The work was financed by Slovenian Research Agency grants J6-9372 (Terminology and Knowledge Frames across Languages - TermFrame).

1. Introduction

1Adopting geographical terminology into other languages implies a comparison of different terrain configurations, but also of the cognitive processes that affect the perception and understanding of the space and environment. Burenhult and Levinson (2008) and Levinson (2008) studied the terms for landforms through an interlinguistic perspective emphasizing different ontological relations between concepts with regard to languages ​​and the important role of culture in the construction of relief concepts. Although concepts such as river, lake, mountain were expected to be universal, contrastive multilingual analysis confirmed differences in the categorization and selection of specific denotational features for geographical concepts.

2Multilingual glossaries are usually considered as a reliable source of knowledge for establishing equivalence between the terms in the chosen domain. Consulting such glossaries, one gets the sense of uniformity of terminology and concepts between languages that corresponds to the principle of univocity in terminological theory. However, interlingual conceptual differences or the absence of denotations need to be taken into account when comparing terminologies in different languages. Bearing in mind the existing literature on the conceptualization and categorization of geographical terms, one should be mindful of the importance of a cognitive approach to geography and the need to introduce ethnophysiography (Mark and Turk, 2004).

3In this paper we start from the premise that the complexity of specialized meaning or specific differences at the conceptual level between languages is, in a multilingual perspective, only noticeable at the phraseological level. Our analysis is primarily oriented towards interlingual discrepancies in a comparative context, which directly excludes the choice of the source language and target language and allows for parallel observation of three language systems as three coexisting functional systems, three lexical systems, three morphological structures, three semantic-discursive structures, three paradigmatic, and three syntagmatic plans.

4Historical-scientific interaction between the French language and the language of the Dinaric countries in the field of karstology was described by French authors E.A. Martel and J. Nicod who conducted karst research in Croatia and the former Yugoslavia. They made a great contribution to the creation of karst terminology by recording expressions such as dolina, hum, kamenica, polje, ponor, ponikva, or jezero, which have become permanently established in international karst terminology. Using some of these terms as examples, we will present a method of reviewing the mutual conceptual similarity in English, French and Croatian.

5Our research is part of the TermFrame project dedicated to the creation of the trilingual (English-Slovenian-Croatian) interactive termbase for karstology available at the website https://termframe.ff.uni-lj.si/​. The digital termbase was developed as a result of the collaboration of the interdisciplinary project team of researchers from the fields of terminology, cognitive and computational linguistics, as well as geography, including karstology and speleology. In addition to the terms’ definitions carried out in accordance with the annotation model (Vintar and Stepišnik, 2020; Vintar and Martinc, 2022), the termbase also contains visualizations (fig. 1) that display hierarchical, partitive and associative relationships in the conceptual system of the karstology ​​(Vintar et al., 2019; Pollak et al., 2020).

6First results of the bilingual comparative approach to multi-word terms (MWTs) are described in Grčić Simeunović et al. (2021) where adjectives were extracted in English and Croatian corpora and categorized into semantic attributes. In this article we continue to explore the distribution of attributes for a particular conceptual category by comparing multi-word terms in English, French and Croatian.

2. Methodology

7For the purposes of this research, we used the English and Croatian parts of the TermFrame corpus, which otherwise also contains Slovene as the third language. We additionally compiled a specialized comparable corpus in French. For each language the corpus consisted of two sub corpora, one containing scientific texts (doctoral dissertations, scientific papers, conference proceedings) and the other, didactic texts (textbooks and lecture notes). All three corpora are comparable in terms of domain and text types included, but the English is slightly larger in size (tab. 1).

8Selection of the candidate terms at the macrostructure level as well as the classification of terms, required collaboration between terminologists and domain experts. For this research, we relied on semantic categorization from the TermFrame project, which contains top categories and conceptual relations to describe various geomorphological forms shaped as a result of processes such as corrosion (karst erosion), accumulation, abrasion, mechanical weathering and fluvial, glacial, aeolian and other processes (Stepišnik, 2011). The TermFrame project relies on the semantic hierarchical framework of the EcoLexicon project (Faber et al., 2011), which was later adapted for karstology in compliance with Ford and Williams (2007) and Jennings (1985). Conceptual system was divided into five categories and 14 subcategories that include surface and underground landforms, hydrological forms, processes, typical karst environments or geomes, properties, and typical methods and instruments in karstology (fig. 2).

