Jump to ContentJump to Main Navigation
Show Summary Details
In This Section

Cognitive Linguistics

Editor-in-Chief: Newman, John

4 Issues per year


IMPACT FACTOR 2016: 2.135

CiteScore 2016: 1.29

SCImago Journal Rank (SJR) 2015: 0.592
Source Normalized Impact per Paper (SNIP) 2015: 1.277

Online
ISSN
1613-3641
See all formats and pricing
In This Section
Volume 28, Issue 1 (Feb 2017)

Issues

Inside in French

Philippe Gréa
  • Corresponding author
  • Université Paris Nanterre & MODYCO CNRS-UMR7114, France
  • Email:
Published Online: 2017-01-07 | DOI: https://doi.org/10.1515/cog-2015-0127

Abstract

This article concerns five French prepositions that mark an inclusion relation and are ordinarily considered to be synonyms of dans ‘in’. The first is a simple preposition: parmi ‘among’. The other four are complex prepositions built from nouns of internal location (au centre de ‘at the centre of’, au milieu de ‘in the middle of’) and from names of body parts (au cœur de ‘at the heart of’ and au sein de whose word-for-word translation is ‘at the breast/bosom of’ but stands for ‘within’ ‘in’ or ‘among’ depending on the context). We will examine them from two different, yet complementary points of view. As part of a statistical approach, we use an association measure to determine the distributional trends of each preposition. In addition, we explore in greater detail the semantic mechanisms brought into play using the theoretical framework of Cognitive Grammar. This double analysis leads us to demonstrate that these prepositions are far from synonymous and involve very different types of constraints.

Keywords: corpus linguistics; association measure; Cognitive Grammar; location; spatial prepositions

1 Introduction

This paper focuses on five French prepositions that mark an inclusion relation (Langacker 1987: 225–228) and are generally considered to be synonyms of dans ‘in’ (Fagard 2012). The first is a single preposition, parmi ‘among’. Several studies have been devoted to it and compare parmi to entre ‘between’ (Ashino 2007; Franckel and Paillard 2007; Gréa 2015; Guentchéva 2003; Hilgert 2007, 2009, 2010, 2013; Kwon-Pak 2006; Van Goethem 2009). The other four prepositions are complex units. The first two use what, in French linguistics, we call noms de localisation interne, meaning ‘internal location nouns’ (henceforth ILN) (Aurnague 1991, 1996; Borillo 1988, 1999): milieu ‘middle’ and centre ‘centre’. These nouns are used to refer to part of a whole which is located inside that whole (hence the name ‘internal location’). For example, le centre or le milieu d’une table (‘the centre’ or ‘the middle of a table’) refers to part of the table, but also to a part that is approximately at an equal distance from the edges of the table. Two other prepositions are constructed using nouns referring to parts of the body: cœur meaning ‘heart’ and sein meaning ‘breast/bosom’. While cœur has become grammaticalized to give an ILN (le cœur de la ville ‘the heart of the town’ means the part of town the furthest from the boundaries of the town), the same is not true of sein whose use as an ILN is very restricted and applies mostly (and with an archaic flavour) to administrative bodies: dans le sein du conseil général ‘within the general council’. In combination with the preposition à ‘at’, these nouns produce four complex prepositions which denote a relationship between two participants, the trajector (henceforth tr) which is the primary focal participant (the located entity), and the landmark (lm) which is the secondary focal participant (the reference entity): au milieu de ‘in the middle of’ or ‘amidst’, au centre de ‘at/in the centre of’, au cœur de ‘in the heart of’ and au sein de (‘within’, ‘in’ or ‘among’ depending on the context).

Most studies on spatial location have the drawback of using spatial criteria such as the dimension and direction of the lm or the relative position of the tr in relation to the lm (Borillo 1998). From this perspective, centre ‘centre’, milieu ‘middle’, cœur ‘heart’, sein ‘breast/bosom’ and parmi ‘among’ are generally put into the same category and contrasted with another series which, on the contrary, profiles the limits of the lm (bout ‘tip’, extrémité ‘end’, périphérie ‘periphery’, marge ‘margin’, etc.). Implicitly or explicitly, this manner of proceeding supports the idea that the prepositions in each series are synonymous. Below, however, we show that the relationships they construe are not comparable, and that they impose constraints which go well beyond the spatial criteria ordinarily used.

To do so, we partly rely on a statistical method that has a long-standing tradition in French lexicométrie 1 ‘lexicostatistics’ called the calcul des spécificités meaning ‘specificity calculus/calculation’ (Habert 1983, 1985; Labbé and Labbé 2001; Lafon 1980, 1984; Lebart and Salem 1994; Salem 1987). It is primarily intended to highlight vocabulary that is over- or underrepresented (cf. Section 2.2 for these terms) in political, literary or philosophical discourse (see Loiseau [2006] for an application to the work of Deleuze, and Habert [1985: 129] for a review). However, it is perfectly applicable to semantics and can be used, unchanged, to measure the attraction or repulsion between lexemes and a given preposition (as opposed to one or more other prepositions). This method is presented in Section 2. We will focus on the nominal lm of each preposition in comparison to the other four, i. e., nouns that are both frequent after a given preposition as well as infrequent after the other four. The “specificity calculation” is therefore a statistical association measure (henceforth AM) which does not fundamentally differ from the methods that exist today. As with these methods, the specificity calculation generates a word ranking revealing the major distributional trends and forming the starting point of a strictly semantic analysis. In Section 3, we use the theoretical framework of Cognitive Grammar (henceforth CG, [Langacker 1987, 1991b, 2008]) to propose a schematization of the prepositions. As we will see, the diagrams reflect the distributional trends brought to light by the statistical approach.

2 Representativeness

2.1 Corpus and retrieval process

To conduct our study, we used the FrWaC corpus developed within the framework of an international research project called WaCky (Web as Corpus kool yinitiative) (Baroni et al. 2009). This is a very large Web-derived corpus considered over time (approximately 1.6 billion tokens) and its construction is described in Ferraresi et al. (2010). It has been designed to maximise the variety of genres and contents, and therefore contains ‘public sphere’ documents (academic and journalistic texts addressing socio-political issues and the like), and ‘personal interest’ pages (personal pages, blogs, or postings in forums). Of course, there is little doubt that the distributions of prepositions are sensitive to textual genres (novel, legal text, etc.) or influenced by socio-linguistic variations. For example, we will have an opportunity to raise the case of some constructions (au cœur du monde ‘in the heart of the world’) which occur mainly in religious discourse (see Section 4). It should be noted, however, that these variations are outside the scope of the present study.

As Table 1 shows (column 1), an exact search term on FrWaC (for example, centre) returns a large number of matches. Occurrences were therefore filtered using several criteria such as searching for au ‘at’ before the match, excluding the cases in which a punctuation mark is just before or after the match, etc. (Table 1, column 2). 2 The main drawback of this kind of corpus lies in the fact that it is built automatically and therefore contains a lot of noise consisting of a significant proportion of duplication on different web pages (14 % of our corpus, taking a window of approximately 220 characters). For example, Le soleil brillerait mieux au cœur de nouvelles cellules (‘the sun would shine better inside new cells’ in the context of solar panels) is the title of a single article cited 1,042 times on other sites, which results in artificially overrepresenting (cf. Section 2.2) cellules after au cœur de. We therefore decided to remove such duplicates (Table 1, column 3).

The next step consisted in using the Nooj software programme (Silberztein 2003, 2004) to extract all the tokens of the five constructions of interest to us (1). 3 Table 1, column 4, indicates the number of tokens of each construction.

(1)

  • a.

    <au centre de (Det) (AP) N1>

    ‘at the centre of (Det) (AP) N1’ (we will henceforth refer to this construction using simply centre)

  • b.

    <au cœur de (Det) (AP) N1>

    ‘in the heart of (Det) (AP) N1’ (cœur)

  • c.

    <au milieu de (Det) (AP) N1>

    ‘in the middle of (Det) (AP) N1’ (milieu)

  • d.

    <parmi Det (AP) N1>

    ‘among Det (AP) N1’ (parmi)

  • e.

    <au sein de (Det) (AP) N1>

    ‘in/within (Det) (AP) N1’ (sein)

These extractions require some additional selection criteria (Table 1, column 5). First of all, in view of the fact that we are interested in common nouns only, it was decided to exclude all the proper nouns from the N1 position in the five constructions examined. In order to do that, we used both NooJ dictionaries (for annotated proper noun and unknown words) and the capital letter criterion. Second, we have had to deal with problems specific to each preposition. The case of centre involves a particular difficulty as this lexical unit is frequently used to construct a proper noun serving to refer to a place. When it is preceded by the proposition à ‘at’ (2a), nothing allows it to be distinguished from the complex preposition (2b):

(2)

  • a.

    Cette discussion a lieu [au [Centre de Réflexion de l’Université]Proper N]PP.

    ‘This discussion takes place at the University’s Centre for Thinking.’

  • b.

    Cette discussion est [au centre de [la réflexion de l’université]NP]PP.

    ‘This discussion is a central feature of the university’s thinking.’

Place names (2a) take a capital letter in only half the cases. This formal criterion is therefore not sufficient to exclude this case. The second effective criterion for identifying proper nouns is the absence of a determiner after au centre de. It has the advantage of excluding a lot of place name such as au centre de loisirs ‘at the leisure centre’ (546 tokens), au centre de formation ‘at the training centre’ (480), au centre de documentation ‘at the documentation centre’ (373), au centre de rétention ‘at the internment centre’ (353), etc. But it also has the drawback of excluding a number of cases which fall within our study (3).

