Prepositions can be found with and without adjacent complements in many forms of popular spoken French. The alternation appears in main clauses (il veut pas payer pour ça ~ il veut pas payer pour “he doesn't want to pay for [it]”) and, though with a more restricted social and geographic distribution, in relative clauses (j'avais pas personne avec qui parler ~ j'avais pas personne à parler avec “I had no one to whom to talk ~ I had no one to talk to”). In main clauses, the variant lacking the adjacent complement is said to have an orphaned preposition (il veut pas payer pour); in relatives, it is said to have a stranded preposition (j'avais pas personne à parler avec). In popular spoken French in Canada, stranding appears to be much more frequent than in other Francophone areas. Because so many French speakers in Canada are bilingual, because of the high frequency of stranding, and no doubt also because stranding violates prescriptive norms, stranded prepositions in French in Canada are widely believed to be instances of English influence (e.g. j'avais pas personne à parler avec is regarded as modeled on I had no one to talk to). But in a masterly variationist treatment, Poplack, Zentz and Dion (Reference Poplack, Zentz and Dion2011, this issue) argue that Canadian stranding is not of English origin. Stranded Canadian prepositions represent, instead, the expansion to relative clauses of the ordinary main-clause orphans. The historical source for Canadian stranding is thus analogy-induced and internal (French orphans), not contact-induced and external (not English stranding).
Poplack et al. construct a linguistic variable whose scope is the placement of prepositions in relative clauses. In this variable, the stranding variant alternates with the variant where the preposition does take a complement, which is called pied-piping (e.g. j'avais pas personne avec qui parler), and with the variant where the preposition is missing altogether, which is called absorption (e.g. j'avais pas personne à parler). For Poplack, as for many variationists, conditions on the selection of the variants of a variable are part of the structure of the language, part of the grammar. Poplack et al. show that, in Canada, French stranding is structurally unlike English stranding (the conditioning factors are different) but like French orphaning (the conditioning factors are similar), leading to the conclusion of internal French analogy rather than external English contact. As part of a strong empirical base, the authors base their research for the most part on corpora of naturalistic English and French speech gathered through sociolinguistic interviews.
The differences between French and English stranding that Poplack et al. bring out are indeed striking. They center primarily on stark differences between the languages with regard to lexical conditioning. In spoken English, most prepositions strand; but in spoken French, most prepositions do not (they pied-pipe or are absorbed). The only preposition that strands consistently in French is avec “with”, whereas many prepositions regularly strand in English (for example, in English, to, at, and of are stranding prepositions, whereas in French, their close counterparts à “to” and de “of” are almost never stranded). In the corpora under study, nine of the English prepositions are stranded categorically (the very frequent with, to, about, at, of, into, around, and through), whereas only three of the French ones are (the very infrequent dessus “above”, à travers de “through”, and après “after”). Stranding in French is limited to relative clauses introduced by à “to”, que “that”, pour “for”, and Ø, whereas stranding in English is much more general. In English, stranding is categorical with that-relatives, whereas in French no relatives involve categorical stranding. There is less corpus variability in English than in French; in the corpora under study, 97 percent of English relative clauses show stranded prepositions, whereas only 12 percent of French relatives do. And the relative clause prepositional placement variable even has different variants in the two languages. In French, stranding alternates with pied-piping and absorption; in English, according to Poplack et al., only with pied-piping.
The claim of a language-internal origin for French stranding in Canada receives considerable support from the plain fact that stranding is found also in French in Europe, where French–English bilinguals are not likely to have played a significant role. (In Europe, stranding appears to be less frequent than in Canada, but still frequent enough to attract the attention of prescriptivists.) Poplack et al. further strengthen their case against English influence by pointing out that the generativist analyses of orphaning and stranding are different for English and French. Further support comes from the fact that Canadians who code-switch liberally between French and English do not engage in stranding any more than do the more austere code switchers. Based on these considerations, the authors’ position is unequivocal for the diachronic analysis (French stranding did not arise from external contact with English, but rather from internal analogy with autochthonous French orphaning) as well as for the synchronic one (“the bilingual speakers studied here have different grammars for preposition placement, one for French and another for English” (under section heading: “Preposition placement in the English of bilingual francophones”).
