Search

This paper explores speakers’ ratings of figurativeness of three Croatian ingesting verbs – jesti ‘eat’, žderati ‘devour’ and gutati ‘swallow’ – in SVO constructions. We hypothesize that a meaning will be rated as more figurative if the object is a noun which does not refer to food, and if the subject is inanimate or abstract. We also hypothesize there will be differences among the three verbs as to how their meanings are rated. One hundred native speakers of Croatian filled in a questionnaire with 60 sentences, 30 of which contained the three ingesting verbs. Sentences with ingesting verbs comprise four groups based on their common collocates in the hrWaC corpus. The participants rated a verb’s meaning in each sentence from 1 (‘completely literal’) to 5 (‘very figurative’). The cutoff threshold for figurativeness was set at 3.5. The results show that the degree of figurativeness of the three verbs’ meanings depends on the type of SVO pattern in which they occur, with those containing the least prototypical subject (abstract noun) and object (human) being rated as the most figurative. Furthermore, there are significant differences in the ratings of the three verbs’ literal meanings. The findings suggest that gradability of figurative meanings has psychological reality.

Croatian is a language with a flexible word order, case-marking, gender-marking and animacy-marking. However, the canonical order is considered to be SVO (Greenberg 1966). While the incremental argument interpretation in transitive structures in Croatian predominately relies on case marking (MacWhinney and Bates 1989), case syncretism can help differentiate the role of morpho- syntactic, semantic and information-structural principles in word ordering. We use temporarily ambiguous input to test the effects of (i) morphological marking vs. sentential position, (ii) animacy, and (iii) givenness in processing SVO and OVS orders. We have conducted 3 self-paced reading experiments followed by acceptability judgement tasks (scale 1-5). Self-paced reading task explores to what extent speakers use sentential position, animacy and givenness to incrementally assign arguments. Acceptability judgment task is used to test the overall preference for (i) S>O, (ii) animate > inanimate and (iii) given information > new information.

Given the recent discussions on replicability, the author will compare experimentation in two closely related languages, Croatian in Slovene, aiming at answering the question of relative comparability of conceptual domains.

Suvremena semantička literatura ne slaže se u vezi s izjednačavanjem kategorijalnoga razlikovanja konkretnosti i apstraktnosti s jedne i visoke i niske predočivosti (engl. imageability) s druge strane. U tom se kontekstu predočivošću leksičke jedinice smatra lakoća kojom se kreira vizualna ili auditivna slika referencije. Za razliku od konkretnosti, koja pretpostavlja mogućnost osjetilnog doživljavanja referenta, predočive riječi nisu uvijek dostupne senzornom iskustvu. Rezultati nekih psiholingvističkih istraživanja s tim u vezi ukazuju na to da leksemi nekoga jezika istovremeno mogu biti visoko apstraktni i relativno visoko predočivi, osobito kad je riječ o specifičnim grupacijama leksema kao što su leksemi za izražavanje emocionalnih stanja. Ta istraživanja, kao i istraživanje o kojem ćemo izlagati, polazi od teorije o dvostrukom kodiranju (Dual coding theory, Paivio, 2010). Osnovna je hipoteza te teorije da se kod visoko predočivih riječi uspostavlja dvostruki put (vizualni i jezični) kojim se uspostavlja značenje, dok se kod nisko predočivih riječi uspostavlja samo jezični put. U izlaganju se izlažu rezultati i uspostavlja korelacija između procjene razine apstraktnosti i konkretnosti s jedne i predočivosti s druge strane. Istraživanje je provedeno na uzorku od 30 studenata preddiplomskog studija koji su procjenjivali predočivost i konkretnost 200 leksema na ljestvici od 1 do 5, i to imenica, glagola, prefigiranih glagola, glagolskih imenica i neriječi. Takav je leksički odabir rezultat nastojanja da se pokaže kako, unatoč tvrdnjama o višoj predočivosti imenica u odnosu na druge vrste riječi, na predočivost, osim kategorijalne pripadnosti, utječu i drugi faktori. Izlaganje je dio projekta «Modeliranje mentalne gramatike hrvatskoga: ograničenja informacijske strukture», u okviru kojega se provodi veliko istraživanje suodnosa apstraktnosti/konkretnosti i visoke/niske predočivosti u hrvatskom, a u kasnijim će se fazama provesti dodatna istraživanja utjecaja procjene prema navedenim kategorijama na ograničenja u rečeničnom slaganju vezana uz pripadnosti određenoj semantičkoj kategoriji. O tome će također kratko biti govora u izlaganju.

