Abstract
Aims and objectives:
This study addresses fundamental questions about how grammatical gender systems accommodate foreign lexical items by examining English loanwords in contemporary Italian. I investigate whether gender assignment operates through active cross-linguistic semantic networks or default mechanisms, and test the hypothesis that marked (feminine) assignments exhibit stronger semantic motivation than unmarked forms.
Methodology:
Employing corpus-based quantitative analysis, I observe loanword gender patterns using naturalistic spoken data in the domain of Italian workplace discourse. To investigate whether semantic constraints significantly influence gender assignment, I analysed 779 tokens of 112 unassimilated English loanwords from the KIParla corpus (2017–2021), comprising 560 speakers and over 150 hr of recorded speech. Each loanword was systematically coded for (1) assigned gender, (2) availability of native semantic equivalents, and (3) gender alignment with these equivalents.
Data analysis:
Statistical analysis employed Fisher’s exact test and chi-square test to examine semantic gender correspondence patterns.
Findings:
Results reveal significant gender correspondence between loanwords and their Italian associates (p < .05, Cramer’s V = 0.28). Among loanwords with identifiable semantic equivalents, 82.1% exhibit semantic gender correspondence, suggesting active cross-linguistic influence in assignment patterns. Notably, feminine assignments show perfect alignment (100%), while masculine assignments show more variation (75.8%).
Originality:
To my knowledge, this is the first corpus-based study of loanword gender assignment in contemporary Standard Italian using naturalistic spoken data. In particular, this research proposes the semantic accessibility constraint, suggesting that feminine assignment requires clear semantic activation, whereas masculine assignment relies on both semantic and default mechanisms.
Significance:
Results suggest that contemporary unassimilated loanword integration operates primarily through active cross-linguistic semantic networks rather than surface-level processing. In particular, the semantic accessibility constraint offers novel insights into theories of feminine as a marked gender requiring specific semantic justification.
Introduction
Since at least the mid-twentieth century, English has been the primary donor of loanwords in Italian. For much of this period, English terms primarily featured in the specialised lexical domains of technology, politics, finance and entertainment. A more recent development, however, is the influx of English in everyday language, particularly in workplace settings. By what criteria, though, do Italian professionals acquire le skill to improve la performance (in the feminine) by completing un workshop in un open space (in the masculine)? As Corbett (1991, p. 71) notes, ‘borrowings of nouns into languages with gender systems [. . .] are like a continuously running experiment, which allows us to verify the assignment system in the languages in question’.
For Italian, a Romance language with a two-gender system, the assimilation of English loan nouns has long been a subject of linguistic inquiry. While there is a general perception that foreign words entering Italian are almost always assigned masculine, which acts as the ‘neutral’ or ‘common gender’ (Klajn, 1972, pp. 59–60), closer examination reveals a more complex picture. Early studies of gender assignment (GA) focusing on necessity-driven borrowings of American Italian varieties (Correa-Zoli, 1973; Rabeno & Repetti, 1997) emphasised the role of phonological and semantic analogy with L1 words, and suggested a general trend towards ‘default’ masculine gender when these cues were absent or weak. However, the contemporary linguistic landscape presents new conditions. In the domain of current Italian office discourse, with increasingly globalised communication and enhanced English proficiency and cross-linguistic awareness among Italian professionals, the cognitive connections between recently adopted anglicisms and existing lexical items may be significantly strengthened. This new context challenges previous models by suggesting that semantic factors might play a more prominent role in the GA of modern borrowings, potentially contributing to a stronger adherence to the semantic assignment criterion, whereby loanword gender mirrors that of its closest translational equivalent. To explore this hypothesis, this research draws upon empirical data from the KIParla corpus (Mauri et al., 2019), a new collection of contemporary spoken Italian recorded in naturalistic settings between 2017 and 2021, offering a unique window into the evolving dynamics of grammatical gender assignment and cross-linguistic transfer.