Fig. 2 – Semantic categorization of terms in ​​karstology.
Fig. 2 – Catégorisation sémantique des termes en karstologie.

9The conceptual hierarchy also includes 16 conceptual relationships following the recommendations of the subject experts and the geomorphological analytical approach according to Pavlopoulos et al. (2009). They describe distinctive properties of terms that refer to spatial distribution (LOCATION), morphography (FORM), morphometry (SIZE), morphostructure (COMPOSITION), morphogenesis (CAUSE), morphodynamics (RESULT, FUNCTION) and morpho chronology (AGE) of karst forms.

10As features in taxonomies are usually considered attribute terms, so are the features expressed with adjectives in this paper. The introduction of adjectives allows to expand the understanding of the contextual frame in which the observed entity can be placed, which is also a precondition for finding an equivalent in the target language. We therefore view multi-word terms as one of the mechanisms for creating specialized meaning while understanding the phraseological level of terms in a broader sense.  The established correlations between the attributes and semantic categories in karstology aim to offer new perspectives on the dynamic modeling of specialized knowledge.

11The research described here focuses on the most frequent key terms from the three conceptual categories: surface landforms (terms: doline, depression, karren, valley), hydrological forms (terms: spring, aquifer, ponor, conduit) and processes (terms: denudation, weathering, erosion). The degree of equivalence between key terms in three languages was further verified by analyzing their combinatorial potential on the phraseological level. For that purpose, we chose to limit our analysis on the adjectives expressing semantic relation inside multi-word terms.

3. Research results

12Our trilingual analysis was focused on comparing distribution of multi-word terms for each of the key terms from three conceptual categories. This methodology allowed us to illustrate not only the equivalents in three languages but also the near synonymes that can lead to misunderstandings in the specialized communication. For example, the synonymous relation between the terms doline and sinkhole in English became obvious after comparing their combinatorial potential and the actual multi-word terms solution sinkhole – solution doline, collapse sinkhole – collapse doline and suffosion sinkhole – suffosion doline. Two synonymous terms ponikva and vrtača were also confirmed in Croatian according to the shared multi-word terms urušna ponikva – urušna vrtača, bunarasta ponikva – bunarasta vrtača, ljevkasta ponikva – ljevkasta vrtača. However, when comparing results in trilingual analysis it became obvious that the term doline in English and French is equivalent exclusively to the Croatian terms ponikva and vrtača (Skračić and Jurić, 2004) and should not be associated with the Croatian term dolina. According to Pahernik (2012) dolina should be distinguished from ponikva because the predisposition for the emergence of doline in Cretaceous and Jurassic limestone rocks was created by tectonic activity when dolomite-limestone rocks break more intensively creating numerous karren of different dimensions and directions. Furthermore, considering that karst relief or doline cannot be formed without the carbonate basis, the primary factor in the development of dolines is lithological basis, i.e. the existence of carbonates. The shape, size and intensity of occurrence (density) of the dolines are therefore the result of structural, morphometric and climatic-biological factors on such a lithological basis. In addition, the formation of dolines can also be the result of the collapse of endokarst forms, suffusion and subcutaneous corrosion of unbound clastites on carbonates (Sauro, 2003). Taking into account the linguistic and conceptual level of the correspondence, our comparison of terms on the phraseological level will illustrate the challenges of establishing equivalents in three languages.

3.1. Terms from the category surface landforms

13For the terms doline – doline – ponikva numerous multi-word terms have been confirmed in all three languages, however, the distribution of certain distinctive features or attributes is not completely uniform. The corpus results confirmed several equivalents expressing CAUSE in all three languages like suffosion doline – doline de soutirage – sufozijska ponikva, collapse doline – doline d'effondrement – urušna ponikva, and alluvial doline – doline sous-alluvial – aluvijalna ponikva. These examples actually illustrate some of the dominant processes in the formation of dolines. According to Ford and Williams (2007) processes such as dissolution, collapse, subsidence and suffocation can act simultaneously or separately and result in the formation of dolines. While English and French equivalents solutional doline – doline de dissolution and subsidence doline – doline de subsidence are confirmed in the corpora, Croatian equivalent solucijska (or disolucijska) ponikva is missing. One of the possible explanations could be that in Croatia, dissolution is the most common cause of the emergence of doline so the adjective (di)solucijski may be considered redundant.

