Quelle folle aventure! Adjective positioning in native French and translated American TV shows and movies

Abstract

This paper analyzes adjective positioning in French through the lens of language contact. Though some adjectives are nearly always either before their noun (pre-nominal) or after (post-nominal), others – usually adjectives that depict a positive or negative value – can be found in either position without any detectable semantic change. Our aim is to investigate the possible influence of English on the positioning of these adjectives by using a corpus of native French TV media as well as media translated from English. The categorical pre-nominal pattern of English might be causing a higher rate of pre-nominal adjectives to be used in translated media in comparison to native media. Our findings indicate that the translated corpus shows a preference for pre-nominal adjectives compared to the native French corpus. It was also revealed that some adjectives seem more flexible in translated media. Thus, while French media prefer one position over the other for a specific adjective, the translated media will more easily place the same adjective on either side of the noun. Therefore, we conclude that language contact has a role to play in the positioning of adjectives in TV media in French.

Résumé

Le but de cet article est d’analyser la position des adjectifs en français par le biais du contact des langues. Alors que certains adjectifs sont presque toujours placés soit avant le nom (pré-nominaux), soit après (post-nominaux), d’autres – surtout les adjectifs dénotant une valeur positive ou négative – peuvent être placés dans les deux positions sans qu’il y ait de changement notable de sens. Notre but est d’étudier l’influence possible de l’anglais sur la position de ces adjectifs en utilisant un corpus de média télévisés français et traduit de l’anglais. La position catégoriquement pré-nominale des adjectifs en anglais pourrait provoquer une utilisation plus importante d’adjectifs pré-nominaux dans les médias traduits. Nos résultats montrent que les médias traduits de l’anglais ont une préférence pour les adjectifs pré-nominaux par rapport aux médias français. Nous avons également découvert que certains adjectifs sont plus flexibles dans le corpus traduit. En effet, alors que les médias français préfèrent une position ou l’autre pour certains adjectifs spécifiques, les médias traduits placent le même adjectif plus facilement des deux côtés du nom. Ainsi, nous avons conclu que le contact des langues joue un rôle dans le positionnement des adjectifs dans les médias en langue française.

1. Introduction

Adjective positioning in French is a topic that has been studied from many different perspectives: diachronic (Ménard, 1976; Rickard, 1995; Buridant, 1997; Langford and O’Connor, 2005; Buridant, 2019), syntactic (Abeillé and Godard, 1999; Laenzlinger, 2005) and semantic (Weinrich, 1966; Waugh, 1976; Rowlett, 2007; Benzitoun, 2014). In allowing both pre-nominal (Pre-N) and post-nominal (Post-N) positions, French – along with most Romance languages – is typologically unusual (Langford and O’Connor, 2005: 484). This phenomenon has led many linguists to explore the question of why an adjective might be placed before or after the noun. One aspect that has not been studied extensively is whether language contact is affecting the positioning of adjectives in French. Though it has been mentioned in passing that English, being a prominent language on the international stage, could have affected French syntax and more specifically the position of adjectives (Etiemble, 1964: 186; Wilmet, 1986: 133; Rickard, 1995: 140), there are very few empirical studies that assess this issue.

Schöntag (2009) and McLaughlin (2011) have both studied various written corpora in order to explore the question of English influence on French adjectives. Their hypothesis was that because English adjectives are always Pre-N, if French is being affected by English, then perhaps we would see more Pre-N adjectives in corpora that are translated from English. However, neither study was able to confirm this (see Section 3). In the current study, we pursue this line of work, investigating the translation of TV media. Though the media has been said to be a vehicle for the dissemination of anglicisms (Etiemble, 1964: 186; Pergnier, 1989: 146), it has seldom been studied as a vector for linguistic change. We believe that American movies and TV shows, which are very popular in France, may be an interesting medium to investigate.

The question we pose for this study is: do the translations of American TV shows display a higher number of Pre-N adjectives than native French TV shows and movies? In order to answer this question, we have analyzed multiple corpora of TV media:

• Ten episodes of four translated American TV shows

• The scripts of 12 episodes of one other translated American TV show

• Ten episodes of four French TV shows

• The scripts of 19 French movies

Because not all adjectives are subject to changing positions, in all of the above-mentioned corpora we only analyzed adjectives of positive and negative value since these adjectives can be both Pre- and Post-N without meaning change; this follows Larsson’s (1994) methodology (see Section 2). We hypothesize that these adjectives will more often be used pre-nominally in the translated media than in the native French media because the English Pre-N adjectival pattern will influence the translation.

In Section 2, we will introduce the issue of adjective positioning in French, Section 3 will cover English influence on French syntax as well as translation and media as an influence on language change, Section 4 will develop our study in more detail, and Section 5 will discuss the results.

2. Adjective positioning in French

French allows for both Pre-N and Post-N adjectives, but not all adjectives can be in either position in any context. Post-N adjectives are the most frequent (Grévisse, 1993: 498). Older studies concluded that adjectives were postposed approximately 65% of the time and this figure could go up to 90% of the time in technical contexts (Weinrich, 1966: 82). What is more, Wilmet (1986: 152–153) asserts that Post-N is the unmarked position in French. Most grammar manuals agree on a list of categories of adjectives that are always found in Pre-N position and another list of categories that are categorically Post-N (Grillaert, 1976; Judge and Healey, 1990; Hawkins and Towell, 2010). The categories of post-nominal adjectives include: colour, nationality, shape, religion, place of origin, past participles, adjectives that are followed by a complement. Pre-N adjectives are often listed individually since they do not fit into such neat categories. They are short, frequent adjectives such as grand ‘big’, petit ‘small’, joli ‘pretty’, and nouveau ‘new’. These are sometimes referred to as elementary adjectives (Larsson, 1994).