From Formal Hierarchies to Sociolinguistic Models
Existing scholarship on gender assignment has explored the importance of semantic (meaning-based) versus formal (phonological or morphological) cues, often emphasising the role of a ‘default’ gender where other factors fail to apply (cf., among others, Corbett, 1991; Nesset, 2006; Poplack et al., 1982; Rice, 2006; Thornton, 2009). However, a consensus on the relative weight of these criteria remains elusive, particularly concerning the impact of sociological factors. As Poplack et al. (1982, p. 4) aptly state, ‘there is little consensus in the literature on whether formal or semantic influences predominate, even in the study of the same language’. The integration of genderless nouns into Italian’s gendered system offers valuable insights in this regard, allowing us to examine how different rules and constraints come into play, interact and potentially even compete with one another. This section reviews the gender assignment criteria most frequently cited in loanword studies, providing the theoretical framework for the present research.
Loans With Animate Referents
There is broad agreement that the gender of animate referents supersedes all other factors and constraints when assigning grammatical gender in the Romance languages. Nesset (2006, p. 1386) goes as far as to say that there is one dominating criterion, which applies universally: the Core Semantic Override Principle, whereby ‘rules referring to biological sex take precedence in gender assignment’. This theory is supported by the empirical data of multiple language-specific studies: Poplack et al. (1982, pp. 23–24) found ‘physiological gender’ to be a ‘knockout factor’ in determining the gender of lexical borrowings in both Puerto Rican Spanish and Montreal French, while DuBord’s (2004, p. 32) analysis of English loanwords in the Spanish of Southern Arizona determined that ‘biological sex’ is the greatest motivator of gender assignment, followed by phonological gender and ‘analogical’ [semantic 1 ] gender. Also in the case of Italian, the English noun ‘manager’, for example, will become il manager (m.) or la manager (f.), depending on whether it denotes a male or female referent. 2
Phonological Gender
The phonological criterion posits that a loanword’s gender is determined by its sound shape, particularly its final segments for Italian, such that words ending in -a and -o can generate feminine and masculine, respectively, in line with the host language’s existing gender paradigms. As noted by Klajn (1972), cases of phonologically motivated GA are rare for anglicisms in Italian, given that English nouns are typically consonant sound-final. We do, however, see compelling evidence of this criterion at work in research conducted on orally acquired English borrowings into varieties of Italian spoken by first- and second-generation Italian immigrants in North America (Correa-Zoli, 1973; Rabeno & Repetti, 1997). These loanwords can show a tendency of assimilation to the phonological constraints of the speakers’ L1, 3 resulting in forms that aligned more closely with typical Italian word endings. Correa-Zoli (1973, p. 125) illustrated a broad trend for final position -er in English borrowings to be phonologically reinterpreted as /a/, in turn generating feminine gender assignments, for example:
1. (a) freezer > frisa (f.)
(b) quarter > quora (f.)
(c) cellar > sella (f.)
(d) sweater > suera (f.)
Such patterns provide a useful insight into the role of perception and production in this criterion: given the tendency of English speakers in the areas where many Italians settled (e.g., New York and Boston) to pronounce a word-final /r/ as /ə/, it is quite plausible that first-generation Italian Americans would have perceived a final schwa sound in [frizə], which they then would have produced as [friza] (assigned feminine gender), with /a/ being the closest approximate in Italian’s vowel inventory (Correa-Zoli, 1973, p. 128). 4
It is also a compelling example of the influence of sociolinguistic context (to which we will return) in gender assignment processes, given that the same loanword in modern-day Standard Italian remains unassimilated (freezer) and is instead assigned masculine, aligning with the gender of its native synonym (congelatore). Indeed, this reflects a broader shift, as quantitative research has shown that since the mid-twentieth century, modern borrowings are increasingly non-adapted (Ferrari, 2017). 5 While earlier waves of borrowings underwent substantial phonological and morphological adaptation to avoid consonant-final outputs (e.g., shilling > scellino, beefsteak > bistecca, wagon > vagone), 6 recent loanwords tend to preserve their original English form, or ‘foreignness’, as a symbol of prestige (Rando, 1970, p. 139; Winter-Froemel, 2011, p. 317). Romaine (2004, p. 51) observes that for languages with ‘great prestige, words borrowed from it may be pronounced in a phonetic form as close as possible to the original [. . .] as a mark of status and education’. Crucially, however, unassimilated English loanwords are largely consonant-final, lacking the canonical vowel endings (-a, -o) that signal gender in Italian. The absence of such inherent phonological cues may therefore elevate the importance of other GA factors, such as semantic analogy.
