How do language models disambiguate semantically and pragmatically complex and polysemic meanings? In this study, we present a computational approach to the analysis of Spanish's polysemic literalmente 'literally, ' an adverb whose meaning and pragmatic functions range from strict word-by-word denotation to (inter)subjective intensification and emphasis. Focusing on the Spanish pre-trained bert model -beto-, two objectives are pursued: i) to shed light onto how artificial language processors interpret pragmatically polyfunctional and semantically polysemic words, and ii) to showcase how the contextual cues drawn on by an artificial language processor can help elucidate semantic polysemy and change in natural language. Using Local Interpretable Model-Agnostic Explanations (lime), our results show that more innovative and grammaticalized uses exhibit a higher degree of syntactic polyfunctionality. We discuss parallelisms and cross-pollination potential between the uncovered computational dynamics of polysemic literalmente and theories of grammaticalization and semantic change.

This study analyzes to what extent communicative accounts, in contrast to production ease ones, can predict the variable coding length of conditional constructions in Spanish. If coding length is determined by communicative efficiency, we anticipate more accessible antecedents to be more parsimoniously encoded and less accessible ones to be more explicitly encoded. Insofar as coding length is determined by production-internal processes, longer coding is expected when speakers experience production difficulties. A Bayesian mixed-effects multinomial regression of 977 conditional constructions indicates that coding length is chiefly regulated by the speaker's communicative goals in that highly accessible antecedents are realized with shorter forms, and more inaccessible antecedents involve longer coding. We propose a speaker-hearer efficiency cline of conditional constructions that ranges from maximally explicit coding, which is less efficient for the speaker but easier to parse for the hearer, to maximally inferential coding, which is more efficient for the speaker but increases the inferential cognitive load of the hearer. Moreover, we find support for the view that pragmatically indirect conditionals are frequently redundantly coded. Because indirect speech act resolution is cognitively heavy for the listener, speakers seem to counterbalance the higher processing costs exerted by them through longer coding.

This cross-dialectal diachronic study focuses on actualization-the gradual mapping out of structural innovations across linguistic environments-and asks: to what extent can the route by which actualization unfolds be explained by appealing to common cognitive constraints, and to what extent are pathways of actualization shaped by community-specific patterns in language use? I analyze real-time syntactic change in Spanish oblique relative clauses (ORCs), which are undergoing an enhancive change towards increasing use of antecedent-agreeing definite articles (e.g., la casa en [la] que nací ‘the house in which I was born’). Over 8,000 occurrences of ORCs were compiled from three Latin American varieties. A Bayesian mixed-effects logistic regression analysis shows that various constraints representing the cognitive accessibility of the antecedent influence actualization similarly across varieties. The discussion addresses the cognitive realism of historical corpus data and formulates testable predictions that present profitable lines of inquiry in future cognitively oriented cross-dialectal diachronic research.

Levshina (2022, p. 230) introduces the Hypothesis of Construction–Lexeme Accessibility and Formal Length, positing that the accessibility of a lexeme is negatively correlated with the coding length of a construction (i.e., low accessibility renders longer coding, and vice versa). Analyzing morphosyntactic variation in contemporary English, Levshina (2022, p. 244) lends support to the hypothesis and suggests that this tendency may account for the evolution of existing constructions and the emergence of new ones. However, empirical testing on non-English languages and diachronic verifications are lacking. The present talk addresses this gap by examining two ongoing processes of language change in Spanish: one involving constructional complexification/lengthening (the conventionalization of a definite article in oblique relative clauses, illustrated in [1]), and one involving constructional reduction through conventionalized omission (the constructionalization of the exclamative construction la de que, emerged as a reduced variant of la cantidad de que, presented in [2]).

(1) a. la casa en que nac-í

DET.FEM.SG house in REL born-1SG-PRET

b. la casa en la que nac-í

DET.FEM.SG house in DET.FEM.SG REL born-1SG-PRET

‘The house in which I was born’

(2) a. la cantidad de vec-es que

DET.FEM.SG amount of time.PL REL

b. la de vec-es que

DET.FEM.SG of time.PL REL

‘The number of times that’

The study analyzes constructional attraction (Schmid, 2000, 2010; see also Levshina, 2018; Schmid & Küchenhoff, 2013), focusing on the predictability of a lexeme given a construction.

