Italian Phrasemes as Constructions: How to Understand and Use Them

Article history Received: 07-05-2015 Revised: 17-06-2015 Accepted: 17-06-2015 Abstract: The present article describes the objectives and methods for a learner-centered description of Italian idioms based on the theoretical principles of Construction Grammar (CxG). The aim of the underlying research project is to develop a new way of looking at idioms, taking into account all linguistic aspects that could help to fully understand and usethem in a formally and functionally adequate manner, including situational and discursive features. By phrasemes we understand different kinds of word combinations characterized by idiomaticity and/or entrenchment. I will focus here on (a) “predicative phrasemes” (typically figurative and containing an inflected verb as a predicate, also called expressions idiomatiques verbales), (b) “phraseotemplates” (lexically open or formal idioms, in German phraseology Phraseoschablonen) and (c) “pragmatemes” (according to the French term pragmatème, i.e. pragmatically highly conventionalized phrases, also called expressions-énoncés). Idiomaticity is characterized by non compositionality of its components and unpredictability of the whole structure. Italian examples for each of the three types are: (a) Tenere il piede in due staffe (‘to run with the hare and hunt with the hounds’), (b) Cosa me lo chiedi a fare? (‘Why are you asking me at all?’) and (c) In bocca al lupo! (‘Break a leg’). In chapter 2 I will first discuss basic concepts of Construction Grammar and Cognitive Grammar. In the following part (chapter 3) I will present my ideas about CxG and its ability to create a greater awareness of how many word combinations (in this context called constructions) in a language are idiosyncratic and unpredictable and about how to use fundamental theoretical issues of usage-based CxG (mainly Goldberg, Croft) and unification-based approaches in order to describe idioms in a new holistic way. Chapter 4 will deal with some important classifications of phrasemes and their practical aptitude for phraseodidactics. Chapter 5 will finally present my idea of applying methods of Construction Grammar and Fillmore’s semantics of understanding to build a new digital lexicographical format for phrasemes which is going to be called phraseoframe. This will be illustrated for the three types of phrasemes mentioned above. Each phraseme will be described by means of a simple meta language which is easy to understand and has links to prosodic, morphological, syntactic, semantic and pragmatic and discourse functional properties. As my approach is corpus based empirical evidence will be given for each of the three types of idioms by using several corpora of Italian spoken and written language (e.g., BADIP, PAISÀ and WEBBIT).


Introduction
The main purpose of the present paper is to sketch out an approach to the learner-centred treatment of phrasemes within the framework of Construction Grammar. I will first sum up similarities and differences between the concepts 'symbolic unit', 'construction' and 'phraseme' and define the latter (chapter 2). In the third part, I will develop my phraseoframe model, mainly based on U-semantics and Construction Grammar, whose primary objective is to describe phrasemes in a comprehensive and learnerappropriate way by transposing the modified attributevalue method of Frame semantics and unification-based Construction Grammars to my purposes. Chapter 4 will deal with various classifications of phrasemes and will propose my own typology. I will finally (chapter 5) apply the theoretical issues discussed in the preceding part to three types of phrasemes: Predicative phrasemes (attaccare un bottone), phraseotemplates (the sì (che) construction) and pragmatemes (a buon rendere, e poicolpo di scena), using the phraseoframe schema.

Symbolic Units, Constructions and Phrasemes
The similarities between central theoretical issues in Cognitive Grammar (CG), Construction Grammar (CxG) and phraseology have recently been pointed out by Gries (2008). They can be summarized by the author's claim that phrasemes (or, in his terminology, phraseologisms) as well as constructions are symbolic units (which is the basic term in CG). A symbolic unit is, according to Langacker (1990, 16), "'bipolar', consisting of a semantic unit defining one pole and a phonological unit defining the other". A construction follows the same theoretical principles: Each construction will be a form-meaning pair (F, M), where F is a set of conditions on syntactic and phonological form and M is a set of conditions on meaning and use (Lakoff, 1987, 467).
The concepts 'symbolic unit' and 'construction' are indeed very similar 1 , if we part from Lakoff's definition of construction, which does not include noncompositionality as one of its integral features. This is different in Goldberg's early definition of construction (1995), quoted here from one of her later articles, according to which: [a]ny linguistic pattern is recognized as a construction as long as some aspect of its form or function is not strictly predictable form its component parts or from other constructions recognized to exist (Goldberg, 2003, 219).
In comparison to her previous opinion Goldberg then made a somewhat surprising alteration to her definition, bringing into play the criterion of frequency and combining it with the feature of nonidiosyncrasy, thus abandoning the original defining criterion of non-compositionality: In addition, many constructionalist approaches argue that patterns are stored even if they are fully predictable as long as they occur with sufficient frequency (Goldberg, 2003, 219f.).
In her second book about constructions this amendment is formulated even more clearly: " […] Patterns are stored as constructions even if they are fully predictable as long as they occur with sufficient frequency" (Goldberg, 2006, 5). Thus Gries (2008) is right to postulate that a construction, which, all in all, he considers a synonym for symbolic unit, "must have occurred frequently enough for it to be entrenched in a speaker/hearer's linguistic system" (Gries, 2008, 13). This means that frequency has been added subsequently as one of the prerequisites for the existence of a construction, which is certainly due to the fact that idiosyncrasy (or non-predictability) is not only caused by non-compositionality 2 (idiomaticity) but also by entrenchment (which, in Gries' opinion, is the result of high frequency).
Here, I would like to add two observations: First, a symbolic unit does not necessarily have to be frequent. Defining a morpheme, lexeme (simple or complex) or a combination of lexemes as a symbolic unit only means that the signifiant and the signifié of this unit (that we might also call linguistic sign) are, by convention, inextricably linked to one another, as Saussure's sheet-of-paper metaphor so aptly described it (CGL,246). Second, entrenchment does not only have to do with frequency, it may also be due to a fixed or even strange word order (recalling Fillmore, Kay and O'Connor's 'familiar and unfamiliar pieces unfamiliarly arranged') or to a situational restriction (e.g., formulae that refer to one specific situation).