Semantic Analogy
On the semantic side, much attention has been devoted to borrowed nouns that are assigned gender on account of equivalences in meaning or concept association. To refer to this process, previous literature has employed a variety of terms – ‘analogical gender’ (Poplack et al., 1982); ‘synonymic gender’ (Smead, 2000), or ‘gender copy’ (Onysko et al., 2013), whereby a loanword’s gender is inherited from that of a pre-existing L1 near-synonym or closest translational equivalent. It is widely assumed, for example, that a (presentation) slide in Italian is feminine, since its Italian associate is diapositiva (f.). When examining loanwords with well-established genders, however, it is often difficult to determine whether such cases of semantic gender alignment reflect genuine GA motivations or post hoc observations, particularly when chance correspondence remains plausible. New loans and neologisms, on the other hand, such as COVID-19 or Brexit, offer a unique opportunity to observe emergent GA strategies in real time, often accompanied by explicit metalinguistic commentary drawing directly on cross-linguistic semantic mapping. In the early stages of the COVID-19 pandemic, there was frequent oscillation between masculine (il Covid) and feminine (la Covid) in the press, even within the same article. 7 The clear skew of news outlets towards the masculine gender, however, directly contradicts the recommendation of the Accademia della Crusca, a leading authority on the Italian language, which advocated for the feminine gender based on semantic equivalence with the Italian feminine noun malattia (disease), since the original acronym in the donor language stands for Coronavirus Disease (Giovine, 2020). 8 The same language institution used similar semantic reasoning to support its recommendation for a feminine GA of the portmanteaus Grexit and Brexit, based on their derivation from -exit, for which the corresponding Italian word is uscita (f.) (cf. Thornton, 2016). Ultimately, however, the emergence of semantic considerations at the forefront of metalinguistic debate in cases where the feminine is urged as the ‘correct’ gender for new lexical integrations is consistent with theories of markedness (Battistella, 1990), whereby the assignment of marked gender requires more explicit motivation than less-marked ‘default’ alternatives.
Default Gender
Finally, the literature makes regular mention of the tendency to assign a default gender in the absence of other prominent factors. In previous GA studies on English borrowings into Italian, a general bias has been shown for a masculine default (Correa-Zoli, 1973; Klajn, 1972; Rabeno & Repetti, 1997; Thornton, 2003a), following trends observed for other Romance languages (e.g., Prado, 1982; Sánchez, 1995; Smead, 2000). Where these studies have sought to establish hierarchical relationships between GA criteria, the default criterion is typically positioned as a fallback mechanism, taking effect where semantic or formal mechanisms fail to generate one gender or another. Rice’s (2006, p. 1395) Optimal Gender Assignment Theory proposes a universality whereby ‘all constraints referring to gender-relevant features are equally ranked’, but are subject to a ‘markedness hierarchy’, such that where there is a conflict between two criteria, the least marked gender will prevail. In the case of American Italian, Rabeno and Repetti (1997) posit that most English loanwords default to masculine gender, since it is the unmarked gender. Outside of theoretical linguistics, the perception of masculine as the default gender for English > Italian borrowings is perhaps reinforced by grammars recommending the masculine gender for consonant-final foreign items, such as bar, rock and tram (Serianni et al., 1997, p. 79). Yet, as we shall see, the GA patterns observed from our corpus of unassimilated loanwords in current usage contradict this simplified view.