There are also some adjectives whose semantics change according to their position with respect to the noun (Spence, 1976; L’Huillier, 1999; Hawkins and Towell, 2010; Benzitoun, 2014):

Beyond these lists, there are other reasons that have been proposed to explain why an adjective would be placed before or after the noun it modifies. Some authors believe there are syntactic reasons (e.g. Abeillé and Godard, 1999; Laenzlinger, 2005; Cinque, 2010), others believe there are semantic reasons (e.g. Weinrich, 1966; Waugh, 1976; Rowlett, 2007; Benzitoun, 2014), morphological reasons (e.g. McLaughlin, 2011), stylistic reasons (e.g. Spence, 1976; Grillaert, 1976; Judge and Healey, 1990; L’Huillier, 1999; Hawkins and Towell, 2010), phonological reasons (e.g. Chevalier et al., 1964; Larsson, 1994; Forsgren, 1997) and more. The following discussion of the existing literature will present some of these theories. But we will also explain why they cannot always be applied in every situation and at the end of the section we will provide a list of adjectives that have been known to have a flexible position without any meaning change.

2.1. Semantic independence of the adjective

The most common explanation that has been given to this issue is the inherent vs. intersective theory. According to many linguists (Waugh, 1976; Wilmet, 1986; Abeillé and Godard, 1999; Langford and O’Connor, 2005; Rowlett, 2007), adjectives that express an inherent quality of the noun will be placed pre-nominally whereas adjectives whose meaning intersects with that of the noun will be post-nominal. Waugh (1976) explains that Post-N adjectives retain their full meaning and remain semantically independent whereas Pre-N adjectives do not. Examples (2) and (3) are taken from Waugh (1976: 86).

In example (2), the person depicted is both a liar and furious. The noun and the adjective are understood separately from each other, they simply intersect. On the other hand, the semantic interpretation of a Pre-N adjective is not fully independent. Instead, it serves to modify the lexical meaning of the noun and it cannot be broken down into the parts of the sum, as shown in example (3).

Forsgren (1978) takes this even a step further. According to him, adjectives in Pre-N position not only lose their independence, but they also have a reduced semantic meaning and are consequently synonymous with the elementary adjectives. Thus, the Pre-N position does not allow the adjective to be semantically full and the information it carries is reduced drastically. In another article, he even goes so far as to compare adjectives in this position to function words that have very little content (Forsgren, 1997).

Though this theory of intersective vs. inherent quality of the adjective makes a strong case for many adjectives, not all linguists are convinced. Spence (1976) highlights that there isn’t always a way to prove that the quality of the adjective is more inherent to the noun just because it is Pre-N. This distinction often relies on judgements of interpretation, which are subjective. Examples (4) and (5) are from Spence (1976: 46):

How can one prove that the greasiness of the shoulders is more inherent in example (4) than in example (5)? Some adjectives create a clear semantic contrast with their position in the sentence (see ancien ‘old/former’ in examples 1a and 1b), but this isn’t the case for all adjectives. The clearer the semantic distinction, the less freedom of movement the adjective has. Though anyone is free to use ancien in either position, one cannot expect the same semantic outcome. However, with examples (4) and (5), the distinction is not so clear, thus the position of the adjective is much freer.

In his study, Larsson (1994), who analyzed adjective positioning in several novels, non-literary prose, newspapers, travel catalogues and tour guides, also points out the limits to the theory that Pre-N adjectives have reduced meaning. He mentions that there are some adjectives (e.g. excellent ‘excellent’) where no distinction was found between the full meaning and the reduced meaning. What is more, this analysis cannot explain why synonymous words do not behave in the same way. He mentions that immense ‘huge’ is often Pre-N and if we assume it is because its meaning is reduced to that of the elemental adjective grand ‘big’, then why are gigantesque ‘gigantic’ or géant ‘huge’ not found pre-nominally as often for the same reason (Larsson, 1994: 69–71)? Thus, Larsson points out that this analysis may be a tendency more than a rule.

2.2. Intension and extension of the adjective

Larsson (1994: 217–218) shows that the intension and the extension of the adjective affect its position. The intension is the amount of information contained in the adjective, while the extension is the number of nouns the adjective can modify. If the intension is narrow and the semantics of an adjective is vague, its extension is wide because it can be applied to many different nouns. For example, bon ‘good’ is a very general adjective that can be applied to nearly any noun; thus, its extension is great. On the other hand, very specific adjectives, such as adjectives of nationality or shape, have a small extension because they carry a lot of information and can only modify a limited number of nouns. He concludes that adjectives of great extension have a tendency to be Pre-N, which explains why elementary adjectives are usually pre-nominal, as well as other generic adjectives such as superbe ‘great’, excellent ‘excellent’, vaste ‘vast’.

Weinrich’s (1966) theory is quite similar. He argues that Pre-N adjectives act like “morphemes” whereas Post-N adjectives act like “lexemes” (Weinrich, 1966: 85). Weinrich’s definition of morphemes includes articles, demonstrative adjectives, possessive adjectives, interrogative adjectives, numeral adjectives and prepositions; whereas his definition of lexemes includes nouns, complements, relative clauses, apposition and verbal adjectives. The intension of morphemes is narrow, but their extension is wide. For example, the article le ‘the’ (definite, masculine, singular) has a very restricted meaning and carries little information. However, it can be applied to any noun whose grammatical gender is masculine. On the other hand, the intension of lexemes is larger, but their extension is smaller. For example, the word poète ‘poet’ refers to something with more semantic content than le, but it can only refer to things that are poets. This is how Weinrich views adjectives. If the adjective is placed before the noun, its interpretation is less narrow and can apply to many nouns; thus, it has become part of morphology. However, if the adjective is placed after the noun, its semantic interpretation is much fuller, and its denotation has more specifications.