(3)

Les actions des grandes banques centrales sont [au centre de [réflexions sur la vie économique et financière]NP]PP.

‘The actions of the major central banks are the focus of thinking on economic and financial life.’

However, such cases are sufficiently infrequent for them to be dispensed with: they have no significant impact on the word ranking. 4 Conversely, choosing to keep occurrences of the place names introduces significant noise. For example, loisirs ‘leisure’ becomes artificially overrepresented in the centre construction, because of the 546 tokens of au centre de loisirs.

The names of body parts cœur and sein pose a similar problem. A distinction must be drawn between their use as a component of a complex preposition (4) or as a part of the body (5).

(4)

  • a.

    Le jardin est [au cœur de [la maison]NP]PP.‘The garden is inside the house.’

  • b.

    La peinture pénètre [au cœur de [la pierre]NP]PP.

    ‘The paint penetrates inside the stone.’

  • c.

    [Au sein de [la famille]NP]PP, les divergences s’accroissent.

    ‘Within the family, divergences are growing.’

(5)

  • a.

    L’admiration est si chère [au [cœur du poète]NP]PP.

    ‘Admiration is so dear to the poet’s heart.’

  • b.

    Un homme [au [cœur de pierre / d’or / d’artichaut / de glace]NP]PP

    ‘A coldhearted man/A man with a heart of gold/A soft-hearted man/A coldhearted man’

  • c.

    Dieu a mis les enfants [au [sein de la mère]NP]PP.

    ‘God put children at the mother’s bosom’

To exclude the latter case, we therefore removed all the tokens in which the N1 denotes an animated singular individual (poète ‘poet’, mère ‘mother’, père ‘father’, enfant ‘child’, etc.), and a number of names of matter (pierre ‘stone’, but also or ‘gold’, glace ‘ice’, etc.) when they are used without a determiner.

Table 1:

Retrieval process and frequency table.

2.2 Hypergeometric distribution

There are several methods for measuring the attraction (or repulsion) between a word and a given preposition compared to other prepositions (for example, an N1 occurring in the construction cœur, compared to the number of tokens of N1 in centre, milieu, parmi and sein). Collostructional analysis (henceforth CA) is today the most frequently used method in corpus-linguistic studies. It is a family of methods which breaks down into three sub-classes (Gries 2015: 2): collexeme analysis (Stefanowitsch and Gries 2003), distinctive collexeme analysis (Gries et al. 2004b), and co-varying collexeme analysis (Gries et al. 2004a; Stefanowitsch and Gries 2005). In the context of French lexicométrie, however, a similar method has existed since the 1980s, in the seminal work by Lafon (1980) and is known as the calcul des spécificités. It has been discussed and evaluated at length (Habert 1985; Labbé and Labbé 2001; Salem 1987), and is presented in detail below. 5

The calcul des spécificités is based on a probability law, hypergeometric distribution. This is a law of distribution used to describe the probabilities of a result of sampling without replacement. Three parameters are involved: T is the total length of the corpus (in our case, T=356,380, cf. Table 1); t is the length of a sub-part of this corpus (for example, for the sub-part corresponding to cœur, t=66,501); and f is the frequency of a given word (an N1 taken from among the set of N1 in the five constructions). To understand how calcul des spécificités functions, we will take the example of conflit ‘conflict’. There are 193 tokens of this word in the FrWac corpus (f=193). Table 2 presents the distribution of these 193 tokens among the five constructions.

Table 2:

Distribution and representativeness of conflit.

We want to know whether this distribution conforms to the null hypothesis (the expected situation if the distribution of conflit were random) or if, on the contrary, conflit is overemployed (or underemployed) in a given construction, taking into account the number of tokens of each construction in the corpus (Table 1, column 5). For example, we want to know the probability of having 138 tokens of conflit after au cœur de, given that, firstly, this noun also appears 55 times after the other four prepositions, secondly, that the cœur construction has 66,501 tokens and thirdly, that the size of the total corpus is 356,380 tokens. This is what hypergeometric distribution enables us to calculate. In the situation just described, it emerges that the most likely result is 47 (this is what we call the mode value), whereas the probability of having 138 tokens of conflit (almost three times more) is close to zero.

As Lafon (1980: 141) pointed out, our intuition should nonetheless be specified. The observed value is 138. It therefore appears high and could have been even higher (139, 140, etc.). The real point of interest to us is not the probability of conflit occurring exactly 138 times in the cœur construction, but of conflit occurring therein at least 138 times. In other words, the sum of the probabilities: Prob138+Prob139+Prob140, etc. 6 This point is the main difference between the calcul des spécificités and multiple distinctive collexeme analysis. To obtain this measurement, we must go from the density to the distribution function (Appendix A, Figure 12(a)). The figure thus obtained is not however the best way of representing our problem: we are in the (right) “tail” of the curve. One way of “thickening” this tail consists in using a logged value. This value has the characteristic of tending towards – ∞ when a value is close to 0, and of being equal to 0 when its value is equal to 1. We thus obtain Figure 12(b) (Appendix A). The last operation consists in distinguishing two cases (Appendix A, Figure 12(c)). The first relates to results above the mode value (there are more tokens of conflit than expected) of which we give the absolute value and which will henceforth be positive: in the context of French lexicométrie, we call them spécificités positives ‘positive specificities’ (Lafon 1980: 142). In this paper, and to avoid the misunderstanding that the term “specificity” can involve, we will use the terms “representativeness” or “typical” (we will thus say that conflit is overrepresented or typical in the cœur construction). The second case concerns results below the mode value (there are fewer tokens of conflit than expected) called spécificités negatives ‘negative specificities’ (Lafon 1980: 142), but which we will translate by ‘underrepresented’. As Figure 13 shows (Appendix A), the latter case is illustrated by the lexeme conflit for the parmi and sein constructions (conflit is underrepresented after parmi and sein). 7 These results are interpreted as follows: regarding the lexeme conflit, it is highly probable that the cœur, parmi and sein constructions will be different to a random draw. We find, on the contrary, a high overrepresentation of conflit after cœur (in other words, a strong attraction between cœur and conflit) and conversely, a significant underrepresentation of conflit after parmi and sein (a strong repulsion).

If we now focus on the lexeme conflit compared to the other two remaining constructions, i. e., centre and milieu, the situation is quite different from the above. In both cases, the value observed is close to the mode value, i. e., the expected value in the case of a random draw (cf. Appendix A, Figure 14(a) and (d)). The number of tokens of conflit in the centre and milieu constructions is approximately the same as the one expected in the null hypothesis situation. We will say that conflit is a “commonplace” word for centre and milieu. In this context, the separation between commonplace words and typical words is a question of threshold, which is set arbitrarily. Below, we will consider a word to be commonplace when its representativeness is within the interval [–15, 15]. 8 The column “representativeness” in Table 2 summarises the results of those calculations. In Appendix B, we present the exact method for calculating the representativeness of two examples: conflit for cœur and for sein.

Let us now examine the limits of the method. They concern words having a low frequency. As an illustration, we will use the N1 poussière ‘dust’. There are only 18 tokens in our corpus and 16 are combined with milieu ‘middle’ (cf. Table 3). This results in a significant overrepresentation of poussière in the milieu construction. But owing to the smaller size of centre (cf. Table 1), and to a lesser extent, of cœur and sein, the N1 poussière also emerges as a commonplace word for these three constructions. 9 In these three cases, poussière is therefore commonplace for the wrong reason: the number of tokens is not sufficiently high. Only our intuition as speakers tells us that such wordings are odd in French, not to say impossible. This limit is examined by Salem (1987) who suggests setting a “specific absence threshold” and Labbé and Labbé (2001) who suggest “associating [with this method] a minimum frequency threshold below which the calculation will not be done”. Here, we restrict our analyses to the most frequent N1’s and only examine low-frequency N1’s when they can be grouped together in a large semantic class that is relevant for our analysis. 10

Table 3:

Distribution and representativeness of poussière.

2.3 Results and preliminary conclusion

In Appendix C, Tables 1216 respectively present the twenty most typical N1’s for the constructions centre, cœur, milieu, parmi and sein. For information, we also give the representativeness of the same N1 for the other four constructions. The representativeness of an N1 compared to a given construction is therefore a good way of quantifying its exceptional (overrepresentation or underrepresentation) or, conversely, its unexceptional nature (expected). It thus gives a measure of the degree of attraction between an N1 and a given preposition, within the framework of a given corpus and compared to other prepositions having a similar meaning. As a conclusion, the general philosophy of this method is no different to the one at work in multiple distinctive collexeme analysis.