Imageability of lexical units can be defined as degree of ease with which the visual or auditive picture of a reference is created. Previous research on imageability of verbs and nouns point to higher imageability of nouns (Bird, Franklin, Howard 2001 ; Simonsen, Lind, Hansen, Holm, Melvik 2012). However, in the course of preliminary research on imageability carried out on the sample of 200 Croatian verbs and nouns (selected from the list of most frequent 1000 verbs and nouns from the hrWaC, Croatian web corpus), containing the pseudowords formed as nouns and pseudowords formed as verbs as well, the participants (students of the University of Zagreb) rated Croatian pseudoverbs significantly higher than pseudonouns (Peti- Stantić & Tušek 2015). Pseudowords by definition lack lexical meaning, having form only. This means that they actually represent a string of phonemes that would be acceptable according to phonotactic rules of a given language, without any semantic meaning attached. Without any semantic content, however, pseudowords should not be imageable. Pseudowords were composed by replacing first letter in the Croatian word with another differing from the original one in four distinctive features. Their resemblance to the real word has been tried to reduced to minimum by testing their possible meanings with a group of native Croatian speakers. The participants were given a list of pseudowords and were asked to answer whether the word resembles to any word from their native dialect. If more than 70% of the participants answered negatively, the pseudoword was considered adequate for the research. If in the remaining 30% who answered positively, more than 20% gave the same answer, the pseudoword has not been used. In this presentation we start from the results of this preliminary research. Consequentially, our hypothesis is that the pseudowords in/with verbal form– although the verbs themselves are supposed to be less imageable than the nouns – will be rated higher in imageability than the nouns. The reason for such a hypothesis is in the recognizeability of word-formation suffixes for infinitive which we used in composing pseudowords. Namely, in Croatian there are only two infinitive suffixes, -ti and –ći, and therefore the effect of familiarity should be very high (ie. pseudoverb njolati signalizes the verb through morpheme -ti). As a result, we expect these two verbal word-formation morphemes to be more transparent than, for example, null-morpheme in nominative of pseudonoun snub, or -a in the feminine form snuba. The report on the results of an experimental study in which the participants evaluated the imageability will be given. Firstly, they rated imageability of pseudoverbs and pseudonouns using the scale from 1 to 5. In the next part of the experiment, they got the list of the same pseudoverbs, but with additional verbal prefixes added, for example raz-, pred-, nad-, po- etc. Then they ranked pseudoverbs both without a prefix and with prefixes according to their imageability. Using this technique, we tested the impact of prefixes to the imageability of verbs in an isolated context, without the influence of root imageability.

In Croatian, as well as in some other South Slavic languages, SV and VS conditions in gender agreement of coordinate structures differ substantially. Results of our previous psycholinguistic experiments within the Coordinated Research in the Experimental Morphosyntax of South Slavic Languages (EMSS) project confirm statistically significant differences in agreement types between these two positions. Although prominent claims in the literature favor the structural (pertaining to the syntactic hierarchy) rather than the linear factors in grammatical agreement, our findings from the controlled psycholinguistic production task experiment, as well as the follow-up grammaticality rating study and non-words study, contradict these claims. The baselines for our research are the coordinated structures of the same and mixed gender, the most interesting condition being the NF and FN conditions. In VS condition, the first conjunct after the verb is at the same time the highest hierarchically (HCA), first in a linear structure (out of two or more conjuncts, FCA) and closest to the verb (CCA). This triggers the agreement with the highest/first/closest conjunct so strongly that the agreement with the lowest/second/furthers is almost non-existing. In the SV condition, however, there is much higher variability of results the variability of results is much higher, but at the same time there is a visible tendency towards the CCA in mixed gender conditions not containing M as one of the conjuncts. In our current research, we test only the SV condition, as it’s more informative. We also focus on two linear features that we labeled “first comes first” and “proximity for all”. Our hypothesis is that the rate of the FCA in SV condition with coordinated structure placed outside the focus position would be even lower than previously attested, despite the fact that the first conjunct is hierarchically highest out of two in both conditions. In our previous experiments, the coordinated structure in the SV condition was initially placed in the linear sentence structure. Since our goal was to disambiguate between the hierarchically first and linearly closest agreement, we did not control for the factors pertaining to the information structure. In our current experiment, the coordinated structure is placed in the second position, while the very first position is reserved for adverbs. (Stimulus: Dogovor je pomaknut na petak. Target: sjednice i vijeća vs. stimulus: Iznenada je dogovor pomaknut na petak) As a result, we avoid the effect of focus position. We will present the combination of results from some of the offline comprehension studies, grammaticality judgment ratings and online production tasks supporting the view that linearity plays not only a more prominent role than what has been claimed in the literature, but a crucial one. Our findings are in line with the Parallel Architecture and the Simpler Syntax Hypothesis. The implications of our findings go far beyond the gender agreement itself by opening the doors to a possible theory of psycholinguistically informed agreement processing.