What we ought to bear in mind, however, is that all three concepts, 'symbolic unit', 'construction' and 'phraseme', do not exclude one another but are, on the contrary, related by irreversible inclusion: Symbolic units include constructions which include phrasemes (cf. Gries, 2008). The main difference lies in the treatment of 2 Non-compositionality is discussed in Svensson (2008).
lexically unspecified patterns such as abstract grammatical patterns (e.g., argument structures or word order like SVO), which are definitely part of the first concept (symbolic unit), not necessarily part of the second (construction), depending on the particular constructionalist approach and in no case part of the third concept (phraseme). As far as noncompositionality is concerned, it is irrelevant for CG and its symbolic units, it may be relevant for CxG and its constructions (see the difference between Lakoff (1987;Goldberg, 1995, pass.) and it may also be characteristic of a phraseme (see my definition below). With regard to frequency, there is no unanimity about its being relevant for the definition of phraseme. Gries claims that frequency is crucial, not only for phrasemes, but also, as we discussed above, for constructions and symbolic units. But do constructions really have to be frequent in order to be considered constructions? Is let alone a frequent pattern? What about the construction "afraid/worried /concerned/anxious + lest + subject + subjunctive verb form" (e.g., She worried lest he should tell someone what had happened, LDOCE5), or the idiom to take the gilt off the gingerbread? I think frequency is optional 3 . It is rather entrenchment (in a broader sense) that counts, on the one hand: Entrenchment due to strange or fixed word order, or to pragmatic restrictions. On the other hand, it may be noncompositionality that makes a linguistic pattern something opaque. Both of them mean nonpredictability and thus idiosyncrasy. In the end, Gries' definition of phraseme (phraseologism) as: co-occurrence of a form or a lemma of a lexical item and one or more additional linguistic elements of various kinds which functions as one semantic unit in a clause or sentence and whose frequency of cooccurrence is larger than expected on the basic of chance (Gries, 2008, 6) Is very similar to mine, but more narrow on the one hand and broader on the other. I therefore suggest the following: Any combination of at least two separate elements, i.e., at least one lexical morpheme (word-form or lexeme 4 ) or one grammatical morpheme plus at least one 3 I do in fact very much appreciate usage-based approaches, but I doubt whether "anything that has been encountered often enough to be accessed as a whole [ought to be] considered a construction, even if it exhibits no idiosyncrasy of form and meaning" (Hoffmann and Trousdale, 2013a;2013b, 5). 4 The term lexeme is understood here, to mean 'fundamental unit of the lexicon' (cf. Matthews, 1991;22, or Schafroth, 2014a, 3f.) and not 'free morpheme' in general, used as a generic term for lexical and grammatical morpheme (as in more lexical morpheme or one more grammatical morpheme, can be defined phraseme when the whole construction is either entrenched (in some way) 5 or idiomatic (non-compositional) or both of them 6 . Every phraseme is a construction and every construction is a symbolic unit, but not the other way round 7 .
Reversely, any linguistic pattern consisting of two free morphemes that is neither entrenched nor idiomatic (non-compositional) cannot be called phraseme. Thus compositional (transparent) compounds like motor vehicle are no phrasemes, although they are symbolic units and may be, depending on the theoretical approach, constructions. Red herring is noncompositional (it is no 'red' 'herring'), thus a phraseme (as well as a construction and a symbolic unit). Far and wide, back and forth are phrasemes (and constructions) because they are entrenched, as well as collocations like to lay the Derivations like driver or uncomfortable are symbolic units and maybe constructions, too, still depending on the theoretical framework, but they are no phrasemes. A derivation like womanizer may be qualified as a construction, because it is non-compositional, but it is no phraseme, because it is one (complex) lexeme. The situation is more difficult, however, with lexically open phrasemes, i.e., when the entrenchment is due to a fixed syntactic structure, for instance the (more/better/bigger, etc.) the (better/more/worse, Kortmann, 2014, 88). Complex lexemes are usually derivations and compounds. 5 In some way excludes however mere frequency. As already noted, I do not consider co-occurrences as police and fire brigade or piano and violin constructions (or even phrasemes) only because they have a frequency of occurrence that is greater than expected (cf. Gries, 2011). For a critical comment on frequency as a defining characteristic of a construction see Traugott and Trousdale (2013, 5). 6 Using the term combination (instead of co-occurrence, deriving from corpus linguistics) I want to emphasize that is not the parole level that is decisive but the langue level. This clarification is consistent with the concept of symbolic unit and it is consistent, too, with the basic premise of usage-based Constructions Grammars and also of my approach: That linguistic knowledge is understood as an emergent product of linguistic use (cf. Ziem, 2008, 55). 7 Verbs like It. ricordarsi di qualcosa/qualcuno or Fr. se souvenir de quelqu'un/quelque chose ('to remember something/someone') are indisputably polymorphous constructions, but as the reflexive pronoun se (or mi in the finite form mi ricordo) and the preposition de/di belong to the lexeme they are a matter of argument structure and valency, not of phraseology. etc.), or the famous incredulity-response construction Him be a doctor? (which is X be a Y?). These phraseotemplates, as I call them, are certainly borderline cases-they are more constructions than phrasemes (cf. Stathi, 2011)-but they are included in the definition of phraseme given above 8 .
Having given the necessary definitions of construction and phraseme, I will now suggest a usage-based approach to an adequate linguistic description of phrasemes.

Construction Grammar and the Necessity of a Holistic Description of Phrasemes
As far as the form-meaning paradigm of constructions is concerned Croft specifies his interpretation of meaning: [t]he term meaning is intended to represent all of the CONVENTIONALIZED aspects of a construction's function, which may include not only properties of the situation described by the utterance but also properties of the discourse in which the utterance is found (Croft, 2001, 19).
The complexity of the levels involved, both on the form and the meaning side-here again I would like to recall Saussure 9 -is illustrated in Fig. 1 and will play a vital role in my understanding of phrasemes as constructions to be described as comprehensively as possible. An idiom like fareil terzo grado a qualcuno ('to question, cross-examine someone') is such a form-meaning pair characterized by particular morphological and syntactic properties: The verb fare has to be inflected, qualcuno as well as the grammatical subject are represented by a noun and we are dealing with a ditransitive argument structure (Subj V Obj Obj 2 ) with an agent and recipient (or patient) bearing the feature [+ human] and being grammatically expressed by a 8 According to Gries' definition, phraseotemplates could also be considered phrasemes, as they represent a linguistic unit in terms of a semantic unit (cf. the author's example of the intocausative, which he himself regards a phraseologism; Gries, 2008, 5). Collocations, however, would not be covered by his definition of phraseme. 9 "La langue est encore comparable à une feuille de papier: La pensée est le recto et le son le verso; on ne peut découper le recto sans découper en même temps le verso" (CGL, 246). Quoting Saussure here does not mean that I adhere to the central structuralistic assumption that there is something like "a strictly semantic structure that exists independently of the general cognitive organisation of the human mind" (Peeters, 2000, 4). Moreover, the linguistic sign in Sign-Based CxG "embodies 'at least phonological structure, (morphological) form, syntactic category, semantics and contextual factors, including structure' (Sag, 2012, 71)" (Traugott and Trousdale, 2013, 4). subject and an indirect object. But there are also properties on the meaning side of the pair, referring to semantic and pragmatic contents, if one is inclined to separate them at all 10 , but also to discourse-functional properties specifying the situation in which something is uttered.
