2.1 The relevance of phraseology in language teaching

Phraseology is a discipline concerned with the study of word combinations which are bound together by very specific combinatory restrictions and form a semantically and/or pragmatically meaningful unit. The terms phraseme or phraseological unit are used in this article as umbrella terms for all kinds of multi-word expressions which fulfil these conditions (in the literature, there has been debate about the inclusion of single word units within the remit of phraseology, see Burger 2015 and Targońska 2022, but addressing this issue is beyond the scope of the present article). This broad notion of phraseology, which is not only confined to the study of idiomatic and fixed expressions, is in line with current trends in the field. Today, the area of study of phraseology includes many units “which display a high degree of syntactic variability and semantic compositionality” (Granger / Meunier 2008a: xx).

Over the last decades, the contribution of phraseology to language teaching research has been growing in importance. There are two main reasons for this. Firstly, recent research has highlighted the frequency of phrasemes both in the language system – in terms of number of items in the lexical stock – and in language use. Based on an empirical corpus linguistic study, Erman and Warren (2000: 29) arrive at the conclusion that, on average, around half of the language used in spoken and written texts consists of what they describe as “prefabricated patterns”. Secondly, there is evidence that phraseological units play several key functions in the development of linguistic competence. As Nesselhauf (2005) explains, two of these functions, the facilitation of fluency and of comprehension, are of special relevance for non-native learners (see Nesselhauf 2005 for an overview of previous research on this topic).

2.2 Phraseology from a plurilingual perspective

In the PhraseoLab project, the development of materials and activities for teaching-learning phraseo-logy is approached from a plurilingual perspective. Pluralistic approaches are “didactic approaches that use teaching/learning activities involving several (i.e. more than one) varieties of languages or cultures” (Candelier et al. 2010: 5). These approaches stand in contrast to ‘singular’ approaches, in which a particular language or culture is dealt with in isolation from other languages and cultures of which the student may already have some knowledge. A central concept in a plurilingual approach is the notion of plurilingual competence. This is seen “not as the superposition or juxtaposition of distinct competences, but rather as the existence of a complex or even composite competence on which the user may draw” (Council of Europe 2001: 168; see also Cenoz / Gorter 2013). The interconnection of the learner’s knowledge of two or more languages is conceived of – and taken advantage of – as a constitutive element of a composite linguistic competence rather than as a source of interference between separate linguistic competences.

The specific constellation of languages selected for PhraseoLab is English (L2) and German (L3). In this framework, the learners’ knowledge of English, and more specifically their phraseologic-al competence in English, is exploited as a facilitating factor for developing phraseological competence in German (as L3). German and English form an especially suitable pair of languages for a plurilingual approach. Similarity in language type and intensity of language contact are two factors which play a key role in the construction of “transfer bridges” between L1/L2 and L3 (the term “transfer bridge” is credited to Meissner 2000, as cited in Neuner 2004). The German-English pair meets these two conditions: they are very similar types of languages – in fact, they stem from the same close ancestor, Germanic – and there is a very intense contact between them at present. Moreover, the fact that English today is so widely learned as an L2 – certainly more than German – reinforces its potential as a suitable bridge language.

In the framework of this approach, equivalence relations among languages are endowed with a significant pedagogical potential, and the analysis of these relations represents necessarily an important step that must precede the design of the teaching-learning materials and activities. However, inter-linguistic equivalence and phraseological equivalence are intricate relationships involving multiple dimensions and levels of analysis. In the following sections we will seek to elucidate some aspects of equivalence relations that are of special relevance to the implementation of a plurilingual approach.

3.1 Operational levels of phraseological equivalence

The origin of the concept of inter-linguistic equivalence can be traced back to contrastive linguistics and linguistics-oriented translation studies. Contrastive linguistics aims at describing the features of two compared linguistic systems, identifying both convergent and divergent aspects. The insights gathered from this discipline have enriched both translatology and foreign language didactics, though in distinct ways. Since translation studies have developed more intensive research on phraseological equivalence, this section starts with the concepts and models of phraseological equivalence proposed within translatology. The next section will deal with the use that didactics has made of this concept.

In the realm of translation studies, particularly influenced by mathematical or formal logic, equivalence assumes a paramount role, and it points to the idea of faithfully conveying the entirety of a text’s message – encompassing its meaning, stylistic attributes, and effects – into another language. Hence, it should not come as a surprise that the intersection of the concept of equivalence with phraseology arises from works on contrastive phraseology and phraseological translation.