Sociolinguistic Factors and Impetus for Borrowing
While much of the debate over the relative importance of gender assignment criteria has centred around rules that may apply universally (Corbett & Fraser, 2000; Nesset, 2006; Rice, 2006) or language-specifically (e.g., Poplack et al., 1982), I argue that even within a given language, the salience of different gender assignment criteria is conditioned by the sociolinguistic contact scenario that gave rise to the borrowing itself. Indeed, studies exploring the dynamics influencing GA have devoted relatively little attention to the effect of context-specific differences, speaker characteristics and borrowing motivation. With regard to the latter, one important parameter is the traditional distinction between need- and prestige-motivated borrowings.
It has long been accepted that ‘necessary’ loans fill lexical gaps by naming novel referents for which a domestic term is not available, while ‘prestige’ or ‘luxury’ loans are adopted for their symbolic value, even when a competing synonym is available in the host language.
9
If English loanwords borrowed out of necessity by first- and second-generation Italian immigrants to North America served to address genuine lexical needs as speakers encountered concepts or objects for which they had no Italian equivalents (Carnevale, 2009), then it is unlikely that their gender assignment would be semantically motivated. As Haugen (1969, p. 449) argues for Norwegian spoken in North America,
[English borrowings] were used precisely because the native word escaped the speaker or because they had never heard a native word for the idea in question. There is no reason to suppose that his subconscious should have whispered the gender of the native ‘equivalent’ to him when it failed to deliver the equivalent itself.
Indeed, given the high illiteracy rates in southern Italy at the turn of the twentieth century (around 70%) and the fact that many first-generation immigrants had limited formal education and primarily oral language acquisition, it is not surprising that phonological criteria or a default gender assignment might rank more highly than semantic cues (cf. Zamora, 1975). 10
In contrast, ‘luxury loans’ (Pulcini, 2023) used in modern-day Italy proper often enter the lexicon in contexts where Italian equivalents already exist. Lexical choices such as call and calendar (instead of chiamata and agenda) can therefore convey a sociopragmatic rather than purely referential meaning (cf. Pinnavaia, 2005; Winter-Froemel & Onysko, 2012). Such stylistic and pragmatic choices between competing expressions, following Myers-Scotton’s (1993, p. 6) markedness model, represent a ‘skilled performance’ with communicative intent, allowing speakers to leverage socio-psychological values. 11 With the perceived prestige and utility of English in the contemporary Italian context for advancing one’s career potential, contemporary borrowings can serve as an indexical linguistic choice for the social signalling of modernity, sophistication, professionalism and success (Aiello, 2017; D’Achille, 2003; Gazzardi & Vásquez, 2020; Pulcini, 2024; Vettorel & Franceschi, 2019). Importantly, this shift in motivation has occurred alongside a marked increase in English proficiency among the current cohort of Italian professionals, owing to educational reforms, international mobility opportunities, globalised media, and demands of the job market (Berruto, 2012; Carlucci, 2017a; Pulcini, 2024). Such socio-cultural change, therefore, invites new perspectives on the integration of borrowings in contemporary Italian.
Terminology
Before presenting our hypotheses, it is important to clarify the use of key terminology, given the definitional divergence in existing scholarship. A common distinction treats borrowing as referring to foreign insertions adapted to the morphological, syntactic, and phonological patterns of the host language, while loanwords denote ‘well-established’, high-frequency lexical items with wide acceptance, often displacing a native synonym (Poplack et al., 1988, p. 52; Poplack & Meechan, 1995, p. 200). Pulcini (1999, p. 361), on the other hand, uses the term loanword more broadly to indicate English words borrowed ‘either without any formal change, or with adaptation to the orthographic and morphological rules of Italian’.