The use of statistical association measures provides an excellent empirical aid. 11 In addition, there is little doubt that such association measures are correlated with fundamental cognitive mechanisms such as entrenchment, learning or prototype formation (cue-validity) (Gries 2012: 483). Thus, Gries (2012: 479) is right when he states that association measures “reveal functional patterns (where functional encompasses ‘semantic’, ‘pragmatic’, ‘information-structural’, etc.)”. But it should be noted that the emergence of functional patterns is not at all straightforward. To illustrate these, it could be useful to follow Stefanowitsch (2005) where an analysis is conducted in order to determine the difference between two constructions: <{in/at} the {center/heart} of NP>. This analysis is performed in the framework of conceptual metaphor (Lakoff and Johnson 1980) and this choice leads the author to assume that “the collexemes of the metaphorical expression [in/at the heart of NP] should refer to more complex (i. e., less directly experiencable and thus less easily understandable) concepts than the collexemes of the literal expression [in/at the center of NP]” (2005: 169). Unfortunately, this explanation doesn’t seem to apply easily to our data. For example, consider the word ranking of centre (Appendix C, Table 12). The first two words in the list are préoccupations ‘concerns’ and village ‘village’. The first is a plural stative (abstract) noun, whereas the second refers to a single thing with a spatial extension (concrete). 12 It is difficult to group together two such different nouns and explain their rank using the notion of “experiencability”. The situation becomes all the more complicated when we see that these two nouns are also in the top ten ranking words of cœur (Appendix C, Table 13, rank 2 and 8). Should we conclude that cœur and centre have identical functional patterns just because their distributions overlap? This would be unwise, since some N1’s very clearly make the distinction between cœur and centre (au cœur de l’hiver ‘in the heart of winter’ vs. *au centre de l’hiver *’ in the centre of winter’). 13 So if préoccupations and village are typical of centre and cœur, it is no doubt for good reasons, but these reasons may be very different from those set out in Stefanowitsch (2005). To interpret our results, we rely on a schematization of the constructions within the framework of CG. As we will see, this choice leads us in turn to investigate several variables (collective vs. atomic noun, distributive vs. collective plural, mass vs. count noun, etc.) that do not seem, at first glance, directly related to the inclusion relationship but which are crucial. This also means engaging in an exercise that contrasts totally with any statistical approach, but which we believe must be used to complement it, that is data handling validated by intuition. For example, we will use examples with a very low (or non-existent) frequency but which reveal certain semantic properties (cf. [18] for an illustration). In Section 3, such data handling will help us to discover empirical evidence that does not necessarily have any significant statistical weight, but which nonetheless sheds light on the distributional trends observed by means of the AM rankings.

3 Five different types of inclusion

3.1 Sein: A question of collective nouns

As Table 16 in Appendix C shows, sein primarily selects collective N1’s (entreprise ‘company’, équipe ‘team’, groupe ‘group’, etc.), i. e., nouns that denote something mainly characterised by its internal plurality. 14 Conversely, nouns which have no salient internal plurality are underrepresented in sein (see Table 4). This is confirmed by (6), where the N1’s, whether they are one-dimensional (6a), two-dimensional (6b) or three-dimensional (6c), sound odd with sein whereas they sound fine with centre and milieu. 15

(6)

  • a.

    au {centre / milieu / *sein} de {la / une} {rue / piste}

    ‘{in the centre of/in the middle of/*among} {the/a} {street/track}’

  • b.

    au {centre / milieu / *sein} de {la / le / d’une / d’un} {assiette / terrain / table}

    ‘{in the centre of/in the middle of/*among} {the/a} {plate/land/table}’

  • c.

    au {centre / milieu / *sein} du {couloir}

    ‘{in the centre of/in the middle of/*among} the {corridor}

Table 4:

sein and non collective nouns.

It should be added that some of these phrases can become acceptable again in an appropriate context, i. e., a context that activates a certain internal plurality in the N1, particularly via a metonymy. This is the case of (7a), where rue ‘street’ refers to the street’s inhabitants, and (7b) where table refers to the persons seated at it.

(7)

  • a.

    L’enquête repose […] sur l’observation des interactions sociales se tenant au sein de la rue de la République.

    ‘The survey is based [] on observing social interactions taking place in rue de la République.’

  • b.

    Chaque joueur au sein d’une même table a sa propre vision des choses.

    ‘Each player at the same table has his own view of things.’

However, it is evident that the sein construction is preferred for a collective lm when the N1 is polysemous (église ‘church’, corps ‘body’, espace ‘space/area’). For example, église ‘church’ has two meanings in French. The first one is synonymous with “all the members of the church” (collective meaning), the second with “building” (in this case, church denotes an object). In this configuration, au sein de systematically selects the collective meaning (8a) which is not the case for the other constructions (9a).

(8)

  • a.

    au sein de l’église‘within the church’ [“All the members of the church”]

  • b.

    au sein du corps {enseignant / diplomatique}

    ‘within the teaching staff/the diplomatic staff’

  • c.

    au sein de l’espace {économique / francophone}

    ‘within the economic/French-speaking area’

(9)

  • a.

    au {centre / cœur / milieu} de l’église

    ‘{in the centre of/inside/in the middle of} the church’ [“building”]

  • b.

    au {centre / cœur / milieu} du corps (humain)

    ‘{in the centre of/inside/in the middle of} of the (human) body’

  • c.

    au milieu de l’espace

    ‘in the middle of space’

The sein construction therefore preferably attracts collective lm, and particularly collective nouns consisting of human beings. 16 However, it repels plural landmarks, whether they comprise human or inanimate beings. They are indeed greatly underrepresented in sein, whereas they do not pose any problem in milieu and parmi ([10], Table 5).

(10)

  • a.

    {au milieu des / parmi les / *au sein des} {gens / hommes / enfants /amis} 17

    ‘{in the midst of/among/*within} {people/men/children/friends}’

  • b.

    {au milieu des / parmi les / *au sein des} {arbres / livres / fleurs / pierres}

    ‘{in the midst of/among/*within} {trees/books/flowers/stones}’

Table 5:

sein and Npl.

In CG framework, a collective noun is a singular count noun which designates one discrete entity (an inherently bounded region). A plural count noun, on the other hand, designates a replicate mass, i. e., an unbounded region (Langacker 1991b: 78). We thus assume that the sein construction attracts bounded lm. This assumption is confirmed by the fact that a plural landmark becomes natural again with sein when its lexical meaning or the context is capable of activating a certain bounded region itself. This is the case of salariés ‘employees’ in (11a), where the employees are construed as the employees of a given company (salariés is a commonplace word for sein). It is also the case for arbres ‘trees’, amis ‘friends’, gens ‘people’ or hommes ‘men’ where they are determined by a possessive (12a) or modified (by a prepositional phrase, for example) in (12b–c). In this specific context, we see that au sein de can swap with parmi without any significant change of meaning.

(11)

  • a.

    {au sein des / parmi les} salariés (de l’entreprise)

    ‘{among the/among} employees (of the company)’

(12)

  • a.

    Il y a une rébellion {au sein de / parmi} {mes amis / ses hommes}.

    ‘There is a rebellion {among/among} {my friends/his men}.’

  • b.

    {au sein des / parmi les} magnifiques arbres du parc de la Villette

    ‘{among/among} the splendid trees in the parc de la Villette’

  • c.

    {au sein des / parmi les} gens de science

    ‘{among/among} scientific people’

We may now propose a prototype diagram for sein. It must account for two constraints: sein favours an lm (i) that is bounded and (ii) has a salient internal plurality. We depict it here in Figure 1, using the schematization of collective nouns proposed by Langacker (2008: 342–343, 2009: 59). The small circles represent the elements of a plurality (for example, a plurality of team mates), the large one a collective entity (the team). Recall that Langacker makes also an important distinction between object (for example, a team which is intrinsically bounded) and unbounded mass (a mass noun or a plural such as team mates). We use the CG notational device for representing this distinction, using a circle for the former, and an ellipse for the latter (Langacker 2008: 131). The identity (or “coincidence” in Langacker’s terminology) between the plurality and the collective entity is symbolized by the equal sign ‘=’ (Langacker 2009: 59). All of these components form the conceptual base of the sein construction, but not its meaning. As pointed out by Langacker, “an expression’s meaning is a function of both the content [i. e., the conceptual base] it evokes and a particular way of construing that content. Construal includes such factors as the perspective taken on a situation, the level of specificity at which it is conceived, and the relative prominence accorded various substructures. […] An especially important kind of prominence is what I call profiling.” (Langacker 1999: 59–60). On the issue at hand, the sein construction profiles a relationship between a tr (a small circle in the diagram) and a lm (the large one). In Figure 1, as in all the others, the profile is indicated with heavy lines.

Figure 1:

Cognitive Grammar diagram for au sein de.

One may ask whether favouring bounded landmarks (the largest circle in Figure 1) prevents the sein construction from being combined with plural landmarks (an unbounded replicate mass). Table 16 in Appendix C shows that this is not the case and that many plural collective nouns (Ncollpl) are typical complements in the sein construction (entreprises ‘companies’ rank 8, établissements ‘establishments’ rank 16). In this specific case, however, each element of the plurality (which respects the bounded constraint) specifies the lm of sein, and not the plurality (which does not respect the bounded constraint). We represent this phenomenon in Figure 2, where a composite structure (au sein des équipes ‘within the teams’) is construed in relation to its symbolic components (the complex preposition au sein de and a plural collective noun équipes). 18 Composite structure and components make up a complex symbolic assembly by virtue of a set of correspondences (represented by dotted lines) that links the components with the composite structure (vertical correspondences) and the two components together (horizontal correspondences). We have chosen to depict only the latter so as not to overload the schemas. As we can see in Figure 2, the lm of sein corresponds to each team (which denotes a bounded entity), and not the plurality of teams (an unbounded plural mass). Lastly, it should be mentionned that component structures do not have the same weight in the integration process. Most of the time, the composite structure inherits the profile of one of its components, the profile determinant, which corresponds, in the case of au sein des équipes, to the complex preposition (indicated by a heavy-line box). This analysis is confirmed by the distributive reading systematically seen when au sein de introduces an Ncollpl (13).

(13)

au sein des {entreprises / équipes/etc.}

‘within {companies/teams/etc.} [→ in each company/→ in each team]’

Figure 2:

Complex symbolic assembly corresponding to au sein des équipes.