At this point it seems appropriate to go back some decades in time in order to remember Fillmore's semantics of understanding (or U-semantics). According to Fillmore: a sentence can only be fully interpretedif we know something about the situation in which it has been used; in many cases, then, understanding a sentence involves knowing the class of situations in which it could be appropriately uttered and knowing what effect it could be expected to have in that situation (Fillmore 1975, 16). In Construction Grammar semantics and pragmatics are not clearly separated (Ziem, 2008, 134ff.). Gazdar (1979, 2) even claims that pragmatics is nothing else than meaning minus truth conditions and this is, as Busse (2009, 72) points out, almost everything semantics is generally interested in: Namely aspects depending on the context, the use and the users of linguistics signs for communicative purposes, things that cannot be explained by logical semantics. Thus semantics could, according to this point of view, be equated with pragmatics. We can recognize the shortcomings of "pure" semantics in the functions of discourse markers like but, well or actually, because these elements do not deal with truth conditions at all but with pragmatic aspects as contrast, insecurity, coherence, politeness or involvement (cf. also Fried and Östman, 2005). As phrasemes have at least as much pragmatic as semantic relevance  if we think of phraseotemplates and pragmatemes they have primarily pragmatic potential -there will be no categorical distinction between semantic and pragmatic features in the phraseoframe (chapter 5).
The consequence of this view is to claim a holistic encyclopaedic approach which reflects the "necessity to study lexical concepts as integral part of human cognition in general" (Peeters, 2000, 4). This perspective has already been adopted by Langacker (1987), one of the founding fathers of Cognitive Linguistics, who insisted that the distinction between linguistic and extra linguistic (encyclopaedic) knowledge is as false and impossible as the distinction between semantics and pragmatics (cf. Stathi, 2011). Although the non-separation of semantics and pragmatics is one of the theoretical principles of usage-based CxG (cf. Ziem and Lasch, 2013, 51) the construction list practice of describing constructions does not always fulfil it. Up to now (Schafroth, 2013a;2013b; 2014b) I have not made this distinction in my phrase frames. Although even authors postulating the holistic paradigm (e.g., Croft, 2001;Fried and Östman, 2004) do not always distinguish between semantics and pragmatics I have recently come to the decision not to separate the semantic from the pragmatic features. I thus fully agree that: a. "semantics is based on speakers' 'construals' of situations, not on objective truth conditions" (Goldberg, 2013, 16) and b. "Semantics, information structure and pragmatics are interrelated" (ib.), And that linguistic knowledge therefore is c. "not merely semantic in the traditional […] sense of that term", but that it also "represents conventional pragmatic or discourse-functional or information-structural or even stylistic or social dimensions of the use of a grammatical form or construction" (Croft, 2001, 93).
The basic assumptions of the holistic paradigm, already outlined in Fillmore, have been substantially developed by the German linguist Dietrich Busse (2012, pass.) in his theoretical framework of interpretative semantics and linguistic epistemology. Busse's theory of an explicative semantics is linked to a communication-theory in which the expressive and the appellative function of linguistic signs are on the same level as the referential function (cf. Ziem, 2008, 151). Busse's postulate of epistemic relevance goes hand in hand with the holistic paradigm: Linguistic understanding always takes place in a conceptual space (cf. Busse, 1988;Ziem, 2008;Croft, 2011) in which a distinction of linguistic and extra linguistic knowledge is not possible. Individuals being part of this space construe their reality by language, drawing on knowledge by inferences during the interpretation of linguistic signs: Knowledge they acquired by previous experiences. As a result, what should be considered in a holistic description of constructions is not only linguistic knowledge but also "how speakers implicitly anchor what they say to extralinguistic factors and how these factors, in turn, constrain the usability of a specific linguistic expression" Östman 2005, 1760) 11 . To sum up these theoretical observations, we can say that: [u]nderstanding [a] word requires knowing the whole scenario; understanding a sentence containing this word requires knowing the scenario and using the lexical contents and grammatical structure of the rest of the sentence to fill in some of the details; understanding a large text containing such a sentence may require situating a scene described by this text as a well-defined part of a larger story or state of affairs (Fillmore, 1976, 28).
Why I call my description of phrasemes (phraseo) frame will be explained in the following part: In Frame semantics, a linguistic unit, here, a word (in just one of its senses) evokes a frame. The frame is the structure of knowledge required for the understanding and appropriate use of lexical items or phrases (Fillmore, 2004, 406).
The phraseoframe is a frame of knowledge representation that tries to register the relevant aspects to be considered in order to describe a phraseme in a holistic manner. "[E]ach frame is associated with descriptions of roles of the entities that participate in them (individuals, props, setting features" (Fillmore, 2012, 284). These frame elements reflect linguistic properties as well as the conceptual links to the frame evoked by a phraseme, summarized as "thesaurus". The format of representation is a frame, literally (typographically) speaking, elaborated for digital lexicographical purposes, as it is currently realized in the learners' platform project (cf. Schafroth, 2014a). In this lexicographical description, I of course concentrate on the representation of epistemically relevant linguistic knowledge about phrasemes (rules of linguistic usage in the broadest sense, including communicative parameters) -I have so far indeed restricted my constructionist 11 Speakers' interpretations refer to contexts like (a) culture, tradition, history, (b) interactional and conversational constraints and (c) feelings and opinions (Fried and Östman, 2005;Östman, 2005, 1760). Other levels and types of knowledge are discussed in Busse (1991;summarized in Ziem, 2008, 154ff.). activities on these-not ignoring, however, extralinguistic knowledge such as cultural or psychological items (expressed by remarks on the particular use of a construction in a particular situation).