The predictions that the study tests are:

I. The less accessible (predictable) an antecedent given an oblique relative clause, the greater the chances of the definite article.

II. The more accessible (predictable) the lexeme given the la [cantidad / Ø] de que construction, the greater the chances of the shorter variant (la Ø de que).

The first case study (1) analyzes data from Peruvian, Colombian, and Argentinian Spanish extracted from the historical CORDE (1850-1974) and CREA (1975-2000) corpora. The second study (2) compiles European Spanish data from the CORDE, CREA, and CORPES XXI (2001-2023) corpora. Using mixed-effects logistic regression analyses, the influence of the constructional predictability is analyzed in interaction with real time. Results consistently show a significant influence of constructional predictability on synchronic variation and diachronic change. As predicted in (I), the definite article emerges in oblique relative clauses with inaccessible antecedents, spreading over time to high-accessibility contexts. Regarding (II), the study reveals the initial omission of cantidad in highly accessible contexts, with the shorter variant spreading to increasingly inaccessible contexts over time. This study provides diachronic support for the Hypothesis of Construction–Lexeme Accessibility and Formal Length, offering a unified account of the influence of accessibility constraints on efficient coding and the relationship between efficiency and the actualization of language change.

Este estudio indaga en la gramaticalización de los posesivos tónicos cuyo uso se documenta desde hace algo más de un siglo en el sintagma adverbial (por ejemplo, delante mío) y en el sintagma verbal (por ejemplo, habla mal tuyo). Se arroja luz sobre la difusión del posesivo tónico “no posesivo” de un ámbito lingüístico a otro, cómo los cambios observados se relacionan entre sí, y qué condiciones y mecanismos posibilitan la extensión del posesivo tónico en su función argumental de persona. Los hallazgos ponen de manifiesto cómo la actualización del cambio avanza sigilosamente a través de extensiones analógicas sucesivas.

This study deals with the grammaticalization of Spanish tonic possessives as person-referential complements. Recently, these possessives have spread to the AdvP (e.g., delante mío 'in front of me') and to the VP (e.g., habla mal tuyo 's/he speaks badly of you') where they replace prepositional complements (e.g., de mí > mío; de ti > tuyo; etc.). The paper scrutinizes the development and diffusion of the possessives’ new function, and the interconnectedness of the changes in the AdvP and the VP. The findings support the view that actualization hinges on similarity-based generalizations, i.e. analogy.

The actualization of language change unfolds gradually across linguistic environments. Over the past decade, there has been growing recognition that actualization is largely driven by similarity-based generalizations-that is, by analogy (De Smet, 2012b). Moreover, frequently used structures often serve as analogical models, exerting a 'gravitational pull' that attracts other structures (De Smet, 2012a, p. 72). In other words, actualization through similarity-based generalizations is largely steered by frequency effects, and actualization hinges on usage and experience with language use. In this talk, I address a somewhat overlooked aspect of actualization theory, namely that the input structures that are most frequent in one community may differ from the most frequent structures in another community (see also Fischer 2013, p. 528). Granted that analogy is sensitive to frequency effects, this discrepancy in input frequency of structures could, hypothetically speaking, result in variations in the substructures guiding the course of actualization across different varieties of a language (see e.g. Grafmiller et al., 2018; Röthlisberger et al., 2017; Verhagen et al., 2018; Barking et al., 2022). Since the cross-varietally potentially variable composition of probabilistic grammars has not yet been incorporated into actualization theory, the extent to which a determined course of actualization is community-specific cannot be predicted by actualization theory in its current form.