3.2 The constructions milieu and parmi

3.2.1 Milieu: A question of coincidence

As (10) show, milieu differs from sein in that it may be combined with any type of plurality (cf. Table 5). We also see that mass nouns are overrepresented in the milieu construction (cf. Table 6). By mass nouns, we mean all nouns that cannot follow a numeral determiner (or which change meaning when they do) and which take the partitive determiner (du / de la). They may therefore be collective mass nouns (foule ‘crowd’), concrete mass nouns (eau ‘water’, pelouse ‘lawn’, verdure ‘greenery’, [14a]) or abstract mass nouns (chaos ‘chaos’, désordre ‘disorder’, vacarme ‘din’, tumulte ‘commotion’, confusion ‘confusion’, [14b]), plural concrete mass nouns (terres ‘lands’, eaux ‘waters’, [14c] [Acquaviva 2008: 109]) or pluralia tantum (ruines ‘ruins’, décombres ‘remains’, [14d]).

Table 6:

milieu and mass nouns.

(14)

  • a.

    au milieu {de l’eau / de la poussière / de la pelouse / de la verdure / du sable / du feu / de la lande}

    ‘in the middle {of the water/the dust/the lawn/the greenery/the sand/the fire/the moor}’

  • b.

    au milieu du {chaos / vacarme / silence}

    ‘in the middle of/amidst the {chaos/din/silence}’

  • c.

    au milieu des {terres / eaux / airs / herbes / sables / glaces}

    ‘in the middle of the {lands/waters/airs/grasses/sands/ice+pl}’

  • d.

    au milieu des {vestiges / débris / ordures / décombres}

    ‘in the middle of the {vestiges/debris/waste+pl/remains}’

As mass nouns, these nouns are not inherently bounded. In this respect, they are similar to plural landmarks. We deduce that the milieu construction does not bound its lm, which we represent using an ellipse in Figure 3. This schematic representation explains both the combination of milieu with a mass N1 (Figure 4) and the combination with a plural landmark (Figure 5). In these two cases, indeed, the lm of milieu is felicitously specified by an unbounded mass (represented by an ellipse in the two figures): the mass noun water in Figure 4, or the plurality of stars in Figure 5. In the latter case, our analysis is confirmed by the collective reading obtained when milieu is combined with a plural landmark (15).

(15)

au milieu des étoiles

‘in the middle of the group formed by the stars’

Figure 3:

Cognitive Grammar diagram for au milieu de.

Figure 4:

Complex symbolic assembly corresponding to au milieu de l’eau.

Figure 5:

Complex symbolic assembly corresponding to au milieu des étoiles.

Yet, milieu can also be combined with a noun denoting a bounded thing, as examples 6 and Table 4 show. In this context, au milieu de is generally regarded as a synonym of au centre de and differs simply by its greater imprecision. But in fact, the sense of imprecision stems from a more subtle principle. The milieu construction is the only one that can describe situations in which the tr and the lm have what we might call a relation of superposition. By “superposition”, we mean a relation in which the tr is located inside the lm, but in such a way that it remains isolated from the lm and has no salient functional link with it. This profile is well illustrated by the unusual behavior that milieu displays with certain collective nouns typical of sein, but greatly underrepresented in milieu: entreprise ‘company’, équipe ‘team’, association ‘association’, etc. In most occurrences of au milieu de l’entreprise ‘in the middle of the company’, for example, milieu is not an ILN but takes on the meaning of “social group” (16).

(16)

Je m’intéresse au milieu de l’entreprise.

‘I am interested in the corporate world.’

When milieu is an ILN, the collective meaning of entreprise ‘company’ systematically gives way to the spatial meaning (not the collective one). This is the case in (17), where the boss is not construed as a member of the company, but merely located in the space the company occupies while remaining free of connection to it. The result is quite odd.

(17)

*Le patron se trouve au milieu de l’entreprise.

*‘The boss is in the middle of the company.’

The reason is that when it specifies the lm of an inclusion relation, entreprise ‘company’ requires strong functional dependence of the tr on the lm. Being included in a company implies taking part, playing a role, holding a position in it. Now, this is exactly the kind of relationship that milieu does not seem able to profile. Rather, it construes a functional independence between the tr and the lm which poses a problem with all nouns having a salient internal structure and for which the inclusion of the tr implies functional dependence between the tr and the lm. This intuition is confirmed by (18) which sound quite odd, as is the case for (17).

(18)

  • a.

    *La Chine est au milieu de l’échiquier asiatique.

    *‘China is in the middle of the Asian scene.’

  • b.

    *Peter est au milieu {de l’organisation politique / du réseau}.

    *‘Peter is in the middle {of the political organisation/network}.’

  • c.

    *(Replacer) la victime au milieu de l’institution judiciaire.

    *‘(Put) the victim back in the middle of the judicial institution.’

  • d.

    *(Replacer) l’élève au milieu du système scolaire.

    *‘(Put) the pupil back in the middle of the education system.’

These examples give the impression that we are simply seeking to locate the tr (La Chine ‘China’, Peter ‘Peter’, etc.) inside the lm without any functional relationship being established between the two (i. e., a simple superposition). But establishing a superposition relationship between the tr and the lm poses a problem insofar as échiquier asiatique ‘Asian scene’, organisation politique ‘political organisation’, réseau ‘network’, etc. also require a functional integration of the tr in the lm. Conversely, the superposition relationship disappears in (19), in which milieu is replaced by centre. Section 3.3.1 will indeed show that centre differs from milieu by profiling a salient relationship of functional dependence between its tr and its lm.

(19)

  • a.

    La Chine est au centre de l’échiquier asiatique.

    ‘China is at the centre of the Asian chessboard/scene.’

  • b.

    Peter est au centre {de l’organisation politique / du réseau}.

    ‘Peter is at the centre {of the political organisation/network}.’

  • c.

    (Replacer) la victime au centre de l’institution judiciaire.

    ‘(Making) the victim the focus of the judicial institution.’

  • d.

    (Replacer) l’élève au centre du système scolaire.

    ‘(Making) the pupil the focus of the education system.’

One of the main differences between milieu and centre is therefore that the former implies a superposition relationship (absence of functional relations between the tr and the lm), while the latter construes a functional dependency between the two participants. This established, we can now go back to (6) and put forward another explanation concerning the impression of imprecision that milieu produces (compared to centre). We assume that the relation of superposition, which (18) clearly illustrate, is inherent in the meaning of milieu and is maintained for (6). In this context, the tr is located in relation to the lm, but it is construed as functionally independent from the lm. To put it simply, in la salade est au milieu de l’assiette ‘the salad is in the middle of the plate’ (cf. [6b]), the salad is there a bit “by chance”, whereas in la salade est au centre de l’assiette ‘the salad is in the centre of the plate’, the salad has a functional relationship with the rest of the plate (for example, a very precise arrangement of the food in accordance with rules of culinary presentation).

Another important characteristic distinguishes milieu from the centre construction, but also parmi and sein. 19 Au milieu de can be combined with nouns having a temporal profile and itself take a temporal meaning (Appendix C, Table 14, rank 1, 3, 4, 8 and 14, [20a]). To do so, the temporal noun must merely have two salient bounds, i. e., a start and an end, unlike for example, commencement ‘beginning’ and fin ‘end’ which only have one (20b). We also find this temporal profile with all events which denote, by definition, a temporal relational profile. Thus, in (20c), the time of departure occurs at a time midway between the beginning and end of the period denoted by N1 (réunion ‘meeting’, conférence ‘conference’).

(20)

  • a.

    Il est né au {*au centre de / au milieu de / *parmi / *au sein de} {des années soixante / de la nuit / de la journée / de cette année / du siècle dernier}.

    ‘He was born {*in the centre of/in the middle of/*among/*within} {the sixties/the night/the day/this year/the last century}.’

  • b.

    *au milieu {du commencement / de la fin}

    *‘in the middle {of the beginning/the end}’

  • c.

    Il est parti {*au centre de / au milieu de / *parmi / *au sein de} {la réunion / la conférence}.

    ‘He left {*in the centre of/in the middle of/*among/*within} {the meeting/the conference}’.

As we have said, milieu profiles a superposition relation when N1 has a spatial extent. That kind of relation is obviously not suitable in the case of a temporal N1. However, we consider that both situations share an identical schematic relation, i. e., a relation of coincidence between the tr and the lm. In the space domain, this coincidence translates into a superposition relation, in which the tr is functionally separated from the lm although located within it. In the temporal domain, the coincidence corresponds to a relation of concomitance, so that the event denoted by the tr occurs within the time interval denoted by the lm but without any functional relation with it. In both cases, the relation between the tr and the lm implies that there is no functional link between the two (that’s a coincidence).

3.2.2 Parmi: A question of plurality

Parmi differs from milieu on two points: it does not have any temporal meaning (20), and cannot easily introduce concrete or abstract mass nouns. 20 However, parmi and milieu share the fact that they are frequently combined with plural landmarks ([10] and Appendix C, Table 15). In addition, pluralia tantum and some plural concrete mass nouns are relatively frequent in parmi (Table 6, rank 20, 40, 42, 95, 154, 171, 249, 324). We can deduce two main constraints for the schematization of parmi: this preposition selects an lm (i) having an internal plurality and (ii) which is not necessarily bounded. We therefore represent parmi in Figure 6. It differs from milieu in that its lm is prototypically heterogeneous (this heterogeneity corresponds to the small circles in Figure 6) whereas that of milieu is prototypically homogeneous (cf. Figure 3 in which there are no such circles).

Figure 6:

Cognitive Grammar diagram for parmi.