The form of representation of a construction's features roughly follows the attribute-value structure as it is used in Construction Grammar, simplifying, however, the notational conventions: Attributes and values are neither formalized in boxes-within-boxes diagrams nor in the form of bracketed pairs organized into sets (attribute-value matrices) (cf. Fried and Östman, 2004) but as a sort of phraseme sketch containing the relevant types of information in prose, distributed on the domains prosody, morphology, syntax, semantics and pragmatics, discourse functions 12 . The attributes indicate the types of knowledge that are or may be relevant for the frame evoked by the phraseme: e.g., within the domain semantics and pragmatics: 'Speech act', 'situational setting', 'motivation'. As the unification of linguistic properties is focussed in this approach, I will follow here the methods of unification-based Constructions Grammars, making a clear difference between internal and external characteristics (internal/external syntax of a construction). On the other hand, the phraseoframe is also usage-based, following the premise that "experience with language creates and impacts the cognitive representations of languages" (Bybee, 2013, 49). As a consequence, if corpus data reveal "idiomatic variation" (Wulff, 2013, 285) the formal and semantic variants will be recorded as well 13 .
In order to give a quick overview of the structure and the elements of the eclectic phraseoframe model we can bear in mind the following items: • Theoretical fundamentals: Fillmore's semantics of understanding, Busse's epistemically relevant semantics, Frame semantics, Construction Grammar (Cognitive Construction Grammar, Radical 12 A similar way of treating phrasemes, here pragmatemes, in terms of morphosyntactic, semantic and pragmatic description is suggested by Kauffer (2011). According to the author, there are pragmatemes expressing a particular communicative function, such as threat, doubt, resignation, incredulity or, even more specifically, the speaker's refusal to continue his communication. My phraseoframe thus tries to put into practice a phraseological desideratum: "Mais il est nécessaire que toutes les dimensions du figement soient prises en compte: Les critères morpho-syntaxiques, ceux de nature sémantique et également la dimension pragmatique, plus intéressante, encore plus analysée, mais particulièrement pertinente pour les phraséologismes pragmatiques" (Kauffer, 2011, 46). 13 In fact, when examining phrasemes as they occur in texts we should not speak of constructions but of constructs instead, because the former represent the general abstract structure and the latter are their instantiations licensed by constructions.
Construction Grammar and, as far as the attributevalue structure is concerned, as well unification-based Construction Grammars) (cf. Ziem and Lasch, 2013) • Theoretical premises: Holistic paradigm (trying to capture all facets of linguistic knowledge relevant for understanding and adequately using linguistic signs); non-separation of semantics and pragmatics; usage-based analyses • Methodological essentials: Learner-appropriate metalanguage; breakdown by linguistic domains; attribute-value structure in prose (without CxG formalism of boxes and brackets) reflecting linguistic knowledge types (attributes) and their descriptive elements (values); inclusion of formal and semantic variants; broad documentation of corpus data • Technical features: digital representation; hyperlinks to other phrasemes of the same semantic fields and frame; hyperlinks to audio-and video files to illustrate prosody and gesture

Classification of Phrasemes
What does this mean for phraseology? And what about the criteria usually taken into account when we speak about phrasemes? Before giving an answer to these questions I will briefly treat some phraseological classifications before presenting my own typology 14 .
Let us start with Makkai (1972; who has made some important distinctions. Firstly, he classifies "idioms" (phrasemes) in subgrammatical idioms (phonetic, phonemic, morphological idioms), lexemic idioms (nominal combinations, phrasal verb idioms, tournure idioms, phraseological idioms of encoding and sememic idioms), proposing then even hypersememic or pragmemic idioms (referring to conversational implicatures, one of several techniques of pervasive idiomaticity). It is remarkable that among all imaginable types of phrasemes Makkai does not mention syntactic idioms, which would correspond to our phraseotemplates. Makkai's tournure idioms are predicative phrasemes in my terminology (e.g., to spill the beans), whereas phrasemes like not written in stone, to make a long story shortor to cut to the chase are called phraseological idioms of encoding, because they display their idiomatic character, according to Makkai, only in speech production (encoding). Without having learned them, speakers would not able to produce them. On the other hand, "their meaning is clear the first time they are heard" (Makkai, 2011, 222). 14 Literature on phraseology is abundant. Summaries of different states of research, theoretical approaches, terminology and concepts can be found in Moon (1998;Burger, 2007;Ellis, 2008;Granger and Meunier, 2008;Ettinger, 2009;Benigni et al., 2015). Meunier and Granger (2008) focus on phraseology in foreign language learning and teaching.
(Besides, this may be true for native speakers but not for foreign learners). By using this example I have already mentioned the second important type of phrasemes, the idioms ofencoding and idioms ofdecoding, picked up by Fillmore et al. (1988) who call them encoding and decoding idioms and whose definitions seem to me clearer than the one given in Makkai (1972): A decoding idiom is an expression which the language users couldn't interpret with complete confidence if they hadn't learned it separately. With an encoding idiom, by contrast, we have an expression which language users might or might not understand without prior experience, but concerning which they would not know that it is a conventional way of saying what it says (Fillmore et al., 1988, 504f.).
I will illustrate this with Italian examples. Idioms of decoding are darsi la zappa sui piedi ('to shoot oneself in the foot'), alzare il gomito ('to have one too many'), arrampicarsi sugli specchi ('to grasp at straws'), idioms of encoding rosso acceso (bright red), fin qui tutto bene (so far so good), più che altro (above all), per così dire (so to speak), in linea di massima (basically). All of them are idiosyncratic constructions which a L2 learner of Italian would either not understand at all or not be able to produce in exactly the same way. Even if one encoding idiom or the other might be transparent to a foreigner learner, let us say fin qui tutto bene, it is doubtful that any of them could be produced on the basis of the open-choice principle-learners would probably say fino a qui, tutto bene. So, in the end, the distinction between encoding and decoding idioms is not really helpful for learner-centred phraseology and confirmation has come once again: In a foreign language everything is different (cf. Hausmann, 1997).