The conclusions are strongly warranted. And as in previous work by Poplack, they constitute a welcome corrective to misapprehensions that distort the assessment of the extent of contact in many bilingual settings, a problem that, with reference to Spanish in the United States, has occupied some of my own efforts (Lapidus & Otheguy, Reference Lapidus and Otheguy2005; Otheguy Reference Otheguy, Contini-Morava and Sussman1995, Reference Otheguy and Díaz-Campos2011; Otheguy & Stern, Reference Otheguy and Stern2011; Otheguy & Zentella, Reference Otheguy and Zentella2011; Otheguy, Zentella & Livert, Reference Otheguy, Zentella and Livert2007). Still, Poplack et al.'s treatment of stranding in French in Canada leaves a number of important questions open.
First, the presentation remains somewhat ambiguous as to whether what is to be noted in French in Canada is a qualitative or a quantitative trait. Based on the evidence brought out by the authors, it is almost certain that, in the mental grammars of the majority of speakers of French everywhere, stranding has never been explicitly ruled out. We know this from the fact that, as the authors remind us, it has long since been explicitly ruled out by prescriptive grammars, always a clear indication of widespread availability in the socially unguarded vernacular registers of all speakers. It thus seems clear that the question is not how the French grammars of Canadian bilinguals came to allow stranding, but how they came to allow it to be so frequent. Here we are up against Poplack's conviction, eloquently championed in many publications (e.g. Poplack & Levey, Reference Poplack, Levey, Auer and Schmidt2010), that differences of occurrence rates are of little relevance to understanding group differences, or to understanding change, including contact-induced change. But in a variationist grammar such as Poplack's, where data come from natural speech, the postulated grammars are best conceptualized as mechanisms that contribute to guiding speech behavior. It is not clear why such performance grammars should in principle have nothing to do with the differences of frequency that are registered in speech. This is not to deny that frequency differences can in many cases be spurious and misleading. And one would welcome, of course, the pertinent demonstration that the structural environments that favor French stranding, or the expressive choices associated with it, happen by chance to have greater incidence in the speech of Canadians than in the speech of other Francophones, and that consequently the greater frequency of stranding in Canada is accidental and contingent, merely a matter of communication, and not structural and systemic, a matter of grammar. Absent demonstrations of this sort, frequency differences should remain as relevant to variationist grammar as to the study of any human faculty conceived as a guide to behavior. Thus it remains unclear why occurrence rates are to be dismissed as epiphenomenal, or considered off limits for the analytical ascertainment of synchronic differences and diachronic change. The fact that French–English bilinguals show some of the highest rates of stranding in the Francophone world appears to be regarded by Poplack et al. as a sort of untutored layman's observation, theoretically uninteresting and unrevealing, when it may very well be central to the analytical alternatives under consideration (analogy from an internal or external model).
A second question remains open, namely why the issue of the English origin of French stranding in Canada has to be posed as a Yes–No question (with a No answer), allowing little room for an analysis involving French origins and English support. We learn from Poplack et al. that the antecedent monolingual French grammars of today's Canadians always provided, as we have mentioned, not only for stranding but also, and even more clearly, for orphaning. It is thus quite sensible to argue, as do the authors, that Canadian bilinguals used French orphaning as their primary model for today's frequent stranding. But there is no reason to deny that they could very well have been encouraged in their stranding behavior by a secondary English model. In this view, the fact that stranding is so much more limited in French than in English would be due precisely to the fact that French stranding is but an imperfect approximation to its secondary English model. One does not have to engage in teleological or futuristic speculation to benefit from a conception that sees French stranding, always internally licensed, later externally fomented, advancing through a grammar into which it has made only limited headway, having reached only some of the prepositions and relative types that so widely sponsor stranding in English. The very generality of the phenomenon in English and its limited range in French, which Poplack et al. see as evidence against contact, would be compatible with a role for English stranding that, as an external model, abets the analogies of the internal model provided by French orphans.