The question arises of whether milieu and parmi are synonymous when they are combined with a plural landmark (10). The answer is no, for a simple reason but one that is impossible to see based solely on a word ranking. As (21a) shows, parmi has a partitive reading (Hilgert 2007; Kleiber 2001; Kwon-Pak 2006; Theissen 2003) that milieu does not (21b).

(21)

  • a.

    Il y a des victimes parmi les hommes.

    ‘There are some casualties among the men.’

  • b.

    Il y a des victimes au milieu des hommes.

    ‘There are some casualties in the middle of the group of men.’

A partitive reading consists in “extracting a sub-set, including the singleton, from a bigger set” (Theissen and Schnedecker 2003: 4). This is illustrated in (21a), where the victims are construed as a sub-set of the set of men. Conversely, those same victims are construed as a separate set from the set of men in (21b) (even though they occupy the same space as the one occupied by the men). This distinction reinforces our conception of the meaning of milieu (cf. previous section), which construes a relation of superposition, or more generally, of coincidence between the tr and the lm.

One final remark about the difference between parmi and sein is in order. We note that, like the sein construction, parmi may be combined with certain collective nouns (Table 7, [22]). However, it turns out that parmi can only precede a reduced sub-class of collective nouns, i. e., those with a lexical meaning the closest to a real plurality and devoid of any salient internal organisation: élite ‘elite’, population ‘population’, assistance ‘audience’, etc. While such collective nouns may be relatively frequent in the sein construction, the latter tends to favour Ncoll that have a salient internal structure, with established hierarchies and functional relationships between each component which make it difficult for such collective nouns to be reduced to a mere plurality of elements (entreprise ‘company’, association ‘association’, système ‘system’, cf. Section 3.1). Such “cohesive” collective nouns are considerably underrepresented in the parmi construction (cf. Appendix C, Table 16), suggesting that parmi does not combine with nouns having a salient internal structure and that it favours a lm which can be easily reduced to a plurality of elements. 21

Table 7:

parmi and Ncoll.

(22)

  • a.

    La propagation de la maladie parmi {la population / le personnel}

    ‘The spread of the disease among {the population/the staff}’

  • b.

    Le retour parmi l’élite.

    ‘The return to the elite’

  • c.

    Il s’assoit parmi l’assistance.

    ‘He sat down among the audience.’

3.3 The centre and cœur constructions

3.3.1 Centre: A question of functional dependence

As (23a) and Table 6 show, centre differs from milieu in that centre cannot be combined with mass nouns whereas milieu can. 22 The only possibility to combine centre with a mass noun is when the context imposes a boundary on the lm ([23b] where the water is bounded by the sink).

(23)

  • a.

    au {*centre / milieu} de l’eau

    ‘in the {*centre/middle} of the water’

  • b.

    au centre de l’eau du lavabo

    ‘in the middle of the water in the washbasin’

When N1 has two meanings, a mass meaning and a countable meaning, as is the case with terre ‘earth/soil’ or peinture ‘paint/painting’, centre systematically selects the countable meaning as seen in (24) and (25), whereas milieu may also select the mass meaning (25).

(24)

  • a.

    au centre de la terre

    ‘at the centre of the Earth’

  • b.

    au centre de la peinture

    ‘in the centre of the painting’

(25)

  • a.

    au {*centre / milieu} de la terre

    ‘in the {*centre/middle} of the soil’

  • b.

    au {*centre / milieu} de la peinture

    ‘in the {*centre/middle} of the paint’

Centre does not combine either with pluralia tantum (26) or plural concrete mass nouns (27), which are also uncountable (hence, unbounded), while this combination is perfectly natural in the case of milieu.

(26)

au {*centre / milieu} des {ténèbres / flots}

‘in the {*centre/midst} of the {darkness+pl/waves}’

(27)

au {*centre / milieu} des {sables / neiges / eaux}

‘in the {*centre/midst} of the {sands/snow+pl/waters}’

However, centre easily combines with a bounded entity (6), which leads us to conclude that centre requires a saliently bounded lm. One reservation must nonetheless be made regarding certain collective nouns which also denote bounded entities but are underrepresented in centre (and typical of sein). In fact, a statement like Peter est au centre de l’équipe ‘Peter is at the centre of the team’ is not impossible. However, it induces a very particular construal (therefore more marked and hence, much less frequent), in which the tr (Peter) has a salient functional relationship with the lm (équipe ‘team’). In this case, Peter is construed as the person who ensures the team’s cohesion. The same is true of (19) (which are more frequent), where, as we have seen, centre requires a functional relationship between the tr and the lm, when, on the other hand, milieu induces a disconnection between the tr and the lm, by profiling a superposition relationship that makes (18) quite strange. This functional dependence between the tr and the lm is therefore one of the main characteristics that distinguishes centre from milieu. It is included in our schematization of centre as a double vertical line (Figure 7). 23 Figure 7 thus accounts for the two characteristics of centre: (i) a bounded lm (notated by a circle), and (ii) a relation of functional dependence between a tr and an lm (notated by a double vertical line). This diagram also explains the distributive reading seen when centre introduces a plural noun (28).

(28)

Les conditions physiques qui règnent au centre des étoiles sont terribles.

‘The physical conditions that reign at the centre of stars are terrible’ [→ inside each star]

The principle is identical to the one highlighted for sein (Figure 2). As a plurality (a replicate mass) is by definition unbounded, only the elements of the plurality (which are bounded entities) are capable of specifying the lm of centre. Thus, in Figure 8, the lm of centre is specified by each star and not the plurality of stars. It should nevertheless be noted that this distributive reading may give way to a collective reading, provided that the plurality is construed as a collective (a bounded entity). This is the situation we represent in Figure 9. The stars are profiled as being the stars of a cluster, and therefore the collection (the large circle) is able to specify the lm of centre.

Figure 7:

Cognitive Grammar diagram for au centre de.

Figure 8:

Complex symbolic assembly correspondong to au centre des étoiles.

Figure 9:

Complex symbolic assembly corresponding to au centre des étoiles (cluster).

Just like the parmi and sein constructions (and unlike milieu), centre has no temporal reading (20). But as Table 8 shows, it can be combined with (singular or plural) nominalizations, such as préoccupations ‘concerns’, réflexion ‘thinking’, apprentissage ‘learning’, (29a) or nouns formally linked to a verb, such as débats ‘debates’, démarche ‘approach’ (29b). 24

Table 8:

Centre and nominalizations.

(29)

  • a.

    Cette question est au centre {des préoccupations / de la réflexion / de l’apprentissage}.

    ‘This question is {a core concern/a core concept/at the core of learning}.’

  • b.

    Ce problème est au centre {du débat / de la démarche}.

    ‘This issue is the main focus {of the debate/of the approach}.’

Such nouns are semantically heterogeneous. They may denote both states (préoccupation ‘concern’) and events (débat ‘debate’). 25 But all these cases are based on an identical principle that we are going to address with the example of N1 apprentissage ‘learning’. In (30), learning is used in a periphrastic construction (Langacker 1991b: 35–50) and profiles a relationship between a tr (the pupil) and a lm (grammar).

(30)

L’apprentissage de la grammaire par l’élève

‘The learning of grammar by the pupil’

If we now consider (31) we see that learning can also specify the lm of centre.

(31)

au centre de l’apprentissage

‘at the centre of learning’

In this situation, the question arises as to which participant could, in the concept of learning, specify the tr of centre. The answer stems from our characterisation of centre. The tr of centre must merely have a relationship of functional dependence with the lm learning. Now, what could be more functionally dependent, for a concept like learning, than the person who learns (i. e., pupil, tr of learning) or further, the subject being learnt (grammar, lm of learning)? As we see in (32), they are both capable of specifying the tr of centre. In Figure 10, we represent the complex symbolic assembly corresponding to the second case (grammar at the centre of learning) in which the lm of centre is specified by learning, and its tr by the lm of learning (i. e., grammar). 26

(32)

Il faut remettre {l’élève / la grammaire} au centre de l’apprentissage.

‘{The pupil/Grammar} should become the main focus of learning.’

Figure 10:

Complex symbolic assembly corresponding to au centre de l’apprentissage.

The functional dependence between the tr and the lm required by centre goes much further however in its application. While, as we have just seen, it may concern the tr or the lm of a state or an event, it may also involve other secondary participants. This is the case for the instigator ([33a], where the trainer is behind the dealing), the theme ([33b], where the debate concerns State aids), or even the manner ([33c], where reading text refers to a method of learning). We also see some ambiguities where, despite the specification of its tr, centre does not create a distinction between the instigator and the theme. This is the case of (34), where Prince Charles may be the cause or the victim of the conspiracy. In both cases, however, Prince Charles is construed as having a relationship of functional dependence with the conspiracy.

(33)

  • a.

    Le soigneur de l’équipe est au centre d’un trafic de produits dopants.

    ‘The team’s trainer is at the centre of a drug dealing affair.’

  • b.

    Les aides de l’état sont au centre du débat.

    ‘State aids are at the centre of the debate.’

  • c.

    La lecture de texte est au centre de l’apprentissage.

    ‘Reading text is at the centre of learning.’

(34)

Le prince Charles est au centre du complot.

‘Prince Charles is at the centre of the conspiracy.’

3.3.2 Cœur: A question of intensity

The cœur construction is a more sophisticated puzzle than the others. As we mentioned in Section 2.3, there is a class of nouns typical of centre and cœur that are underrepresented in milieu and sein. Such is the case, in particular, for nouns such as ville ‘town’ or village ‘village’ (cf. Appendix C, Table 13, rank 1 and 8). In addition, cœur can be combined with states or events (Appendix C, Table 13, rank 2, 4, 7 and 9) and seems to easily replace centre in (29), (32), (33) and (34). In other words, cœur and centre appear to have a similar distributional profile that contrasts with the other prepositions. But there are also some striking semantic differences between centre and cœur. Unlike centre, for example, any entity that has no intrinsic functional relationship with the lm is capable of specifying the tr of cœur, as seen in (35).