Inter-linguistic equivalence functions at two distinct levels: the systemic and the textual level (cf. Dobrovol’skij’s 2011 systematic and translational equivalence). The first level, intrinsic to contrastive linguistics and lexicology, regards language as an abstract system of linguistic signs devoid of contextual influence. Applied to phraseology, at this level the search for equivalence entails identifying the phraseme with a corresponding meaning in another language. For example, the corresponding phraseological unit of the German jmdm. das Wort aus dem Mund nehmen is the English take the words (right) out of someone’s mouth. In contrast, the textual level, pertinent to translation, considers all the contextual, cotextual and even extralinguistic cues in the pursuit of equivalence, identifying various potential expressions whose adequacy depends on the specific situation. Consequently, the equivalence of a particular phraseological unit in a text may deviate from the systemic equivalence of this unit. Similarly, the same phraseme may correspond to different equivalents in different texts.

In addition to these two levels, Mellado Blanco (2015), following Hessky (1987), distinguishes a third level – the lexicographic level – traditionally subsumed by the systemic level. According to Mellado Blanco (2015), the level of dictionaries is a level on its own situated between the systemic and the textual levels. This level should take the communicative function of the phraseme as its starting point, with a special focus on the functional equivalence and the communicative effect of the unit. It follows that lexicographic equivalence “debe obtenerse mediante el análisis del comportamiento prototípico de las unidades fraseológicas en el nivel del discurso, para lo cual es esencial saber discriminar los usos típicos de los periféricos o poco representativos”² (Mellado Blanco 2015: 155). In accordance with our approach, we will not delve into the textual level, which also encompasses translation strategies, but will focus on the systemic and lexicographic levels.

3.2 Models of interlinguistic phraseological equivalence

Several models have been proposed to identify the parameters involved in interlinguistic phraseo-logical equivalence. While these models may differ in terms of the number and types of parameters, most of them cover lexical, semantic, morphosyntactic, and pragmatic aspects. The lexical and semantic traits involve phraseological meaning, imagery and lexical components. The syntactic function and potential syntactic transformations are included within the morphosyntactic features. Finally, the pragmatic dimension encompasses genre, speakers’ features, modifications, and pragmatic functions, among others.

While all these parameters of comparison are relevant for determining the degree of equi-valence between phrasemes of different languages, their specific role depends on the type of equi-valence being considered in each case. In addition to the well-known distinction between ‘systemic’ and ‘textual’ equivalence, it is also important to note, with Mellado Blanco (2015), that the lexicographic level should not be subsumed within the systemic one. Of the six parameters of equivalence defined in Mellado Blanco’s (2015) model (see below), the last three bear particular significance at the lexicographic level: (1) lexical components, (2) morphosyntactic structure and syntactic structure, (3) image, (4) semantic scope and structure, (5) connotative and pragmatic components, and (6) syntactic and semantic valency. As regards the systemic level, one of the most traditional and influential models of equivalence distinguishes three main groups: full equivalence – also known as complete, absolute or total equivalence – partial equivalence, and null equivalence or zero equivalence (cf. Korhonen 2007). Corpas Pastor (2003) situates these categories on a continuum and argues that total equivalence is attained when phrasemes share denotative and connotative meanings, metaphorical bases, distribution and frequency of use, conventional implicatures, pragmatic load, and connotations, for instance, Ger. Lügen haben kurze Beine and Eng. lies have short legs. However, due to the strict requirements that the units must fulfil, total equivalence is rare. There seems to be consensus that this type of equivalence is difficult to obtain (cf. e.g. Corpas Pastor 2003; Dobrovol’skij 2000, 2011; Fiedler 2007; Mellado Blanco 2015). In fact, even in cases where there is complete similarity in form and meaning, asymmetries in the pragmatic parameter may arise, related to frequency, conditions of use or functional features. Nevertheless, linguistic historical proximity and language contact, as observed in the German-English pair, may favour these equivalents. Typical examples of this group include loan translations, that is, calques, such as Ger. bis an die Zähne bewaffnet sein and Eng. to be armed to the teeth, or Ger. das Kind mit dem Bade ausschütten and Eng. to throw the baby out with the bathwater, and the so-called internationalisms and europeanisms³, such as Ger. Achillesferse, Eng. Achiles’ heel.

Partial equivalence is the group which comprises pairs of phrasemes with the same or nearly the same meaning, but with a divergent lexical and syntactic structure. Examples of this equivalence type include Ger. von Kopf bis Fuß and Eng. from head to toe. Corpas Pastor (2000) argues that this is the prevailing type of equivalence, and she highlights the diversity of differences revealed by the contrast of phrasemes. In her view, the divergence may range from differing metaphorical basis, frequency of use and dialectal/geographical restrictions to asymmetries in the semantic scope of the units. It should be noted, however, that Corpas Pastor’s concept of partial equivalence includes a type of equivalence which other scholars classify into a separate category, i.e. the so-called phraseological parallels (cf. Dobrovols’kij 2011). Phrasemes in this category share a core meaning but differ in the underlying image. The units Ger. zwei Fliegen mit einer Klappe schlagen / Eng. to kill two birds with one stone, Ger. aus einer Mücke einen Elefanten machen / Eng. to make a mountain out of a molehill, and Ger. um den heißen Brei herumreden / Eng. to beat around the bush illustrate this group.

