Indirectness is a common feature of daily communication, yet is underexplored in NLP research for both low-resource as well as high-resource languages. Indirect Question Answering (IQA) aims at classifying the polarity of indirect answers. In this paper, we present two multilingual corpora for IQA of varying quality that both cover English, Standard German and Bavarian, a German dialect without standard orthography: InQA+, a small high-quality evaluation dataset with hand-annotated labels, and GenIQA, a larger training dataset, that contains artificial data generated by GPT-4o-mini. We find that IQA is a pragmatically hard task that comes with various challenges, based on several experiment variations with multilingual transformer models (mBERT, XLM-R and mDeBERTa). We suggest and employ recommendations to tackle these challenges. Our results reveal low performance, even for English, and severe overfitting. We analyse various factors that influence these results, including label ambiguity, label set and dataset size. We find that the IQA performance is poor in high- (English, German) and low-resource languages (Bavarian) and that it is beneficial to have a large amount of training data. Further, GPT-4o-mini does not possess enough pragmatic understanding to generate high-quality IQA data in any of our tested languages.
Large language models routinely hallucinate APIs and mislocalize edits, while language servers compute verified, IDE-grade facts about real code. We present Lanser-CLI, a CLI-first orchestration layer that pins and mediates a Language Server Protocol (LSP) server for coding agents and CI, exposing deterministic, replayable workflows. Our position is that language servers provide not only structural information (definitions, references, types, diagnostics) but also an actionable process reward: machine-checked, step-wise signals that align an agent's planning loop with program reality. In this work, Lanser-CLI contributes: (i) a robust addressing scheme beyond brittle "file:line:col" via a Selector DSL (symbolic, AST-path, and content-anchored selectors) with a principled relocation algorithm; (ii) deterministic Analysis Bundles that normalize Language Server responses and capture environment/capability metadata with stable content hashes; (iii) a safety envelope for mutating operations (rename, code actions) with preview, workspace jails, and Git-aware, transactional apply; and (iv) a process-reward functional derived from Language Server facts (diagnostic deltas, disambiguation confidence, and safe-apply checks) that is computable online and replayable offline. We formalize determinism under frozen snapshots and establish a monotonicity property for the process reward, making it suitable for process supervision and counterfactual analysis. Project Page: https://github.com/yifanzhang-pro/lanser-cli
Data profiling is critical in machine learning for generating descriptive statistics, supporting both deeper understanding and downstream tasks like data valuation and curation. This work addresses profiling specifically in the context of code datasets for Large Language Models (code-LLMs), where data quality directly influences tasks such as code generation and summarization. Characterizing code datasets in terms of programming language concepts enables better insights and targeted data curation. Our proposed methodology decomposes code data profiling into two phases: (1) an offline phase where LLMs are leveraged to derive and learn rules for extracting syntactic and semantic concepts across various programming languages, including previously unseen or low-resource languages, and (2) an online deterministic phase applying these derived rules for efficient real-time analysis. This hybrid approach is customizable, extensible to new syntactic and semantic constructs, and scalable to multiple languages. Experimentally, our LLM-aided method achieves a mean accuracy of 90.33% for syntactic extraction rules and semantic classification accuracies averaging 80% and 77% across languages and semantic concepts, respectively.
Node-based programming languages are increasingly popular in media arts coding domains. These languages are designed to be accessible to users with limited coding experience, allowing them to achieve creative output without an extensive programming background. Using LLM-based code generation to further lower the barrier to creative output is an exciting opportunity. However, the best strategy for code generation for visual node-based programming languages is still an open question. In particular, such languages have multiple levels of representation in text, each of which may be used for code generation. In this work, we explore the performance of LLM code generation in audio programming tasks in visual programming languages at multiple levels of representation. We explore code generation through metaprogramming code representations for these languages (i.e., coding the language using a different high-level text-based programming language), as well as through direct node generation with JSON. We evaluate code generated in this way for two visual languages for audio programming on a benchmark set of coding problems. We measure both correctness and complexity of the generated code. We find that metaprogramming results in more semantically correct generated code, given that the code is well-formed (i.e., is syntactically correct and runs). We also find that prompting for richer metaprogramming using randomness and loops led to more complex code.
