The authors of ‘Phrase-final prepositions in Quebec French: An empirical study of contact, code-switching and resistance to convergence’, Poplack, Zentz & Dion (Reference Poplack, Zentz and Dion2011, this issue), henceforth cited as PZD, make a strong case for showing that, in spite of surface similarities, preposition stranding in Canadian French relative clauses cannot be qualified as a case of grammatical convergence due to language contact with English, but that it rather turns out to be a result of analogical extension of a native French strategy, preposition orphaning, to a new context. The application of a particularly sound and accountable methodology, the comparative method of variationist sociolinguistics (Poplack & Meechan, Reference Poplack and Meechan1998; Tagliamonte, Reference Tagliamonte, Chambers, Trudgill and Schilling-Estes2002), allows them to invalidate the hypothesis of a causal relationship between contact and the phenomenon under study.
When two languages are in close contact with each other and when one of them features a grammatical phenomenon which very much resembles a supposedly related phenomenon in the other language, it is certainly tempting to search for an explanation in mechanisms such as structural borrowing or other effects of contact. As a result, research on the outcomes of language contact abounds with a wide array of examples where linguistic phenomena are allegedly freely transferred from one language to the other, be it on the lexical, phonological or even morphosyntactic level. The impression is created that ‘anything goes’ (Thomason & Kaufman, Reference Thomason and Kaufman1988, p. 91). Yet, many studies on the effects of language contact exhibit considerable empirical flaws. It is not rare to find cases in which contact-induced change is proclaimed only on the basis of similar rates of occurrence or even based on the mere resemblance of the phenomena under consideration. It is, hence, likely that among all putative instances of contact-induced change, a certain amount can actually be ascribed to contact-independent, e.g. language-internal, triggers. Poplack & Levey (Reference Poplack, Levey, Auer and Schmidt2010), building on earlier work by Poplack & Meechan (Reference Poplack and Meechan1998), Poplack & Tagliamonte (Reference Poplack and Tagliamonte2001) and Tagliamonte (Reference Tagliamonte, Chambers, Trudgill and Schilling-Estes2002), present a pioneering work in which they propose a multidimensional methodology to assess the actual role of language contact in concrete cases of language change. The phenomenon under consideration needs to be compared with its putative counterpart in a number of benchmark varieties: First of all, a comparison with a pre-contact stage of the language shows whether the phenomenon is undergoing an actual change or whether it is subject to stable variation. Secondly, a contact-based account of language change predicts that the phenomenon under study is absent from those parts of the language which have not been exposed to the presumed source language. Finally, the phenomenon should of course be present in the presumed source language. Crucially, the standard of comparison is neither the presence or absence of a phenomenon nor its relative frequency, but rather its grammatical profile as defined by the intra- and extralinguistic constraints conditioning variant choice in usage. Very few studies on contact-induced change comply with these stringent requirements, and it is one of the most notable strengths of the PZD study that it uses them to put to the test an example case of alleged syntactic borrowing, preposition stranding in French.
In a first step and in accordance with the principle of accountability (Labov, Reference Labov1972), the authors define the variable context in which preposition stranding is embedded. Apart from pied-piping of the preposition, they identify a third form involved in variant choice, the absorption of the preposition. Absorption is the majority variant in the French data, but it is absent from the variable system featuring preposition stranding in English. This already is an important result since the actual role a particular variant plays in the grammar of language use can only be understood when contrasted with its main opponents, the other variants. PZD's study hence does not only account for those contexts in which preposition stranding was actually used, but also for those where it could have been used but wasn't (see Wolfram, Reference Wolfram and Preston1993); and these are quite different in the two contact languages considered here. Although this procedure has since long been quite common practice in variationist sociolinguistics, it is still much too often neglected in general linguistic research.