Addressing a third question, one notes that the authors did not choose to study the plain difference between retention and omission of prepositional complements in all syntactic environments taken together. Were we to conceive of the variability in such natural terms, we would very likely see omission of the complement favored by main clauses more than by relative clauses (since orphans are more frequent than strandeds). With this in mind, the authors’ rejection of the possibility that the bilinguals’ English stranding may have played a role in the growth of their French stranding appears to rest on a theoretical position that may be correct but is not fully articulated. The position appears to privilege, even in bilinguals, analogical modeling within a language over such modeling across languages, and to regard differences of syntactic conditioning as less of a barrier to analogy than differences of lexical conditioning. Canadian bilinguals are seen by Poplack et al. as quite capable of analogizing the existing variable omission of adjacent prepositional complements in the French models (their French orphans) onto the existing variable omission of these complements in the French replicas (their French strandeds). The new high rates of the replica are, on the authors’ account, due to the previously available high rates of the model, despite the syntactic differences (the model omissions were in main clauses, the replica omissions are in relative clauses). But why should such analogies overcome syntactic barriers but not lexical ones, and why should they be limited by language boundaries in the bilingual's mind? On an account parallel to the one offered by the authors, Canadian bilinguals could be seen as analogizing the existing variable omission of adjacent prepositional complements from their English model (their English strandeds) onto the existing variable omission of these complements in the French replica (their French strandeds). The new high rates of their French replica would be due to the previously available high rates of their English model, despite the lexical differences (the model omissions, which are common, are found with for, of, to, with, etc., whereas the replica omissions, which are uncommon, are found mostly only with avec “with”).
These questions acknowledge the requirement by Poplack that, if contact is to be postulated, the constraints on the impacting variable of the source language should be the same as those on the impacted variable of the host language. In the absence of this structural identity (in the absence of the same or highly similar conditions on variability in the relevant variables in the two languages) the putative external source is discarded. But the question remains: Why is this requirement of structural identity imposed across languages, but relaxed within the same language? Why is it that constructions of the general type I have no one to talk to do not qualify as models for the general type j'avais pas personne à parler avec because the favoring factors that condition variation, especially the lexical ones, are so different, yet constructions of the type il veut pas payer pour can be the model of j'avais pas personne à parler avec, even though the strength of the favoring factors here are, in at least one respect, also quite different? If models that are imperfect because of syntactic differences can trigger analogies within languages, then it is not clear why, in the mind of the bilingual, models that are imperfect because of lexical differences cannot trigger analogies across languages.
Finally, and in more general terms, one wonders whether comparative sociolinguistics (Poplack & Tagliamonte, Reference Poplack and Tagliamonte2001; Tagliamonte, Reference Tagliamonte, Chambers, Trudgill and Schilling-Estes2002), which has served the field so well in addressing other issues, is equally applicable here. The approach thrives on comparisons between conditions on variability in different groups of speakers of the same or different languages. Yet the comparison is not as clearly available in the case at hand. The authors do offer three constraint hierarchies for French stranding, pied-piping, and absorption (involving clausal type, proximity to the verb, and prepositional weight). But the hierarchies for stranding and pied-piping for English are not made available, probably because in English corpora there is less variability to study (recall the 97-to-3 ratio of stranding vs. pied-piping in English, and the categorical stranding of that-relatives), rendering the comparative technique less useful.
The absence of a straight out comparison of constraint hierarchies may have contributed to an exaggeration of the differences between English and French that Poplack et al. use to dismiss any contribution from English stranding to French stranding. To see this, we note that the French hierarchies are to a considerable extent, according to the authors, functionally grounded: “Situated with respect to the entire system of preposition placement, the role of stranding becomes clear: it is selected most often in contexts where no intervening element might hinder the interpretation of the discontinuous prepositional phrase as a single constituent” (under section heading: “Comparing preposition placement strategies in French relative clauses”). It would have been useful to learn whether similarly grounded constraints are operative in English, where discontinuities may have a similar dampening effect on the frequency of stranding. Were this to be the case, the similarities between English and French would be greater, and the possibility of cross-linguistic contact, on the authors’ reasoning, all the more likely. It may also be useful to learn more about the structural analysis that leads the authors to assert that there is no absorption in English. While the assertion may be justified (for Poplack's English corpora or in general), it does seem that a case could be made for absorption in English constructions like that's a place I go a lot. If absorption were an available variant in English in Canada, then one of the arguments against contact (that the actual variants of prepositional placement are different in the two languages) would be weakened.
In Poplack, Zentz and Dion (Reference Poplack, Zentz and Dion2011) we have, then, a most enlightening analysis that offers convincing evidence for the primacy of internal factors in the rise of prepositional stranding in relative clauses in French in Canada. The anticipated answers to some of the remaining questions that will come from the leading authority in the variationist analysis of contact and change are certain to be more enlightening still.