(35)

  • a.

    Le journaliste est au centre du conflit.

    ‘The journalist is at the centre of the conflict’.

  • b.

    Le journaliste est au cœur du conflit.

    ‘The journalist is at the heart of the conflict.’

In (35a), the journalist is construed as being functionally dependent on the conflict (he is the instigator for instance) which implies a specific situation (a controversy triggered by the journalist). In (35b), the journalist is not necessarily a participant in the conflict: he may also merely be doing his job, i. e., reporting what he sees. We can thus conclude that while cœur does not preclude cases of functional dependence between tr and lm, it does not require such dependence either.

Other evidence leads us to establish a link between cœur and milieu (and distinguish it from centre). For example, cœur may be combined with certain unbounded N1, such as concrete mass nouns (36a-b) and pluralia tantum (ex. 36c).

(36)

  • a.

    Peter habite au {cœur / *centre / milieu} de la {nature / campagne}.

    ‘Peter lives in the {heart/*centre/depths} of {nature/the countryside}’

  • b.

    au {cœur / *centre / milieu} de l’obscurité

    ‘in the {heart/*centre/midst} of darkness.

  • c.

    au {cœur / *centre / milieu} des ténèbres

    ‘in the {heart/*centre/midst} of darkness+pl.

It is true that the preposition au centre de can appear in the context of (36a), but it then forces a very precise construal. When preceded by centre, nature becomes synonymous with “creation” and the tr becomes a functional component of the lm nature. In our corpus, that tr is always the same, namely homme ‘man’, as in il faut remettre l’homme au centre de la nature ‘man should be put back at the centre of nature’ (in the sense that man should control and channel it). The same applies to the lm campagne ‘countryside’, which takes, after centre, the meaning of “election campaign” or “military campaign” and where the tr corresponds to the person in charge of the campaign (this configuration is much more frequent as these meanings are lexicalized for campagne). 27 In both cases, as we see, nature and campagne are bounded (they can be counted) when they are the complement of centre, whereas they are not necessarily so after au cœur de ([36a] where nature and campagne are mass nouns).

Like the milieu construction, cœur can also be combined with nouns having a temporal profile, some of which (particularly the seasons) are even typical for cœur (nuit ‘night’, Appendix C, Table 14, rank 3, hiver ‘winter’, été ‘summer’, etc., as seen in [37]).

(37)

au {cœur / milieu / *centre} {de la nuit / de l’hiver / de l’été / du mois d’aout / de l’année 1971}

‘in the {heart/middle/*centre} of {the night/winter/summer/1971}’

The meaning of cœur must nonetheless be quite different to milieu, since in the case where N1 designates a plurality of things, cœur behaves in exactly the same way as centre. Indeed, (38) shows that cœur requires the same distributive reading as centre ([28], Figure 8), whereas milieu prefers a collective reading ([15], Figure 5).

(38)

Les conditions physiques qui règnent au cœur des étoiles sont terribles.

‘The physical conditions that reign inside stars are terrible’

Another interesting fact is the existence of a sub-class of things (already mentioned with sein, cf. [6], Table 4) that are underrepresented in cœur, but are typical (or commonplace) for centre and milieu. These things may be one-dimensional (39a), two-dimensional (39b) or three-dimensional (39c).

(39)

  • a.

    au {centre / *cœur / milieu} de la piste‘in the {centre/*heart/middle} of the track’ 28

  • b.

    au {centre / *cœur / milieu} {de l’assiette / de la table / du cercle / du plat}

    ‘in the {centre/*heart/middle} of the {plate/table/circle/dish}’

  • c.

    au {centre / *cœur / milieu} {de la pièce / du couloir / du bureau}

    ‘in the {centre/*heart/middle} of the {room/corridor/office}’

In summary, our schematization of the meaning of cœur must therefore account for several facts: the possibility of introducing states or events (au cœur {de la préoccupation / du conflit} ‘at the heart {of the concern/of the conflict’}) and its attraction for certain temporal nouns (au cœur de l’hiver ‘in the heart of winter’), certain uncountables (au cœur {de la nature / des ténèbres} ‘in the heart {of nature/of darkness’}), and certain things (au cœur de la ville ‘in the heart of town’). In addition, the underrepresentation of things that are nonetheless natural in centre and milieu must be explained: assiette ‘plate’, table ‘table’, couloir ‘corridor’. Lastly, we must examine the causes of the distributive reading that cœur requires when N1 denotes a plurality of things (au cœur des étoiles ‘inside stars’).

Table 9 provides an important clue. It shows that certain abstract mass nouns (which take the partitive article and do not accept the plural without changing meaning) are typical in cœur: vie ‘life’, foi ‘faith’, mystère ‘mystery’, tourmente ‘turmoil’, engagement ‘commitment’, etc. Such nouns were studied in detail in Van de Velde (1995) who calls them noms intensifs meaning ‘intensive nouns’. They have the distinctive feature of designating qualities or states of which the primary domain is neither space nor time, but intensity. One way of demonstrating this uses the concept of expansibility which characterizes mass nouns (Langacker 1991a: 71–72, 2008: 142). In French, the mass phrase de la marche ‘walking’ has an expansibility which comes within the primary domain of time (de la marche plus de la marche gives walking that lasts longer). The expansibility of de l’eau ‘water’ is in the space domain (if we add de l’eau to de l’eau ‘water to water’, the result takes up more space). But the expansibility of du mystère ‘mystery’ fits into a scale of intensity: adding mystery to mystery increases the degree of mystery. 29 The overrepresentation of intensive nouns in cœur thus prompts us to make the following assumption: the lm of cœur elaborates an intensity scaling subject to variation, of which cœur detects the maximum. A schematic diagram of this principle is provided in Figure 11, where the lm corresponds to an intensity scale, and where the tr is located in relation to its maximum. From this perspective, au cœur du mystère ‘at the heart of the mystery’ means the area in which the mystery is the most intense.

Table 9:

Cœur and intensive nouns.

Figure 11:

Cognitive Grammar diagram for au cœur de.

It can be mentioned in passing that while intensive nouns are greatly overrepresented in cœur, this does not prevent them from occurring after the other prepositions. For example, vie ‘life’ (Table 9, rank 19) and foi ‘faith’ (rank 30) are respectively typical and commonplace words of centre. However, the ensuing construal is not at all the same. When vie ‘life’ or foi ‘faith’ are combined with centre, the functional relationship between the tr and the lm becomes salient. In (40), for example, life and faith are devoted to the bible or guided by the bible. The same is true with pouvoir ‘power’ (Table 9, rank 129) which is a commonplace word for centre and sein. 30 When pouvoir is combined with centre, the functional relationship between the tr and pouvoir becomes salient as in (41a), where the monarch is the person who holds power. In the context of sein, pouvoir takes a collective meaning, as in (41b), where it means “group of people who run a country”.

(40)

Il a placé la bible au centre de sa {vie / foi}.

‘He has put the bible at the centre of his {life/faith}.’

(41)

  • a.

    Le monarque est au centre du pouvoir administratif et financier.

    ‘The monarch is the keystone of administrative and financial power.’

  • b.

    Il y a une lutte de clans au sein du pouvoir soudanais.

    ‘There is a struggle between clans within power in Sudan.’

Our hypothesis thus accounts for the combination of cœur with an intensive noun. But as we have said, cœur also (and especially) combines with countable nouns which have no intrinsically intensive meaning (ville ‘town’, village ‘village’, étoile ‘star’). How could this second case be reflected? We argue that cœur has the ability to bring to the forefront intensive values that are not part of the meaning of the N1, as is the case of intensive nouns, but which are prototypically associated with it. When we scan the surface area of a town, we see that certain parts differ from others by their greater qualitative intensity. For example, au cœur de la ville ‘in the heart of the town’ means the liveliest place. This part is generally located in a place that is farthest from the edges, and this causes the confusion between cœur and centre. But while the place located by means of these two constructions appears to be the same (au {centre / cœur} de la ville ‘in the {centre/heart} of the town’), it is important to bear in mind that they are the result of two different construals. Conversely, such qualitative variations are more difficult to highlight in the case of a plate or a corridor. By scanning the surface of a plate or the inside of a corridor, we do not discover any place that is qualitatively different from the others. They cannot therefore felicitously specify the lm of cœur, which explains their underrepresentation. Some temporal nouns may also have such qualitative variations. For example, au cœur de l’hiver ‘in the heart of winter’ means the time when the weather is coldest, au cœur de la nuit ‘in the heart of the night’ means the time when it is darkest. Conversely, temporal nouns such as mois ‘month’ or année ‘year’ do not have any obvious qualitative variation (they are underrepresented in cœur and overrepresented in milieu). They only occur after au cœur de when they are modified: au cœur du mois d’aout ‘in the heart of August’ ([37], where August is prototypically associated with summer), au cœur de l’année 1971 ‘in the heart of 1971’ ([37], where the quantitative variations may concern the fashion of the time or the political situation, etc.). So it is only once they are modified that month or year may have the expected qualitative variations. The case of stative nouns (préoccupation ‘concern’), events (conflit ‘conflict’) and concrete mass nouns (nature) is less problematic insofar as intensive quantities are more readily accessible: au cœur de la préoccupation ‘at the heart of the concern’ locates the tr in relation to a maximum degree of concern; au cœur du conflit ‘at the heart of the conflict’ (35b) refers to the place (in a spatial context) where the conflict reaches its greatest intensity; au cœur de la nature ‘in the heart of nature’ (36a) means the place where nature is the most luxuriant. To finish, we must explain the reason for the distributive reading of (38), which is simple. A plural is a replicate mass and precisely because it is a replication, a plural is the configuration the least capable of expressing the qualitative variations required by cœur. Thus, au cœur des étoiles ‘inside the stars’ has a distributive reading because as a replication of conceptually identical stars, a plurality of stars does not have any qualitative variation enabling it to be combined with cœur. Therefore, it is each star that offers the expected qualitative variation, hence the distributive reading. Only certain pluralia tantum (ténèbres), which are not replicate masses, may have such qualitative variations: au cœur des ténèbres ‘in the heart of darkness+pl’ means the place where the darkness is the most intense.