In a second step, the authors conduct a variationist analysis of the conditioning of these variants in a corpus of spontaneous French speech stemming from the language contact region. They find that even though each of the variants may in principle occur in the relevant contexts, every variant has its particularly favorable context of occurrence. Stranding, in fact a minor variant within the variable system of preposition placement in French, seems to be favored when the stranded preposition is not separated from the verb it complements. When an additional complement intervenes, the preposition is rather pied-piped. The strongest predictor of variant choice, however, turns out to be the preposition itself: the semantically strong prepositions avec “with” and, to a minor extent, dedans “in”, pour “for”, and dessus “on” represent an extremely favorable context for stranding, whereas de and à, both semantically weak, virtually never strand. But instead of favoring pied-piping, they form a preferential context for absorption.
A comparison of stranding in French and in its alleged source variety, Canadian English, shows strikingly different patterns of conditioning. Stranding is by far the majority variant and contextually not restricted. It occurs with all prepositions, strong and weak, and with all relativizers. The only context where pied-piping, an extremely rare variant in English, occurs with a noticeable rate is the relativizer which. This finding already makes the involvement of language contact between English and French a highly unlikely candidate to account for stranding in French.
Yet, PZD do not content themselves with this result. They add two analyses, one comparing the variable system of preposition stranding in the speech of sparse and copious code-switchers and another considering preposition placement in the English speech of francophone speakers, the presumed agents of structural borrowing. With the former of these analyses, the authors intend to disprove that code-switchers are in fact the agents of contact-induced change, at least in the present case. In my view, however, this is a rather weak argument, as the groups are defined solely on the basis of the total number of switches per recording. This division appears to be arbitrary and ad hoc rather than linguistically motivated. If PZD want to challenge the hypothesis that bilingual fluency might lead to a corrosion of one (or both) of the respective grammars in the bilingual speaker, e.g. as a result of structural borrowing, I doubt that the distinction between sparse and copious code-switchers will serve this aim. It is also not at all clear what the causal connection might be between bilingual fluency and the vulnerability of the participating languages to contact-induced change.
The analysis of preposition placement in the English speech uttered by francophones seems to me much more meaningful. It turns out that these informants follow the same constraints as had been observed in the analysis of Canadian English. This shows that bilingual speakers strictly separate their grammars in usage. When producing French preposition stranding, they do it in compliance with the French constraints of variant choice; when uttering stranded prepositions in English, it is according to the English patterns of conditioning. They clearly do not show signs of interference such as remnants of their English system in their usage of French. This is, in my view, the strongest evidence ruling out language contact as the cause of French preposition stranding. It may well be the case that contact-induced change can only take place when each language is represented by a large enough number of speakers. The existence of two speech communities with different languages or dialects, one whose speakers share a certain linguistic feature and another one which is about to adopt this feature in its language, as well as a certain degree of interaction and exchange between these groups is, of course, a prerequisite for contact-induced change. But a change always and necessarily originates in the language use of the individual speaker. Either a child uses a form different from that his or her parents do, which would be a case of generational change, or the language use of an individual is subject to a change during his or her own adult lifetime (see Sankoff & Blondeau Reference Sankoff and Blondeau2007). In any case, the crucial entity is the individual speaker, and scenarios of contact-induced change inevitably rest upon the assumption of a restructuring of the linguistic or grammatical knowledge of the individual speaker. However, the proponents of contact-based scenarios of language change are often stuck for an answer when it comes to provide a cognitively plausible explanation of how this restructuring may take place in the individual mind. In spite of the significance of this problem, it is rarely discussed and the grammar-altering influence of contact is widely taken for granted. PZD show that contact with English is not only an unlikely cause of French preposition stranding in the speech of the language community from the contact region, but also in the speech of individual, bilingual, speakers. This is, in my opinion, the most important merit of the study.
PZD's methodological approach, comparing the conditioning of variant choice not only in the language variety exhibiting the alleged effect of language contact, but also in the relevant benchmark varieties (the putative source variety, Canadian English, the varieties of French spoken by more and less fluent bilinguals, the variety of English spoken by speakers of the French contact variety, and a pre- (and non-)contact variety of French), presents a case of extraordinary empirical accountability and sets a precedent for future studies assessing the role of language contact as regards phenomena of language variation and change. It makes the theoretical conclusions particularly robust and shifts the burden of proof again to those who draw on the causal role of language contact much too hastily.