Building on a longstanding debate of the role of semantics in syntax and grounded in the analysis of the results of controlled psycholinguistic production experiment conducted within the EMSS project, we investigate the influence of semantics in agreement and, in particular, the sensitivity of various agreement strategies to semantic guidance in Croatian. Furthermore, favouring the constraint model in selecting agreement features, under which the features from the mental lexicon are selected under semantic guidance, over the control one, we investigate (i) how the deprivation of semantic features influences the selection process, and (ii) to what degree morphological markers on preverbal subjects can be treated as reliable and sufficient predictors of agreement on the verb. We test for this by fully excluding the semantic component - gender, from the recent Marušič et al. (2015) stimuli. On-line production experiment with non-words was created adapting the Marušič et al. factorial stimulus design to be tested on Croatian native speakers (n=30). Stimuli from the experiment were further used in a written grammaticality judgments task administered to another group of Croatian native speakers (n=30). In contrast to the original stimuli Coordination—Adverb—Auxiliary—Participle structured items were created, in order to reduce the potential of a blind suffix copy strategy. Non-words were created to substitute nominal roots in all 9 gender combinations (MM, FF, NN, MF, MN, FM, FN, NM, NF). Non-words were controlled for length, number of syllables and plausible syllable structures in order to reduce to a minimum the frequency and the word length as potential predictors of agreement. Three sets of non-words were created in total, one set for subject in model sentence and two for the coordination in stimulus. The two latter sets were created primarily due to syncretism between the nominative Npl suffix –a and Fsg suffix –a, leading to ambiguous -e agreement on the verb in FN and NF condition. Second set consisted of coordinated structures containing non-words and a prototypical plural- gender suffix (-i for Masculine, -e for Feminine and –a for Neuter) (n=26). Third set consisted of non-words created in the first set, with inserted frequent gender infix (-et- for Masculine, -ic- for Feminine and –enj- for Neuter) between the non-word and the suffix (n=26). Although this disambiguation is needed only with the suffix –a, we employed it systematically for all stimuli in all genders. Experiment consisted of 6 sentences in 9 conditions: (in total 54 sentences). 54 sentences of varying structures were added as fillers. By using gender, position, and presence of infix as main factors in analysis, we predict: 1. that the agreement in examples with non- words already in NN sequences, to which only ambiguous suffixes are attached (that can be understood as Npl and Fsg), would exhibit three-fold pattern of agreement (syntactic M and both N and F), and that the variation would be even greater in NF and FN sequences ; 2. agreement with masculine gender has been proven valid, but should not be treated as a default, rather as a syntactically driven agreement ; 3. greater proportion of agreement in coordinated structures with mixed gender driven by linear, rather than narrow syntactic hierarchical feature, points to the lack of semantic guidance ; 4. deducing all the factors (semantic and attraction) to a minimum a true highest conjunct agreement should appear.

We conducted a comparison of preverbal and postverbal subjectverb agreement in an elicited production study carried out with 60 speakers of different language varieties in South Slavic, spanning Bosnian, Croatian, Serbian, and Slovenian at six local universities. These languages have three genders, and we measured the response type and total reaction time for production for the nine possible combinations of two plural noun phrases. Elicited production experiments for the preverbal and postverbal versions were conducted separately, with 54 items and 54 fillers. A sample target item involved a model sentence, immediately followed by a replacement phrase, and participants had to produce the entire sentence that resulted from combining the new replacement subject noun phrase (a coordination, such as the books and the magazines), with the remaining predicate, making morphosyntactic adjustments where necessary (e.g. in agreement). Our analysis of the gender agreement reveals that preverbal versus postverbal positioning make a large difference in the availability with the linearly furthest conjunct: while highest agreement is possible preverbally (e.g. the FN and NF columns), lowest-conjunct agreement is essentially unattested postverbally. In terms of reaction times, we find longest latencies for the conditions where speakers have the most grammatical options to choose among, suggesting that all three strategies are in principle available, although constrained by syntactic (hierarchical) and morphological factors (default masculine and relative markedness). The consequences of these results will be discussed with respect to three recent theoretical models of closest-conjunct agreement in South Slavic (Boskovic 2009, Puskar & Murphy 2014, Marusic, Nevins & Badecker 2015).