Another typology was proposed by Fillmore et al. (1988), who distinguish firstly lexically open (or formal, schematic) and lexically filled (or substantive, specific) phrasemes-the term idioms used by the authors comprises all kinds of phrasemes-and secondly three types of usual or unusual combinations to be qualified as idiomatic: (a) Unfamiliar pieces unfamiliarly arranged (kith and kin, with might and main), (b) familiar pieces unfamiliarly arranged (all of a sudden, in point of fact), (c) familiar pieces familiarly arranged (to pull someone's leg, to tickle the ivories). I have always wondered why they did not add the fourth type: Unfamiliar pieces familiarly arranged. One might think of phrasemes like uscire dai gangheri ('to fly off the handle'), mettersi in ghingheri ('to put on one's finery'). The two poles, lexically open and lexically filled phrasemes, are represented in Fig. 2. Is it reasonable, following Nunberg et al. (1994), to claim the existence of two types of idioms, (a) idiomatically combining expressions and (b) idiomatic phrases? The first type permits an interpretation according to the argument structure, inherited by the hierarchically superior general structure of transitive verbs (Ziem and Lasch, 2013, 58), as in the bivalent monotransitive construction alzare il gomito ('to have one too many'). The relationship between the verb and its direct object is determined in terms of semantics, however, not in terms of syntax: "Collocational dependencies are in fact the overt manifestation of semantic relations, not syntactic relations" (Croft, 2011, 176). In the predicative phraseme (idiom) we are looking at, alzare allows only the lexeme gomito. "This collocational dependency is called an idiom chunk and it is evidence for a relation between" (ib.) alzare and gomito: [OBJECT] According to one of the principles of Construction Grammar, what we have to look at are not the single components of a construction, but the construction on the whole and consider them non-derivational and monostratal (Fried and Östman, 2004, 25). Only the categories and roles (i.e., argument structure here) are derivative, constructions themselves, however, must be taken as basic (Croft, 2001, 175). In the idiomatically combining expression alzare il gomito, following Croft (2001,181), "the syntactic parts of the idiom" (alzare and gomito) "can be identified with parts of the idiom's semantic interpretation" ('bere' and 'troppo alcool'), according to which alzare corresponds to bere ('drink') and il gomito to troppo alcool ('too much alcohol').
As far as idiom typology is concerned only a few linguistic proposals turn out to be applicable from a foreign language teaching point of view 15 . I thus do not consider it necessary to distinguish between idiomatically combining expressions and idiomatic phrases (Nunberg et al., 1994), because neither dare del filo da torcere ('to cause a lot of trouble'), representing the first category, nor tirare le cuoia ('to die', similar to kick the bucket), standing for the second one, are semantically transparent for a foreign learner of Italian. Both of them have to be described more or less in the same way in a dictionary. The typology set up in Fillmore et al. (1988) seems to be suitable for linguistics, because they make us understand how phrasemes are structured, but it is nonetheless of little help for L2 learners, who have different needs.
My own classification of phrasemes is simple. Based on the definition of the term phraseme given above and taking into consideration the fundamental definition of a construction (as a form-meaning pair) I will distinguish the following types, moving down, according to Fig. 2, from lexically open phrasemes at the top to lexically filled phrasemes at the bottom, passing over, however, phenomena which are between the main types as well as cases with more or less idiomaticity or entrenchment: • Phraseotemplates (più/meno X più/meno Y, Che (OBJ 1 ) (OBJ 2 ) VERB infl a fare?, … sì (che)…) • Predicative phrasemes (darsi la zappa sui piedi, fare mente locale, alzare il gomito, menare il can per l'aia) 16 • Collocations 17 (apparecchiare/sparecchiare la tavola, capelli radi, scapolo impenitente, digrignare 15 Notable among them is however the onomasiological approach in Bárdosi et al. (2003). 16 Predicative phrasemes are characterized by an inflectional verb and a nominal part, often reflecting the structure "(direct, indirect or prepositional) object (+ adverbial)": Alzare il gomito, uscire dai gangheri, darsi la zappa sui piedi. Furthermore they are typically figurative. Some authors call them idioms (e.g., Moon, 1998), but this term is too ambiguous because it is also used in a broad way in the sense of set phrases or phrasemes/phraseologisms (Makkai, 1972;Fillmore et al., 1988). Others authors refer to terms like expressions idiomatiques verbales (González Rey, 2001), which is very precise, by the way, while yet others split the concept into two entities they call idiomatically combing expressions and idiomatic phrases (Nunberg et al., 1994, see above). The terminological confusion is considerable . 17 Collocations are understood here as semi-fixed lexeme combinations consisting of a cognitively primary and Predicative phrasemes are known to be especially difficult to understand (let alone use), particularly when they are fully non-compositional, as essere alla frutta ('to be broke' or 'to be on one's last legs'). Even in the case of idioms whose meaning might be motivated by metaphoric interpretation, such as darsi la zappa sui piedi ('to shoot oneself in the foot') or, using one's imagination, even alzare il gomito ('to have one too many'). On the other hand, we have to deal with expressions like gettare olio sul fuoco, being rather transparent and having its equivalent in to add fuel to the fire, but showing also pitfalls because semantically autonomous base, e.g., ipoteca ('mortgage') and a semantically dependent collocator, e.g., the verb accendere, whose meaning 'to take out' depends on its base mortgage, but which in other contexts (e.g., accendere una sigaretta 'to light a cigarette') may be completely different. 18 It is not relevant here whether or not the collocator is idiosyncratic, e.g., verde in numero verde ('toll-free telephone number') or un mondo in divertirsi un mondo ('to have a lot of fun'), or di seconda mano in una macchina di seconda mano ('a second-hand car'). Cf. Konecny and Autelli (2012-). Complex attributive adjectives (adjectival phrases) as di seconda mano (synonymous with usato) are treated here as adjectival collocators. 19 As far as word formation is concerned, we have a theoretical problem. Adhering to the above definition of phraseme, derivations may be constructions, but they are not to be regarded as phrasemes. The problem with compounds is that they may be fused in one form (e.g., cacciavite 'screwdriver', tergicristallo 'windshield wiper') and thus not be considered phrasemes. Beyond that, due to pure orthographical conventions, compounds might be written together as cassaforte ('safe') and thus be outside the phraseme category, or they might be inside (carro armato 'tank') and outside (carrarmato) at they same time. I however stick to my decision to call compounds phrasemes only if separate lexical units are taken into account, whether they are partially noncompositional (lettera aperta 'open letter', mercato nero 'black market') or entirely non-compositional (colletto bianco 'whitecollar', pesce d'aprile 'April fool'). Compositional compounds (e.g., motore a benzina 'gasoline engine', cavallo di razza 'thoroughbred') may be constructions (based on the criterion of entrenchment), but they are no phrasemes. fuel would rather be translated by carburante in Italian and to add s.th. to s.th. would generally not be rendered by the verb gettare. So the obvious solution, departing from an English-to-Italian perspective would be *aggiungere (del) carburante al fuoco.