4 Conclusion

At the end of this study, the constructions centre, cœur, milieu, parmi and sein, which most studies regard as synonymous, in fact emerge as very different. They construe inclusion relations that differ from one another on the basis of numerous criteria: internal plurality of the lm, bounded (or unbounded) lm, degree of functional dependence between the tr and the lm (or conversely, simple coincidence), degree of internal structure of the lm, homogeneity or heterogeneity of the lm, and to finish, variation of an intensive magnitude associated with the lm (or absence of any intensive magnitude). The reader may have noticed that none of these criteria are geometrical or physiological. For example, we have not taken the property of equidistance into consideration, even though it is contained in the meaning of centre. Neither have we relied on the connotations that cœur has (as the centre of emotions). Yet, such values are naturally not extraneous to the semantic functioning of the prepositions. The mathematical concept of equidistance is directly linked to the bounded requirement highlighted in Section 3.3.1. Similarly, as a prototypical substratum of emotions, it is not surprising that au cœur de is used to express intensive magnitudes (Section 3.3.2). However, these criteria have one crucial drawback as they do not allow generalisations to be made beyond their scope of application (respectively, geometry and feeling). It would be very difficult to analyze examples such as l’élève est {au centre / cœur} de l’apprentissage ‘the pupil is {at the centre/heart} of learning’ in terms of equidistance (with centre) or emotion (with cœur). This is because such constructions are today grammaticalised and feature significant schematicity.

The schematic constraints identified in the above sections allow us to predict which facets or meanings of a given N1 are activated by a given preposition. 31 An N1 like conflit, for example, is construed very differently depending on the mode of inclusion chosen. The phrase au centre du conflit ‘at the centre of the conflict’ tends to profile a tr that is functionally dependent on the conflict (the issue or the instigator of the conflict). Au cœur du conflit ‘at the heart of the conflict’ profiles the place or time at which the conflict reaches its height. The phrase au sein du conflit ‘within/amidst the conflict’ is less frequent (cf. Table 2) as it forces conflit to take on a collective value that is not inherent in its meaning (see [7] for similar cases). The phrase au milieu du conflit ‘in the middle of the conflict’ can be understood in two ways: a temporal reading (Peter est parti au milieu du conflit, ‘Peter left in the middle of the conflict’, relation of concomitance), and a spatial reading (Peter se retrouve au milieu du conflit ‘Peter is in the middle of the conflict’). In the latter case, Peter’s position is inside the conflict zone, but Peter is not actively involved in the conflict (superposition relation). Only parmi cannot be combined with conflit (*parmi le conflit *‘among the conflict’), because the meaning of conflit cannot amount to a simple plurality of individuals and does not therefore satisfy the requirements of parmi.

The polysemous N1 monde ‘world’ is another example that illustrates these changes of meaning. The phrase au centre du monde ‘in the centre of the world’ systematically establishes a functional relationship between its tr and monde. In astronomy, the Earth frequently plays the role of tr (La terre est au centre du monde ‘The Earth is at the centre of the world’ where monde means ‘universe’). But other domains are also involved, such as geopolitical area in La Chine est au centre du monde ‘China is at the centre of the world’. When monde is introduced by au sein de, it is almost always modified by an adjective or a prepositional phrase and takes a collective meaning: au sein du monde {éducatif / administratif / associatif} ‘within the world of {education/administration/associations}’; au sein du monde du travail ‘within the world of work’. Au cœur du monde ‘at the heart of the world’, on the other hand, is a phrase of which half the tokens are in the religious sphere: le fils de dieu descendu au cœur de ce monde, pour sauver tous les hommes ‘the son of god descended to the heart of this world, to save all men’. In this case, world takes on intensive values such as “humanity” or “materiality”. The same is true of au milieu du monde ‘to the middle of the world’ which appears principally in the religious domain and where this time the believers are construed as being disconnected from a world that does not necessarily share the same values (superposition relation). Lastly, parmi le monde ‘among/amidst the world’ (which is much more infrequent) construes world as a plurality of individuals (and becomes synonymous with “people”). This is why it favours the presence of quantifiers not found with the other prepositions: parmi {tout ce / tant de / beaucoup de} monde ‘among/amidst {all these/so many/a lot of} people’.

In conclusion, this study demonstrates that the “inclusion” label, which is usually used in semantics for its convenience, is not, despite its apparent simplicity, easy to handle. The inclusion relation is not a formal relation that applies regardless of the conceptual content of its complements. Evidence shows, on the contrary, that the inclusion relation encompasses very diverse construals and involves multiple semantic parameters.

Acknowledgement

I would like to thank Sylvain Loiseau for feedback on the R script and three anonymous reviewers for their comments. This article owes much to the editing of John Newman. I thank him for his advice and comments. All remaining errors are of course my own.

References

  • Acquaviva, Paolo. 2008. Lexical plurals. Oxford: Oxford University Press.

  • Ashino, Fumitake. 2007. A propos de la préposition parmi : Étude contrastive avec entre. Cahiers d’Études Françaises 12. 1–16.

  • Aurnague, Michel. 1991. Contribution à l’étude de la sémantique formelle de l’espace et du raisonnement spatial: La localisation interne en français, sémantique et structures inférentielles. Toulouse: Université Paul Sabatier.

  • Aurnague, Michel. 1996. Les noms de localisation interne: Tentative de caractérisation sémantique à partir de données du basque et du français. Cahiers de Lexicologie 69. 159–192.

  • Baroni, Marco, Silvia Bernardini, Adriano Ferraresi & Eros Zanchetta. 2009. The WaCky wide web: A collection of very large linguistically processed web-crawled corpora. Language Resources and Evaluation 43(3). 209–226.

  • Borillo, Andrée. 1988. Le lexique de l’espace: Les noms et les adjectifs de localisation interne. Cahiers de grammaire 13. 1–22.

  • Borillo, Andrée. 1998. L’espace et son expression en français. Paris: Ophrys.

  • Borillo, Andrée. 1999. Partition et localisation spatiale: Les noms de localisation interne. Langages 13. 53–75.

  • Bybee, Joan. 2010. Language, usage and cognition. New York: Cambridge University Press.

  • Fagard, Benjamin. 2012. Prépositions et locutions prépositionnelles: La question du renouvellement grammatical. Travaux de linguistique 64(1). 161–183.

  • Ferraresi, Adriano, Silvia Bernardini, Giovanni Picci & Marco Baroni. 2010. Web corpora for bilingual lexicography: A pilot study of English/French collocation extraction and translation. In Richard Xiao (ed.), Using corpora in contrastive and translation studies, 337–359. Newcastle: Cambridge Scholars Publishing.

  • Flaux, Nelly & Danièle Van de Velde. 2000. Les noms en français: Esquisse de classement. Paris: Ophrys.

  • Franckel, Jean-Jacques & Denis Paillard. 2007. Grammaire des prépositions, tome 1. Paris: Ophrys.

  • Goldberg, Adele E. 1995. Constructions: A construction grammar approach to argument structure. Chicago: The University of Chicago Press.

  • Gréa, Philippe. 2015. Entre et parmi: Deux perspectives sur la pluralité. Travaux de linguistique 70(1). 7–38.

  • Gries, Stefan Th. 2012. Frequencies, probabilities, and association measures in usage-/exemplar-based linguistics: Some necessary clarifications. Studies in Language 36(3). 477–510.

  • Gries, Stefan Th. 2015. More (old and new) misunderstandings of collostructional analysis: On Schmid and Küchenhoff (2013). Cognitive Linguistics 26(3). 505–536.

  • Gries, Stefan Th. & Anatol Stefanowitsch. 2004a. Co-varying collexemes in the into-causative. In Michel Achard & Suzanne Kemmer (eds.), Language, culture, and mind, 225–236. Stanford, CA: CSLI.

  • Gries, Stefan Th. & Anatol Stefanowitsch. 2004b, Extending collostructional analysis: A corpus-based perspective on <alternations>. International Journal of Corpus Linguistics 9(1). 97–129.

  • Guentchéva, Zlatka. 2003. Entre: Préposition et préfixe. In Peter Blumenthal & Jean-Emmanuel Tyvaert (eds.), La cognition dans le temps: Ètudes cognitives dans le champ historique des langues et des textes, 59–74. Tübingen: Max Niemeyer Verlag.

  • Habert, Benoit. 1983. Etude des formes spécifiques et typologie des énoncés (Les résolutions générales des congrès de la CFTC-CFDT de 1945 à 1979). Mots 7(1). 97–121.

  • Habert, Benoit. 1985. L’analyse des formes “spécifiques” [bilan critique et propositions d’utilisation]. Mots 11(1). 127–154.

