Explaining and teaching systematic false friends via metonymy 1
2 How false are false friends?
For a period of time the foreign language scene had been dominated by the idea of language transfer, in particular the interference or negative transfer between languages had been blamed as the major, if not the sole source of er-rors in SL/FL production (cf. Dulay and Burt 1972). It inspired a large number of cross-linguistic studies, i.e. contrastive linguistic comparisons that were expected to uncover the degree of similarity and/or difference between lan-guages in particular areas selected for study in the hope of identifying points of difficulty in the learning process. Detection of false friends, as a special type of lexical contrasts, was a natural part of this enterprise. Prefiguring classical contrastive analyses, Lado (1957: 84) thus established six classes of words obtaining in a contrastive linguistic situation based on their form and mean-ing in an attempt to predict the degree of ease or difficulty in learnmean-ing. False friends are in the category Lado calls deceptive cognates, which occupies the highest position on the scale of difficulty. They are labelled “sure-fire traps.”
This overly naïve, i.e. too optimistic and simplifying view on language trans-fer, was in due time replaced by a more realistic picture. Juhász (1970a and b) demonstrates that cases exhibiting smaller differences may cause more difficulties that blatant contrasts. On a more general level, Sharwood Smith and Kellerman (1986) demonstrate that accounting for learning or acquisition of a language must take into account many more variables than just transfer or interference.
Despite this gradual evolution of views on the role of cross-linguistic sim-ilarities and difference in the process of second or foreign language learning, hardly anything has changed concerning false friends. Admittedly, the picture has become more differentiated, as several new types and subtypes have been recognized, which has also called for more elaborate classifications of false friends.
As pointed out in the introductory section, a series of definitions have been put forward, along with proposals for making finer distinctions based on distinguishing several subtypes. An extreme type of false friends are so-called pseudo-anglicisms (or false anglicisms), pseudo-italianicisms (or false itali-anicisms), i.e. words in language other than English that leave the impression
of being borrowed from English, where no such words are actually current (cf. Breitkreuz 1976; Furiassi 2003, 2010, 2012; Görlach 1996; Sočanac 1994;
Zergollern-Miletić 1995; Antunović 1996).
False friends are often defined as lexical units belonging to different languag-es that sound same or similar, but mean utterly different things (Perl/Winter 1972; Chalker/Weiner 1996: 149), whose spoken or written form is similar, but whose meaning is different (Gottlieb 1972), and which are often etymologically related (Emericzy 1980). Chamizo Domínguez (2007) distinguishes chance or coincidental false friends, which are pairs without a common etymological ancestor, from semantic false friends, sharing the same etymological origin.
Some examples of chance false friends are:
(1) Croatian ekser ‘nail’ (derived from Turkish) – Hungarian ékszer
‘jewellery’ (from ék ‘decoration’ + suffix -szer)
(2) English coin ‘a flat piece of metal used as money’ – French coin ‘corner’
(3) Spanish chumbo ‘cactus pear’ – Portuguese chumbo ‘lead’
In this article we will be concerned with the latter type, called semantic false friends by Chamizo Domínguez. Let us illustrate them with a couple of exam-ples for which we suggest an alternative designation: systematic false friends.
We are of the opinion that this label is more useful as it captures the fact that they are not random sets, but rather regular in partly overlapping and in exhibiting some common stages in the development, and also as often as not attaining the status of internationalisms, being part of common lexical and conceptual stock of a number of language more or less closely related in genetic terms.
It will be seen that the three defining criteria mentioned in the above works (actually, these are representative of the majority of studies of this phenome-non)—formal (phonetic and/or orthographic), semantic and etymological—
are clearly correlated, but are subject to gradation. Saying that something sounds same or similar, is spoken or written in a similar way, or that they are (utterly) different in terms of their meaning, beg questions like
Still, although most authors state that false friends have different meanings in different languages, what do they mean by “having different meanings”?
Definitions seem to suggest that the semantic differences in false friends are in the denotative or referential meaning, but there are some false friends which are slightly different in their “associative meaning” (Leech 1974: 20), that is, in their register, style, frequency and/or geographical distribution. (Roca-Varela 2015: 32).