14When it comes to the FORM, doline’s morphologies are territory specific and depend on the natural factors such as the intensity and shape of denudation on the one side and the dispersion, strength and different directions of the karren on the other. Given that the formation of karren along the tectonic routes enables the denudation effect of rainwater, the form of dolines varies from round to oval or irregular with steep or slightly sloping sides. The equivalent pairs were found in French and Croatian: doline étoilée – zvjezdasta ponikva, and English and Croatian: funnel-shaped doline – ljevkasta ponikva, pan-shaped doline – tavasta ponikva, and bowl-shaped doline – zdjelasta ponikva. Only English corpus confirmed examples such as rounded doline and flat-floored doline. Possible reason can be that these adjectives are redundant as sinkholes are, by definition, funnel-shaped circular flat-bottomed depressions in karst. Other examples expressing FORM that were found only in Croatian are presented in Glossary.

15Apart from their mutual cause and form, dolines differ in size, location, structure, and composition of the floor. The SIZE attribute, which in this case implies depth and width, is confirmed in all three languages, for example, deep doline – doline profonde – duboka ponikva. However, in the absence of a conceptual categorization that defines size ranges it is not always clear whether they are complete equivalents or only close synonyms, as in the case of small doline in English and plitka ponikva in Croatian, which indicates the subjectivity of the description.

16Considering that the terms depression – dépression – depresija ​denote generic concepts for all areas that are lower in altitude than the surrounding areas, discrepancies between languages ​​were expected. Namely, depressions can be of different locations and origins: depressions in karst, depressions in deserts, tectonic depressions, etc. The LOCATION of depression is expressed in all three corpora, but without equivalent pairs: in English surface depression, in French dépression désertique, dépression périphérique and in Croatian jezerska depresija, morska depresija, oceanska depresija, sinklinalna depresija. These examples refer to different terrains in three geographical areas. The same is true for the FORM attribute for which two pairs of equivalents have been recorded: closed depression – dépression fermée – zatvorena depresija, and circular depression – dépression circulaire, but in addition to these, each language also has its specific examples (Glossary).

17As for dolines, the SIZE of depressions is expressed by similar but not identical equivalents in the three languages illustrating the subjective description of SIZE: shallow depression – petite depression – plitka depresija, large depression / wide depression – prostrana depresija and deep depression – vaste dépression.

18While for the CAUSE attribute only one pair of equivalents was recorded in all three languages tectonic depression – dépression tectonique – tektonska depresija, more numerous examples have been confirmed in the English corpus collapse depression, erosional depression, paleokarst depression, solution depression, subsidence depression and suffosion depression. The results from the corpus are in accordance with the fact that the term depression in English denotes an area that is lower in altitude than the surrounding areas. When the term is not strictly defined through the dimensions, it is important to notice that the English term depression, in these cases, can correspond to Croatian terms depresija but also ponikva. The presence of the adjective indicates in these cases the precise equivalent term. For example, collapse depression can be equivalent to the Croatian urušna ponikva, while tectonic depression can be equivalent to tektonska depresija, depending on whether an individual author uses a generic term or defines a particular landform more precisely.

19COMPOSITION of depressions is expressed in the English and French corpora, but without the equivalent pairs: sediment-filled depression in English and dépression argilo-marneuse, dépression à évaporite, dépression marno-gréseuse in French. The AGE of a depression is confirmed by one example in each language: paleokarst depression in English, dépression permienne in French, and mlađa depresija in Croatian. Those are examples of possible associations of karstic objects with periods.

20For the third term karren – lapiaz, lapiez, lapiés – škrapa, the equivalent pairs for LOCATION have been found in all three languages: cavernous karren – lapiez de fissure – pukotinska škrapa, denuded karren – lapiaz dénudé, littoral karren / coastal karren – priobalna škrapa, and lapiaz de surface – površinska škrapa and lapiés de paroi – zidna škrapa. Many multi-word terms expressing FORM have been recorded in all three languages but without the equivalent pairs (Glossary). Given that these landforms were created in different geographical areas, it is understandable that the forms do not necessarily coincide.