Lexicographical Treatment of Phrasemes within the Framework of Construction Grammar
As a permanent (and eternal) learner of several foreign languages, I have always regretted being rather limited in my ability to understand let alone produce phrasemes. Even university teachers advised us against trying to master the field of phraseology because of its being too difficult for L2 learners, as they argued. I have never been convinced, however, of this explanation. Later, as a linguist and university professor, I tried to figure out new ways to enhance the methodological situation for learners in the field of phraseology (cf. Ettinger, 2009;. Construction Grammar helped me to understand what I would have to do. This is how the phraseoframe came into being. In the following section, some innovative ideas on L2 phraseology will be described (cf. also Bolly, 2009). As indicated above, as it is impossible to treat all phraseological types, three kinds of phrasemes, predicative phrasemes, phraeseotemplates and pragmatemes, will be illustrated by the phraseoframe. The objective is to elucidate the importance of a comprehensive description of phrasemes for learners of Italian as a foreign language. The entirely corpus-based description will be realized according to the types of features on the form and meaning side of these constructions: Prosodic, morphological and syntactic properties will be followed by semantic-pragmatic and discourse-functional properties. The mode of representation will be an attribute-value structure in prose.

Predicative Phrasemes
The first phraseme is a somewhat particular case. Attaccare un bottone a qualcuno (literally 'to sew on a button to somebody') would be in English something like to talk somebody's ear off. If we want to know how it is used we do not learn much when we consult traditional Italian dictionaries like Zingarelli, Treccani and Garzanti.
(3) (a) attaccare bottone, iniziare un discorso; tentare un approccio: Ha attaccato bottone con tutte (b) attaccare un bottone, ( fig.) iniziare un discorso lungo e noioso: Mi (= a me) ha attaccato un bottone che non finiva più! What we can conclude form these three lexicographical explanations are two things: First, there seem to exist two different constructions with attaccare and bottone, which are syntactically different and, second, Italian dictionaries obviously do not agree on the meaning of the two phrasemes. This remarkable result allows the conclusion that the two constructions are mixed together in everyday language.
Let us first describe the two syntactic patterns and try to grasp the difference in meaning: These are indeed two different constructions, each of them having different syntactic, semantic and pragmatic properties (bearing in mind the indefinite article in A and the zero article in B). Here once again we can see that the semantic content cannot be separated from the pragmatic one. A and B imply different goals and motivations and are associated with different prototypical semantic features, which can be verified on the basis of corpus documents: (i) Mauro ha poi trovato subito un ottimo argomento per attaccare bottone con Sarah: La Spagna, meta di viaggio di entrambi; la bella napoletana infatti ha vissuto per cinque anni a Madrid e lì ha conosciuto un modello spagnolo con il quale ha avuto una storia durata sei anni, ma ora, è di nuovo single. (PAISÀ) (ii) Fermarsi a guardare superficialmente questo evento si vedono solo dei gran bevitori di birra e gente che va li' solo per attaccar bottone con gente del sesso opposto (anche dello stesso come s'e' visto prima...), [...]. (WEBBIT) (iii) Neanche Berlusconi sa bene cosa sia meglio fare, in questa giornata che segna il trionfo di un ex comunista. Entra, esce a farsi fotografare con una comitiva di turisti, rientra, si siede accanto a Fini e gli attacca un bottone su chissà cosa (il presidente di An a un certo punto non riesce a trattenere uno sbadiglio) (CORIS) (iv) Mi scusi se le ho attaccato un bottone, vedo che lei sta partendo. Beh, spero che vada in un bel posto e che passi una bella vacanza. (WEBBIT) A should indeed not be confused with B. Whereas B may express (a) coming into contact or starting a conversation with someone or (b) trying to flirt with someone, A unambiguously means a negative judgment from the speaker's point of view who sees himself engaged in a conversation against his will 20 . Thus A is often accompanied by the features [+disturbing] and [+boring], which can be reflected on the linguistic surface (by adjectives, adverbs and other modifiers): Attaccare un bottone infinito, for instance. Linguistic context is particularly important for understanding the semantic-pragmatic core of the respective constructions. It is not always easy to grasp the meaning without paying careful attention to the context. When the direct object is used in the plural (bottoni), to mention another difficult case, A and B could be possible.
In the following phraseoframe only one of the two similar phrasemes, namely A (attaccare un bottone a qualcuno), will be illustrated 21 : to engage someone in a conversation considered excessively long by the interlocution semantic field(s) irritation semantic roles subject is agent, indirect object (Obj 2 ) is addressee (and patient) reference to hearer yes, but less frequent than reference to third person reference to speaker no reference to third person very often speech act(s) informative 22 What ought to be done, if we want to consistently follow Construction Grammar's theory, is to distinguish between several degrees of abstraction: (1) macro-constructions (abstract meaning-form pairings, such as Subj V Obj 1 Obj 2 / X CAUSES Y to RECEIVE Z), (2) meso-constructions ("similarlybehaving specific constructions"), (3) micro-constructions ("individual construction-types"), and (4) constructs ("the empirically attested tokens, which are the locus of change") (cf. Traugott, 2008, 136). This will be done in future studies. 23 As far as adjuncts like mica da ridere ('[which is] not funny') is concerned, it could be argued that this syntactic enlargement is a matter of external syntactic, but we subsume this phenomenon under internal syntax, because it still modifies the noun (Obj 1 ) syntactically as an adjective does. Another possibility is univerbation based on a phraseme, both cases representing plays on words. Phrasemes are hence constructions to be placed between compositional phrases and morphology (cf. Stathi 2011, 151). This is exactly what often happens in texts. The metalinguistic function is activated when a phraseme is "dephraseologized" or used in another syntactic structure or in a word cluster, e.g. to burn the midnight oil (to be understood literally) spiller of the beans (as a nonce formation). Aspects of idiomatic creativity, including idiom variation and variability, are treated by Langlotz (2006

Phraseotemplates
The next type of construction, which is about half way between a lexically open and a lexically filled idiom, might be called "phraseotemplates" in English-in German there are several terms like Phrasem-Konstruktionen (Dobrovol'skij, 2011) or Phraseoschablonen (Fleischer, 1997), in French, for instance, expressions-énoncés (González Rey, 2001 What is astonishing about this construction is that there is not a single Italian dictionary in which you can find it. Monolingual dictionaries as well as specialized phraseme dictionaries do in general ignore this type of phrasemes, probably because before Construction Grammar they had not been identified as such, neither in English books (cf. Götz and Lorenz, 2002) nor in other languages' phraseological dictionaries (cf. Bárdosi et al., 2003for French phrasemes, or Lurati, 20112002 for Italian phraseology) 25 . 25 This is however not entirely true for other languages. In Mel'čuk (1987) there are indeed phenomena (of Russian) that might be compared to phraseotemplates.
Beyond that, how should all these constructions be described in a dictionary in order to provide a maximum of knowledge about its meaning and its use?