All the phrasemes mentioned above share the same meaning and have a common phraseo-logical nature. However, there are instances in which phraseological expressions do not find corresponding phraseological units in the other language, as exemplified by Ger. eine Extrawurst braten, which means ‘to give special treatment’. These cases of zero equivalence have been termed “phraseological gaps” or “lacunary units” (Szerszunowicz 2013). Although socio-cultural aspects embedded in the phraseological unit are sometimes responsible for this lack of interlinguistic equivalence, these elements do not invariably cause null equivalence. For instance, consider Eng. to carry coals to Newcastle, which might be erroneously assumed to lack equivalence in German due to the image component of the lexical constituents. The German unit Eulen nach Athen tragen is a phraseological parallel of the English idiomatic expression. According to Fiedler (2007: 118), examples of null equivalence between German and English phraseology are uncommon due to the status of English as a lingua franca and the high degree of language contact between them⁴.

In addition to these four categories, instances of cross linguistic phraseological homonymy give rise to phraseological false friends. These are defined as “two or more identical expressions that evoke almost identical or very similar mental images but show significant differences in the actual meaning” (Dobrovolskij / Piirainen 2005: 109), as seen in Ger. Kopf hoch, which is an “aufmunternder, ermutigender Zuspruch bei Niederlagen, Enttäuschungen und in schwierigen Situationen” (“Kopf hoch”, provided by Digitales Wörterbuch der deutschen Sprache⁵, and Eng. heads up, which in one of its senses is defined by the Cambridge Dictionary as “a warning that something is going to happen, usually so that you can prepare for it” (CDO; s.v.). Such false friends need to be classified separately from all the types of equivalence described above.

4.1 The role of equivalence in translation and language teaching

As stated in section 3., most of the research on inter-linguistic equivalence has originated within contrastive linguistics and translation, where the textual level is both the starting point and the end point. In contrast, in phraseodidactics, the levels under consideration are mainly the systemic and the lexico-graphic level.

In translation studies, the emphasis has always been on the similarity of the linguistic units, given the goal of achieving the highest degree of equivalence in terms of communicative function. However, the field of language teaching has traditionally approached contrastive analysis differently. Here, the focus has not been on determining similarities but rather on exploring divergences as the latter reveal aspects that demand careful consideration during the processes of teaching and learning a second language (L2). The inability to foresee all potential areas of interference between languages and the uneven relationship between differences and difficulties lead to the development of Error Analysis. This method takes students’ errors as a starting point to predict possible problems for learners. Nonetheless, other methods such as performance analysis and interlanguage studies direct attention towards the learning process, observing not only errors, but also instances of non-errors.

Still, errors have had a strong influence on the creation and evolution of teaching methods, which have tended to place more emphasis on warning about the differences than on highlighting and capitalizing on interlinguistic similarities. Accordingly, until the 20th century, language transfer was often viewed as a negative phenomenon (cf. Jarvis / Pavlenko 2008), and so has been interference. A consequence of this has been the avoidance of the learner’s native language (L1) in the language classroom, with native speaker competence being established as the ultimate goal in L2 and FL learning. To avoid the negative connotations associated with the term ‘language transfer’, some researchers (see Jarvis / Pavlenko 2008) opt for using ‘cross-linguistic influence’. However, both terms are often used interchangeably and coexist with ‘interference’, which specifically denotes negative cross-linguistic influence.

4.2 The role of equivalence in phraseodidactics

Several scholars have highlighted the value of contrastive analysis in L2 phraseology teaching. Penadés Martínez (1999), for instance, underlines the advantages offered by contrasting mother tongue and L2 phrasemes as a useful approach to enhance phraseological comprehension, memorization and use. Beyond the comparative aspect, attention has naturally been directed towards the judicious selection of phraseological units and the development of instructional materials. In this regard, Dobrovols’kij (2011) has stressed the need for learners to have access to the functional meaning of phrasemes. Providing more specific insights into selection criteria, Korhonen, like Penadés Martínez, highlights the significance of considering the learners’ mother tongue in this process:

Zu den wichtigsten Fragen in diesem Zusammenhang gehören die Selektion und die didaktische Aufbereitung des Lernmaterials. So erscheint es zweckmäßig, bei der Selektion nicht nur die Geläufigkeit der fremdsprachlichen Phraseologismen, sondern auch die Muttersprache der Lerner zu berücksichtigen: Das Material sollte Beispiele für alle Äquivalenztypen einschließlich falscher Freunde enthalten. (Korhonen 2007: 585)⁶

Korhonen argues for including all forms of equivalence in teaching materials, and he goes further to suggest that “[d]a totale interlinguale Äquivalente geringere Lernschwierigkeiten bereiten und kaum Interferenzfehler verursachen, wäre die Aufmerksamkeit eher auf die übrigen Äquivalenztypen zu lenken” (585)⁷. The idea of focusing on potential interference areas goes hand in hand with the belief that divergence should receive more attention in the L2 classroom. Nevertheless, acknowledging linguistic transfer as not merely a communicative strategy but also a learning strategy (cf. Ellis 1994), we maintain that all forms of phraseological equivalence should be addressed in L2 teaching. The identification of shared phraseological traits in German and English holds the potential to facilitate the acquisition of L2 phraseology.