21One equivalent pair was confirmed for expressing the SIZE, small karren – petit lapiaz. However, the differences between the terms in English large karren, long karren, and French lapiez géant still confirm the subjective determination of this attribute. One of the reasons is that the size and morphology of exokarst forms, hence karren, primarily depends on the proportion of pure CaCO₃ in the rock composition. Moreover, they are the result of the influence of tectonic movements and the resulting cracks, relief (especially morphometric), pedological and climatic and biological characteristics of the area (Perica et al. 1999 ; Veress, 2019). Since the largest forms with a high content of pure calcium carbonate in the Dinaric karst are mostly present in karst deposits formed in the Middle and Upper Jurassic, we can say that the attributes of SIZE and AGE are mutually dependent. Another important aspect is the slope of the relief. On certain parts of higher slopes, the forms are smaller, which is a consequence of faster runoff, i.e., shorter water retention and thus less possibility of corrosive action. Taking into account that most of the water is of atmospheric origin, the amount of precipitation and temperature significantly affect the size of karren and other exokarst landforms (Perica et al., 1999; Veress, 2019). It is obvious that the diversity of geographical areas causes a difference in the average size of large and small forms for one of the three observed areas. In this case, a subjective approach to sizing is logical.

22When it comes to the fourth term in the category of surface landforms valley – vallée – uvala, it is important to say that the Croatian term uvala (when it is not denoting a bay) refers exclusively to karst geomorphology, while the English valley and French vallée cover a much wider meaning, i.e., they refer to karst geomorphology, but also to fluvio-denudation and glacial forms. This becomes obvious when the term is combined with the adjective, for example the Croatian equivalent to the English term karst valley and the French vallée karstique is not krška uvala but (krška) dolina. This distinction becomes even more important because krška uvala is still a debated term in Croatian geomorphology as it is considered as a transitional form of sinkhole and karst field but not related to fluvial, fluviodenudation, or glacial forms (Poljak, 1951; Stepišnik, 2015.). For example, the comparison on the phraseological level, discovers that the French vallée fossile corresponds to fosilna dolina and not fosilna uvala (Buzjak et al., 2019).

23Valley is the only term from the category of surface landforms for which examples expressing FUNCTION have been confirmed in English and French: blind valley – vallée aveugle, dry valley – vallée sèche. In addition to these, active valley and relict valley have been recorded in English. Examples for SIZE have been confirmed in English and French, expressing similar but not identical equivalents: broad valley, deep valley, and wide valley in English, and basse vallée, grande vallée, petite vallée and vallée majeure in French. One equivalent pair express FORM incised valley –vallée encaissée while other examples appear only in one of the languages (Glossary). The LOCATION of the valley is expressed only in French: vallée alpine, vallée supérieure, vallée de surface. There is no equivalent for vallée alpine in Croatian because it is a form specific to the Alps (that is why, for example, it appears in the Slovenian language). According to Crkvenčić (1951) and Lozić et al. (2012) the Croatian equivalent for the French vallée supérieure is not uvala but gornje porječje. Except for one pair of equivalents fluvial valley – vallée fluviatile – fluviokrška uvala, the CAUSE is expressed by different multi-word terms in each language (Glossary).

24The comparative analysis of the combinatorial potential of the 4 terms from the category of surface landforms shows the prevalence of the attribute FORM in all three languages while the attributes CAUSE and SIZE are equally present in English and French and in minor portion in Croatian. LOCATION of the 4 surface landforms is mostly expressed in French and Croatian while English shows less examples. Very few adjectives express the AGE in all three languages while COMPOSITION is present only in English and French. The overall distribution is shown in Figure 3.