The authors of ‘Phrase-final prepositions in Quebec French: An empirical study of contact, code-switching and resistance to convergence’, Poplack, Zentz & Dion (Reference Poplack, Zentz and Dion2011, this issue), henceforth cited as PZD, make a strong case for showing that, in spite of surface similarities, preposition stranding in Canadian French relative clauses cannot be qualified as a case of grammatical convergence due to language contact with English, but that it rather turns out to be a result of analogical extension of a native French strategy, preposition orphaning, to a new context. The application of a particularly sound and accountable methodology, the comparative method of variationist sociolinguistics (Poplack & Meechan, Reference Poplack and Meechan1998; Tagliamonte, Reference Tagliamonte, Chambers, Trudgill and Schilling-Estes2002), allows them to invalidate the hypothesis of a causal relationship between contact and the phenomenon under study.
When two languages are in close contact with each other and when one of them features a grammatical phenomenon which very much resembles a supposedly related phenomenon in the other language, it is certainly tempting to search for an explanation in mechanisms such as structural borrowing or other effects of contact. As a result, research on the outcomes of language contact abounds with a wide array of examples where linguistic phenomena are allegedly freely transferred from one language to the other, be it on the lexical, phonological or even morphosyntactic level. The impression is created that ‘anything goes’ (Thomason & Kaufman, Reference Thomason and Kaufman1988, p. 91). Yet, many studies on the effects of language contact exhibit considerable empirical flaws. It is not rare to find cases in which contact-induced change is proclaimed only on the basis of similar rates of occurrence or even based on the mere resemblance of the phenomena under consideration. It is, hence, likely that among all putative instances of contact-induced change, a certain amount can actually be ascribed to contact-independent, e.g. language-internal, triggers. Poplack & Levey (Reference Poplack, Levey, Auer and Schmidt2010), building on earlier work by Poplack & Meechan (Reference Poplack and Meechan1998), Poplack & Tagliamonte (Reference Poplack and Tagliamonte2001) and Tagliamonte (Reference Tagliamonte, Chambers, Trudgill and Schilling-Estes2002), present a pioneering work in which they propose a multidimensional methodology to assess the actual role of language contact in concrete cases of language change. The phenomenon under consideration needs to be compared with its putative counterpart in a number of benchmark varieties: First of all, a comparison with a pre-contact stage of the language shows whether the phenomenon is undergoing an actual change or whether it is subject to stable variation. Secondly, a contact-based account of language change predicts that the phenomenon under study is absent from those parts of the language which have not been exposed to the presumed source language. Finally, the phenomenon should of course be present in the presumed source language. Crucially, the standard of comparison is neither the presence or absence of a phenomenon nor its relative frequency, but rather its grammatical profile as defined by the intra- and extralinguistic constraints conditioning variant choice in usage. Very few studies on contact-induced change comply with these stringent requirements, and it is one of the most notable strengths of the PZD study that it uses them to put to the test an example case of alleged syntactic borrowing, preposition stranding in French.
In a first step and in accordance with the principle of accountability (Labov, Reference Labov1972), the authors define the variable context in which preposition stranding is embedded. Apart from pied-piping of the preposition, they identify a third form involved in variant choice, the absorption of the preposition. Absorption is the majority variant in the French data, but it is absent from the variable system featuring preposition stranding in English. This already is an important result since the actual role a particular variant plays in the grammar of language use can only be understood when contrasted with its main opponents, the other variants. PZD's study hence does not only account for those contexts in which preposition stranding was actually used, but also for those where it could have been used but wasn't (see Wolfram, Reference Wolfram and Preston1993); and these are quite different in the two contact languages considered here. Although this procedure has since long been quite common practice in variationist sociolinguistics, it is still much too often neglected in general linguistic research.