  • Hilgert, Emilia. 2007. Étude de parmi. Le cas des tours partitifs. Scolia 21. 37–66.

  • Hilgert, Emilia. 2009. Retour sur les prépositions ensemblistes à interprétation spatiale. Scolia 24. 53–69.

  • Hilgert, Emilia. 2010. La partition et ses constructions en français. Genève: Librairie Droz.

  • Hilgert, Emilia. 2013. Les prépositions ensemblistes et la question de leur emploi spatial. In Benjamin Fagard & Dejan Stosic (eds.), Langue, espace, cognition. [Special issue]. CORELA. HS-12. http://corela.revues.org/2808 (accessed 18 December 2015).

  • Kleiber, George. 2001. Indéfinis: Lecture existentielle et lecture partitive. In George Kleiber, Brenda Laca & Lilianne Tasmowski (eds.), Typologie des groupes nominaux, 47–97. Rennes: Presses Universitaires de Rennes.

  • Küchenhoff, Helmut & Hans-Jörg Schmid. 2015. Reply to “More (old and new) misunderstandings of collostructional analysis: On Schmid & Küchenhoff” by Stefan Th. Gries. Cognitive Linguistics 26(3). 537–547.

  • Kwon-Pak, Song Nim. 2006. Entre vs. parmi: Deux prépositions au centre de la partition. In Georges Kleiber, Catherine Schnedecker & Anne Theissen (eds.), La Relation partie-tout, 651–668. Leuven: Peeters.

  • Labbé, Cyril & Dominique Labbé. 2001. Que mesure la spécificité du vocabulaire? Lexicometrica 3. http://lexicometrica.univ-paris3.fr/article/numero3.htm (accessed 18 December 2015).

  • Lafon, Pierre. 1980. Sur la variabilité de la fréquence des formes dans un corpus. Mots 1. 127–165.

  • Lafon, Pierre. 1984. Dépouillements et statistiques en lexicométrie. Genève-Paris: Slatkine-Champion.

  • Lakoff, George & Mark Johnson. 1980. Metaphors we live by. Chicago: The University of Chicago Press.

  • Lammert, Marie & Michelle Lecolle. 2014. Les noms collectifs en français: Une vue d’ensemble. Cahiers de lexicologie : Revue internationale de lexicologie et lexicographie 105. 203–222.

  • Langacker, Ronald W. 1987. Foundations of Cognitive Grammar: Theoretical prerequisites. Stanford: Stanford University Press.

  • Langacker, Ronald W. 1991a. Concept, image, and symbol. Berlin: Mouton De Gruyter.

  • Langacker, Ronald W. 1991b. Foundations of Cognitive Grammar: Descriptive application. Stanford: Stanford University Press.

  • Langacker, Ronald W. 1999. Correspondences, compositionality, and grammar. In Fernando Toda Iglesia, Juan A. Prieto Pablos, Mariá José Mora & Teresa López Soto (eds.), Actas del XXI Congreso International de AEDEAN (Asociación Española de Estudios Anglo-Norteamericanos), 55–74. Sevilla: Univesidad de Sevilla.

  • Langacker, Ronald W. 2000. Grammar and conceptualization. Berlin: Mouton de Gruyter.

  • Langacker, Ronald W. 2008. Cognitive Grammar. A basic introduction. Oxford: Oxford University Press.

  • Langacker, Ronald W. 2009. Metonymic grammar. In Klaus-Uwe Panther, Linda L. Thornburg & Antonio Barcelona (eds.), Metonymy and metaphor in grammar, 45–71. Amsterdam: John Benjamins.

  • Lebart, ‎Ludovic & André Salem. 1994. Statistique textuelle. Paris: Dunod.

  • Loiseau, Sylvain. 2006. Sémantique du discours philosophique: Du corpus aux normes: Autour de G. Deleuze et des années 60. Nanterre: Université de Paris X Nanterre.

  • Loiseau, Sylvain. forthcoming. The lexicométrie school: History, theories and methods. In Jacqueline Léon & Sylvain Loiseau (eds.), Quantitative linguistics in France. Lüdenscheid: Ram Verlag.

  • Pincemin, Bénédicte. 2012. Sémantique interprétative et textométrie. Texto! XVII(3). http://www.revue-texto.net/index.php?id=3049 (accessed 18 December 2015).

  • Salem, André. 1987. Pratique des segments répétés: Essai de statistique textuelle. Paris: Klincksieck.

  • Schmid, Hans-Jörg & Helmut Küchenhoff. 2013. Collostructional analysis and other ways of measuring lexicogrammatical attraction: Theoretical premises, practical problems and cognitive underpinnings. Cognitive Linguistics 24(3). 531–577.

  • Silberztein, Max. 2003. NooJ manual. http://www.nooj4nlp.net (accessed 18 December 2015).

  • Silberztein, Max. 2004. NooJ: An object-oriented approach. In Claude Muller, Jean Royauté & ‎Max Silberztein (eds.), INTEX pour la Linguistique et le Traitement Automatique des Langues, 359–369. Cahiers de la MSH Ledoux. Besançon: Presses Universitaires de Franche-Comté.

  • Stefanowitsch, Anatol. 2005. The function of metaphor: Developing a corpus-based perspective. International Journal of Corpus Linguistics 10(2). 161–198.

  • Stefanowitsch, Anatol & Stefan Th. Gries. 2003. Collostructions: On the interaction between verbs and constructions. International Journal of Corpus Linguistics 8(2). 209–243.

  • Stefanowitsch, Anatol & Stefan Th. Gries. 2005. Covarying collexemes. Corpus Linguistics and Linguistic Theory 1(1). 1–43.

  • Theissen, Anne. 2003. Un des N et un N en lecture partitive. Langages 151. 67–86.

  • Theissen, Anne & Catherine Schnedecker. 2003. Présentation. Langages 151. 3–8.

  • Van de Velde, Danièle. 1995. Le spectre nominal. Des noms de matières aux noms d’abstraction. Bibliothèque de l’information Grammaticale. Louvain: Peeters.

  • Van Goethem, Kristel. 2009. L’emploi préverbal des prépositions en français: Typologie et grammaticalisation. Bruxelles: De Boeck Duculot.

  • Vandeloise, Claude. 2001. Aristote et le lexique de l’espace. Stanford: CSLI.

  • Vandeloise, Claude. 2004. Quatre relations fondamentales pour la description de l’espace. Histoire Épistémologie Langage 26(1). 89–109.

Appendix A

Figure 12:

Representativeness of conflit for cœur.

Figure 13:

Representativeness of conflit for parmi and sein.

Figure 14:

Representativeness of conflit for centre and milieu.

Appendix B

To calculate the representativeness, we used an R script implemented by B. Desgraupes (Modal’X, Université de Paris Ouest Nanterre la Défense) and S. Loiseau (Université de Paris 13). In order to permit the replication of our results, we illustrate this method with two examples. The first concerns the lexeme conflit in the cœur construction, and the second, conflit in the sein construction. In what follows, we summarize the data in the 2x2 cooccurrence table format.

Table 10:

Cooccurrence table (conflit and cœur construction).

This first example illustrates the situation in which the observed frequency is greater than the mode value. In this case, we compute the cumulative probability for the upper tail of the distribution.

phyper(obsVal-1, f, T-f, t, log.p=TRUE, lower.tail=FALSE)

> pObsVal <- phyper(138–1, 193, 356380–193, 66501, log.p=TRUE, lower.tail=FALSE)

> pObsVal

[1] –130.4534

With this function, however, the mode is different from 0:

phyper(modVal, f, T-f, t, log.p=TRUE)

> pModVal <- phyper(36, 193, 356380–193, 66501, log.p=TRUE)

> pModVal

[1] –0.6099561

This is not the result that we should try to achieve. On the contrary, the representativeness of the expected frequency should be held at zero. In order to do that, the value of the mode is subtracted from all values:

> representativeness <- abs(pModVal - pObsVal)

> representativeness

[1] 129.8435 (cf. Appendix A, Figure 12(c)

Table 11:

Cooccurrence table (conflit and sein construction).

In the case of the lexeme conflit for the sein construction, the observed frequency is below the mode value (cf. Appendix A, Figure 13(d)). We therefore compute the cumulative probability for the lower tail of distribution.

phyper(obsVal, f, T-f, t, log.p=TRUE)

> pObsVal <- phyper(13, 193, 356380–193, 148001, log.p=TRUE)

> pObsVal

[1] –62.48986

Again, the mode is different from 0:

phyper(modVal, f, T-f, t, log.p=TRUE)

> pModVal <- phyper(80, 193, 356380–193, 148001, log.p=TRUE)

> pModVal

[1] –0.6501442

Hence the subtraction:

> representativeness <- -abs(pModVal - pObsVal)

> representativeness

[1] –61.83971 (cf. Appendix A, Figure 13(f))

A more detailed description of this method may be found at this link:

https://r-forge.r-project.org/R/?group_id=1547

Appendix C

Table 12:

List of centre’s 20 most typical N1.

Table 13:

List of cœur’s 20 most typical N1.

Table 14:

List of milieu’s 20 most typical N1.

Table 15:

List of parmi’s 20 most typical N1.

Table 16:

List of sein’s 20 most typical N1.

About the article

Received: 2015-12-20

Revised: 2016-06-05

Accepted: 2016-06-09

Published Online: 2017-01-07

Published in Print: 2017-02-01



Citation Information: Cognitive Linguistics, ISSN (Online) 1613-3641, ISSN (Print) 0936-5907, DOI: https://doi.org/10.1515/cog-2015-0127. Export Citation

Comments (0)

Please log in or register to comment.
Log in