For the purposes of this study, I adopt Pulcini’s inclusive definition, using loanword as the primary term to encompass both adapted and unassimilated English-origin nouns in Italian. While our data focus specifically on unassimilated forms (items retaining their English phonological shape), I acknowledge that distinguishing established loans from idiosyncratic or ‘nonce borrowings’ (spontaneous single-item switches by speakers with L2 knowledge, cf. Poplack, 2004, 2018) remains challenging. Recent workplace terms such as call, task, calendar, desk and vacancy may well fall into this transitional category of loans that have not yet achieved widespread acceptance, particularly as their usage still faces notable pushback, including among Italian professionals themselves. 12 However, as others have argued (Field & Comrie, 2002; Myers-Scotton, 2002; Poplack & Meechan, 1998, among others), such items are operationally indistinguishable from established loanwords for the aim of the present study. I therefore use loanword and borrowing interchangeably throughout, avoiding strict categorical distinctions.
The Present Study: Hypotheses
Operating from a functionalist perspective, whereby linguistic structures and changes are shaped by their communicative and social functions within a given community, I propose that the relative influence of gender assignment criteria may be modulated by the sociolinguistic context and specific borrowing impetus. The present study examines one such scenario: contemporary educated Italian speakers with higher L2 proficiency than previous generations, borrowing for sociopragmatic rather than necessity-driven purposes. This new context presents a valuable opportunity to revisit gender assignment mechanisms, potentially favouring semantic analogy effects, as speakers possess cross-linguistic knowledge that might activate semantic GA cues. Indeed, research in psycholinguistics and bilingualism consistently demonstrates that speakers with inter-lingual proficiency are more attuned to the lexical relationships between their linguistic systems, supported by enhanced metalinguistic awareness (Bialystok, 2021; Cenoz, 2009; de Bot & Bülow, 2020; Grosjean, 2007). Thornton (2003a, p. 81) hypothesises that cross-linguistic semantic connections can actively influence grammatical gender processing, positing, for example, that the reason all borrowed nouns with the Germanic suffix -ship (partnership, leadership, readership, etc.) are feminine is because these terms are predominantly used by educated speakers who are able to ‘grasp the qualitative value of the English suffix’ and, in turn, the feminine gender is assigned since qualitative nouns in Italian are feminine, for example, altezza, intelligenza, bellezza (height, intelligence, beauty).
The present research specifically investigates the lexical domain of Italian aziendese (office discourse), which features a large number of borrowings adopted for their sociopragmatic value (cf. Winter-Froemel & Onysko, 2012). This scope was selected on the assumption that speakers’ semantic and associational knowledge is more actively leveraged in loans that represent a more conscious stylistic choice for the signalling of innovation and professionalism. I propose two main hypotheses:
To test these hypotheses, the present research employs a quantitative, corpus-based methodology, allowing us to assess the extent to which loanword genders are, in fact, “whispered” to speakers by their L1 lexicon (to borrow Haugen’s metaphor) in our corpus of contemporary spoken Italian from the domain of professional discourse.
Methodology
Corpus
The KIParla corpus is a recently developed and publicly accessible resource for the study of contemporary spoken Italian (Mauri et al., 2019). The corpus consists of over 150 hr of audio-recorded and transcribed discourse, collected between 2017 and 2021. It comprises 560 speakers, 61% female and 38% male, the majority of whom are aged between 16 and 40 (75.2%). With regard to level of education, 82.9% of speakers are university-educated (16.6% PhDs). While the corpus does not include individual L2 proficiency assessments, our speaker sample aligns closely with populations documented to have enhanced English proficiency compared to previous generations (Carlucci, 2017a, 2017b; Pulcini, 2024). In turn, this profile supports our research objectives, representing the demographic predicted to leverage cross-linguistic semantic connections when assigning loanword gender. Given that, as we have discussed, Italian grammars and language academies may recommend gender assignments for new loanwords that do not always align with actual usage, the use of naturalistic spoken data from the KIParla corpus of oral Italian in use provides an optimal testing ground for investigating the spontaneous GA patterns of contemporary Italian speech, rather than prescriptive norms.