Patterns are words with terminals and variables. The language of a pattern is the set of words obtained by uniformly substituting all variables with words that contain only terminals. Length constraints restrict valid substitutions of variables by associating the variables of a pattern with a system (or disjunction of systems) of linear diophantine inequalities. Pattern languages with length constraints contain only words in which all variables are substituted to words with lengths that fulfill such a given set of length constraints. We consider membership, inclusion, and equivalence problems for erasing and non-erasing pattern languages with length constraints. Our main result shows that the erasing equivalence problem - one of the most prominent open problems in the realm of patterns - becomes undecidable if length constraints are allowed in addition to variable equality. Additionally, it is shown that the terminal-free inclusion problem, a prominent problem which has been shown to be undecidable in the binary case for patterns without any constraints, is also generally undecidable for all larger alphabets in this setting. Finally, we also show that considering regular constraints, i.e., associating variables also with regular languages as additional restrictions together with length constraints for valid substitutions, results in undecidability of the non-erasing equivalence problem. This sets a first upper bound on constraints to obtain undecidability in this case, as this problem is trivially decidable in the case of no constraints and as it has unknown decidability if only regular- or only length-constraints are considered.
Qubit allocation is a process to assign physical qubits to logical qubits in a quantum program. Since some quantum computers have connectivity constraints on applications of two-qubit operations, it is mainly concerned with finding an assignment and inserting instructions to satisfy the connectivity constraints. Many methods have been proposed for the qubit allocation problem for low-level quantum programs. This paper presents a type-based framework of qubit allocation for a quantum programming language with first-order functions. In our framework, the connectivity constraints are expressed by a simple graph of qubits called a coupling graph. We formalize (1) the source language, whose type system verifies that the number of qubits required for a given program to run does not exceed the number of nodes of the coupling graph, (2) the target language, whose qualified type system verifies that a well-typed program satisfies the connectivity constraints, and (3) an algorithm to translate a source program into a target program. We prove that both languages are type-safe and that the translation algorithm is type preserving.
Im toten Winkel der Kulturkritiker*innen findet die Anschlusskommunikation Jugendlicher heute zum großen Teil im Internet statt und zeichnet sich u.a. dadurch aus, dass verschiedenste Medienprodukte gestaltet werden. Soziale Netzwerke, Video- und Literaturportale, in denen diese Produkte geteilt, kommentiert, diskutiert und überarbeitet werden, prägen mittlerweile die literarische Sozialisation. Der folgende Beitrag setzt sich mit der Frage auseinander, inwiefern ein digitaler handlungs- und produktionsorientierter Literaturunterricht die häufig festgestellte unproduktive Aufspaltung zwischen Schul- und Privatlektüre vermindert und dabei Interpretationskompetenz fördern kann. Hierfür werden nach einer kurzen Einführung (1) zunächst die digitalen Lesepraktiken Jugendlicher charakterisiert (2), um aus diesen Problemstellungen für den schulischen Literaturunterricht abzuleiten (3) und diese für einen digitalen handlungs- und produktionsorientierten Literaturunterrichts zu adaptieren. Unter Rückgriff auf Spinners und Köpperts Konzeption handlungs- und produktionsorientierten Literaturunterrichts wird ein Phasenmodell entworfen (4) und an einem Unterrichtsbeispiel veranschaulicht (5).
Abstract (english): Digital reading culture goes school! On the conception of digital action- and production-oriented literature lessons
In the blind spot of cultural critics, young people‘s follow-up communication today largely takes place on the Internet and is characterized, among other things, by the fact that a wide variety of media products are created. Social networks, video and literature portals, in which these products are shared, commented on, discussed and revised, now shape literary socialization. The following article addresses the quest-ion of the extent to which digital action- and production-oriented literature instruction is able to reduce the unproductive split between school and private reading and can promote interpretive competence in the process. For this purpose, after a short introduction (1), the digital reading practices of young people are first characterized (2) in order to derive problems for school literature lessons from these (3) and to adapt them for digital action- and production-oriented literature lessons. Using Spinner‘s and Köppert‘s concept of action- and production-oriented literature lessons, a phase model is designed (4) and illustrated with a lesson example (5).
Emnet for artikkelen er drøftande skriving, slik skrivemåten vert forstått i skriveopplæringa. Det å få fram ulike sider ved ei sak er ein mykje brukt karakteristikk av drøfting. I skriveopplæringa vert dette forstått som å legge fram argument både for og mot eit standpunkt, altså drøfting forstått som «pro og kontra»-struktur, noko ein ikkje minst ser operasjonalisert via skriverammer. Formålet med artikkelen er å få fram eit breiare bilete av den drøftande teksten.