2.3. Other factors

It has also been suggested that Pre-N adjectives are more subjective or abstract than Post-N adjectives (Gillaert, 1976; Judge and Healey, 1990; Grévisse, 1993; L’Huillier, 1999). Larsson (1994: 171) looked into some adjectives that have both a concrete and a metaphorical meaning. For example, the adjective brillant can mean both ‘bright’ (e.g. a bright light) and ‘brilliant’ (e.g. a brilliant student). Larsson found a correlation in his data between the position of the adjective and the meaning it carries: the metaphorical meaning (brilliant) was more often pre-posed to the noun and the concrete meaning (bright) was more often post-posed. That being said, he did not find the same correlation with the adjective merveilleux that can mean both ‘supernatural’ and ‘wonderful’. Though the meaning ‘supernatural’ may seem more concrete, it is not found more often in Post-N position in his data. Thus, the theory that the position of the adjective can carry specific semantics is only a tool to help comprehension, it is not categorical (Larsson, 1994: 206).

The idea that newness of information can affect the position of the adjective has also been mentioned by many linguists and the general consensus is that adjectives expressing new information are Post-N and adjectives referring to previously mentioned information are Pre-N (Forsgren, 1978; Grévisse, 1993; Larsson, 1994). Forsgren’s (1978: 196) corpus study shows that this is an important factor in the positioning of adjectives. He adds to this that when the noun is used with a definite article, the modifying adjective is preferred in Pre-N position. This happens because the use of the definite article usually indicates that the information carried by the noun is already known. This was not tested specifically in our own study.

Another reason that adjectives might be Pre-N or Post-N is because of prosody. According to prescriptivist manuals, shorter elements are preferred before longer elements in French (Chevalier et al., 1964; Larsson, 1994; Forsgren, 1997); which is a common factor influencing word order. Thus, if the noun is longer than the adjective, then the adjective is more likely to be Pre-N. This is also true for sequences of adjectives. If more than two juxtaposed Pre-N adjectives are needed, they must be placed after the noun (L’Huillier, 1999: 414). That being said, Larsson’s (1994: 186–187) study shows multiple counterexamples to this theory. Thus, in practice, this aspect may not be so categorical.

Another factor that can affect adjective positioning is whether the adjective is modified by an adverb or a complement. Though sometimes, adjectives modified by an adverb can be placed pre-nominally – usually when the adverb is short – Post-N remains the preferred position in these cases. According to grammar manuals, if the adjective is modified by a complement, then it is mandatory to place the adjective after the noun. Examples (6) to (10) are taken from Chevalier et al. (1964: 205).

The elementary adjectives can usually stay Pre-N but only if the adverb is short:

Nevertheless, if the noun has a complement, then the adjective can optionally be Pre-N:

This seems to go in line with the prosodic constraints mentioned earlier: longer segments are Post-N. This factor is also confirmed by descriptivist studies (Forgren, 1978; Larsson, 1994).

Many have mentioned that there can be stylistic reasons for the choice of the adjective’s position (Spence, 1976; Grillaert, 1976; Judge and Healey, 1990; L’Huillier, 1999; Hawkins and Towell, 2010). Pre-N adjectives can be used to create a more literary effect or for emphasis. However, not all situations of flexible adjective positioning can be explained by style.

2.4. Flexible adjectives

Despite all of this, it is still agreed upon by most linguists that some adjectives seem to have a flexible position with no identifiable semantic change. Larsson (1994) found several instances of this. Examples (11) and (12) are from Larsson (1994: 118) demonstrating minimal pairs with the same adjective modifying the same noun.

Table 1 lists adjectives that have been mentioned in the literature as being able to alternate positions without any semantic change.

Table 1. Adjectives with flexible positions

Most of these adjectives refer to positive qualities. These are the types of adjectives that Larsson (1994) specifically focused on in his study because he claims that they are more subject to change than others. He left out their negative counterparts because a large part of his corpus consisted of travel catalogues and tour guides, a genre that doesn’t usually have many negative adjectives. Thus, we will be focusing on adjectives of positive value as well as adjectives of negative value in our study since our corpus is not constrained in that matter.

The current study aims at observing the behaviour of these positive and negative value adjectives and at investigating whether contact with English might cause them to be used pre-nominally more often. However, one of the main issues with making a case for variation and change due to contact is proving that the phenomenon is not simply an internal variation. If Pre-N adjectives are becoming more common in French, it could simply be a syntactic change happening independently of contact with English. Nevertheless, Thomason and Kaufman (1988) mention that language variation and change due to contact can often be argued if there are other structural borrowings in the same recipient language that can be attributed to the same donor language. For our purposes, this means that if French syntax has been proven to be affected by English before, it could potentially be affecting adjective positioning as well.

3. English influence on French syntax and the influence of translation

Since WWII ended, English has become a major international language and the American culture has been massively spread across the world. Therefore, even countries that do not have direct borders with the US are still in contact with English (Schöntag, 2009: 19). This could have led to syntactic changes in French and other languages.

3.1. Language contact and French syntax

Pergnier (1989: 95) conjectures about how the English pattern that places the modifier before the element modified in compound words has been borrowed into French. Examples like libre-service ‘self-service’ or compact-disque ‘compact disc’ illustrate his point. In French morphosyntax, the main noun (service or disque) should be placed before the modifier, but these French compounds show a different pattern.

