Teaching Affects in French as a Foreign Language . A Trial Focusing on Intensifiers in Collocations

This article suggests creating networks of lexical items in the field of affects, in the context of teaching French as a foreign language. After a lexico-semantic study of some intensifiers present with certain lexical items from the field of affects, we will look at what kind of networks should be developed in order to help students memorise lexicon and associated collocates of intensity. The underlying hypothesis is that building these networks should enable learners to better memorise the lexical associations of the field in question.


INTRODUCTION
Teaching French lexicon to students who are non-native speakers (nns) is at the heart of our questions in the context of French as a Foreign Language (ffl). This contribution focuses on a specific lexical form: the lexicon of affects and their intensifiers, at the level of phrase structure and the syntactic combination of collocations. It should be noted that the authors of the Common European Framework of Reference for Languages 2 recommend the teaching of phraseology (a set of elements including collocations) almost to the same degree as the rest of the lexicon. Several observations, linked to the teaching of these elements, will allow us to put forward hypotheses concerning their choice and use in language classes: 1. The teaching of affects is now commonplace in ffl, however, the forms taught are often restricted to simple lexical units (peur [fear] 3 , joie [joy] or mépris [contempt], for example). I have already noted the absence of phraseology in the teaching of lexicon, and particularly of affect-related lexicon 4 . It, therefore, seems crucial to me to try and address this deficiency.
2. Choosing which collocations to teach is always a difficult exercise for the teacher who must try and sort through often disparate and yet equally important elements. I do not have any preference regarding types of collocation, however, the criteria of frequency guides me in my choices as a teacher and I cannot, initially, teach elements that are rare for native speakers. Therefore, the choice of collocations linked to affects is not insignificant: this semantic characteristic is frequent in native speakers' language -avoir très peur [to be very scared], faire une grande joie [to delight], etc. It is therefore necessary to see how these elements are characterised linguistically so as to make choices guided by this description.
3. The use of various IT tools, and notably of corpora, is now omnipresent in the teaching of lexicon in ffl 5 . Corpora -and above all concordancers -have been used as teaching tools in efl didactics for a long time. ffl has not escaped this wave of interest in corpora and interesting attempts in the field have already been published 6 . Part of the Frantext corpus, selected for the beginning of the Emolex project 7 , will be used here for the extraction of data and implementation of teaching activities.
4. Concordancers meet with unanimous approval within the teaching community, however, their limitations on the didactic front become clear as soon as learners' ability to memorise the data they have been taught fails to meet expectations. Indeed, we can note that learners do not always retain how to redeploy a linguistic element that they have studied with the help of a concordancer. This observation led me to use complementary tools to help the learner to memorise elements and their syntactic combinations so as to be able to use them again in an appropriate fashion.
These few observations give rise to several questions and hypotheses that this article will try to answer and confirm: how can we classify the intensifier adjectives that are present alongside nouns of affect? What representation would help learners to memorise these lexical associations with a view to using them again? The underlying hypothesis here is that presentation in varied contexts helps memorisation (corpora can be of help here) and so a classification into semantic networks could be relevant. It, therefore, seems to me that it would be interesting to get learners to build semantic networks.

The choice of affects
In the context of the Emolex project, the team is working on the lexicon of emotions in five Indo-European languages. This contribution focuses on French as taught to nns. For this initial attempt at description for teaching, we will look at the lexis of emotions in three semantic fields: Surprise, Respect and Dépit (respectively surprise, respect and resentment/disappointment). The study will be carried out around the three grammatical categories carrying the main meaning for the collocation base: nouns, verbs and adjectives (cf. Diagram 1). Diagram 1. Distribution of lexical items according to semantic field and grammatical category Source: Author's own elaboration.
The lexical units in Diagram 1 were extracted from an initial corpus (Frantext) linked with intensifiers found alongside them. The collocations were then sorted according to how frequently they appeared in the corpus.

