Romanian has a peculiar rule which prohibits the overt realization of definiteness marking in DPs consisting only of D0 and N0 in the complement position of most prepositions, called “article drop”. We investigate this phenomenon from a diachronic and comparative point of view. Similar phenomena are found in Albanian, Macedonian and some Rhaeto-Romance varieties, but in these varieties the rule is more limited in scope (it is restricted to locative PPs, and also to certain nouns in Rhaeto-Romance). Article drop is a common Romanian phenomenon, being found in the South-Danubian Romanian dialects as well as since the oldest attestations of Romanian. In the other Balkan languages, it may be a structural borrowing from Aromanian (Prendergast 2017). In Romanian, the prepositions exempt from article drop are those that never had locative uses or had predominant non-locative uses, which indicates an origin in locative PPs. Moreover, person-referring nouns are sometimes exempt from article drop, to a larger extent in Old Romanian and Aromanian than in Modern Romanian (where only name-like definites are exempt from the rule). We propose that article drop emerged from the reanalysis of a semantically-conditioned phenomenon into a syntactically-conditioned rule. We identify the origin of article drop in an oscillation between definite marking and zero marking in ‘weak definites’ (in the sense of Carlson & Sussman 2005; cf. Engl. at school, in jail). Weak definites often occur in locative PPs, tend to disallow modification and are usually inanimate. These properties correspond to the conditions of application of the article drop rule, as evinced by our diachronic and diatopic survey.

This paper proposes a cognitive view on sentence stress patterns to discuss focus elements in both broad and narrow focus contexts. The cognitive perspective is based on the hypothesis that prosodic phrases correspond at the cortical level to cognitive binary relations between speech objects of utterances. On this view, cognitive relations are produced by a generic information packaging (IPk) mechanism that pairs constituents with different cognitive functions. At the utterance level, cognitive relations are implemented by prosodic phrases (relations) where different pitch features mark their two functional constituents. Our proposal is to assign sentence stress patterns with corresponding cognitive structural patterns of utterances. One of the two constituents of cognitive and prosodic relations is nuclear and projects its cognitive function to the whole cognitive unit which it belongs to. The paper proposes a nuclear accent analysis by connecting the cognitive functions of constituents with their phonetic/phonological features. The contours analyzed in the paper as hierarchies of cognitive/prosodic relations are selected from those used by Ladd (2008) to exemplify sentence stress patterns in broad focus statements with ascending and descending contours, and in contrastive focus statements. We conclude that, in the new perspective, different cognitive structural patterns can be assigned to contrastive/broad focus statements in different semantic contexts.

A longstanding debate in L2 research focuses on how syntactic complexity needs to be operationalized to account for L2 performances. Whereas many studies investigated this issue in L2 production, very few studies focused on L2 sentence processing. The present study aimed at investigating the effect of syntactic complexity on gender agreement processing in L2 French, while controlling for the learners’ working memory capacity. We tested 37 Dutch learners of French by means of a self-paced reading technique. The results showed decreased sensitivity to gender agreement in embedded structures, but increased sensitivity to gender agreement in non-embedded structures. We concluded that the number of clauses in gender agreement constructions accounts for the effect of syntactic complexity on gender agreement processing in L2 French and that this measure is negatively correlated to sensitivity to gender agreement. We furthermore concluded that (non-verbal) working memory does not affect L2 gender agreement processing.

Differential object marking (DOM) has been shown, in an impressive number of production studies, to be acquired by monolingual children at around age 3. The picture which emerges from comprehension data, however, reveals that DOM is an area of vulnerability in L1 acquisition. This study investigates the acquisition of DOM by monolingual Romanian children using a preference judgment task. 80 monolingual Romanian children (aged 4;04-11;04) and a control group of 10 Romanian adults took part in the study. Results show that DOM is vulnerable and trace this vulnerability to the animacy feature. Romanian children incorrectly overgeneralize DOM to inanimate proper names and inanimate descriptive DPs until age 9. The vulnerability of animacy is predicted by its variable behaviour with respect to object marking as well as by the current increase in the use of clitic doubling, a DOM marker less sensitive to animacy. On the learnability side, we account for the findings in terms of Biberauer & Roberts’ (2015, 2017) Maximize Minimal Means model. We suggest that, in accordance with the Feature Economy bias, Romanian children first identify only the role of referential stability (which has more robust cues in the input) and consider the possibility of animacy as a relevant feature later. In line with the Input Generalization bias, children maximize the role of referential stability which results in overgeneralization of DOM to inanimate objects, especially to inanimate proper names.