Loock (2013) focuses on a similar phenomenon but has different conclusions. He specifically looks at N+N phrasal structures in which the head noun is to the right (regressive order), which is typologically unusual for French, yet completely natural in English. He claims that though some may have been borrowed from English (e.g. tennis club, boat people), others have developed on their own, independently of English borrowings. He mentions examples like Star Academy (popular French reality TV show) or la grève attitude (a positive behaviour towards striking). What is more, Loock (2013: 263) believes that this regressive order has become a morphological innovation that is now available, and possibly even productive: brochette party ‘kebab party’, Gaz dépannage ‘emergency gas-related repairs’… Thus, this order may not be a borrowing after all. However, it is not because the structure is available to French speakers that it must be native. If enough structures with a regressive order have been borrowed from English into French, then a syntactic reanalysis could have occurred. This would lead to this structure now being used in French for innovative purposes without ever going through English lexical items. Millar (2007: 23) claims that this has happened with the -ing suffix. According to him, there are so many English borrowings with this suffix in the French language (e.g. parking, camping, shopping…), that -ing has been reanalyzed as a full-fledged suffix and is now used productively to create new French words, both with English bases (e.g. zapping ‘changing the channel’) and French bases (e.g. bronzing ‘tanning’). Thus, sometimes structures or morphological units can be used in a way that does not involve lexical borrowing and yet still be the product of borrowing.

What is more, the influence of English on adjective positioning in Guernésiais has been shown (Jones, 2002). The island of Guernsey is located in the English Channel off the coast of Normandy. The language spoken there is a variety of French that has been greatly influenced by English because of its proximity to England as well as its history and its political status. In traditional Guernésiais, unmarked qualifying adjectives are Pre-N (this mostly includes adjectives of colour), all others as usually Post-N. However, in modern Guernésiais, 70% of adjectives are used pre-nominally. In fact, this tendency has become so strong, that even compound adjectives and superlative structures are found Pre-N (longer, heavier sequences are usually not pre-nominal in Guernésiais) (Jones, 2002: 154). Thus, if this can happen in one variety of French, it could be happening in other varieties as well. The author does not exclude the possibility that this could be due to multiple causations. Pre-N has always existed in Guernésiais, even if it is supposed to only be used in restricted contexts. Therefore, the linguistic system could be trying to converge both patterns into one word order through a process of simplification. Since the Adjective + Noun order is the pattern used in English, it could be a factor that is helping the Pre-N word order win out.

This concept was also brought up by Heine (2008: 38); he calls it narrowing. Narrowing may happen when the recipient language has multiple options for the same linguistic feature, but the donor language only has one. The redundancy of the two patterns in the recipient language is narrowed down to only one option, usually the same one as the donor language. Though the narrowing of adjective positioning in French has not yet been proven, it is possible to hypothesize that this could be happening in French. Some adjectives have two possible positions without any meaning change, but Pre-N is the unmarked option in English. Therefore, French could be trying to model the English pattern because of contact. Though it is true that Post-N adjectives do exist in English, they have a more restricted reading than their Pre-N counterparts. Used pre-nominally, adjectives in English can convey anything a Post-N adjective can, but the opposite isn’t true (Cinque, 2010: 23–24). Examples (13) and (14) are from Cinque (2010: 18):

Example (14) is in fact a reduced relative clause: Mary interviewed every candidate that is possible for her to interview. However, (13) is ambiguous. Though it can have the same reading as (14), it can also be synonymous with potential in this context. Thus, in English, Pre-N adjectives appear in more contexts and are therefore more likely to be used as a pattern to model. McLaughlin (2018) mentions that the closer the variant is to its counterpart in the other language, the more likely it is to be influenced by contact. The case of adjective positioning involves two very similar patterns; consequently, contact could possibly be at play.

The influence of English in Quebec French and the other varieties of French in Canada has also been studied extensively. Because of the proximity of these French-speaking communities to English speaking areas (in Canada and in the US) and because of a strong bilingual presence, these varieties have often been said to be influenced by English. The use of the future tense is one example. The French language has different ways of expressing the future: the present tense, the periphrastic future (aller ‘to go’ + infinitive of a verb) or the synthetic future (verb stem + future ending). The periphrastic future is typologically closer to the English pattern will + verb than the synthetic future. Though the synthetic future is disappearing from all varieties of French, studies that have compared variants of French that are geographically close to English-speaking communities to variants of French that are more distant have found some interesting results. Tremblay et al. (2019: 85) have shown that Quebec French speakers use the periphrastic future more frequently than Belgian French speakers while Grimm and Nadasdi (2011: 174) have confirmed that Ontario French uses the periphrastic future more often than Quebec French or Hexagonal French. The latter study has also come to the conclusion that the more participants spoke English, the more likely they were to use the periphrastic future while speaking French (Grimm and Nadasdi, 2011: 185). However, Poplack and Turpin (1999: 158) in their study, found no correlation between the choice of future tense in French and the participants’ proficiency in English.

The large number of bilingual speakers in Canada has also led linguists to study the language development of children who were raised speaking both French and English. Nicoladis (2006) has investigated the cross-linguistic transfer of syntactic patterns in bilingual children in Alberta, Canada. She compared the adjectival use of children who spoke both languages to that of children who only spoke English. She found that bilingual children reversed Post-N adjectives more often than monolingual children, placing them Pre-N to match the English pattern. This proves her theory that linguistic influence can happen during the developmental stages of children who speak more than one language.

Making a claim for language contact to be the cause of variation and change is always difficult. Nevertheless, it remains possible that the syntax of French has been affected by English in different ways. According to Matras and Sakel (2007), replication of patterns (PAT) from one language to another – as opposed to replication of matter (MAT) – is more subtle and less likely to be fought against by the general population, or even by prescriptivism. Therefore, though linguistic purism is usually quite strong in France, this sort of syntactic borrowing could more easily have made its way into the language than any lexical borrowing. If this is happening to the adjectival pattern of French, and English is the cause, it could be coming from media translations. In France and around Europe, American produced media is a very popular platform. In 2010, 75% of the best rated shows in Europe were from the United States (Biermé, 2018). This means that many people are consuming this media and, if it can be shown that translated media is in fact displaying a different adjectival pattern than native media, then it might be hypothesized that people are hearing this pattern as they watch the shows and movies and could be absorbing it and possibly reusing it.