The values of intensity
Intensity is one of a series of semantic dimensions (polarity, manifestation, verbalisation…) used in Emolex to describe lexical units.
Two grammar manuals were initially of help in defining intensity, with linguistic studies then confirming and refining these descriptions. According to Chevalier et al. 8 , strong intensity "has a number of stylistic means at its disposal" while, according to Riegel et al. 9 , the classification of qualifying adjectives uses a three-tiered scale: low-medium-high. However, the authors of these manuals admit that they do not take everything into account for these descriptions, which are therefore further refined by linguists confirming the two unavoidable values of "strong" and "weak" 10 . Romero puts forward a typology (Diagram 2) that allows elements to be classified according to intersected criteria.  What I shall retain from this diagram are the two criteria "quantitative" and "qualitative" that, as we shall see, allow my own classifications to be refined and seem relevant for teaching as they can be approached didactically. Furthermore, explains that the criteria of quantity is frequently associated with intensity "and that there is no real difference in principle between quantifiers and intensifiers" 11 . The criteria of quality is harder to explain and we will, therefore, see that its use remains less frequent, and sometimes vague, on the semantic level. It could be seen as approximating a form of pleasant judgment. My examples will illustrate and clarify these points.
Intensifiers, therefore, fit into two categories: "weak" and "strong", linked with "quantitative" and "qualitative" criteria. I initialled retained the "medium" intensity category described notably by Chevalier et al. 12 , but elements of this category could fit into the two others depending on the context (être un peu étonne [to be a little surprised] (often weak) or être assez contrarié [to be fairly annoyed] (often strong)). Therefore, I chose to remove this criterion that was difficult to use on a didactic level.
The aforementioned authors, thus, enabled me to formulate the following definitions in order to teach intensifiers around nouns, verbs and adjectives of affect. In general, this dimension allows the collocations providing indications about the intensity of the affect expressed to be grouped together using more or less metaphorical means.
1. The collocations of "strong" intensity express the fact that the affect experienced is of greater intensity with regard to an average that is relative but recognised by native speakers (often quantitative): -French: Être très énervé [to be very annoyed]; Il la respecte profondément [He respects her deeply], -English: Absolutely amazed; To appreciate fully. 2. The collocations of "weak" intensity express the fact that the affect experienced is of moderate or low intensity compared to an average that is relative but recognised by native speakers (often quantitative): -French: Être légèrement surpris [to be slightly surprised]; être peu fâché [to not be very cross], -English: Less respected.

Intensity in the corpus
Intensity is one of the dimensions that has the specificity of cutting across the three grammatical categories studied in the project. I began by extracting examples from Frantext (21,600 words; 1960-2007; categorised database) in order to see the quantity of intensifiers present with the lexical units studied (see: Diagram 1 and Table 1).
We can observe that the collocations in the field of Surprise call more upon intensifiers than those in the other two fields. This adds intensity to a field that already carries intrinsic intensity. Surprise is more intense as an affect than Respect or Dépit. The tlfi 13 definition of Surprise describes it as a sudden and unexpected action that creates an état de trouble [state of agitation or turmoil] accompanied d'émotions qui en découlent [by the resulting emotions]; this sudden aspect means Surprise fits into the category of transitory emotion, contrary to the feelings of respect and disappointment, which have a more lasting aspect. It should, however, be noted that Respect resembles Surprise in its use of quantifying intensifiers such as grand [big] or beaucoup [a lot], which is rarer where Dépit is concerned. However, it would seem that when grand is used with Respect, it does not have quite the same meaning as when used with Surprise. This appears to be an example of the distinction between quantitative and qualitative mentioned earlier: grande surprise would, therefore, be more quantitative than grand respect, in which the quality intrinsic to Respect would seem to be extended to the adjective (perhaps by related seme).
The figures also reveal a lower use of intensifiers around adjectives, in all three fields. Overall, the field of Dépit displays the least intensifiers. This observation could be explained by the fact that Dépit has a less quantitative aspect. As Gaatone specifies, this criteria is largely present for intensifiers, it would, therefore, seem that Dépit is less quantifiable than Respect or Surprise. This remains to be confirmed with more extensive corpora and statistics, which is not the aim of the present article. The following few examples summarise the intensifiers encountered in the French corpus in question: STRONG 1. Quantitative collocates ("going beyond a norm" Riegel et al. 14  "Strong" intensifier adjectives are more numerous than "weak" ones; this corroborates the analysis of Chevalier et al. 16 on the subject. Apart from the intensifiers of "strong" or "weak" types, we also encounter intensifiers that appear to be specific to each field. These specificities are semantically interesting and enable a more targeted form of teaching to be envisaged when these characteristics only appear with certain lexical items from the field in question.