Artikkelen bygger på ein retorisk analyse av 16 drøftande debattartiklar frå ungdomssteget der stasislæra har vore hovudinnfallsvinkel. Stasislæra er ein typologi over grunnleggande spørsmål ein kan stille til ei sak, ofte vist til som argumentasjonsfasar. Resultata syner at det i alle elevtekstane vart stilt fleire typar grunnleggande spørsmål, og at det dessutan kunne etablerast visse mønster for distribusjonen av argumentasjonsfasane. Artikkelen konkluderer med at opplæringa i drøftande skriving bør fokusere meir på kva spørsmål den drøftande teksten skal svare på.
Philology. Linguistics, Germanic languages. Scandinavian languages
I revisjonen av Bokmålsordboka og Nynorskordboka treng leksikografane reiskapar som effektivt og etterretteleg finn ordsekvensar som seier noko om typisk bruk og kva frekvente uttrykk ord inngår i. Artikkelen tek føre seg korleis ein definerer og studerer kollokasjonar i ulike tradisjonar, då fleire tilnærmingar til omgrep som kollokasjon og fleirordsuttrykk møtest i eit praktisk felt som leksikografi. Hovudvekta i artikkelen ligg på korleis ulike reiskapar kan nyttast for å finne kollokasjonar som er relevante å ha med i ordbøkene. Korpuskel-leks er eit korpusgrensesnitt utvikla for leksikografisk bruk. Der kan brukarane søkje etter kollokasjonar ved hjelp av statistiske assosiasjonsmål og ved regulære uttrykk på tvers av mange korpus. NorGramBank er ein trebank med søkjetemplat utvikla for leksikografisk bruk. Der er kollokasjonar ikkje berre eit reint statistisk fenomen av ord som førekjem saman og kan søkjast etter lineært, men søket kan òg referere til syntaktisk struktur og syntaktiske relasjonar mellom orda.
Philology. Linguistics, Germanic languages. Scandinavian languages
We introduce Qunity, a new quantum programming language designed to treat quantum computing as a natural generalization of classical computing. Qunity presents a unified syntax where familiar programming constructs can have both quantum and classical effects. For example, one can use sum types to implement the direct sum of linear operators, exception-handling syntax to implement projective measurements, and aliasing to induce entanglement. Further, Qunity takes advantage of the overlooked BQP subroutine theorem, allowing one to construct reversible subroutines from irreversible quantum algorithms through the uncomputation of "garbage" outputs. Unlike existing languages that enable quantum aspects with separate add-ons (like a classical language with quantum gates bolted on), Qunity provides a unified syntax and a novel denotational semantics that guarantees that programs are quantum mechanically valid. We present Qunity's syntax, type system, and denotational semantics, showing how it can cleanly express several quantum algorithms. We also detail how Qunity can be compiled into a low-level qubit circuit language like OpenQASM, proving the realizability of our design.
The article is dedicated to the investigation of the motivational features of the linguocultural concept TEXTILE in the English language. The investigation has been done within the framework of the cognitive scientific paradigm, namely, linguoconceptology and linguoculturology. Despite the large number of researches devoted to the study of motivational features of concepts in the English language, it should be admitted that the concept of TEXTILE still has not been properly investigated. According to the fact that the motivational features are understood as the internal form of the word, etymological analysis based on the comparative-historical method was used to identify them. The research was based on etymological dictionaries, general explanatory dictionaries of the English language and dictionaries of textile terminology. An etymological study of the English lexeme ‘textile’ revealed that its original form was a derivative from a Pre-Indo-European root meaning “to weave”. As a result of the study of the etymological dictionaries entries of the English language, 7 main motivational features of the studied concept were identified, namely: ‘canvas’, ‘cloth’, ‘fabric’, ‘texture’, ‘tissue', ‘web’ and ‘woven’. The source of the aforementioned English lexemes was both Germanic and Romance languages. Three of these lexemes come from Old French, namely ‘canvas’, ‘fabric’ and ‘tissue’; English lexemes ‘cloth’, ‘web’ and ‘weave (woven)’ come from Proto-Germanic; and the lexeme ‘texture’ is derived from Middle French. There is also a gradual increase in the volume of the concept structure due to the emergence of new components that reflect the current understanding of the concept under study. The prospect of the research is seen in further investigation of the verbalised concept of TEXTILE in the English language.