3.2. The influence of translation

Translated texts have been known to be vectors of cross-linguistic influence. It can remain limited to translated texts, but it can also spread to native written corpora (Danchev, 1984; Baumgarten and Özçentin, 2008; Kranich et al., 2011). Sometimes, linguistic structures from the source language are translated into the target language with structures that are close to the original even if they are marginal. These structures are grammatical but less frequent. This can lead to the marginalized features becoming more prominent and even becoming the norm (Baumgarten and Özçentin, 2008: 294). Thus, the new pattern starts to become normal for native speakers. In some cases, the new pattern introduced by translations is simply a stylistic innovation that is only used in certain contexts or certain registers. Kranich et al. (2011: 11) mentions this to be the case in legal documents for example. What is more, syntactic transfer can be less salient than other types of transfer, such as lexical or even morphological transfer (Danchev, 1984: 49). This means that syntactic borrowings may be more likely to go beyond the realm of translation and make their way into native speakers’ speech.

English texts translated into German have been studied from a diachronic perspective in order to see if there was any change over time. The results show that not only is the syntax in German translated texts influenced by English, but also that, over time, native German texts have been influenced by the same features (Kranich et al., 2011: 34). For example, the use of sentence-initial but to create a dialogical effect. Starting a sentence with aber ‘but’ is not typical in German but has seen a rise over time (Baumgarten and Özçentin, 2008: 295).

In the introduction, we mentioned Schöntag (2009) and McLaughlin’s (2011) studies on adjective positioning in French. Schöntag (2009) studied a small corpus of two magazines, two translated novels, two newspapers, one scientific report and a collection of oral texts. He analyzed the position of adjectives in all of these texts to see if genre or translation affected their positioning. His study did not yield any positive results. First, he did not find an unusually high rate of Pre-N adjectives and second, he could not prove that the Pre-N adjectives were, in fact, caused by contact with English.

McLaughlin (2011) wanted to test the influence of English on the positioning of adjectives in French in translations of news articles. Her results show that only 13% of the tokens were preposed, which is below the average of Pre-N adjectives in French (according to the McLaughlin 2011: 44, the average is 35%). In fact, she even noticed a negative effect of English on the positioning of adjectives in the French translations. In her data, there were more Pre-N structures in French when the original English structure did not contain an Adjective + Noun sequence. If that sequence was present in English, the translations almost always used the unmarked French structure Noun + Adjective. This indicates that the translators were aware of the structural difference between French and English and they consciously reversed it (McLaughlin, 2011: 42–43). She thus concludes that English does not influence the positioning of adjectives in French in her corpus. However, she did observe that there is a correlation between the speed of translation and the frequency of preposed adjectives: the faster the text was translated, the more Pre-N adjectives were used in the translation (McLaughlin, 2011: 48). This is due to there being less conscious attention to cross-linguistic differences, thus the translators use word orders that may be more intuitive regardless of prescriptivism. Media translation is a fast-moving world, and clients are requesting translations to be completed with tighter deadlines than ever before (CNC, 2021). Thus, it is possible to imagine that faster translations might be affecting the positioning of adjectives in this type of corpus. Finally, McLaughlin also noticed that the tokens were more likely to be preposed when they were used in translations of reported speech, which she posits is because there is “less narratorial control” (McLaughlin, 2011: 50). This is relevant to our study because TV shows and movies are meant to imitate oral speech, thus somewhat escaping the prescriptivist rules and the style constraints of written corpora.

Even though her study on adjectives did not yield any positive results, McLaughlin also analyzed two other syntactic features in the same type of corpus. The first was the use of passive structures (she also mentions this in an earlier article, McLaughlin, 2010). In French, the passive is not commonly used, and alternative structures are usually preferred. However, in English the passive is quite common, especially in news headlines. McLaughlin found that the rate of passive forms rose in the French translations she studied while the use of the alternative structures decreased (McLaughlin, 2011: 72). She even noticed that this happened in situations where the original English sentence did not have a passive structure (McLaughlin, 2010: 1934). She concluded that English was influencing the use of the passive in French in her corpus. The second feature she analyzed in her 2011 study was the verbal -ant forms. Adding this suffix to a French verbal root can create a present participle or a gérondif (a verbal form that expresses simultaneous actions). Though this can be related to the verbal -ing in English, the French uses are not as common as the English ones. However, contact with English has been cited before as reasons for the expansion of the progressive construction in both Italian and Spanish (see Durante, 1981 and Poutain, 1994). McLaughlin’s results show that the distribution of the two verbal -ant forms has been altered by influence from English. Though the gérondif is a more commonly used structure in Modern French than the present participle, the translated texts showed that 83% of the -ant forms were present participles, whereas only 17% were gérondif (McLaughlin, 2010: 83). Thus, her study shows that translated texts are subject to being influenced by the source language.

3.3. Audio-visual medium

Standardization is also an important factor in the process of translation influence. If the target language is highly standardized, it might be more inclined to resist the transfer. Though Benzitoun (2014) did not work on translated works, he did study adjective positioning in French in oral corpora. He found that many adjectives that are more often Post-N in written corpora were used pre-nominally in his oral corpus (Benzitoun, 2014: 2339). This shows that the medium that is under investigation can have an effect on syntax. We infer that written and formal corpora are less likely to show innovative forms than oral and informal corpora.