What is more, it is very likely that formal similarities may be attributed to common origin, i.e. etymology, in which case their semantic difference may (or must) be less than absolute. In the light of all this, Zethsen (2004: 139) concludes that
The phenomenon of “false friend” is not merely a question of two formally identical words having completely different meanings, but one that can operate at more subtle levels too.
O’Neill and Casanovas Catalá (2007: 108–109) address the problem of se-mantic differentiation of false friends and distinguish three possibilities. When none of the meanings of one member of the pair coincide with the meanings of the other member, we have a case of segregation. When some meanings are shared (but not the rest), they talk about intersection. Finally, the situation in which all the meanings of one member coincide with the meanings of the other member, but one of the members also exhibits additional meanings, is referred to as inclusion. Abou-Khalil, Flanagan and Ogata (2018: 2), add true friends to these three types, but also distinguish two possibilities in the case of inclu-sion, depending on whether all the meanings of the native word are included in the semantic structure of the word belonging to the target language, or the other way around. They visualize the possibilities as in the following figure:
Figure 1: Types of false friends, according to Abou-Khalil, Flanagan and Ogata (2018: 2).
The complexity of the phenomenon of false friends is reflected in the classifi-cation of false friends by Gouws, Prinsloo and De Schryver (2004: 805), which is apparently based on the idea of continuum or cline and various degrees of false friendness and semantic resemblance, as reproduced in the figure below:
Types of friends Visualization Example
F1: True friends
The words in the native language and target language have similar meanings.
F2: False friends
The words in the native language and target language have totally different meanings.
F2: Partial false friends N⋂T
Some meanings of the words are similar in the native language and the target language. Other meanings are different.
F2: Partial false friends T⊂N
Meanings of the word in the target lan-guage are included in the meanings of the word in the native language.
F2: Partial false friends N⊂T
Meanings of the word in the native lan-guage are included in the meanings of the word in the target language.
The area of the circle represents the breadth of meanings of a word
adventure (eng.) aventure (fr.)
constipated (eng.) constipado (sp.)
ancient (eng.) ancien (fr.)
driver (eng.)
ドライバ (doraiba) (jp�)
demand (eng.) demander (fr.) native
language target
language
Types of false friends between a student’s native language and the learned target language.
Figure 2: A semantic continuum of false friends, according to Gouws, Prinsloo and de Schryver (2004: 805)
3 On metonymy
In order to fully understand the potential of metonymies in explaining and teach-ing false pairs, it seems apposite to shed some light on the nature of metonymy as a basic cognitive process, i.e. define it and introduce its main types. Most classic definitions of metonymy mention that it is a figure of speech, or a poetic figure.
This would imply that metonymy is a mere ornament because it is just another way of saying something that could be stated in a more usual way, simpler or more complex, as the case may be. While the latter is basically true, it cannot be simply an ornament in speech, parasitic on the literal meaning. At the very begin-nings of cognitive linguistics Lakoff and Johnson (1980) explicitly challenge this view as they consider both metaphor and metonymy to be basic and ubiquitous cognitive processes that pervade all our thinking, speaking and acting.
Despite a whole volume devoted to the problem of defining metonymy (cf. Benczes/Barcelona/Ruiz de Mendoza 2011), we are still far from a con-sensus view. As Barcelona (2011: 8) stresses, the fact that there is something
Total absence of semantic resemblance
Absolute
false Friends Absolute
True Friends strong Partial False Friends weak
Homonymic convergence
Related, but opposing meanings
Different polysemous
senses
Different usage levels
Influence of language
dynamics Degree of false friendness
Degree of partial false friendness
Degree of semantic resemblance
most researchers would agree to call a “standard” cognitive-linguistic notion of conceptual metonymy that contains core elements of the cognitive view of metonymy, “it is by no means a completely uniform notion, as there is some disagreement among these authors over a number of issues.” Various definitions of metonymy that have been proposed can only be appreciated if we first consider how it was treated in traditional rhetoric as well as how it differs from metaphor. Within the cognitive linguistic framework, both these processes have been contrasted with respect to five central points of difference, although it has been repeatedly claimed that the borderline between the two is blurred (cf. Barcelona 2000a and 2000b; Ruiz de Mendoza 2000).