The editorial team greatly appreciates the reviewers who have dedicated a considerable amount of time and expertise to the journal’s rigorous editorial process over the past 12 months, regardless of whether the papers have been finally published or not [...]
We introduce a new notion of C-simple problems for a class C of decision problems (i.e. languages), w.r.t. a particular reduction. A problem is C-simple if it can be reduced to each problem in C. This can be viewed as a conceptual counterpart to C-hard problems to which all problems in C reduce. Our concrete example is the class of non-regular deterministic context-free languages (DCFL'), with a truth-table reduction by Mealy machines (which proves to be a preorder). The main technical result is a proof that the DCFL' language $L=\{0^n1^n; n\geq 1\}$ is DCFL'-simple, which can thus be viewed as the simplest problem in the class DCFL'. This result has already provided an application, to the computational model of neural networks 1ANN at the first level of analog neuron hierarchy. This model was proven not to recognize $L$, by using a specialized technical argument that can hardly be generalized to other languages in DCFL'. By the result that $L$ is DCFL'-simple, w.r.t. the reduction that can be implemented by 1ANN, we immediately obtain that 1ANN cannot accept any language in DCFL'. It thus seems worthwhile to explore if looking for C-simple problems in other classes C under suitable reductions could provide effective tools for expanding the lower-bound results known for single problems to the whole classes of problems.
Giorgio Papitto, Luisa Lugli, Anna M. Borghi
et al.
According to the embodied cognition perspective, linguistic negation may block the motor simulations induced by language processing. Transcranial magnetic stimulation (TMS) was applied to the left primary motor cortex (hand area) of monolingual Italian and German healthy participants during a rapid serial visual presentation of sentences from their own language. In these languages, the negative particle is located at the beginning and at the end of the sentence, respectively. The study investigated whether the interruption of the motor simulation processes, accounted for by reduced motor evoked potentials (MEPs), takes place similarly in two languages differing on the position of the negative marker. Different levels of sentence concreteness were also manipulated to investigate if negation exerts generalized effects or if it is affected by the semantic features of the sentence. Our findings indicate that negation acts as a block on motor representations, but independently from the language and words concreteness level.
Ob zum Austausch von zwischenmenschlichen Bedürfnissen, wie dem Gedankengut oder zu Geschäftszwecken, wie am Beispiel der Lingua Franca, galt Sprache seit jeher als das Kommunikationsmittel der menschlichen Spezies. Durch den wachsenden Bedarf an Mehrsprachigkeit im 20. Jahrhundert, rückte die Spracherwerbsforschung zunehmend ins Forschungszentrum. Die Erforschung von Deutsch als Zweitsprache (L2) anhand verschiedenster Muttersprachen (L1s) findet ihre Anfänge Mitte der 1980er Jahre (vgl. MEISEL 1975). Bereits zu Beginn des 19. Jahrhunderts erkannten Forscher, wie SCHLEGEL (1818), die Wichtigkeit, Sprachen nach morphotypologischen Klassifikationen zu schematisieren, um geneologische Zusammenhänge zwischen Sprachen im Kontext analysieren zu können. Die Frage, wie und ob sprachliche Transferprozesse auf eine sprachtypologische (Nicht)-Verwandtschaft zwischen L1 und L2 zurückzuführen sind, ist Gegenstand dieser Studie. Die untersuchten L1s sind Spanisch (flektierend), Vietnamesisch (isolierend), Türkisch (agglutinierend) sowie Deutsch (flektierend) als L2. Praktisch werden Daten zu zwei Grammatikalitäten analysiert: die Subjekt-VerbKongruenz sowie die Realisierung der grammatischen Artikel im Deutschen. Die Ergebnisse zeigen, dass in beiden Bereichen eine enge Verbindung zwischen strukturellen Ähnlichkeiten sowie Differenzen für den Transfer zwischen L1 und L2 ausschlaggebend sein kann, da die spanischsprachigen TeilnehmerInnen die geringsten Interferenzquoten, die vietnamesischsprachigen TeilnehmerInnen hingegen die höchsten Fehlerquoten aufzeigen.
German literature, Germanic languages. Scandinavian languages
Julian Berger, Maximilian Böther, Vanja Doskoč
et al.