The audiovisual corpus that we have chosen to analyze is interesting because it is meant to imitate natural, oral speech, but the dialogue used in the shows and movies comes from a written script. In one of his articles, Barra (2013) gives a detailed, step-by-step review of the translating and dubbing process of a TV show. He mentions several times that fidelity to the original script is very important, along with making the show culturally accessible to a foreign audience, adapting it all to fit the technical constraints of the medium while still “keeping the characters and their way of speaking realistic and natural in the target language” (Barra, 203: 106). Thus, an effort is made to maintain the orality of the genre (assuming that is the style conveyed in the original show). Barra also mentions that, though the translators are the first people to work on the show – they work with both the script and the visual content – the adaptor (who is not a translator) often rewrites the scripts to fit the technical specifications of the medium. Finally, the dubbing director still has the right to make minor modifications during the recording sessions (adjective uses are listed as a modification they may make). Thus, the final product is often far from the first draft offered by the translating professionals. What is more, adherence to the prescriptive rules of the target language is never mentioned in Barra’s article. Not only are the professionals involved in this process faced with too many constraints (time, financial, cultural, technical…) for this to be a priority, but also, the script is modified by many different people who may not all be aware of prescriptive linguistic features. Yet all of them have access to the original version at some point in the process, which could lead to it influencing the choices they make in the translations.

This leads us to hypothesize that Pre-N adjectives could be used more often in translated media. Since French is a highly standardized language, it is unlikely that this change is happening to the same degree in all genres and in all contexts, but it is most likely happening in contexts of lower prestige. Thus, the media might be a good place to see this change in progress. In the same way that standardization affects linguistic transfer from translated texts, genre and register also play a role. Genres that are less standard and less prestigious can more easily be influenced by translations (Kranich et al., 2011: 28). Though it has been mentioned that lexical anglicisms are often spread to the general population via the media (Etiemble, 1964: 186; Pergnier, 1989: 146), this platform is still an understudied language context when it comes to language change and language contact. It was long believed that the media cannot affect the speech of viewers because there is no interaction involved, but this view is now challenged. It has been found that passive watching isn’t enough, motivation for the consumption of TV shows or movies as well as engagement with the characters and the story are essential to the development of language change; however, the change is still there (Androutsopoulos, 2014: 16; Stuart-Smith et al., 2013: 502). Androutsopoulos (2014: 16) also mentions that portraying stigmatized variants on TV has been known to prompt speakers of those variants to change their linguistic behaviours to be closer to the standard. Tagliamonte and Roberts (2005) studied the use of intensifiers in the popular American TV show Friends. They observed that the uses of so (vs. really or very) over the years correlates with its uses in the US by the general population (Tagliamonte and Roberts, 2005: 296). Thus, the TV show correlates with what was happening linguistically in real life. However, the reverse seems to be true as well. Stuart-Smith et al. (2013) analyzed TH-fronting and L-vocalization in Scottish vernacular of Glasgow. Though these features are not usually found in this dialect, they are present in London English, which led them to wonder if people in Glasgow who watched TV shows originating from London were more inclined to use these features than others. Their results show that the younger participants were using these features more than the adults and they found that engagement with TV shows was the second strongest contributing factor in this trend (Stuart-Smith et al., 2013: 528). Thus, there is evidence that linguistic change due to media consumption is possible.

What is more, American TV shows are often translated fast because episodes come out regularly and must be aired quickly (Barra, 2013; CNC, 2021). As mentioned earlier, McLaughlin (2011) found that the faster a text was translated, the more Pre-N adjectives were used in the translation. Therefore, we can reasonably speculate that translated American TV shows might demonstrate a higher rate of Pre-N adjectives than native French shows.

4. The study

4.1. The corpus

We started by analyzing the first ten episodes of the first season of four American TV shows dubbed in French and four native French TV shows. We have also read the scripts of the first 12 episodes of the first season of one extra American show dubbed in French. The reason for this is that two of the selected American shows have episodes that only run for 20 minutes each whereas all the French shows run for at least 45 minutes. A shorter running time means there are less opportunities to collect adjectives, thus we have decided to add one more translated show to even out that gap. Though this study was not meant to be diachronic, an effort was made to select shows ranging from the late 1990s to the present day for both types of shows. We also took care to pick genres that would be conducive to much dialogue in order to ensure greater opportunities for adjectives to be used (comedies and romantic comedies). Tables 2 and 3 present a list of these shows, the years they first aired and the name of the companies that translated the American shows into French.

Table 2. List of American shows in our corpus

Table 3. List of French shows in our corpus

For big budget movies, most dubbing professionals adapt the translation of visual media so the translated dialogue fits with the lip movement of the actors. Though this can be the case for TV shows, it is the author’s opinion as a native speaker of both French and English, that the lip movements of the actors were not taken into account during the translation of these specific shows, thus, only the content was conveyed via the translation.

In addition to these shows, a few French movies were analyzed as well. Though TV shows are particularly interesting to analyze because they usually expand over time, and their translation is sometimes rushed to meet their scheduled airing time, movies can also be of interest because their longer running time offers more opportunities for adjectives to be used. In an attempt to bring additional data to our study, we have collected the adjectives from 19 scripts of native French movies on a website called Lecteurs Anonymes (n.a.) (http://lecteursanonymes.org/scenario/).