Rječničke definicije, odrednice (kojima se, između ostalog, implicitno označuju i sociolingvističke jezične informacije i odnosi), uputnice i primjeri, koji se nalaze uz rječničku natuknicu, te sam izbor rječničkih natuknica, trebali bi čim preciznije ilustrirati značenje, kao i upotrebu određene riječi za potrebe korisnika rječnika, bilo vlastitog, bilo stranog jezika. S obzirom na niz društvenopolitičkih promjena u Hrvatskoj i Sloveniji od zadnjeg desetljeća 20. stoljeća do danas, a koje su svoj odraz imale i u jeziku, zanimalo nas je kako su se te promjene odrazile na razlike u (rječničkom) opisu dvaju u jezika u odnosu na rječnike izdane prije 90-ih godina 20. stoljeća. Analiza rječnika provest će se na odabranom korpusu hrvatskih i slovenskih rječnika te će se prikazati kako opis jezika u tim priručnicima prati određene društveno-političke promjene. Istraživanje će biti provedeno na leksičkom fondu za označavanje istog sadržaja u oba jezika s posebnim naglaskom na sljedeće kategorije: 1. rječnički status leksika za označavanje društveno-političkih pojava u Sloveniji i Hrvatskoj (npr. omladinac/(o)mladinec), 2. rječnički status i obilježavanje terminologije kod koje je došlo znatne promjene (zamjene) izraza u jednom ili oba jezika u proteklih 20 godina (npr. major/major), 3. leksik za označavanje elemenata popularne kulture (npr. stojadin/stoenka), 4. rječnički status kratica koje su svoje mjesto našle u rječnicima. Usporedo s analizom opisa tih riječi u rječnicima, provest će se i analiza korpusa te će se na konkretnim primjerima utvrditi učestalost korištenja i situacije u kojima se analizirani izrazi koriste (ako se još uvijek koriste). Konkretnije, analizom će se pokušati utvrditi koliko žîvi jezik žívi u analiziranim rječnicima, a što je od živog jezika „proglašeno“ neživim i izbačeno iz rječnika tj. opisa jezika.

U radu su predstavljene mogućnosti izrade slovensko-hrvatskog dvojezičnog leksičkog resursa korištenjem metode automatskog prikupljanja slovensko-hrvatskih prijevodnih ekvivalenata iz usporedivog slovensko-hrvatskog korpusa. Prikupljanje prijevodnih kandidata temelji se na tzv. distribucijskoj pretpostavci po kojoj riječi sličnog značenja imaju i slične kontekste. Ta se pretpostavka može proširiti i na dva jezika pri čemu se za riječ iz ishodišnog jezika kao moguće prijevodne ekvivalente može smatrati one riječi u ciljnom jeziku koje imaju najsličniji kontekst riječi u ishodišnom jeziku. Pretpostavka primjenjivosti ove metode jest postojanje početnog dvojezičnog leksikona kojime se prevode konteksti ciljnog jezika nepoznatih leksema. Time se postiže preduvjet za mogućnost usporedbe konteksta ishodišnog i ciljnog jezika. Dodatno, kako se radi o hrvatskom i slovenskom koji su vrlo bliski jezici, u ovom se istraživanju kao “početni leksikon” koristi postojeće leksičko preklapanje između tih jezika. Popis slovenskih leksema korištenih u istraživanju je 6, 000 najčešćih imenica iz slovenskog internetskog korpusa slWaC. Svakoj od odabranih imenica pridruženo je 10 prijevodnih kandidata iz hrvatskog internetskog korpusa hrWaC s najsličnijim kontekstima. U prvoj su fazi evaluacije studenti diplomskog studija slovenskog jezika označavali točnost dobivenih 10 kontekstualno najsličnijih prijevodnih kandidata. Prijevodne su kandidate označavali bilo kao prijevodne ekvivalente, bilo kao riječi koje su semantički vrlo bliske slovenskom leksemu, bez konzultiranja s postojećim dvojezičnim slovensko-hrvatskim rječnicima. Pri početnom označavanju svaki je prijevodni kandidat označen od strane dvaju studenata što omogućuje mjerenje dogovora među označivačima te lakšu kontrolu rezultata. Dodatno su dio rezultata njihovog označavanja ručno provjerili studenti završnog semestra studija koristeći se raspoloživim rječnicima. Svim provedenim analizama dobiven je podatak o točnosti označavanja od strane studenata, te točnost metode za automatsko prikupljanje prijevodnih kandidata iz usporedivog korpusa. Rad će prikazati rezultate ručne evaluacije rezultata dobivenih metodom automatskog prikupljanja, postotak točnih prijevodnih ekvivalenata koji se pojavljuju na prvom mjestu popisa kandidata, broj prijevodnih ekvivalenata među ostalim kandidatima, te koliko njih se ne može smatrati prijevodnim ekvivalentima, već semantički povezanima s izvornom riječi. Time će se pokazati mogućnost izrade dvojezičnih leksičkih resursa metodom automatskog prikupljanja te testirati razlike u označavanju prijevodnih kandidata od strane označivača u svrhu analiziranja semantičkih podudarnosti i razlika među sličnopisnicama između bliskosrodnih jezika kakvi su slovenski i hrvatski.