In language learning in the limit, the most common type of hypothesis is to give an enumerator for a language. This so-called $W$-index allows for naming arbitrary computably enumerable languages, with the drawback that even the membership problem is undecidable. In this paper we use a different system which allows for naming arbitrary decidable languages, namely programs for characteristic functions (called $C$-indices). These indices have the drawback that it is now not decidable whether a given hypothesis is even a legal $C$-index. In this first analysis of learning with $C$-indices, we give a structured account of the learning power of various restrictions employing $C$-indices, also when compared with $W$-indices. We establish a hierarchy of learning power depending on whether $C$-indices are required (a) on all outputs; (b) only on outputs relevant for the class to be learned and (c) only in the limit as final, correct hypotheses. Furthermore, all these settings are weaker than learning with $W$-indices (even when restricted to classes of computable languages). We analyze all these questions also in relation to the mode of data presentation. Finally, we also ask about the relation of semantic versus syntactic convergence and derive the map of pairwise relations for these two kinds of convergence coupled with various forms of data presentation.
By revisiting interjections in Scandinavian Studies in Language, the explicit goal of the issue is to bring together Scandinavian and global perspectives on interjections. Our volume opens up Scandinavia-based research to the global audience, and at the same time, our analysis is characterized by deep connections with global scholarships in the plural. The seminal work on interjections by Wierzbicka (1991 [2003]); and Ameka (1992) had a crosslinguistic vision that continues to inspire the ongoing work of understanding both local interjections and global comparison.
This issue of the Scandinavian Economic History Review features a list of dissertations published in the Nordic countries in 2017 and 2018 (Appendix). As in our previous surveys (Ojala, Hemminki & Nevalainen, 2016, 2018), we approached Nordic universities and business schools requesting them to provide full lists of dissertations completed in 2017 and 2018 addressing topics in economic and business history. This time our survey yielded altogether 46 dissertations. As in our previous surveys, the topic of the dissertation (not the department nor the discipline) was used to determine if it was eligible for inclusion to our survey. After compiling the list, we approached the authors requesting a summary of their dissertations. These summaries are included as an Appendix to this text. Unfortunately, we did not receive abstracts from all authors. The Scandinavian Economic History Review webpages provide direct links to all dissertations available on the Internet. Considering overall our previous and this latest survey, we can say that economic history research is doing well in Nordic countries: no less than 122 dissertations were produced in the period 2014– 2018 (Table 1). There is some annual fluctuation in these numbers although overall the number of dissertations has ranged approximately between 20 and 30. Yet there are clear differences between the countries. Sweden is the powerhouse of Nordic economic history research: 57% of dissertations in the period 2014–2018 were produced in Sweden, which produces some 13–15 dissertations per annum. Finland follows Sweden with a 27% share, while Norway and Denmark produce roughly ten per cent each of all dissertations in economic and business history. Lund University is the most productive institution, producing some 22% of all dissertations. The second most productive Swedish university is Uppsala (16%), followed by two Finnish universities: Helsinki (11%), and Jyväskylä (10%). English is themost widely used language in economic history dissertations in the Nordic countries: roughly 60 per cent of dissertations arewritten in English.However, there are some striking differences between the countries. While almost 90% of the Danish dissertations are written in English, this share is in Sweden a slightly over 70% and inNorway it is 55%. In Finland, however, the share of dissertations written in English is 30% although this share is clearly on the increase in Finland. If we look at the time periods the dissertations address there are, again, some differences between the countries. In the Finnish case the early modern era (here defined as roughly the time before the 1850s) is clearly prevalent, with a 42% share. In other Nordic countries the bulk of the dissertations are concerned with the industrial era (here defined as lasting from the 1850s to the 1950s). There seems to be a slight trend towards longer time periods: in 2014 roughly one fourth of dissertations still dealt with time periods of 50 years or longer, while this share was 50% in 2018. Moreover, the share of dissertations analysing periods of 100 years or longer has increased from nine per cent (2014) to 24% (2018). Following our previous surveys and the categorisation introduced by Whaples (1991, 2002) we further divided dissertations into 12 partly overlapping categories (Table 2). According to this categorisation the dissertations most typically dealt with topics related to public institutions. These topics are especially popular in Norway and in Finland. Furthermore, business history is still an important topic in Nordic dissertations, although showing some annual fluctuation. Business history