If we part form the construction Che me lo chiedi a fare? as a whole we can recognize on the formal side of the pairing an interrogative clause composed by a whpronoun (che, che cosa or cosa) and a verb phrase consisting of an inflected verb and an obligatory and invariable infinitive complement (a fare). The argument structure of the verb (here chiedere) is fully reflected in the sentence where subject, direct object and indirect object are represented, the latter two by pronouns.
The phraseotemplate I want to describe here is the sì (che) construction used for expressing concession, reinforcement (or insistence) or rectification. The construction is briefly mentioned in some Italian grammar books (e.g., Serianni, 1997, 252, or Patota, 2006, but L2 learners do not find sufficient descriptions in any grammar or textbook. Analysing some linguistic corpora (here PAISÀ, CORIS, BADIP, WEBBIT, La Repubblica Corpus) I could make out the following form-meaning-pairs: ( What we can deduce from these examples is that obviously the sì (che) construction is above all a discursive or even rhetoric strategy to rectify, contradict, strengthen, fine-tune or modify an argument. When used without che the morpheme sì, introducing a sort of concession, is often placed directly behind an inflected verb, which frequently is the copula essere, the verb avere or a modal verb like potere, dovere, volere. If there is a clause containing a concession indicated by sì, then we generally find a coordinate clause joined by ma or, very occasionally, by però.
Isn't this too complicated for a foreign learner of Italian? My answer is: From the European reference level C1 onwards (or even earlier) it is possible to learn a construction like this. Besides, why should such a construction, argumentatively strong and usable in many situations, be restricted only to native speakers? There are only a few linguistic elements involved, the main difficulty consisting in the right position of sì (che) and in combing two clauses.
An epistemically satisfying description could contain the following items (space prevents me from displaying a complete phraseoframe): with a) you may contradict or reinforce an argument or what has been said before; with b) you may rectify what has been said before or you may give the impression to your interlocutor that you want to admit or concede to something which, however, in the following clause introduced by ma is finetuned or modified semantic field(s) underline one's opinion; contradict; rectify; concede reference to hearer possible reference to speaker possible reference to third person possible speech act(s) assertive, directive status of speech act(s) direct motivation you want to be right (a) or to correct someone or something (b) or to convince someone (also b) situational setting unspecific; often used in controversial discussions variety in spoken or written language register neutral frequency no corpus-based frequency results, but very familiar in everyday language DISCOURSE (introducing) new topic no (causing) topic change no discourse function (a) and (b); anaphoric (referring to something known or already mentioned) (b), when conceding something: furthermore cataphoric (indicating that the argument being treated will be carried on) rhetoric value (a) and (b): showing determination and selfconfidence (b), when conceding something: rhetoric strategy used in order to avoid expressing one's opinion directly or too openly RECOMMENDATIONS FOR L2 Both types, Prima pioveva sì and Queste sì che sono rivelazioni, are easy to learn: use just sì and place itdirectly after the verb or use sì che before the verb! CORPUS EXAMPLES see above The last type of phraseme to be treated here are formulae which I shall call pragmatemes, following Mel'čuk (1995), according to which a pragmateme is a phraseme fixed by the situation to which it is related (Mel'čuk, 2008, 4). Form and meaning of pragmatemes represent and evoke the conceptual content of a specific situation to which they are closely linked 27 . Examples of this classical type of construction are Bless you!, Never mind!, Hold the line!, but also formulae like You'll catch your death!, What do you say?, Pardon my French, Whose turn is it to deal? or Wet paint! (cf. Fléchon et al., 2012 with French examples). Pragmatemes, also called expressions-énoncés in French phraseology (González Rey, 2001) 28 , are encoding idioms (according to Makkai, 1972;, familiar pieces familiarly arranged (citing Fillmore et al., 1988) and thus compositional, i.e., the way their elements are combined is regular and their meaning is transparent. However, they are characterized by a high degree of entrenchment so that L2 learners would probably understand them when uttered in a concrete situation, but they would not be able to produce them.
It is a challenging if not impossible task for traditional dictionaries to describe pragmatemes in a satisfactory way. What ought to be done by lexicographers is to provide all information about their meaning and their use. I will try this with respect to the Italian constructions a buon rendere and e poi-colpo di scena.

A Buon Rendere
First of all we would have to determine whether its frequency of occurrence is greater than expected (cf. Gries, 2011). This might be questionable in the case of our pragmateme in Italian. But as we are not able to provide data about this issue we exclude this item.
Let us then turn our attention to Italian dictionaries. In Zingarelli 2015 we find, rather concise, the following paraphrase: 'detto di favore e sim., impegnarsi a restituirlo'. Garzanti 2009 gives a similar explanation: 'Si dice a chi ci ha fatto un favore, promettendo di restituirlo' and in the Nuovissimo Dardano we can read: 'Locuz. per dire che si accetta ql.co con il proposito di ricambiare appena possibile'. None of them enables a 27 Kauffer (2011, 41f.) puts forward the main characteristics of pragmatemes (which he calls actes de langage stéréotypés): "Statut d'énoncé", "idiomaticité", "fonction essentiellement pragmatique". Pragmatemes "ne servent pas à designer un objet, une personne, un procès etc. mais ils participent à la création d'un acte de langage qui a une composante illocutoire, voire perlocutoire" (42). See also Lüger (1999;. 28 Other works refer to the stereotypical character of pragmatemes (Coulmas, 1981;Schapira, 1999;González Rey, 2007;Kauffer, 2011). foreign user to under stand let alone use the construction, although Dardano provides some crucial semantic indications. What is in deed relevant as far as the meaning of a buon rendere is concerned can be expressed in two of my students' lexicographical trials: (a) a buon rendere, si dice a qualcuno che ci ha fatto un favore per sottolineare che siamo pronti a ricambiare il favore nel futuro: Grazie per il tuo sostegno. A buon rendere! (b) a buon rendere, 1. se qualcuno dice a buon rendere vuol dire che vuole ringraziare qualcun altro e ricambiare appena possibile: Grazie per la cena. A buon rendere! (a) and (b) convey the minimum semantic information one ought to have. Actually the description is not sufficient to fully understand how and when to use this formula. The crucial point is that whoever says a buon rendere, after having benefited from a favour (which may be big, but usually is a banal or every day one, like having been offered a coffee in a bar by a colleague) feels obliged to return the favour or at least considers it a formula of politeness to say these words, often combined with some sign of gratitude: Grazie! A buon rendere! Languages differ here considerably. The German (pseudo) equivalent Danke, Du hast was gut bei mir! would not be used if only a little favour is concerned. In Italian, however, the favour is usually small, but may be big, too. Let us have a look at some corpus examples (WEBBIT): (i) «[…] e magari bermi pure io un caffè per bontà vostra ...» «Tutto qua? Ma vieni con me che te lo offro direttamente al bar un caffè!» invitò Paolo, sorridendo a quel tizio che aveva un "non so che" di Pacino e De Niro con l'accento napoletano. «Grazie dottò, a buon rendere. Andiamo!» (ii) volevo kiedervi e se usassi la sendto e recvfrom? potrebbe funzionare? mmm ... AbsoluteBeginner: Scusate se insisto, ma che tipo di server è? Grazie Domy per la tua risposta ma è un po' vaga. Non vorrei rompere le scatole in mezzo a voi che siete tutti esperti, ma è una questione importante. Ho provato a capirci qualcosa leggendo nei motori di ricerca ma ... nun gliela fò! Grazie. E non posso nemmeno dire a buon rendere.