The first of these four points of difference concerns the nature of the rela-tionship involved. A standard ingredient of traditional definitions of metonymy is a statement on what makes it different from a non-figurative expression, viz. that metonymy is a stand-in-for type of relationship of the whole for part type. In other words, a linguistic expression denoting a part of a larger whole is substituted by another expression denoting the whole. The relationship holding between the two is one of contiguity, association, or proximity (Ullmann 1962:
212; Taylor 1989: 122), whereas metaphor is based on similarity. Contiguity is here taken to cover all associative relations except similarity. This view about the nature of the metonymic relationship is also part of the cognitive linguistic approach. This means that metonymies are expressions that are used instead of some other expressions because the latter are associated with or suggested by the former, as illustrated in the following examples:
(4) Buckingham Palace has washed its hands of Prince Harry and Meghan Markle after they ‘told Americans to vote out Trump’.
A spokesman said it refused to comment on ‘not a working member of the Royal Family’ in a stinging response to the couple.3
(5) The first movement, “Quietly Flowing Along,” tapers off with the first violin gently rocking back and forth between two notes, B and D, preserved in the harmonic aspic of soft, sustained chords in the other instruments.4
Metaphors are in fact often considered to be shortened similes, i.e. two entities are brought into correlation as exhibiting some similarity, but there are no
3 https://www�dailymail�co�uk/news/article-8762603/Meghan-Markle-labels-November-vote-important-lifetime�html.
4 https://www�latimes�com/entertainment-arts/story/2020-09-09/john-cage-string- quartet-four-parts-covid-pandemic-music.
function words that would make this comparison explicit. In other words, something is described by mentioning another thing with which it is assumed to implicitly share some features. This can be illustrated on the following paragraph in which COVID-19 is conceptualized as an enemy against which the humanity has to struggle, i.e. battle to stop its spread and thus protect public health:
(6) In the story of humanity, communicable diseases play a starring role.
From the bubonic plague to cholera to HIV, we have been locked in a struggle for supremacy with maladies for millennia. With several Chinese cities under curfew, infections in the hundreds and the potential for significant economic disruption, the efforts to protect public health from today’s novel coronavirus are a striking example of this continuing contest. As governments and health authorities battle to stop the spread of the new virus, they would do well to consider lessons from history.5
The two also differ in terms of the number of conceptual domains involved.
The standard view is that a metonymic shift occurs within a single do-main, while metaphoric mappings take place across two discrete domains.
Conceptual metaphors typically employ a more concrete concept or domain as their source in order to structure a more abstract concept or domain as their target. They typically rest on a whole set of cross-domain mappings, while conceptual metonymies are said to involve only a single link (cf. Ruiz de Mendoza/Peña 2002).
Thirdly, metaphor and metonymy are generally different with respect to the directionality of conceptual mappings involved. Metaphors typically employ a more concrete concept or domain as source in order to structure a more abstract concept or domain as target. In the majority of cases, elements from the physical world are mapped onto the social and mental world. Metaphorical mappings are thus normally unidirectional, and the source and target are not reversible (cf. Kövecses 2002: 6). In some cases we have apparent revers-ibility due to the fact the boundaries between domains may be blurred for a number of historical and anthropological reasons. A triplet of domains, war-sports-politics, may be invoked to illustrate this (cf. Brdar/Omazić/Buljan/
Vidaković Erdeljić 2005). Metonymic mappings can, in principle, proceed in either direction, from the more concrete part of the domain (subdomain) to the more abstract one and the other way round, but of course not simultaneously.
5 https://www�ft�com/content/bd6cad34-3eab-11ea-b232-000f4477fbca.
According to Radden and Kövecses (1999: 22), “[i]n principle, either of the two conceptual entities related may stand for the other, i.e., unlike metaphor, metonymy is basically a reversible process.”
Metaphor and metonymy are also said to have different functions. Lakoff and Johnson (1980: 36) say that metaphor is “principally a way of conceiving of one thing in terms of another, and its primary function is understanding,”
while metonymy “has primarily a referential function, that is, it allows us to use one entity to stand for another.” However, both of the above statements have to be relativized. While Lakoff and Johnson see metonymy as having primarily referential function they are aware of its additional functions and point out not only that metonymy is “naturally suited for focussing” (Lakoff/
Johnson 1980: 37), but that it can just like metaphor have a role in construal.
It makes it possible for us to see and understand things in alternative ways.