This paper considers the relation between prescriptivism and descriptivism in practical lexicography. Using data extracted from two major dictionary projects conducted in different cultural contexts, we argue that these notions are easier to tease apart in theory than in practice. First, we point at the mismatch between the linguistic neutrality aimed for by lexicographers and the authority status conferred on dictionaries by the general public. We then posit that the labels descriptive, prescriptive cannot apply to a dictionary as a whole since, in the process of dictionary making, some tasks are factual and objective, while others call for the lexicographer’s ruling on what is right or wrong in terms of language use. It is such rulings and the factors that determine them that are the focus of attention here, in keeping with the places in which they are likely to occur: the front matter, the macrostructure, and the microstructure of the dictionary.

Review

The current paper investigates the grammatical acceptability of null subjects in the diary register of L2 English by intermediate and advanced students, L1 speakers of Romanian. Overall results indicate performance that does not surpass chance level, and therefore, a strong conservative preference for overtly realized subjects in diary contexts. However, individual analysis has identified a small group of L2 learners who consistenly scored well above chance in the four experimental conditions (overt subject, null subject in root clause, null subject in embedded clause and null subject in yes-no questions). I discuss several factors that relate to the difficulty L2 learners have with accepting on-target null subjects in the diary register. Quite likely, they never had enough exposure to the relevant input. L2 acquisition slows down when multiple grammars (core and non-core) are identified. Also, subject omission is optional, so the study tested a preference, not a mandatory phenomenon.

The goal of this paper is to account for the observation that some Serbian object experiencer anticausatives take instrumental NPs as expressions of the causer participant whereas others take od(‘from’)- PPs. Following a number of authors (Alexiadou et al. 2013, Doron 2014, Anagnostopoulou & Samioti 2014, a. o.), I assume that differences in the licensing of expressions introducing event participants point in the direction of structural differences in terms of presence/absence of certain layers of verbal structure. The observed difference is accounted for by assuming that instrumental NPs are syntactically licensed by VoiceP while od(‘from’)-PPs are rejected by VoiceP owing to a semantic clash. Consequently, full VoiceP structure is present with psych verb anticausatives that license instrumental NPs and absent with psych verb anticauatives that license od(‘from’)-PPs. The analysis presented in the paper has implications for the syntactic and semantic status of SE as well. It is suggested that Chierchia’s (2004) reflexive approach to anticausatives can be extended to psych verb anticausatives which license instrumental NPs whereas the standard approach (Schäfer & Vivanco 2016) should be retained for typical anticausatives with inanimate internal arguments and object experiencers that license od (‘from’)-PPs. Such a “middle-ground” solution follows from the syntactic structures I propose for these two different sets of psych verb anticausatives.

After twenty yearsTwenty years ago, in the fall of 1999, Bucharest Working Papers in Linguistics (BWPL) published its first issue.Many things have changed since then.We began as a combined linguistics and literature journal -University of Bucharest Review, with separate literature and linguistics issues, which was publishing a selection of the papers presented at the annual conference of the English Department of the University of Bucharest.In 2007, the linguistics issues changed their name into Bucharest Working Papers in Linguistics, which became a double-blind peer reviewed journal and in 2011 it obtained its own ISSN.Some things, however, remained constant: the quality of the papers, the enthusiasm of the contributors, the challenging and constructive evaluations of our reviewers, as well as our desire to stay true to our initial goalto create a front-line journal of modern linguistics in the country, featuring both well-established and emerging linguists.When it comes to marking anniversaries, publications usually look back.This is what we did as well.We invited the authors who published in the early issues of BWPL, some of them in the very first issue back in 1999, to contribute to our 21 st volume.This is our way of looking forward as well, since Bucharest Working Papers in Linguistics is coming of age.It has grown more mature, more professional, more international, it has gone open access, it has been indexed in EBSCOhost TM , ERIH PLUS, DOAJ, ROAD and Ulrichsweb.The editorial team thanks all the authors involved in this year's anniversary issues.They all responded to our invitation with enthusiasm in spite of the very tight deadlines.We also express our heart-felt gratitude to all of our authors, reviewers, members of the scientific committee and readers who have supported us since 1999.Without their help BWPL would have never become the respectable journal it is now and we would have nothing to celebrate.