As quite often is the case with lexemes, expressions and phrasemes, speakers tend to invert their meaning by antithesis. This also applies to the pragmateme a buon rendere, which may refer to a situation in which the speaker expresses anger towards his interlocutor, making him understand that sooner or later he will hit back. But this might also be used in an ironic or humorous way. Let me cite one more the student who already proposed (b)  The last case study is to illustrate again the shortcomings not only of learners' lexicography but of monolingual and bilingual (Italian) lexicography in general. Pragmatemes may not be part of the basic learning material of a L2 programme, but they are, as I said, as far as communicative effects are concerned, nonetheless important and should therefore not be limited to native speakers. Such as in the case of epoicolpo di scena(literally plot twist, referring to a "radical change in the expected direction or outcome of the plot" of a story, a novel or a film; OED), which is a rhetoric strategy used in a story or a tale someone tells (or in a newspaper article or novel) in order to increase suspense or to capture the listeners' attention 29 , similar to Italian ta-ta-ta-ta♫, a sort of fanfare mimicry, another construction not mentioned in a dictionary even once.
Uttered with a pause after e poi ('and then') and colpo di scena ('surprising twist') it announces the climax of a story or something unexpected − in English something like surprise! or and this is the surprising part!, woven into an utterance or a written text 30 .
Dictionaries are made for providing linguistic information about lexical units, i.e. lexemes and lexeme combinations, such as collocations, compounds, phrasemes. But they usually stop here. Constructions that go beyond this boundary, especially phraseotemplates and pragmatemes, are often not registered let alone treated in an epistemically relevant way. This is not surprising as far as traditional print lexicography is concerned, but in digital lexicography almost everything is possible 31 . The reasons for not perhaps e poi, colpo di genio) entrenched in a double sense: Formally and with respect to the situation in which it is used. 31 New insights into digital and electronic lexicography are offered by Granger (2012), Granger and Paquot (2012) and L'Homme and Cormier (2014). considering or not holistically describing phrasemes in general are based on the lack of awareness of innovation and new lexicographical techniques. Besides, it involves a lot of hard work to renew a dictionary which goes deeply into the microstructures… The reality looks somewhat different, as we can see when: colpo di scena is, in fact, described in Zingarelli 2015 andGarzanti 2009, but only as a compound, not mentioning the whole pragmateme and not telling us in a satisfactory way when and how it used and what for. Garzanti 2009: (s.v. colpo:) colpo di scena n. 1 (teat., cine.) avvenimento imprevisto che coglie di sorpresa e cambia il corso dell'azione 2 mutamento improvviso e sorprendente di situazione: colpo di scena: arriva il marito.
Without appropriate situational context the example given in Garzanti's meaning no. 2 is not comprehensible. I therefore feel it is of utmost importance to describe phrasemes in a comprehensive manner, the way it could be realized in a digital learners' dictionary of Italian illustrated here by another phraseoframe: Try it when you tell a story. Make a pause after e poi and raise your voice twice (during e poi and colpo di scena), in order to succeed in creating suspense (and a smile) CORPUS EXAMPLES see above, and (vi) (vi) Tutti attendono con ansia quel giorno, perché è l'unico giorno dell'anno in cui ci si possono levare i mutandoni di lana che pizzica e in gabinetto si può usare la carta igienica invece delle ortiche. Fai che ti rifai, ecco che la piccola salta fuori. Ma, oh, colpone di scenona: era tutto un truccone! La vittima sacrificale non era la subdola mocciosetta, ma Edward medesimo! (Il prescelto:Nicolas Cage sfida l'ApeMaia. E le prende di brutto; http://cinescherno.com/2012/05/ 17/il-prescelto-nicolas-cagesfida-lape-maia-e-le-prende-dibrutto/) (Accessed on February 13, 2015).

Conclusion
What are the main results of this article? I have tried to explain why conventional lexicography is not and cannot be able to treat phrasemes in a way that helps foreign languages users to really understand and use them. Whereas there is a long and rich tradition in lexicology as far as research on phrasemes is concerned, lexicography seems to have established a tradition of treating all idiomatic expressions as poor relations: Superficially and without any attempt to elucidate specific linguistic properties and pragmatic conditions, information that is indispensable for the full understanding of phrasemes in order to enable learners to use them like native speakers do. Phrasemes are, as I pointed out, constructions, hence symbolic units. This means that they are first of all conventional, but furthermore entrenched and/or non-compositional (idiosyncratic). If linguistic units are not (fully) predictable, they have to be learned separately. Native speakers usually have no problems decoding or encoding phrasemes (under the condition they have so far come across or learned them), but foreigners have.
As phrasemes, due to their rhetorical or emotional effect, play an important role in language production, in everyday use as well as in formal speech, my concern is to make available this comprehensive knowledge to foreign learners, too. Construction Grammars helps me to do this. Based on the holistic approach of the CxG framework, constructions (i.e., phrasemes, too) can be described by means of attribute-value representations. This is what I wanted to propose here, introducing the phrase frame model that is well suited to applying the attribute-value concept, doing, however, without Construction Grammar's formalism but providing instead explicit information about linguistic properties of all levels. The final objective is a digital phraseme dictionary in which, phrase templates, predicative phrasemes (idioms), collocations, non-compositional compounds, grammatical and adverbial phrasemes, discourse markers, full sentence phrasemes and proverbs as well as pragmatemes can be described in a comprehensive manner, taking systematically into account authentic language (as documented in language corpora). Digital lexicography should at last embark on new ideas. Digital platforms provide enough space to implement all (phrase logical) desiderata of bilingual and bilingual lexicography, including learners' lexicography.