Data Extraction
To identify and extract English loan nouns related to the world of work, a multi-stage approach was employed. The initial search terms were systematically compiled from multiple sources to ensure comprehensive coverage of English-origin nouns associated with the professional environment (e.g., stress, project, deadline, brainstorming, training, skill, etc.). This process involved consultation with established anglicism dictionaries, glossaries, and structured interviews with Italian office professionals to identify commonly used anglicisms in daily workplace discourse. These consultations yielded approximately 240 potential search terms, providing a robust foundation for corpus investigation. This approach served to ensure that term selection was independent of the grammatical patterns under investigation, minimising researcher bias.
Keyword searches were then conducted within the corpus transcripts using the NoSketch Engine interface, allowing for the retrieval of concordances displaying the contexts in which these words appear. Of the approximately 240 terms searched, 112 independent English-origin loanwords met the inclusion criteria (below) and yielded results in the KIParla corpus (totalling 779 tokens). Finally, instances of each loanword were examined to determine its grammatical gender assignment (masculine or feminine) as indicated by articles, adjectives, and other agreeing elements in the surrounding linguistic context. 13
Three types of English-derived nouns were excluded from the data collected and, therefore, not considered in the analysis:
Loan translations, such as estensione (for internet browser extension), since these constitute calques whereby an Italian word takes on the meaning of the English term, displacing its original meaning. In such cases, the gender of the calqued borrowing consistently adheres to the gender of the Italian word, providing little insight for the purpose of our study.
Borrowed nouns which refer to animates, such as babysitter, manager, and so on, as grammatical gender in such cases is consistently dictated by the gender of the human referent above all other parameters in Romance and beyond (cf. DuBord, 2004; Ervin, 1962; Poplack et al., 1982; Smead, 2000).
Loans that are phonologically or morphologically adapted with Italian derivatives, such as sponsorizzazione (sponsorship), as these follow highly predictable patterns in alignment with Italian morphological gender markers (in this case: -zione = f.). 14
By focusing specifically on unassimilated borrowings, this study addresses cases where formal assignment rules are largely inapplicable, 15 allowing semantic factors to emerge more clearly. This category, therefore, enables us to directly challenge assumptions whereby lexical insertions that are not morpho-phonologically adapted in Italian mostly take the masculine form (Di Pisa et al., 2022).
Data Analysis
For each of the 112 independent loanwords extracted from the corpus, the following information was coded: (1) the loanword’s assigned gender; (2) the presence or absence of a close Italian semantic associate, defined as the Italian noun sharing the core denotational meaning and highest degree of functional interchangeability in the observed context; (3) where present, the gender of that associate. 16 To ensure rater reliability in semantic equivalence judgements, Italian associates were identified based on monolingual and bilingual dictionary entries and then independently validated by five native Italian speakers. These judgements confirmed the selected associates’ shared core semantic meaning and functional equivalence in workplace contexts, without significant differences in connotation. 17 Next, semantic matches were operationalised as cases where the loanword’s assigned gender matched that of its Italian associate (if present), coded as Y, while mismatches were coded as N. Table 1 outlines some examples of this coding process.
Example of Semantic Gender Coding.
Descriptive statistics were calculated to determine the overall distribution of masculine and feminine gender assignments among loanwords, as well as the rate of semantic gender correspondence for the subset of loans with identified Italian equivalents. Statistical analysis involved both Fisher’s exact test and a chi-square test of independence to examine the significance of observed match rates. Effect size was calculated using Cramer’s V to quantify the strength of relationships. A significance level of α = .05 was adopted for all tests.
Results
Overview of Corpus Data
Of the 779 tokens of independent English loanwords (N = 112) extracted from the KIParla corpus that met the inclusion criteria for this study, 79.5% were assigned masculine, while 20.5% were assigned feminine gender.