From this follows that transfer, far from being just a source of errors, can also be an advantage in L2 learning (cf. Jarvis / Pavlenko 2008). Besides, it should be noted that cross-linguistic influence is not unidirectional: it does not operate solely from the L1 to the L2, and there are other forms of transfer that play important roles in the learning process, such as L2 to L1 (reverse transfer), and L2 to L3 (lateral transfer).

As a project on tertiary language teaching, and more specifically on the “German-after-English” configuration, the PhraseoLab project is mainly interested in exploiting the potentialities of transfer from L2 to L3. As Neuner (2004) explains, the recognition of transferable elements between L2 and L3 can be used as a facilitating factor in the development of L2 competence. It is essential to emphasise that the pedagogical potential of transferable elements extends beyond the teaching-learning of those elements which are similar in L2/L3. The area of crossovers or bridges created in the learner’s knowledge of the two languages can be used as a basis for a later expansion towards contrasting elements. As Neuner (2004) explains, the area of understanding in L3 developed through the activation of transferable elements “can be further extended by adding on and integrating both what is recognisable and known and what is different and contrary” (Neuner 2004: 25). This has implications not only for the selection but also for the grading of learning materials. In the PhraseoLab project, the degree of equivalence between phraseological units in German (L3) and English (L2) provides a key criterion for the classification of phrasemes into levels of language competence.

Following an adapted version of Korhonen’s (2007) model of phraseological equivalence, PhraseoLab establishes a basic distinction between phrasemes which have full equivalence in L2, on the one hand, and those which have partial equivalence in L2, on the other. The former are introduced in exercises and activities corresponding to A2 level, and the latter are introduced at the B1 level. In total, the database will cover 475–480 multi-word units in L3 (German) plus corresponding full or partial equivalents in L2 (English). The methodology for the selection of these units requires a complex combination of qualitative and quantitative criteria. Qualitative criteria are, of course, indispensable for identifying equivalence relations at various levels of systemic analysis, but the need to take into account the role of frequency makes it necessary to complement qualitative analysis with quantitative measurements. In the following section, we will turn our attention to tools and methods used for this purpose.

5.1 Frequency in phraseodidactics

In the field of phraseodidactics, studies of frequency have three main functions. Firstly, they provide a valuable criterion for the selection of the most useful phraseological material for learners, as various authors have explained (with reference to the teaching of German as a foreign language, see, for instance, Jesenšek 2006 and Hallsteindóttir 2020). Secondly, frequency must be taken into account as one of the parameters of equivalence. Various authors have pointed in this direction. In Farø (2006), frequency is listed as one of the linguistically relevant aspects of idiom equivalence. Corpas Pastor (2000) explains that one of the conditions for full equivalence among phraseological units concerns frequency of use and distribution. Accordingly, phraseological units that differ in this respect are only partially equivalent. In Mellado Blanco (2015), frequency is classified as one of the components of the pragmatic parameter of equivalence. All these remarks underline the importance of frequency as an aspect of phraseological equivalence. This function of frequency is particularly relevant from the perspective of a plurilingual approach, given that, in this framework, the pedagogical value of L2 phrasemes is relative to the degree of comparability with L3 counterparts. The implication is that, given various pairs of L2/L3 phrasemes with equivalent semantic and morphosyntactic properties, those which are also comparable in terms of frequency – in the sense that they are in the same frequency range in L2 and L3 – will be more suitable candidates for building transfer bridges than other phrasemes which are semantically and grammatically similar but have substantially different frequencies of use.

Thirdly, frequency can be considered as one of the necessary criteria for identifying certain types of phraseological units and demarcating them from non-phraseological (free) combinations. Admittedly, this function of frequency has been disputed. The study of frequency has not been accorded the same weight in all approaches to phraseology. This can be explained with reference to the well-known distinction between a linguistic top-down approach to phraseology and a distributional bottom-up approach (for a detailed explanation of the distinction, see Granger / Paquot 2008). In particular, the literature on collocation is a neat reflection of this divide. It is already a commonplace that collocational research can be divided into two main traditions. One is the Firthian tradition, which is characterised by a frequency-based approach to the definition of collocation (cf. Sinclair 1991). The other tradition has been developed by Hausmann (1985, 2006), Mel’čuk (1998) and Benson (1989), among others, and its approach to collocation has been described as a significance-oriented approach (cf. Herbst 1996) or phraseological approach (cf. Nesselhauf 2005; Orlandi / Giacomini 2016a), among other names (for a detailed explanation of the differences between the two traditions, see Herbst 1996 and Corpas Pastor 2001, among others).