3.2. Terms from the category hydrological forms

25For the term spring – source – vrelo largest number of examples is confirmed in French which indicates the possible wider semantic meaning of the term source. Most examples in English and French refer (but not always) to karst geomorphology, given that the terms spring and source cover a wider semantic field than the Croatian term vrelo. That is why the English and French often use the complex terms karst spring / source karstique when referring to vrelo. In cases where the source is described outside the karst area it is important to use the Croatian term izvor. FUNCTION is expressed by the equivalent pairs intermittent spring – source intermittente – intermitentno vrelo, permanent spring – source pérenne while more numerous multi-word terms were found only in English (Glossary). AGE and COMPOSITION only in the French examples source fossile, source jurasienne for the first and source carbogazeuse, source saumâtre, source sédimentaire, source sulfurée, source sulfurique for the second attribute. One example bunarasto vrelo confirms the FORM in Croatian. LOCATION was recorded in three languages but only two pairs of equivalents were established coastal spring – source côtière – priobalno vrelo and source sous marine – podmorsko vrelo, while other examples in French: source latérale, source littorale, source sous lacustre, source sous alluviale differ from those in Croatian pećinsko vrelo, pukotinsko vrelo, špiljsko vrelo. The SIZE is expressed by close synonyms in English large spring and French source majeure, while French also offers the antonymous pair source mineure.

26For the terms aquifer – aquifère – vodonosnik only the FUNCTION is recorded in all three languages with equivalents such as open aquifer / unconfined aquifer – aquifère non confiné – otvoreni vodonosnik, semi confined aquifer – aquifère semi captif – poluzatvoreni vodonosnik, porous aquifer – aquifère poreux – porozan vodonosnik and confined aquifer – aquifère captif / aquifère confiné – zatvoreni vodonosnik. Interesting example is open aquifer / unconfined aquifer which is by its definition bounded upwards by a water surface, and downwards by an impermeable layer (insulator). Thus, it corresponds to the FORM attribute, but such a form affects the specificity of rapid recovery through precipitation and seepage from the surface, which results in significant variation in water level and can also denote the FUNCTION attribute. The same explanation is valid for the semi-closed, closed, and porous aquifer (Riđanović, 1970, 1990; Fetter, 1988). Other examples denoting FUNCTION in English are epigenic aquifer, hypogenic aquifer, and aquifère libre, aquifère potable in French.

27Equivalents for LOCATION are alluvial aquifer – aquifère alluvial, coastal aquifer – aquifère cȏtier / aquifère littoral, and in addition to these examples otočki vodonosnik and podzeman vodonosnik have been confirmed in Croatian. The equivalent pairs aquifère carbonaté – karbonatan vodonosnik, aquifère calcaire – vapnenački vodonosnik confirm the importance of the attribute COMPOSITION in both, Croatian and French, while in French there is also an example aquifère crayeux. Corpus examples confirm that the AGE is expressed in French aquifère liasique and aquifère plio-quaternaire and in Croatian primaran vodonosnik and sekundaran vodonosnik. The FORM is expressed with one pair of equivalents in English and French intergranular aquifer – aquifère granulaire.

28For the term ponor different multi-word terms were confirmed with rare examples of equivalent pairs. COMPOSITION is expressed in all three languages: alluvial ponor – perte sous-alluviale – aluvijalan ponor, and one additional example in French perte sédimentaire. The equivalent pairs for FUNCTION are active ponor – perte active – aktivni ponor and perte temporaire – povremeni ponor. According to Roglić (1961) and Stringfield et al. (1971) multi-word term perte temporaire – povremeni ponor simultaneously describes FORM and LOCATION, which illustrates the complexity of the conceptual classification and the potential multidimensionality of adjectives that should be taken in account. Other examples expressing FUNCTION in French Croatian are available in Glossary.

29Small ponor refers to the SIZE in English, while FORM is confirmed only in Croatian: otvoreni ponor, ponikvast ponor, sitast ponor, and vrtačast ponor. LOCATION is expressed in Croatian and French, but without the equivalent pairs: perte karstique, perte latérale, perte souterraine in French, and jamski ponor, morski ponor, obalni ponor, pukotinski ponor and spiljski ponor in Croatian. There are no examples expressing the AGE of the ponor. The disproportion between the multi-word terms in the three languages ​​can be explained by the fact that the term ponor is a loanword that has synonymous pairs sink, chasm/swallow hole in English and abîme in French.