Prepositions can be found with and without adjacent complements in many forms of popular spoken French. The alternation appears in main clauses (il veut pas payer pour ça ~ il veut pas payer pour “he doesn't want to pay for [it]”) and, though with a more restricted social and geographic distribution, in relative clauses (j'avais pas personne avec qui parler ~ j'avais pas personne à parler avec “I had no one to whom to talk ~ I had no one to talk to”). In main clauses, the variant lacking the adjacent complement is said to have an orphaned preposition (il veut pas payer pour); in relatives, it is said to have a stranded preposition (j'avais pas personne à parler avec). In popular spoken French in Canada, stranding appears to be much more frequent than in other Francophone areas. Because so many French speakers in Canada are bilingual, because of the high frequency of stranding, and no doubt also because stranding violates prescriptive norms, stranded prepositions in French in Canada are widely believed to be instances of English influence (e.g. j'avais pas personne à parler avec is regarded as modeled on I had no one to talk to). But in a masterly variationist treatment, Poplack, Zentz and Dion (Reference Poplack, Zentz and Dion2011, this issue) argue that Canadian stranding is not of English origin. Stranded Canadian prepositions represent, instead, the expansion to relative clauses of the ordinary main-clause orphans. The historical source for Canadian stranding is thus analogy-induced and internal (French orphans), not contact-induced and external (not English stranding).
Poplack et al. construct a linguistic variable whose scope is the placement of prepositions in relative clauses. In this variable, the stranding variant alternates with the variant where the preposition does take a complement, which is called pied-piping (e.g. j'avais pas personne avec qui parler), and with the variant where the preposition is missing altogether, which is called absorption (e.g. j'avais pas personne à parler). For Poplack, as for many variationists, conditions on the selection of the variants of a variable are part of the structure of the language, part of the grammar. Poplack et al. show that, in Canada, French stranding is structurally unlike English stranding (the conditioning factors are different) but like French orphaning (the conditioning factors are similar), leading to the conclusion of internal French analogy rather than external English contact. As part of a strong empirical base, the authors base their research for the most part on corpora of naturalistic English and French speech gathered through sociolinguistic interviews.
The differences between French and English stranding that Poplack et al. bring out are indeed striking. They center primarily on stark differences between the languages with regard to lexical conditioning. In spoken English, most prepositions strand; but in spoken French, most prepositions do not (they pied-pipe or are absorbed). The only preposition that strands consistently in French is avec “with”, whereas many prepositions regularly strand in English (for example, in English, to, at, and of are stranding prepositions, whereas in French, their close counterparts à “to” and de “of” are almost never stranded). In the corpora under study, nine of the English prepositions are stranded categorically (the very frequent with, to, about, at, of, into, around, and through), whereas only three of the French ones are (the very infrequent dessus “above”, à travers de “through”, and après “after”). Stranding in French is limited to relative clauses introduced by à “to”, que “that”, pour “for”, and Ø, whereas stranding in English is much more general. In English, stranding is categorical with that-relatives, whereas in French no relatives involve categorical stranding. There is less corpus variability in English than in French; in the corpora under study, 97 percent of English relative clauses show stranded prepositions, whereas only 12 percent of French relatives do. And the relative clause prepositional placement variable even has different variants in the two languages. In French, stranding alternates with pied-piping and absorption; in English, according to Poplack et al., only with pied-piping.
The claim of a language-internal origin for French stranding in Canada receives considerable support from the plain fact that stranding is found also in French in Europe, where French–English bilinguals are not likely to have played a significant role. (In Europe, stranding appears to be less frequent than in Canada, but still frequent enough to attract the attention of prescriptivists.) Poplack et al. further strengthen their case against English influence by pointing out that the generativist analyses of orphaning and stranding are different for English and French. Further support comes from the fact that Canadians who code-switch liberally between French and English do not engage in stranding any more than do the more austere code switchers. Based on these considerations, the authors’ position is unequivocal for the diachronic analysis (French stranding did not arise from external contact with English, but rather from internal analogy with autochthonous French orphaning) as well as for the synchronic one (“the bilingual speakers studied here have different grammars for preposition placement, one for French and another for English” (under section heading: “Preposition placement in the English of bilingual francophones”).