Specific intensifiers
An example would be the adjectives that carry the meaning of a form of infinite (large or small). These appear alongside certain specific lexical items, but are present in all three fields: immense surprise [immense surprise] and un respect infini/aveugle/sans borne [infinite/blind/boundless respect] for the infinitely large, whereas for the infinitely small we have goutte d'amertume [drop of bitterness], which is a hapax legomenon of the field of Dépit, and lueur d'estime [glimmer of esteem]. We, therefore, find strong and weak intensifiers that are both quantative (lueur) and qualitative (respect aveugle). We can note, however, that the weak intensifiers are more of the order of metaphor than the strong ones (except aveugle, which warrants further attention). The weak intensifiers are metaphorical and some carry the semantic feature of light: lueur [glimmer] or ombre [shadow], for example. Amertume [bitterness] (from the Dépit field) is the only lexical item to be associated with as many weak, metaphorical and lightrelated intensifiers: ombre, lueur, trace, goutte [shadow, glimmer, trace, drop].
In sum, we can observe a large number of intensifiers in the field of Surprise, even though some of the lexical items in this field are only rarely combined with intensifiers -stupeur, stupéfiant and étonné [astonishment, astounding and surprised/amazed]; the question remains as to why surprise and étonnement [surprise and amazement] accept more lexical associations with intensifiers. Many derived adverbs appear in the three fields: profondément, franchement and légèrement [deeply, frankly, slightly]. They can be strong or weak, like other intensifiers. Finally, the presence of many intensifiers carrying the semantic feature /infinite/ alongside certain lexical items is interesting, particularly given that they take varied combinatory forms (derived adverb, without +N infinite ).
This initial insight into these semantic and combinatory specificities thus enables a consideration of the semantic networks that should be built within ffl teaching and raises the question whether it would not be possible to envisage creating collocational networks.
I shall, therefore, now look at a didactic approach that would take these linguistic specificities into account with a view to planning teaching that incorporates these aspects. Given that my aim is to help learners with memorisation, I shall attempt to see how to avoid cognitive overload during the teaching of these lexically and syntactically constrained lexical associations.

TEACHING AFFECTS IN FFL
How can the lexicon of affects be taught in a foreign language, and particularly in ffl? For several years now, the teaching of lexical items in this field have been found in ffl manuals 17 , often with a chapter devoted to them. Affects are broached sparingly perhaps due to the personal aspects that they can reveal 18 .
The authors of these manuals are still cautious regarding approaches to collocations (and not only for affects) despite learners' enthusiasm for all fixed forms of language (fixed expressions, proverbs and other phraseological forms) 19 . It is now recognised that phraseological elements, and in particular collocations, seem complex at first, notably because they are difficult to translate literally, have a supposed oral nature 20 , and are difficult to recognise. Indeed, it would seem that collocations suffer from their frequency and that native speakers (ns) are not aware of using such structures 21 . ns authors of manuals for teaching ffl display this same lack of awareness; and indeed, how can one teach something of which one is not aware?
For nns, semantic deviation, or even metaphors or metonymies on related elements, block access to meaning; many collocations are not predictable for nns.
Pobrane z czasopisma Annales N -Educatio Nova http://educatio.annales.umcs.pl Data: 07/02/2021 22:33:13 U M C S Learners cannot deduce the meaning of the association and are often unable to identify it depending not so much on their level of language as on the lack of help that they receive in this task 22 . These two aspects (identification of forms and access to meaning) are essential when learning a foreign language and must be achieved without translation where culturally marked fixed forms are concerned. I, therefore, envisage simultaneous access to both form (lexical association and syntactical combination) and meaning (this will be developed elsewhere) by using an approach through semantic networks: collocational networks 23 .

Acquisition of lexicon: teaching using semantic networks
The idea of presenting lexicon in networks is not new; for many years now, several authors have used this form of representation 24 , and have thus paved the way for my own use of these networks. The intention is to now use these networks in language classes for reasons linked in particular to the assumed organisation of mental lexicon.
Mental lexicon does not seem to be organised hierarchically but rather in semantic fields. Grammatical category seems to play an important role in the storage of terms, with associations being made preferentially with words from the same category 25 .
On a linguistic level, while storage is preferably by category, teaching shows us that the learner is soon capable of moving from one category to another in various situations (nominalising a verb for reformulation; classifying elements for use…). We, therefore, need both aspects 1) classification according to category, because nouns of affects are mainly combined with intensifier adjectives and verbs of affects are logically associated (in French) with adverbs; 2) classification that creates cross-category links in order to form the expected lexical associations.
On a didactic level, implementing these networks contributes to the consecutive use of two approaches favoured in the teaching of lexicon in ffl: an onomasiological and a semasiological approach. The network paves the way for the onomasiological path, while the list of lexical items thus assembled allows the semasiological approach. Both are relevant in the teaching of lexicon as they are complementary: lists of lexical items (semasiology) linked by meaning to a specific field (onomasiology). Moreover, both approaches should apparently be foregrounded in order to help with memorising 26 .