Various studies have emphasised the pedagogical orientation of the phraseological approach to collocation and its applications in the field of foreign language teaching and pedagogical lexico-graphy (cf. Benson 1989; Herbst 1996; Hausmann 1998). The description of collocations as asymmetrical structures, consisting of a lexically autonomous item (the base) and a dependent one (the collocator or collocate), is motivated by the observation of special encoding difficulties encountered by learners. The selection of the collocator is phraseologically bound by the base, and speakers need familiarity with the lexical context in order to find the acceptable expression. This can be illustrated with the following example of a support verb construction: Spanish-speaking learners of German need to learn that, in combination with Enscheidung, the support verb is treffen and not nehmen (compare: Sp. tomar/?encontrar una decisión, Ger. eine Entscheidung treffen/?nehmen). As Hausmann (2006), Mel’čuk (1998) and other authors have explained, support verb constructions belong to the realm of collocation.

Traditionally, the phraseological approach to collocation has been primarily informed by qualitative parameters of analysis. In fact, the division between quantitative and qualitative approaches to the definition of collocation has been at the heart of an ongoing debate. Hausmann (1985) criticised the Firthian or British contextualist tradition of collocational research for its reliance on the frequency criterion, which, he contended, is insufficient to distinguish collocations from other types of word co-occurrences. A similar criticism of the Firthian concept of collocation has been raised by Alonso Ramos (1994–1995) and Írsula Peña (1994), among others. More recently, this criticism has been relativ-ised. In reference to a collocation dictionary compiled according to a corpus-driven approach, Hausmann (2004) remarked the following: “Er [the Dictionary of English Collocations, by Kjellmer] ist keineswegs uninteressant, denn er liefert Informationen über die Syntagmatik von Einheiten aller Art” (Hausmann 2004: 320)⁸. In this paper, Hausmann suggests that the controversy between the two concepts is essentially ‘a terminological battle’ (“ein Terminologiekrieg”), the underlying discrepancy being which of the two traditions can claim the term collocation for itself. Rather than the validity of a particular concept, the real bone of contention is the (scientific) legitimacy to use a specific term: “der Krieg um die Besetzung des linguistischen Terminus Kollokation” (Hausmann 2004: 320)⁹. Both concepts of collocation inform useful research but with fundamentally different orientation. Nevertheless, the need to use corpora is part of their common ground. Ultimately, both approaches are interested in uncovering the full extent of the phraseological patterning of language, as reflected in idiosyncratic/item-specific combinatory preferences. Corpus linguistic tools are indispensable for this purpose. Thus, despite the aforementioned discrepancies, the following quotation from Hausmann reflects a point of view which is ultimately compatible with the two concepts of collocation: “[D]ie Zukunft gehört den Corpora. Sie lehren uns die Idiomatizität von Sprache, an der die Kollokationen als Bineme, Tripel und Quadrupel teilhaben, mit und ohne zusätzliche grammatische Beschränkungen” (Hausmann 2004: 322)¹⁰. The tendency to bridge the gap between the two approaches and to treat them as complementary to one another has been reinforced by a number of studies that combine qualitative and quantitative (corpus-based) techniques in the description of collocations. There are, in fact, compelling reasons to argue that the phraseological approach to collocation can be enriched with frequency-based parameters of analysis. Koike (2001) emphasises that high co-occurrence frequency is one of the fundamental characteristics of lexical collocations, even though this property is shared with other types of combinations. Thus, even if we concede that co-occurrence frequency is not a sufficient criterion for discriminating collocations and other word co-occurrences, this does not imply the need to exclude frequency criteria from the set of necessary conditions. It is important to underline that some aspects of the formal structure of phrasemes constitute measurable properties. Combinatory predictability, in particular, is a dimension of collocation which can be measured empirically. Thus, it can be argued that a strong syntagmatic association is a necessary but not sufficient condition for collocational status.

A word of caution is needed here: co-occurrence frequencies are necessary to measure the strength of syntagmatic association between the component parts of a collocation, but they do not represent per se an accurate measurement of such strength. Two words may co-occur frequently because they are high-frequency items and not necessarily because their combinatory behaviour shows a strong preference for occurring together. Thus, co-occurrence frequencies are most useful when they are operationalised in lexical association measures. Möhring (2011) provides an interesting example of how lexical association scores (log-likelihood, in his case) can be interpreted in the framework of a phraseological approach to collocation. The role assigned to these scores is, of course, different from the one they have in the Sinclairian approach. In the phraseological approach, lexical association measures cannot provide the decisive criterion for classifying co-occurrences as collocations, because other types of phrasemes and also free combinations can show very high scores too. However, with the appropriate analytical framework, association measures can give us relevant information about the behaviour of collocators. For instance, if we compare the scores of near-synonymous collocators in combination with the same base, it is possible to provide empirical evidence for the existence of lexic-al preferences.