The conclusions are strongly warranted. And as in previous work by Poplack, they constitute a welcome corrective to misapprehensions that distort the assessment of the extent of contact in many bilingual settings, a problem that, with reference to Spanish in the United States, has occupied some of my own efforts (Lapidus & Otheguy, Reference Lapidus and Otheguy2005; Otheguy Reference Otheguy, Contini-Morava and Sussman1995, Reference Otheguy and Díaz-Campos2011; Otheguy & Stern, Reference Otheguy and Stern2011; Otheguy & Zentella, Reference Otheguy and Zentella2011; Otheguy, Zentella & Livert, Reference Otheguy, Zentella and Livert2007). Still, Poplack et al.'s treatment of stranding in French in Canada leaves a number of important questions open.
First, the presentation remains somewhat ambiguous as to whether what is to be noted in French in Canada is a qualitative or a quantitative trait. Based on the evidence brought out by the authors, it is almost certain that, in the mental grammars of the majority of speakers of French everywhere, stranding has never been explicitly ruled out. We know this from the fact that, as the authors remind us, it has long since been explicitly ruled out by prescriptive grammars, always a clear indication of widespread availability in the socially unguarded vernacular registers of all speakers. It thus seems clear that the question is not how the French grammars of Canadian bilinguals came to allow stranding, but how they came to allow it to be so frequent. Here we are up against Poplack's conviction, eloquently championed in many publications (e.g. Poplack & Levey, Reference Poplack, Levey, Auer and Schmidt2010), that differences of occurrence rates are of little relevance to understanding group differences, or to understanding change, including contact-induced change. But in a variationist grammar such as Poplack's, where data come from natural speech, the postulated grammars are best conceptualized as mechanisms that contribute to guiding speech behavior. It is not clear why such performance grammars should in principle have nothing to do with the differences of frequency that are registered in speech. This is not to deny that frequency differences can in many cases be spurious and misleading. And one would welcome, of course, the pertinent demonstration that the structural environments that favor French stranding, or the expressive choices associated with it, happen by chance to have greater incidence in the speech of Canadians than in the speech of other Francophones, and that consequently the greater frequency of stranding in Canada is accidental and contingent, merely a matter of communication, and not structural and systemic, a matter of grammar. Absent demonstrations of this sort, frequency differences should remain as relevant to variationist grammar as to the study of any human faculty conceived as a guide to behavior. Thus it remains unclear why occurrence rates are to be dismissed as epiphenomenal, or considered off limits for the analytical ascertainment of synchronic differences and diachronic change. The fact that French–English bilinguals show some of the highest rates of stranding in the Francophone world appears to be regarded by Poplack et al. as a sort of untutored layman's observation, theoretically uninteresting and unrevealing, when it may very well be central to the analytical alternatives under consideration (analogy from an internal or external model).
A second question remains open, namely why the issue of the English origin of French stranding in Canada has to be posed as a Yes–No question (with a No answer), allowing little room for an analysis involving French origins and English support. We learn from Poplack et al. that the antecedent monolingual French grammars of today's Canadians always provided, as we have mentioned, not only for stranding but also, and even more clearly, for orphaning. It is thus quite sensible to argue, as do the authors, that Canadian bilinguals used French orphaning as their primary model for today's frequent stranding. But there is no reason to deny that they could very well have been encouraged in their stranding behavior by a secondary English model. In this view, the fact that stranding is so much more limited in French than in English would be due precisely to the fact that French stranding is but an imperfect approximation to its secondary English model. One does not have to engage in teleological or futuristic speculation to benefit from a conception that sees French stranding, always internally licensed, later externally fomented, advancing through a grammar into which it has made only limited headway, having reached only some of the prepositions and relative types that so widely sponsor stranding in English. The very generality of the phenomenon in English and its limited range in French, which Poplack et al. see as evidence against contact, would be compatible with a role for English stranding that, as an external model, abets the analogies of the internal model provided by French orphans.