Graphical representations
In order to represent collocational networks, several sources inspire the result sought. Tournier, in particular, developed "recursive lexicograms" in which the nodes and representation of co-occurrent terms of the same grammatical category provide the beginnings of a useful representation for teaching. This kind of lexical network (same grammatical category) can thus be found in the recursive lexicograms (Diagram 3) and representations of semantic proxemy (Diagram 4) of the cnrtl 27 .
These representations have led to the development of the concept maps that are now omnipresent in many economic sectors. The underlying didactic point of view is based on learners' action and meta-linguistic reflection. It is a question of: Several software programs (some even free) allow these networks to be created. Any program that enables lexical items to be entered and links to be created is well suited to the main concerns at stake here (it will no doubt be necessary to further develop applications at a later stage). Here, Cmap Tools-Ihmc 29 have been used -free online tools, created by designers working with universities.
The main aim of my use of these maps is to help learners remember collocations. A trial is currently being carried out with German-speaking learners in Switzerland. The feedback is globally positive regarding the usefulness of the networks they have to put together 30 . It would be profitable to be able to provide learners with a concordancer containing intensifier collocations with selected affects and then ask them to create a diagram of that kind. The creation of the diagram, and the reflection it would entail regarding the links between elements, would be a useful didactic approach to help with memorising and with 29 Concept Map (Cmap) -Florida Institute for Humans & Machine Cognition, http://cmap. ihmc.us [access: 11.10.2017]. 30 Remark from a learner: "I found the glossary and diagrams putting the terms into a network very good. They allowed rapid auto-evaluation and learning at the same time". Pobrane z czasopisma Annales N -Educatio Nova http://educatio.annales.umcs.pl Data: 07/02/2021 22:33:13 U M C S meta-linguistic thought about these elements. It is not a question of providing learners with these diagrams, which could soon become illegible; the aim is rather for learners to create them themselves, on the basis of criteria provided by the teacher. In this way, by adding criteria, we can obtain complex diagrams that may be hard to read, but that are useful to the person who has created them. Diagram 6 offers one such example and used the followed criteria: 1) lexical items from the field Surprise (so several node items), 2) the morphological or semantic links between these lexical items, 3) the links between collocates and bases, specifying 3a) the syntactical place of each elements relative to one another, 3b) the semantic type of intensifier (strong/weak) 4) the groupings of frequent collocates for several bases. Several non-linguistic elements can be taken into consideration in order to understand the links, and, therefore, form or even meaning (an aspect that warrants discussion, particularly where synonymy is concerned), and in order to construct the diagram and make it clear to read. For example, arrows and lines can be present or absent, and the direction of the arrows can also be informative. An arrow indicates a link of lexical association either ante-or postposed according to the collocate's position with regard to the base (syntagmatic link); a line indicates a morphological or semantic link between the elements Pobrane z czasopisma Annales N -Educatio Nova http://educatio.annales.umcs.pl Data: 07/02/2021 22:33:13 U M C S (without association). It should be noted that weak intensifier collocates are systematically placed before the base (irrespective of which one). The number of collocates present with the bases varies according to the base and enables the possible lexical paradigm for each element to be rapidly visualised. We could envisage using different colours in order to further distinguish the elements or adding the frequency next to each collocate; we could also create dynamic (hypertextual) links towards other documents of all kinds (Internet or other): definitions, examples in the corpus, etc.
Such maps can serve as the basis for hierarchical representations of knowledge and can make learners aware of the lexical organisation of a language and of their own personal lexicon. They allow for thought to be structured and organised, and can, therefore, help learners memorise elements and their links. The teacher can use these as formative (or summative) evaluation and, thus, ensure that learners fill in their map as their learning progresses.

CONCLUSIONS
The answers to the questions raised in the introduction can be summarised as follows: the classification of intensifier collocates was first carried out in a quantitative fashion on the basis of the frequency of occurrences around the lexical item; then I opted for a classification organised around the semantic values "strong" and "weak", associated (for some) with an interesting qualitative criterion (which allows collocates' deviation of meaning to be explained).
The observation of the lack of teaching of phraseology in language classes, combined with the relatively few presentations of the lexicon of affects in ffl manuals, gave my endeavour an uncertain future. However, the use of these concept maps (currently been tested) has proved encouraging and widened perspectives for the immediate application of linguistic descriptions. Indeed, description allows elements to be distinguished on the syntagmatic and paradigmatic levels, and answers the questions often asked by learners regarding the choice and position of elements to be associated.
The didactic choice to get the learners to create their own representations of knowledge (and of their own knowledge) is in keeping with approaches to helping memorisation, which are increasingly considered in terms of networks of all kinds 31 . Also, as the authors of the cecrl remind us, making learners act themselves helps in grounding knowledge.