In a second step, the authors conduct a variationist analysis of the conditioning of these variants in a corpus of spontaneous French speech stemming from the language contact region. They find that even though each of the variants may in principle occur in the relevant contexts, every variant has its particularly favorable context of occurrence. Stranding, in fact a minor variant within the variable system of preposition placement in French, seems to be favored when the stranded preposition is not separated from the verb it complements. When an additional complement intervenes, the preposition is rather pied-piped. The strongest predictor of variant choice, however, turns out to be the preposition itself: the semantically strong prepositions avec “with” and, to a minor extent, dedans “in”, pour “for”, and dessus “on” represent an extremely favorable context for stranding, whereas de and à, both semantically weak, virtually never strand. But instead of favoring pied-piping, they form a preferential context for absorption.
A comparison of stranding in French and in its alleged source variety, Canadian English, shows strikingly different patterns of conditioning. Stranding is by far the majority variant and contextually not restricted. It occurs with all prepositions, strong and weak, and with all relativizers. The only context where pied-piping, an extremely rare variant in English, occurs with a noticeable rate is the relativizer which. This finding already makes the involvement of language contact between English and French a highly unlikely candidate to account for stranding in French.
Yet, PZD do not content themselves with this result. They add two analyses, one comparing the variable system of preposition stranding in the speech of sparse and copious code-switchers and another considering preposition placement in the English speech of francophone speakers, the presumed agents of structural borrowing. With the former of these analyses, the authors intend to disprove that code-switchers are in fact the agents of contact-induced change, at least in the present case. In my view, however, this is a rather weak argument, as the groups are defined solely on the basis of the total number of switches per recording. This division appears to be arbitrary and ad hoc rather than linguistically motivated. If PZD want to challenge the hypothesis that bilingual fluency might lead to a corrosion of one (or both) of the respective grammars in the bilingual speaker, e.g. as a result of structural borrowing, I doubt that the distinction between sparse and copious code-switchers will serve this aim. It is also not at all clear what the causal connection might be between bilingual fluency and the vulnerability of the participating languages to contact-induced change.
The analysis of preposition placement in the English speech uttered by francophones seems to me much more meaningful. It turns out that these informants follow the same constraints as had been observed in the analysis of Canadian English. This shows that bilingual speakers strictly separate their grammars in usage. When producing French preposition stranding, they do it in compliance with the French constraints of variant choice; when uttering stranded prepositions in English, it is according to the English patterns of conditioning. They clearly do not show signs of interference such as remnants of their English system in their usage of French. This is, in my view, the strongest evidence ruling out language contact as the cause of French preposition stranding. It may well be the case that contact-induced change can only take place when each language is represented by a large enough number of speakers. The existence of two speech communities with different languages or dialects, one whose speakers share a certain linguistic feature and another one which is about to adopt this feature in its language, as well as a certain degree of interaction and exchange between these groups is, of course, a prerequisite for contact-induced change. But a change always and necessarily originates in the language use of the individual speaker. Either a child uses a form different from that his or her parents do, which would be a case of generational change, or the language use of an individual is subject to a change during his or her own adult lifetime (see Sankoff & Blondeau Reference Sankoff and Blondeau2007). In any case, the crucial entity is the individual speaker, and scenarios of contact-induced change inevitably rest upon the assumption of a restructuring of the linguistic or grammatical knowledge of the individual speaker. However, the proponents of contact-based scenarios of language change are often stuck for an answer when it comes to provide a cognitively plausible explanation of how this restructuring may take place in the individual mind. In spite of the significance of this problem, it is rarely discussed and the grammar-altering influence of contact is widely taken for granted. PZD show that contact with English is not only an unlikely cause of French preposition stranding in the speech of the language community from the contact region, but also in the speech of individual, bilingual, speakers. This is, in my opinion, the most important merit of the study.
PZD's methodological approach, comparing the conditioning of variant choice not only in the language variety exhibiting the alleged effect of language contact, but also in the relevant benchmark varieties (the putative source variety, Canadian English, the varieties of French spoken by more and less fluent bilinguals, the variety of English spoken by speakers of the French contact variety, and a pre- (and non-)contact variety of French), presents a case of extraordinary empirical accountability and sets a precedent for future studies assessing the role of language contact as regards phenomena of language variation and change. It makes the theoretical conclusions particularly robust and shifts the burden of proof again to those who draw on the causal role of language contact much too hastily.