Gender assignments across tokens were found to be remarkably stable. That is to say, where the corpus contained 19 instances of the loanword brand, for example, all 19 tokens were assigned to the masculine gender. 18 Given the potential for inter-speaker vacillation in gender assignment due to social factors (Arndt, 1970), a greater degree of variability may have been expected. The resulting uniformity does, however, cohere with Poplack et al.’s (1982, p. 25) claim that ‘once a borrowed noun is assigned a gender by whatever criteria, there is usually unanimous agreement among speakers’.
A clear semantic associate in Italian could be identified for 75% of the corpus loanwords, allowing for direct testing of the semantic analogy hypothesis (H1). The remaining 28 borrowings (25.0%) represented concepts or terms without direct Italian translational equivalents, reflecting lexical gaps that these loans serve to fill.
As reported by Table 2, feminine-assigned loanwords show remarkably high coverage for the availability of semantic equivalents (95.7% vs. 69.7% for masculine loanwords), suggesting that feminine assignment may be more consistently tied to the availability of clear Italian synonyms.
Gender Distribution, Semantic Equivalent Availability, and Semantic Correspondence Rates.
Semantic Correspondence Patterns
Among borrowed nouns with identifiable semantic equivalents (n = 84), 82.1% exhibit gender assignment that correlates with their respective Italian associates, while only 17.9% show gender misalignment. The data reveal a striking asymmetry in semantic correspondence patterns between masculine and feminine-assigned loanwords:
Masculine loanwords. Of the 62 masculine loanwords with existing semantic equivalents, 75.8% align with the gender of their Italian associates (24.2% mismatches where masculine was assigned despite the presence of an existing feminine associate).
Feminine loanwords. All 22 feminine loanwords having semantic equivalents (100.0%) align with the gender of their Italian associates, with zero mismatches observed.
Inferential Statistics
Fisher’s exact test confirmed a statistically significant tendency for loanwords to match the gender of their Italian associates (p = .009), suggesting that cross-linguistic semantic cues significantly influence gender assignment. A chi-square test of independence (with continuity correction) also demonstrated statistical significance (χ²(1) = 4.94, p = .026), corroborating this finding. 19 The effect size analysis yielded a Cramer’s V of 0.28, indicating a moderate effect, which again suggests that semantic analogy represents a substantial factor in the gender assignments of our corpus loanwords, rather than a marginal influence.
Discussion
At the surface level, the uneven gender distribution of independent loanwords observed in this study (79.5% masculine) reflects the well-documented tendency for English borrowings into Italian and other Romance languages to favour masculine gender assignments. Loans without readily identifiable semantic equivalents were overwhelmingly assigned to the masculine gender (27/28), strongly supporting theories that default assignment operates when translational equivalents are weak or unavailable (Thornton, 2003a). Crucially, however, the high availability of existing semantic counterparts (75% of loans) in our speaker sample created optimal conditions for testing semantic influence mechanisms. As predicted in H1, loanwords with identifiable semantic equivalents showed statistically significant gender alignment with their Italian counterparts (p < .05, Cramer’s V = 0.28; 82.1% correspondence rate), indicating that semantic analogy exerts substantial influence on GA patterns and that factors beyond simple default are operative in our corpus of unassimilated borrowings.
Feminine as Marked Assignment
The perfect semantic alignment observed for feminine-assigned loanwords with native synonyms (100% correspondence rate) suggests the activation of cross-linguistic lexical connections as speakers access existing Italian terms to guide grammatical integration. In particular, it directly supports H2, providing empirical evidence for theories that propose marked elements as requiring more specific semantic justification than unmarked forms, which serve as a ‘conceptual default’ (Battistella, 1990, p. 4). Our results align with Morin’s (2006) findings for English > Spanish loanwords, which indicate semantic analogy as the key predictor of feminine GA, 20 and provide empirical evidence (albeit with small numbers) to challenge Thornton’s (2003a, p. 75) claim that ‘even when a possible Italian associate or translate is feminine, the loanword is predominantly used in the masculine as a result of default gender assignment’ (22 loanwords with feminine associates assigned feminine gender, versus 15 assigned masculine despite feminine associates).
