal syntax and agreement no. 746652 and from the Spanish Ministry of Science, Innovation and semantics 5, CSSP. pp. 179–196. http://www.cssp.cnrs.fr/eiss5/index_en.html Universities under the Ramón y Cajal grant agreement no. RYC2018-025456-I. The Miller PH, Pullum G, Zwicky AM (1997) The principle of phonology-free syntax: funders had no role in the writing of the study and in the decision to submit the article four apparent counterexamples in French. J Linguist 33:67–90 for publication. Nordquist R (2018) Adjective order. ThoughtCo. https://www.thoughtco.com/ what-is-adjective-order-1688972 Competing interests Panayidou F (2013) (In)flexibility in adjective ordering. Queen Mary University of The author declares no competing interests. London, Doctoral dissertation Piantadosi ST (2014) Zipf’s word frequency law in natural language: a critical review and future directions. Psychon Bull Rev 21(5):1112–1130 Ethical approval Pinna B, Deiana K (2015) Material properties from contours: new insights on The Norwegian Center of Research Data reviewed and approved the study protocol object perception. Vision Res 115:280–301 (approval number: 55775/3/LH). Posner R (1986) Iconicity in syntax. The natural order of attributes. In: Bouissac P, Herzfeld M, Posner R (eds.) Iconicity. Essays on the nature of culture. Stauffenburg Verlag, Tübingen, pp. 305–337 Informed consent All participants provided written informed consent prior to their involvement in the R Core Team (2021) R: A Language and environment for statistical computing. study, in accordance with the Declaration of Helsinki. (Version 4.0). (R packages retrieved from MRAN snapshot 2021-04- 01).Retrieved from https://cran.r-project.org Saba W (2020) The Universal Adjective-Ordering ‘Mystery’—not a Mystery. Med- Additional information ium. https://medium.com/ontologik/the-universal-adjective-ordering-mystery- Supplementary information The online version contains supplementary material not-a-mystery-c10614e50761 available at https://doi.org/10.1057/s41599-022-01440-w. Scontras G, Degen J, Goodman ND (2017) Subjectivity predicts adjective ordering preferences Open Mind: Discov Cogn Sci. 1:53–65 Correspondence and requests for materials should be addressed to Evelina Leivada. Scontras G, Degen J, Goodman ND (2019) On the grammatical source of adjective ordering preferences. Semant Pragmat 12(7):1–21 Reprints and permission information is available at http://www.nature.com/reprints Scott GJ (2002) Stacked adjectival modification and the structure of nominal phrases. In: Cinque G (ed.) Functional structure in DP and IP. The cartography of Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in syntactic structures. Oxford University Press, New York, pp. 91–120 published maps and institutional affiliations. 12 HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | (2022)9:436 | https://doi.org/10.1057/s41599-022-01440-w HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | https://doi.org/10.1057/s41599-022-01440-w ARTICLE Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/ licenses/by/4.0/. © The Author(s) 2022 HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | (2022)9:436 | https://doi.org/10.1057/s41599-022-01440-w 13