5.2 Collocation statistics and equivalence from a plurilingual perspective

As stated above, similarities in frequency of use are an integral part of the parameters of equivalence among phrasemes in different languages. In the case of collocations, this parameter has particular relevance, given the role of the frequency criterion as one of the necessary conditions that a word combination must fulfil to be classified as a collocation (see section 5.1). In some cases, this criterion has a decisive impact on the identification of phraseological equivalents. Not all the word combinations which are semantically and grammatically equivalent in a given language pair (L1/L2, L2/L3) can be categorised as collocations in the two languages. In fact, one of the problems that have been observed in the construction of the PhraseoLab database is that L3 word combinations with a high degree of lexical and semantic equivalence with L2 counterparts are not always those with the strongest level of lexical association score in L3.

To illustrate this problem, we will undertake a comparison of specific collocational pairs in two similar corpora of German and English: the DWDS Gegenwartskorpora mit freiem Zugang and the British National Corpus (BNC). The DWDS Gegenwartskorpora is formed by a collection of corpora including various types of written texts and a spoken component made of transcripts. The structure of the corpus is described at the following website: https://www.dwds.de/d/korpora/dwdsxl. One of the main components of the corpus, the German reference corpus of the twentieth century, has been described in Geyken (2007). Like the DWDS, the British National Corpus (BNC) encompasses a variety of written text types as well as a spoken component. The website used to access the BNC in this study has been English-Corpora¹¹.

The collocations compared are Ger. Frieden schließen and Eng. make peace. The meaning of schließen in this collocation matches the definition of sense 8. in the DWDS online dictionary: “etwas eingehen, abschließen” (“schließen”, provided by Digitales Wörterbuch der deutschen Sprache)¹². This entry provides specific paraphrases for each of the two main senses of the collocation Frieden schließen. The first one refers to the achievement of reconciliation after a conflict, quarrel or argument: “sich nach einem Streit versöhnen” (“schließen”, provided by Digitales Wörterbuch der deutschen Sprache)¹³. We will refer to this usage as the ‘reconciliation’ or ‘conflict resolution’ sense of this collocation. The type of conflict referred to by this usage can be understood in the broadest possible sense. It may range from arguments with other people (relatives, friends, colleagues, …) to conflicts between institutions or companies and even psychological conflicts with oneself or intel-lectual struggles with the world and with reality. In an even more abstract sense, it can also refer to the action of restoring harmony. Examples of this usage of the collocation in the DWDS Gegenwartskorpora mit freien Zugang are given in examples (1)-(3). The second sense of the collocation refers to the action of putting an end to war by means of a treaty: “einen Kriegszustand durch einen Friedensvertrag beenden” (“schließen”, provided by Digitales Wörterbuch der deutschen Sprache)¹⁴. We will refer to this second sense as the ‘peace agreement’ or ‘end of violence’ sense of this collocation. This usage of the collocation refers to an agreement to put an end to war or to a violent conflict in a given country or area or between particular groups of people. Examples of the ‘peace agreement’ sense are given in examples (4)-(6) (from Gegenwartskorpora mit freiem Zugang)¹⁵:

In English, the combination make + peace can be used in expressions which convey these two senses. Examples (7)-(9) illustrate the ‘reconciliation’ or ‘restoration of harmony’ sense. The grammatical realisation of this phraseme often includes the use of a possessive determiner, and it can also be followed by a with-headed prepositional phrase, but these elements are optional. The ODE uses the following citation form: make (one’s) peace, and it gives the following definition: “re-establish friendly relations”. Examples (10)-(12) illustrate the use of a lexical collocation with the ‘peace agreement’ sense. All the examples have been obtained from the BNC.

In German, there are two expressions containing the verb machen which are formally and semantically very similar to the aforementioned expressions with the verb make in English. The phraseme seinen Frieden mit jemandem/etwas machen has a very high degree of equivalence with Eng. to make (one’s) peace (with). In addition to the correspondences between the lexical constituents, the possessive pronoun and the prepositional complement with mit/with reinforce their structural similarity. Semantically, these two phrasemes are also comparable. The DWDS defines the German phraseme as follows: “sich mit jmdm. verständigen, versöhnen; sich mit einer schwierigen Situation abfinden, sie akzeptieren” (“seinen Frieden mit jmdm., etw. machen”, provided by Digitales Wörterbuch der deutschen Sprache)¹⁶. Thus, the sense of reconciliation or restoration of harmony described above for the English expression is also conveyed by the German equivalent (see examples (13)-(15), from Gegenwartskorpora mit freiem Zugang¹⁷. What is interesting to note, from an applied contrastive perspective, is that the combination Frieden machen is also used with the ‘peace agreement’ sense (16–18), but the as-sociation strength of this combination is lower than Frieden schließen and, as a collocation, it does not have the same cohesive force.