Finally, due to the fact that metonymy is based on contiguity, while metaphor is based on similarity obtaining between conceptually discrete, and therefore conceptually distant, domains, the type of polysemy these two cognitive op-erations bring about is very different. In the case of a conceptual metaphor, for any domain that can function as the target domain, we may expect to have more than one potential source domain, e.g. time can be conceptualized as movement, commodity, physical object, etc. Conversely, one and the same source domain can be used for different target domains, e.g. we can use the domain of movement, more specifically journey to metaphorically concep-tualize time, love, etc. However, there is not much regularity in what can be used metaphorically to conceptualize something else, and what not, and as a result of this metaphors leads to a more ad hoc type sort of polysemy of lexical items associated with the source domain. The conceptual distance in the case of metonymy is smaller (we remain within a single domain), and the number of choices is relatively restricted. As a result, metonymic shifts within similar specific domains will tend to be very similar, and the lexical items enjoying the same ontological status within these domains will behave in the similar way, i.e. they will function as metonymic vehicles exhibiting the same type of shift (e.g. lexical items denoting some types of minerals, plants, etc. will come to denote some objects made from them), resulting in more regularity (Brdar et al 2009). This is not to deny the systematicity of conceptual meta-phors. We know very well that they can be organized in whole systems, but the dominant organizing principle is hierarchy, i.e. the systematicity is “vertical”:
a general metaphor can be a family of related submetaphors (their source do-main can be quite different), and these can exhibit a number of more specific mappings (which are sometimes considered to be very specific metaphors in their own right) which link to a multitude of lexical items associated with the
domains involved. On the other hand, the systematicity of metonymies is of the “horizontal” type.
Lakoff and Johnson (1980: 35) describe metonymy as the use of “one entity to refer to another that is related to it.” Kövecses and Radden (1998: 39) refine this by explicitly shifting everything to the conceptual level when they say that it is “a cognitive process in which one conceptual entity, the vehicle, provides mental access to another conceptual entity, the target, within the same domain, or ICM [Idealized Cognitive Model].”
In view of the complex nature of conceptual metonymy, it can be seen, taking into account most of the relevant insights in the literature, above all in Panther (2005), Ruiz de Mendoza and Otal Campo (2002) and Ruiz de Mendoza and Díez Velasco (2004) as a cognitive operation of conceptual elaboration based on the part-whole relationship that is triggered by the use of an expression (or metonymic vehicle) that is associated with a certain conceptual cluster (or metonymic source) within a conceptual domain so that the activation of the source conceptual cluster opens up a mental space that is dynamically expanded or reduced so as to come as close as possible to fitting the conceptual frame provided by the co(n)text of use, in the course of which the mental space thus opened and elaborated also comes very close in terms of its contents to another conceptual cluster (or metonymic target) within the same conceptual domain that may be or is typically associated with another expression (Brdar 2017;
Brdar/Brdar-Szabó in press).
It is also assumed in cognitive linguistics that metonymy can involve a range of relationships, not only the whole for part6 relationship (as asserted in the traditional approach), but also the part for whole relationship (in which case we have a subtype traditionally called synecdoche). What is more, the part for part type of relationship has also been assumed, which comes in numerous subtypes, e.g. cause for effect, producer for product, container for the contained, etc., although the viability of this type is disputed by Ruiz Mendoza and his collaborators (cf. Ruiz de Mendoza 2000; Ruiz de Mendoza/Díez Velasco 2002;
Ruiz de Mendoza/Mairal 2007; Ruiz de Mendoza/Pérez 2001). They are better analysed as the outcome of the interaction between two or more successively used metonymies.
Concerning metonymies of the type container for content, which are also claimed to be a subtype of the high-level metonymy part for part, exempli-fied in:
6 It is a notational convention in cognitive linguistics to use small capital letter for conceptual metaphors and metonymies.
(7) The bottle is sour. (‘bottle’ for ‘milk’)
(8) The milk tipped over. (for ‘the milk container tipped over’)
we seem to have two metonymies involving the whole and a part in a se-ries, assuming that the whole is a complex one, consisting of the whole-and-part, i.e. of the container and the content together forming a functional unit.
we seem to have two metonymies involving the whole and a part in a se-ries, assuming that the whole is a complex one, consisting of the whole-and-part, i.e. of the container and the content together forming a functional unit.