Intercultural communication in Croatia is reflected, among other things, in translations of numerous texts as a part of the process of joining the European cultural area. For that reason we aim to investigate the extent to which the main directions in contemporary Croatian terminology regulation reflect the EU standards, but at the same time comply to the tacit knowledge of how one should treat indigenous linguistic tradition. As our main objective, we set up the justification and implementation of terminological principles proposed in different competing documents, for example, in the manual for researchers of the project STRUNA, aimed at systematizing the Croatian terminology, conducted at the Institute for Croatian Language and Linguistic, educational standards and curricula published by the Ministry of Science, Education and Sports, and internal documents published by the Independent Service for Translation in the Field of European Integration of the Ministry of Foreign Affairs and European Integration. Methods used in conducting our research include the sociolinguistic analysis of relevant documents and guideliness produced by three language management institutions in contemporary Croatia that are most influential when speaking of terminology issues. These institutions are the Institute for Croatian Language and Linguistics, the Ministry of Foreign Affairs and European Integration and the Ministry of Science, Education, and Sports. The analyis encompasses all applicable documents published by respective institutions in past 5 years. Building upon the most recent findings on the role of what is to be considered the Language Academy in Croatia (Langston, Peti-Stantic 2011), we will compare data gathered through the extensive inductive research with the situation in several terminological manuals, as well as in the public language represented in media. Hence we intend to present a comprehensive analysis of explicit terminological principles (in particular one of the leading ones that terms coined with domestic (Slavic) lexical material should be prioritized over foreign terminology, both internationalisms and the lexems from the modern European languages), as well as the study of effectiveness of language management aimed at “purifying” the Croatian terminology.

The objective of this research is to analyze Croatian-Slovenian partial false fiends using Croatian and Slovenian national corpora. False friends are pairs of words in two languages that either have the same or similar form while their meanings are completely or partially different. Since Croatian and Slovenian are closely related languages belonging to the western group of South Slavic languages, they share a large number of common lexis (words of Slavic origin as well as internationalisms). While true friends (words that share the same form and meaning) simplify the process of language learning (learners tend to attribute the meaning that a word of the same or similar form has in their language to the word in foreign language), false friends present a negative language transfer for language learners. Such linguistic interference can also be a source of errors in translations. Namely, due to lack of experience in translation and/or source language, translators might not notice that words with similar or same form, do not have the same meaning and cannot always be translated as synonyms between languages. In a language, words can be treated as synonyms if they denote the same object and can consequently be interchangeable. If we understand translation process as identifying synonyms in two different languages, then two words can be treated as synonyms when one can be substituted by the other without changing the meaning of the sentence in target language compared to the meaning of the sentence in the source language. Analysis of partial false friends in this research will be made by collecting Croatian-Slovenian partial false friends, that share the same form and meaning but cannot always be interchangeable without changing the meaning in target language. Using Croatian and Slovenian national corpora these false friends will be analyzed to establish the instances of difference in their meanings in different registers. The analysis will show whether collected Croatian- Slovenian synonyms can be interchangeable in any possible contexts and in what circumstances they become partial false friends and cause errors in translation or communication. The goal of this computer based research is to explore partial false friends and to see the possible mistakes in communication in foreign language or when translating from Slovenian to Croatian.

The paper focuses on some of the challenges translators encounter in the new economy. The objective is to show how translators' competitiveness could be increased by incorporating specific semantic issues in programs for machine translation. Emphasis will be on the problem of false friends (or faux amis) as the pairs of words in two languages that look and/or sound similar, but partially or completely differ in meaning. Due to complexity of the issue, false friends present a serious problem for translators, especially in the contemporary economy where a demand for delivering translation in a timely manner is as important as the quality of the translation itself. The comparative analysis of Slovenian and Croatian, as genetically closely related languages belonging to the western group of South Slavic languages, shows the great number of inter- language homonyms in the two languages. Aim of the analysis of false friends is to provide a typology of false friends for the specific language pair that could have an application in machine translation programs.