Considered from a lexical, semantic and grammatical perspective, Frieden machen/make peace are closer equivalents than Frieden schließen/make peace. The semantic range of the two lexical constituents, the verbs machen/make and the nouns Frieden/peace, are very similar. If we did not take frequency-based information into account, the collocation Frieden machen could be introduced into the earliest level of the PhraseoLab activities (A2) as a result of the existence of an L2 counterpart with full equivalence. However, the results obtained from collocation statistics indicate that Frieden machen (with the ‘peace agreement’ sense) is a weaker collocational pattern than Frieden schließen.

Table 1 displays lexical association scores of these two collocations in a modern sample of the DWDS Gegenwartskorpora. To facilitate comparability with the BNC, only texts from 1980 to 1993 have been selected (the DWDS corpora mentioned cover a broader time span than the BNC, but it is possible to adjust the query to a subset of texts covering a more restricted period). The search entered was: “NEAR(Frieden,schließen,2)” and “NEAR(Frieden,machen,2)”, respectively. This query allows us to obtain instances of various grammatical realisations of the same lexical collocation, such as Frieden schließen, X schließt Frieden, Frieden zu schließen…; X haben den Frieden gemacht, X macht noch keinen Frieden, etc. Manual supervision of the concordances was required in order to filter out unwanted combinations, i.e. those cases in which the words Frieden and schließen/machen co-occur but not as components of the said collocations. The following are examples of concordances that had to be excluded: ein wesentlicher Schritt weg von einem fortdauernden Bürgerkrieg hin zu Frieden gemacht wird; den Frieden sicherer zu machen; was den Frieden stabiler macht; … sowie deren Bedeutung für den Frieden bewußt gemacht werden, etc. The data of joint occurrence displayed in Table 1 – see column F(B,C) – indicates the number of times the base and the collocator have been found to occur together after their non-collocational co-occurrences have been filtered out. F(B) and F(C) indicate the individual word frequency of base and collocator, respectively. This count includes all forms of each lemma in the corpus: Frieden, Friedens; schließen, schließt, schloss…; machen, mache, macht… The association measure used is logDice (cf. Rychlý 2008). With this measure, the maximum score is 14, and any negative value is interpreted as indicating no statistical significance of the combination. The data indicate that the noun Frieden has a stronger preference for schließen than for machen.

In contrast, in English the preferred collocator is make and not conclude. The frequency data in Table 2 have been obtained from the BNC (again, accessed through English-Corpora.org)¹⁸. The search was made using the “Collocates” function and entering PEACE_n as “word” and CONCLUDE_v and MAKE_v as “collocates”. The collocational window was set to L2:R2 (two items to the left and to the right side). As in the German counterpart, this query yields various grammatical realisations of the same lexical collocation (make my peace with…; making peace; peace was made…), but at the same time, manual supervision is required in order to filter out co-occurrences of the same words outside of the structure of the collocation. Some examples of excluded concordances are the following: Examine the diamonds in peace, make your escape; some progress is made in peace talks…; Green Peace are making a formal complaint…, etc. Examples in which peace occurs as modifier and not as head of the object noun phrase were also excluded (e.g. “In 1979 Mauritania concluded a peace agreement with Polisario”; “…the importance of concluding a peace treaty speedily”). The data under F(B,C) shows the joint frequency of the collocation in the corpus after a manual supervision of concordances. The criteria for filtering co-occurrences were informed by the phraseological approach to collocation described above.

The comparison with conclude peace is motivated by the close similarity with one of the two main usages of Frieden schließen, namely, the ‘peace agreement’ sense. The meaning of the verb conclude in this collocation is closely equivalent to the meaning of schließen in the German counterpart. The Oxford Dictionary of English (ODE) offers the following definition for one of senses of conclude: “formally and finally settle or arrange (an agreement)” (s.v.), and it gives a combination with ceasefire as an example: “an attempt to conclude a ceasefire”. This definition shows a high degree of comparability with the semantic contribution of the verb schließen, “etwas eingehen, abschließen” (“schließen”, provided by Digitales Wörterbuch der deutschen Sprache)²⁰, in the German collocation Frieden schließen. However, in the BNC, the as-sociation strength of conclude + peace is remarkably lower. In fact, the raw frequency (2) barely reaches a frequency threshold. The two examples (19–20) are instantiations of the ‘peace agreement’ sense.