Addressing a third question, one notes that the authors did not choose to study the plain difference between retention and omission of prepositional complements in all syntactic environments taken together. Were we to conceive of the variability in such natural terms, we would very likely see omission of the complement favored by main clauses more than by relative clauses (since orphans are more frequent than strandeds). With this in mind, the authors’ rejection of the possibility that the bilinguals’ English stranding may have played a role in the growth of their French stranding appears to rest on a theoretical position that may be correct but is not fully articulated. The position appears to privilege, even in bilinguals, analogical modeling within a language over such modeling across languages, and to regard differences of syntactic conditioning as less of a barrier to analogy than differences of lexical conditioning. Canadian bilinguals are seen by Poplack et al. as quite capable of analogizing the existing variable omission of adjacent prepositional complements in the French models (their French orphans) onto the existing variable omission of these complements in the French replicas (their French strandeds). The new high rates of the replica are, on the authors’ account, due to the previously available high rates of the model, despite the syntactic differences (the model omissions were in main clauses, the replica omissions are in relative clauses). But why should such analogies overcome syntactic barriers but not lexical ones, and why should they be limited by language boundaries in the bilingual's mind? On an account parallel to the one offered by the authors, Canadian bilinguals could be seen as analogizing the existing variable omission of adjacent prepositional complements from their English model (their English strandeds) onto the existing variable omission of these complements in the French replica (their French strandeds). The new high rates of their French replica would be due to the previously available high rates of their English model, despite the lexical differences (the model omissions, which are common, are found with for, of, to, with, etc., whereas the replica omissions, which are uncommon, are found mostly only with avec “with”).
These questions acknowledge the requirement by Poplack that, if contact is to be postulated, the constraints on the impacting variable of the source language should be the same as those on the impacted variable of the host language. In the absence of this structural identity (in the absence of the same or highly similar conditions on variability in the relevant variables in the two languages) the putative external source is discarded. But the question remains: Why is this requirement of structural identity imposed across languages, but relaxed within the same language? Why is it that constructions of the general type I have no one to talk to do not qualify as models for the general type j'avais pas personne à parler avec because the favoring factors that condition variation, especially the lexical ones, are so different, yet constructions of the type il veut pas payer pour can be the model of j'avais pas personne à parler avec, even though the strength of the favoring factors here are, in at least one respect, also quite different? If models that are imperfect because of syntactic differences can trigger analogies within languages, then it is not clear why, in the mind of the bilingual, models that are imperfect because of lexical differences cannot trigger analogies across languages.
Finally, and in more general terms, one wonders whether comparative sociolinguistics (Poplack & Tagliamonte, Reference Poplack and Tagliamonte2001; Tagliamonte, Reference Tagliamonte, Chambers, Trudgill and Schilling-Estes2002), which has served the field so well in addressing other issues, is equally applicable here. The approach thrives on comparisons between conditions on variability in different groups of speakers of the same or different languages. Yet the comparison is not as clearly available in the case at hand. The authors do offer three constraint hierarchies for French stranding, pied-piping, and absorption (involving clausal type, proximity to the verb, and prepositional weight). But the hierarchies for stranding and pied-piping for English are not made available, probably because in English corpora there is less variability to study (recall the 97-to-3 ratio of stranding vs. pied-piping in English, and the categorical stranding of that-relatives), rendering the comparative technique less useful.
The absence of a straight out comparison of constraint hierarchies may have contributed to an exaggeration of the differences between English and French that Poplack et al. use to dismiss any contribution from English stranding to French stranding. To see this, we note that the French hierarchies are to a considerable extent, according to the authors, functionally grounded: “Situated with respect to the entire system of preposition placement, the role of stranding becomes clear: it is selected most often in contexts where no intervening element might hinder the interpretation of the discontinuous prepositional phrase as a single constituent” (under section heading: “Comparing preposition placement strategies in French relative clauses”). It would have been useful to learn whether similarly grounded constraints are operative in English, where discontinuities may have a similar dampening effect on the frequency of stranding. Were this to be the case, the similarities between English and French would be greater, and the possibility of cross-linguistic contact, on the authors’ reasoning, all the more likely. It may also be useful to learn more about the structural analysis that leads the authors to assert that there is no absorption in English. While the assertion may be justified (for Poplack's English corpora or in general), it does seem that a case could be made for absorption in English constructions like that's a place I go a lot. If absorption were an available variant in English in Canada, then one of the arguments against contact (that the actual variants of prepositional placement are different in the two languages) would be weakened.
In Poplack, Zentz and Dion (Reference Poplack, Zentz and Dion2011) we have, then, a most enlightening analysis that offers convincing evidence for the primacy of internal factors in the rise of prepositional stranding in relative clauses in French in Canada. The anticipated answers to some of the remaining questions that will come from the leading authority in the variationist analysis of contact and change are certain to be more enlightening still.