Masculine With Semantic Influence and as Default
The degree of semantic alignment of masculine loanword assignments presents a more complex picture. The 75.8% correspondence rate for masculine loans suggests a primary influence of semantic factors. However, the presence of 24.2% gender mismatches among loans with semantic equivalents supports claims that a (masculine) default gender may operate as a secondary mechanism when semantic cues are weak, ambiguous, or unavailable. Still, the significantly above-chance match rate overall indicates that, even for masculine-assigned loans, semantic analogy is a relevant factor.
Semantic Accessibility
The finding that 95.7% of feminine loanwords have identifiable semantic equivalents (compared to only 69.7% of masculine loans) reveals fundamental asymmetries in gender assignment mechanisms. These patterns suggest that feminine assignment operates under stricter semantic conditions, requiring clear translational relationships. I therefore propose a semantic accessibility constraint: the feminine gender is assigned when speakers can readily access an identifiable feminine Italian equivalent, while masculine assignment occurs under both high and low semantic accessibility conditions. This asymmetry reinforces the marked status of feminine loanword integration, where feminine assignment demands explicit semantic motivation through donor/host-language equivalences.
Conclusions, Limitations and Future Research
Our findings contribute to the understanding of language processing and cross-linguistic influence in lexical integration. In light of the contemporary sociolinguistic landscape, with enhanced L2 proficiency among the current generation of Italian professionals and shifting borrowing motivations, this research offers new insight into the effect of semantic influence in loanword gender assignment. The significant alignment between loanword gender and that of semantic equivalents (p < .05) suggests that when integrating unassimilated loanwords, GA processes operate through active cross-linguistic lexical mapping, with marked gender assignment requiring stronger semantic justification than unmarked alternatives. Central to this understanding is the concept of a semantic accessibility constraint, which posits that feminine gender is primarily assigned when there is clear semantic motivation through equivalence with pre-existing feminine terms. Although our corpus data support this constraint, the correlational nature of these findings means that causal relationships cannot be definitively established. Future research should aim to test the cognitive reality of semantic connections through experimental psycholinguistic methods. In addition, while the perfect semantic correspondence observed for feminine-assigned loanwords is significant, this finding is based on a modest sample of feminine items (n = 22), limiting its statistical power. Even a small number of counterexamples would, therefore, reduce the observed rate, and thus these results should be interpreted as suggestive, pending replication with larger feminine samples.
Several further limitations should be acknowledged. The moderate effect size (Cramer’s V = 0.28) indicates that while cross-linguistic semantic cues are influential, they operate within a complex multi-factorial system. Observed exceptions to the semantic analogy criterion merit further investigation to understand the relative weight of multiple gender-influencing factors, including those not considered in our discussion. Moreover, the limited lexical scope and skewed speaker demographics of the KIParla corpus (82.9% university-educated) constrain the generalisability of findings. The observed patterns may, in turn, reflect the linguistic behaviour of a specific demographic profile and the pragmatic functions that anglicisms serve in workplace discourse, rather than contemporary spoken Italian more broadly. Expanded corpus studies across different borrowing domains, sociolinguistic contact scenarios and speaker populations would therefore provide stronger evidence for the observed patterns and further advance the functionalist approach to research on gender assignment and loanword integration. With socio-cultural change continuing to drive an influx of new borrowings, the Italian language will no doubt continue to serve as an excellent site for investigation into the ways lexicons adapt in response to globalised communication.
Footnotes
Acknowledgements
I would like to thank Cinzia Russi for encouraging me to pursue this line of inquiry. I am also grateful to the anonymous reviewers for their valuable comments on earlier versions of this article.
Abbreviations
A>B borrowed from A to B, or A has transformed into B.
f. feminine
GA gender assignment
L1 first language
L2 second language
m. masculine
Funding
The author received no financial support for the research, authorship, and/or publication of this article.
Declaration of Conflicting Interests
The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.