4.2. Methodology

In the corpus of TV shows, we only focused on adjectives of positive or negative value that, as mentioned previously, are known for having a flexible position. These include the adjectives mentioned in the literature as being flexible but also other adjectives that we considered to fit Larsson’s category of positive or negative value. Therefore, we eliminated adjectives whose position is fixed and would either be ungrammatical or very unnatural if found in a different position. These include:

• Elementary adjectives

• Adjectives of colour, shape, nationality, religion, place of origin

• Past participles used as adjectives

Adjectives that are not of positive or negative value were also eliminated. We recognize that some adjectives are ambiguous as to their semantics and it could be argued that they belong or not in our list. For example, the adjective rare ‘rare’ can be seen as an adjective of positive value to refer to something that is unique, but it can also have a more neutral connotation when it refers to something you don’t come across very often (without any judgement on whether it is good or bad). In some cases, it is not very clear if the context means for this adjective to be used positively or neutrally. We elected to keep this particular adjective, but it only showed up three times in our entire corpus.

We have also chosen to keep adjectives modified by adverbs or complements as well as adjectives modifying nouns that are followed by a complement. As mentioned earlier, these cases are more likely to be Post-N, but there were some counter examples in our data. Also, the number of tokens that fit in this category was not significant enough to affect our results.

In the corpus of movies, only the specific adjectives that had already been gathered from the shows were collected. The main focus was to help identify patterns in adjectival behaviour for specific adjectives that appeared several times across the different media of our corpora.

5. Results and discussions

5.1. The overall results in the TV shows

Table 4 displays the number of Pre-N and Post-N tokens for each TV show individually.

Table 4. Number and percentages of tokens collected in the American and French TV shows individually

In order to get a better look at the effect of translation, we added up all the tokens from the American shows and then all the tokens from the French shows into one table. It will be noticed that the American shows display a higher number of tokens than the French shows. This is mostly due to the style of speech used by the characters in the shows. Though the genres of the American and French shows were similar (comedies and romantic comedies), the characters in the French shows had a tendency to be more vernacular and use shorter sentences than the characters in the American shows, which yielded less adjectives overall. This could be another result of translation (perhaps in the American shows, the characters were also using the same type of language, but the translators chose to change their speech style) or it could simply be a result of the shows we have selected. It is possible to hypothesize that this speech difference could have had an effect on the results instead of or in addition to language contact. This question can be investigated further in the future, our current study does not take this factor into account. Table 5 shows the results.

Table 5. Number and percentages of tokens collected in all the American and French TV shows

The individual results are varied. Post-N adjectives are always more numerous in all the shows, but the difference between the Post-N and Pre-N tokens can be more or less large depending on the show. In the American show How I Met Your Mother the amount of Pre-N and Post-N tokens are almost at 50%, but we cannot conclude that this is a pattern of translated shows because the same can be said of the French show H and the other American shows do not display this pattern (even though Buffy the vampire slayer is close). Dix Pour Cent greatly prefers Post-N, but the percentage difference between both positions in Drôle is similar to the difference in Gilmore Girls and Good Witch. Thus, though it seems that, overall, the French shows have a slight preference for Post-N positioning, from these numbers alone, we cannot discern any strong pattern concerning the behaviour of adjectives in the American shows compared to the French shows.

However, the overall numbers show more interesting results. The raw totals indicate that Post-N adjectives are more common than Pre-N adjectives in both types of shows. But if we look at the percentage differences between the number of Pre- and Post-N in both translated and native shows, we notice that the American shows have a higher rate of Pre-N adjectives than the French shows. The difference between Post-N and Pre-N tokens in American shows is 23.36% and the difference in French shows is 44.02%. What is more, we ran a chi-square test with the values in Table 5 and the results are significant: X² (1, 899) = 10.41, p = .001. Thus, the effect of translation does appear to play a role in the positioning of adjectives.

We further probed the individual adjectives that showed up several times across the shows. This allowed us to better understand how the different adjectives behaved in the different shows. However, the running time of TV shows is relatively short and the number of tokens for each adjective was low. Thus, in order to collect additional data, we decided to look at scripts of films, a medium that has a longer running time and more opportunities for adjective use.

5.2. The specific results in the TV shows and the movies

As mentioned earlier, 19 scripts of French movies were selected and only the specific adjectives that already showed up several times in the TV shows were collected in these scripts.

Below is a table of these adjectives and the number of times they occurred pre- or post-nominally in the American and French media. For sake of space, we have only included here the adjectives that were recorded in both Pre-N and Post-N position. Some adjectives seem to be preferred in one position or the other regardless of whether they are used in the native or translated media. For example, formidable ‘great’ shows up twice in the French media and four times in the American shows and it is always Post-N. Though the number of tokens is low, it is probably safe to assume that this adjective is preferred in Post-N position and requires no further analysis. The selected adjectives have been listed alphabetically in Table 6.

Table 6. Number of Pre-N and Post-N tokens that appear in both American and French shows

We observe that some adjectives are preferred Pre-N in the translated shows and Post-N in the native media (e.g. merveilleux or sérieux). There are also cases where the adjectives in the translated shows seem to be more flexible than in the French shows and movies. For example, important is categorically Post-N in the French media, but does show up twice Pre-N out of 19 tokens in translated shows (this corresponds to 10.53% of the data). Though it may not be much, it could be the beginning of a pattern that would prove more robust with more data. Horrible exhibits the same pattern but in a more convincing way. Though it is categorically Post-N in native media, it is used 10/14 times (71.43%) pre-nominally in translated shows. The same can be said of brilliant. French shows and movies do not place this adjective pre-nominally at all, but it has been found 3/7 time in the Pre-N position (42.86%) in translated shows. Though Larsson (1994) mentioned that the concrete definition of this word (e.g. une lampe brillante ‘a bright light’) is more likely to be Post-N than the abstract definition (e.g. un élève brillant ‘a brilliant student’), this was not the case in our data. In the French shows, this adjective was used twice with the meaning ‘brilliant’ and five times with the meaning ‘bright’, though all were Post-N, and in the American shows, it was used once with the meaning ‘bright’ (Post-N) and six times with the meaning ‘brilliant’ (three times Pre-N and three times Post-N). Other adjectives that could potentially display this pattern are: immense, extraordinaire, merveilleux, stupide, passionnant and violent, but more data would be needed to draw conclusions in these cases. Thus, though our sample of tokens is small, it can be suggested that there is an effect of translation on the flexibility of some adjectives.