To address this issue, I adopt a usage-based perspective on actualization. I hypothesize that if domain-general cognitive mechanisms and processes along with usage factors steer the course of actualization, they should operate similarly across varieties of a language. This, in turn, would mean that the mapping out of change across linguistic contexts should largely align cross-varietally. To test this prediction, I analyze an instance of ongoing syntactic change in Spanish, namely the conventionalization of definite articles in oblique relative clauses (la casa en que nací vs la casa en la que nací, 'the house in which I was born'). Approximately 8000 oblique relative clauses from 19th-20th century Argentinean, Peruvian, and Colombian Spanish are analyzed using mixed-effects logistic regression to test the influence of cognitive and usage-determined factors, viz. the CONSTRUCTIONAL PREDICTABILITY of the antecedent lexeme, the NUMBER and LEVEL OF DETERMINATION of the antecedent, the DISTANCE between the antecedent and the relative clause, and the TYPE OF RELATIVE CLAUSE. Under a usage-based view of the actualization of this change (cf. supra), the prediction is that the new longer form should first be used to convey less accessible (e.g. unpredictable) meanings (Levshina, 2022: 135), and that, over time, the new form diffuses across contexts along similarity-based generalizations by spreading gradually across low-accessibility contexts towards high-accessibility contexts. Crucially, actualization should unfold in a similar manner in different varieties of a language.

The results confirm this hypothesis by showing that, in all of the analyzed varieties, change is initiated and led in contexts involving low accessibility. Over time, the new variant spreads significantly to contexts involving high-accessibility antecedents (e.g. adjacent, definite, singular, higher-frequency antecedents, and restrictive clauses). These findings provide robust support for a usage-based view on actualization with a strong cognitive commitment.

This paper scrutinizes the conventionalization of the Spanish expression la de <noun> que (‘the amount of <noun> that’), a reduced variant of la cantidad de <noun> que. The study seeks to determine the diachrony of and mechanisms underlying the emergence and diffusion of the la de <noun> que expression and whether it has conventionalized to develop into an independent form-function pairing. A Bayesian mixed-effects logistic regression analysis of approximately 2000 observations of diachronic corpus data tests the influence of the conditional probability of lexemes in the noun slot and the register, which both turn out to have a meaningful effect. It is argued that the initial omission of cantidad can be accounted for by appealing to the notion of probabilistic reduction, whereby omission is feasible in contexts involving a high degree of constructional predictability. In the mapping out of change, conventionalization of the innovative la de <noun> que is most observable in contexts involving high constructional predictability and is least prominent in contexts of low constructional predictability. On the grounds that, over time, the la de <noun> que progressively has become stylistically divergent from the longer expression, the two constructions are claimed to be functionally distinct.

The present paper provides a methodological case study on how underlying incipient grammar change might be discerned even when frequencies of the incoming variant are apparently marginal and stable. Analysing the spread of tonic possessive pronouns in complements of locative adverbial constructions in European Spanish from a probabilistic perspective, more than 11,000 locative constructions from 1900 to 2004 were compiled, and probabilistic grammar change was operationalised as an interactive function between language-internal predictors and real time. The results reveal that numerous intralinguistic factors have been and are active in constraining the variation, with the innovation spreading significantly in spite of apparent stability in frequency. Crucially, the findings demonstrate that, even in a relatively standardised written language where the innovation has a considerably low frequency, the innovation grammaticalises along the same pathway as in colloquial vernaculars where the incoming variant is employed much more frequently.

If the same instance of syntactic change is attested across different varieties of the same language, to what extent is the course of actualization noncontingent on regional varieties? This study delves into this question by analyzing approximately 8000 occurrences of Spanish oblique relative clauses from three historical macroregional varieties of Latin American Spanish. A cross-dialectal diachronic variationist analysis using mixed-effects regression modeling tests the hypothesis that, to the extent that actualization is conditioned by cognitive mechanisms and usage factors (such as analogy, frequency, coding efficiency, and accessibility), the trajectories of actualization should unfold cross-varietally in the same manner. The findings corroborate this hypothesis, and furthermore reveal a pattern that demonstrates that part of the variation is entirely functional cross-dialectally, while other constraints act as springboards for actualization. The methodological approach used here thus uncovers dynamics not evident in other variationist approaches to cross-varietal change, such as the comparative sociolinguistic approach.