30For the term conduit – conduit – kanal examples were confirmed in all three languages for FUNCTION, LOCATION, CAUSE and FORM. The equivalent pairs for FUNCTION are active conduit – aktivni kanal, fossil conduit – conduit fossile – fosilni kanal, phreatic conduit – conduit phréatique – freatski kanal and vadose conduit – conduit vadose – vadozni kanal. The LOCATION is confirmed by the equivalent pairs hypogene conduit – conduit hypogène, karst conduit – conduit karstique, fissure conduit – pukotinski kanal, cave conduit – špiljski kanal. In the case of multi-word terms subterranean conduit – conduit souterrain – podzemni kanal it is interesting to note the other three variants in English underground conduit, subsurface conduit, and underwater conduit. The only attribute with numerous equivalent pairs is FORM: elliptic conduit – conduit elliptique, horizontal conduit – conduit horizontal – horizontalan kanal, circular conduit – kružni kanal, labyrinthine conduit – conduit labyrinthique, rectangular conduit – conduit rectangulaire, subvertical conduit – conduit subvertical, vertical conduit – conduit vertical – vertikalni kanal. Only the French corpus confirmed the importance of the attribute SIZE: conduit dédoublé, petit conduit and grand conduit. The only equivalent pair for the CAUSE attribute is submerged conduit – potopljeni kanal. Other examples with no equivalent pairs are shown in Glossary. The distribution of individual distinctive features for the category hydrological forms is shown in Figure 4.

3.3. Terms from the category processes

31For the terms denoting the process of denudation – dénudation – denudacija the largest number of examples was found in the English corpus. All three languages confirm the importance of CAUSE with two equivalent pairs in English and Croatian: fluvial denudation – fluvijalna denudacija and chemical denudation – kemijska denudacija. LOCATION is expressed with two equivalent pairs in English and Croatian karst denudation – krška denudacija and surface denudation – rubna denudacija. Furthermore, podzemna denudacija is recorded in Croatian. AGE attribute is expressed only in English and French e.g. Quaternary denudation and dénudation cénozoique. Other examples with no equivalent pairs are shown in Glossary.

32For the term weathering – altération – trošenje in all three languages, the attributes CAUSE, LOCATION, and COMPOSITION were recorded. The largest number of examples for CAUSE is in English, and the equivalent pairs are biogenic weathering – biološko trošenje, biophysical weathering – altération physico-chimique, chemical weathering – altération chimique – kemijsko trošenje, hydrothermal weathering – hidrotermalno trošenje, physical weathering – altération physique – fizičko trošenje, and mechanical weathering – mehaničko trošenje. The LOCATION is expressed through the equivalent pairs of biokarstic weathering – biokorozija, coastal weathering – abrazija (in the case of wave weathering), periglacial weathering – periglacijalno trošenje, superficial weathering – altération superficielle, and through other examples that differ between languages (Glossary). Regarding the attribute COMPOSITION, the largest number of examples was recorded in English (11), one in Croatian, and four in French with only one pair of equivalents: salt weathering – solno trošenje. The FORM is expressed in English in spheroidal weathering, and in Croatian pločasto trošenje, plošno trošenje. One example for AGE in Croatian is pleistocensko trošenje, while the attribute SIZE was not recorded in any of the three languages.

33For the terms erosion – érosion – erozija, the largest number of equivalent pairs express CAUSE: érosion chimique – kemijska erozija, érosion éolienne – eolska erozija, fluvial erosion – érosion fluviatile – fluvijalna erozija, glacial erosion – érosion glaciaire – glacijalna erozija, marine erosion – marinska erozija, mechanical erosion – érosion mécanique – mehanička erozija, river erosion – riječna erozija, water erosion – vodena erozija. Since in Croatian geomorphological terminology (Roglić, 1959 ; Ford and Williams, 2007) the term erosion refers only to weathering (denudation) by the power of moving water, the adjectives fluvijalni or riječni can be considered redundant. Following the same principle, when weathering is influenced by some other agent (wind, ice, sea waves) the term erosion should not be used as synonym because Croatian terminology offers different terms for each of these forms of denudation: the term deflacija is used instead of eolska erozija, egzaracija instead of glacijalna erozija, korazija for kemijska erozija, abrazija for marinska erozija. The term erozija in Croatian is therefore reserved only for fluvial or river erosion. The LOCATION of erosion is expressed by the equivalents coastal erosion – érosion côtière, lateral erosion – lateralna erozija, littoral erosion – érosion littorale, subsurface erosion – érosion souterraine – podzemna erozija and surface erosion – érosion en surface – površinska erozija. Other examples are presented in Glossary. The FORM was confirmed by one equivalent pair érosion linéaire – linearna erozija, two examples in English downward erosion and upward erosion, one in French érosion verticale and one in Croatian ciklička erozija. The CAUSE of the erosion is expressed only in English: biological erosion, chemical erosion, cosmogenic erosion, paragenetic erosion, as well as the AGE, i.e., the timing: Cenozoic erosion, Pliocene erosion, Quaternary erosion, Tertiary erosion. The attribute COMPOSITION is expressed in one pair of equivalents in English and French sediment erosion – érosion sédimentaire, and in few other examples in English carbonate erosion, karren erosion and travertine erosion. The distribution of individual distinctive features for the category processes is shown in Figure 5.