Nevertheless, without more data it could possibly be argued that these differences between adjectives in French and translated media is simply coincidental. However, we have noticed, in our corpora, a few other English-like adjectival features that point towards the influence of English bleeding through into the translated shows.

5.3. Other adjectival features

First of all, there are a few instances of a noun preceded by several juxtaposed or coordinated adjectives:

These phrases go against the prosodic principles mentioned earlier by grammar manuals. Longer, heavier sequences should be placed after the noun, which is why it is often said that sequences of three adjectives are rare and that if more than two are needed, it is preferable to divide them up Pre and Post-N (Chevalier et al., 1964; Spence, 1976; Judge and Healey, 1990; L’Huillier, 1999). What is more, Forsgren (1978) and Larsson (1994) both found that prosodic weight plays a role in the positioning of adjectives in their descriptive studies. Thus, prosodically, these sequences are unnatural in French, which leads us to believe they may have been influenced by the original English. Even example (14) that only has two adjectives is still quite unnatural in French because both of them are more than three syllables long, whereas the noun only has two syllables, which should indicate that their preferred position is Post-N, especially when coordinated together.

Secondly, the data also contain a handful of superlative structures that seem to be influenced by English:

Grammar manuals explain that the superlative structure generally occurs either before the noun or after the noun according to the natural position of the adjective (L’Huillier, 1999). However, they also mention that, regardless of the adjective’s preferred position, if it is too long, it should be placed post-nominally (Hawkins and Towell, 2010), which is not the case of all of these examples. Larsson’s (1994: 191) study shows that, in practice, comparative and superlative structures seem to favour Post-N positioning. What is more, in our data, important and brilliant were only used post-nominally in the native French shows, which suggests that they are not well accepted in the Pre-N position. As for gentil and aggressive, they are rarely used in our corpus, and they are not mentioned specifically in other studies. Nevertheless, it seems that, prescriptively and descriptively speaking, these examples are not very natural. In the original versions, superlatives were used in each of these cases, therefore, it seems possible that these structures were influenced by English.

Thus, there are a few generalizations that can be gleaned from this study. First of all, as many linguists have concluded, there are some adjectives that can be found in both positions without there being any discernable syntactic or semantic reason for it in the context. These adjectives may be preferred in one position or another, but they can also be subject to outside influence (e.g. English) much more easily than other, more categorical adjectives. Second, though adjectives of positive and negative value are said to be the most flexible (Larsson, 1994), even in this particular group, some adjectives still resist one of the positions, even when the outside influence is at play (see example of formidable mentioned earlier). Third of all, despite this observation, there are still a few adjectives of positive and negative value that seem to display a different pattern when exposed to the influence of English and a few more that might be doing the same, though more data is needed. Finally, many linguists and grammarians have categorically stated the Post-N is the unmarked, preferred position in French (e.g. Weinrich, 1966; Wilmet, 1986; Grévisse, 1993), thus we might have expected a stronger trend towards Post-N in the French media than we actually did. The fact that many of the analyzed TV shows displayed patterns similar to some of the translated shows may suggest that there is a change happening in the media genre that goes beyond the impact of translation.

6. Conclusion

Though the position of some adjectives in French is fixed, others can be placed before or after the noun they modify without there being any semantic difference. Naturally, these are the adjectives that would be the most subject to change. Contact with another language may often condition linguistic change, even if it isn’t always easy to prove. Heine’s (2008) model of structural narrowing could explain what is taking place in French with adjectives of positive and negative value: under this model, contact with English would be putting pressure on the French adjectival system to select only one position for these adjectives. This question goes beyond what we could accomplish in this study, but we believe it is a first step.

Our question was to see if the translations of American TV shows display a higher number of Pre-N adjectives than native French TV shows and movies. Our hypothesis was that dubbed shows would have more Pre-N adjectives than the native French media because they are being influenced by the original English. All of the shows display a higher number of Post-N adjectives, but the percentage difference between Pre- and Post-N adjectives in the translated American media and the chi-square results told a different story. Our study suggests that there may be a higher rate of Pre-N adjectives in the American shows. In other words, translated shows might more easily use Pre-N adjectives than French media.

What is more, with the addition of data from longer, French movies, we were able to look at the behaviour of individual adjectives and it seems that some adjectives may be more flexible when used in shows that are translated from English. In our study, we pointed out a few of these adjectives, but nothing suggests that these specific adjectives are the only ones being influenced by this pattern of flexibility. Thus, this trend might not be restricted to any specific adjective, it could be pervasive to all adjectives of positive or negative value.

Finally, we have also noted two details in the translated shows that may confirm that English influence is at play: sequences of Pre-N adjectives and a handful of superlative structures that are modeled off the English superlative pattern. Since the translations of these TV shows seemed to have been influenced by the original English in terms of Pre-N sequences and superlatives, it is possible to conclude that the more flexible position could also be due to English.

In order to make these results more robust, more episodes of these shows need to be analyzed and more adjectives collected. It would also be useful to add older shows or movies to make a comparison over time. Finally, this is only the beginning of the study. Though we might have shown that English syntax bleeds through in a dubbed TV show, we still do not know if this syntactic change is being absorbed by native French speakers and then reused in everyday life. A study of the adjectival use of the general population (especially the young population that is more likely to be watching these shows and movies) would be necessary to answer that question.