These results suggest that while Frieden schließen has a stronger degree of syntagmatic association than Frieden machen, in English the collocation make peace is stronger than conclude peace (based on data from the BNC). These results dovetail with existing lexicographic information about English collocations. Currently, the verb conclude is not included in the dictionary entry for the noun peace in any of the three main collocation dictionaries of English: LCDT, MCD and OCD. In contrast, the same three dictionaries record the collocation of make with the aforementioned meaning of peace: ‘not war’ (OCD), ‘a situation in which there is no war’ (MCD), ‘a situation in which there is no war or fighting’ (LCDT). In the FWD, both schließen and machen are included in the entry for Frieden, but schließen is marked as especially typical, while machen is not. In terms of register, machen is marked as informal (an analysis of the distribution of these collocations across language varieties was beyond the scope of this study).

In order to obtain a more accurate picture of this comparison, it will be interesting to determine how frequently each of these expressions is used with the ‘peace agreement’ sense. Tables 1 and 2 conflate all the senses of these combinations, but we have observed that the phrasemes make (one’s) peace with sb./sth. and seinen Frieden machen mit jmdm./etwas have distinct features from the lexical collocations Frieden schließen/make peace. Table 3 shows the co-occurrence frequency of Frieden + machen and Frieden + schließen in those cases in which the expression conveys the ‘peace agreement’ sense (thus, excluding occurrences of the phraseme seinen Frieden mit jmdm./etwas machen). The concordances were manually classified according to the sense of the collocation (ideally, an additional comparison should be made with scores that take into account the frequency of different meanings of Frieden/peace, since different word meanings may have different frequencies and may attract dif-ferent collocators, but the scale of the sense disambiguation task required for this purpose lies beyond the scope of the present study). Table 4 shows the results of the same process applied to the English expressions to make/conclude peace. Only those instances in which the combination conveys the ‘peace agreement’ sense were counted. Of course, this requirement was added to the grammatical condition described above: only combinations with Frieden/peace as object head noun with these verbs were taken into account. Again, the results suggest that, once frequency-based criteria are taken into account, Frieden schließen can be identified as the closest equivalent of make peace. In Table 3, we observe that Frieden has stronger preference for schließen over machen as a means of expressing the ‘peace agreement’ sense. In Table 4, the association scores are not significantly different, but we must bear in mind that, with only two occurrences, the combination with conclude is barely above the minimum frequency threshold.

With the analysis of these examples, our aim has been to illustrate how a corpus-based analysis of frequency data can provide us with information which complements a purely qualitative analysis of equivalence. The strongest collocations in a given language pair, such as German (L3) and English (L2) in this case, are not always formed by those combinations which have the closest degree of lexical and semantic equivalence in the two languages. Despite the partial semantic equivalence between schließen and conclude, the combination conclude peace is too weak to provide a good candidate for phraseological equivalence with Frieden schließen. Conversely, the semantic range of make in English is substantially different – significantly broader – than the semantic range of schließen in German, but the collocation make peace provides a stronger collocational counterpart for Frieden schließen.

Nevertheless, there are two questions that remain open and that deserve closer inspection in later research. The first one has to do with the possible influence of grammatical factors in the combinatory profile of conclude. The question that emerges in this respect is whether the weaker association of conclude + peace in the BNC is related to the preference of conclude for countable nouns. A positive answer would imply that the differences between Frieden schließen and conclude peace are not solely determined by lexical preferences and can be derived from grammatical patterning. This question cannot be answered on the basis of the empirical data used in the present study, but in bigger corpora we can find evidence suggesting that conclude can combine regularly with peace in various types of grammatical contexts. Examples (21)-(26) are from the British Web 2007 corpus (ukWaC), accessed through Sketch Engine²².

In the ukWaC corpus, the combination conclude + peace (in verb-object relation) yields a logDice score of 6.7 (the score is calculated automatically by the Word Sketch tool in the Sketch Engine platform), which in fact is higher than the score of make + peace (4.8) in the same corpus. This leads us to the next problematical question: corpus representativeness is still an unresolved issue, especially when dealing with general corpora (for a recent overview of issues involved in attaining representativeness in corpus design, see Kupietz 2016; for a discussion of the limits of corpus representativeness in relation to the extraction of collocational data, see Ďurčo 2010). The same collocation may attain very different association scores in different corpora, especially if these differ significantly in their design. The BNC and ukWaC are general corpora with very different characteristics (for a comparison, see Ferraresi et al. 2008), and it is not a surprise that the same collocational pair can be ranked differently depending on which corpus we use. The results shown in Table 1, 2, 3, 4 could present a dif-ferent picture if they were based on two web corpora instead of the DWDS corpora and the BNC. Logically, the data from the ukWaC corpus should not be compared with those of the DWDS, because their characteristics are not comparable (ukWaC is a web corpus, i.e. it consists entirely of texts collected from the Internet), but a comparison of the same collocational pairs in two web corpora of German and English would be possible and would probably lead to different results.