4. Discussion

34Our trilingual terminological research using corpus analysis brought attention not only to the differences at the phraseological level of terms, but also to cross-cultural differences in the definition of terms and the choice of identifying distinctive features.

35As the research results show, FORM is the most productive attribute for surface landforms in all three languages. This is in line with our expectations since the shape is their most distinctive feature. In the Croatian corpus, a larger number of FORM attributes for surface landforms were recorded, possibly due to the specific characteristics of the relief of that area, equally possibly due to the long tradition of researching precisely such occurrences or maybe something else. Or simply, the corpus of documents in Croatian is oriented towards such occurrences. After FORM, English corpus demonstrates the largest number of multi-word terms expressing CAUSE and SIZE while in French the largest number of examples express FORM, CAUSE and LOCATION. The reason for the fact that the English corpus records a greater number of CAUSE attributes is perhaps that the English approach is more analytical, i.e., it strives more to understand and explain the causes and relationships between karst phenomena, focusing on analysing and on interpreting the phenomena; while, on the other hand, perhaps the French approach is more descriptive, striving for a systematic description of karst phenomena and focusing on observing the phenomena. The COMPOSITION of surface landforms is expressed only in French and English.

36FUNCTION, LOCATION and FORM are the most relevant distinctive features for the four hydrological forms in all three languages. French gives the most numerous examples in all six attributes. The SIZE is not confirmed in Croatian while multi-word terms expressing AGE of the hydrological forms are missing in English.

37All six attributes are expressed in all three languages for the semantic category of processes, but CAUSE and LOCATION are predominant. COMPOSITION is found mostly in English corpus.

38We suggest these preliminary quantitative results serve as the starting point for future more elaborate analysis that could offer possible explanations for the mentioned language and cultural differences.

39It is important to emphasize that the presented corpus results include only terms extracted from our corpus, and it is possible that some local or outdated terms are included among the ones listed. Since this comparative analysis is not prescriptive but descriptive, all the terms are included in the analysis regardless of their pragmatic status. The sources in the corpus were selected according to the criteria of descriptive terminological approach and refer to the date of publication, thematic uniformity, type of texts and language of publication. Only sources available in electronic form were included in the corpus, which unfortunately left out sources in print. For this reason, the criterion of the relevance of the source is only partially met, but considering the extensiveness of the body of literature, we believe that the sample of texts provides an insight into the use of terminology in three languages.

5. Conclusion

40The comparison of the key terms in the triilingual perspective refer to the evoked cognitive frames regarding the distinctive properties or attributes as constitutive parts of conceptual structure of specialized knowledge. Based on the presented results, it can be concluded that the introduction of the adjective collocates expands the understanding of the background frame in which the observed entity can be placed, which is also a prerequisite for finding an equivalent in the target language.

41The results of this analysis provide an insight into the differences in naming specific geomorphological terms in English, French and Croatian, which partly results from differences in terrain configuration. Our analysis of multi-word terms and adjective modifiers also point to different distributions of attributes describing certain geomorphological forms in the three languages and their relevance with respect to physical, geographical, and cultural factors.

42Although a large number of multi-word terms has been recorded, we are aware that the sample of the analyzed terms isn’t sufficient for drawing conclusions about patterns for semantic categories. Therefore, we consider this research to be a pilot study presenting a methodology for further analysis on a larger term sample.

43Glossary

Corresponding author, Tel +385 (23) 20 07 46, lgrcic@unizd.hr (Larisa Grčić)