Many thanks to C. Sibley for re-reading a draft version of this paper and for her encouragement. I also would like to thank my colleagues from the UMR BCL for some useful comments on a presentation of the problems at stake in this paper, as well as the participants in the conference on ‘deceptive’ syntax, more particularly M. De Mattia-Viviès, C. Lacassain-Lagoin, B. Richet and R. Menzies. Still, I remain solely responsible of course for any mistakes or inconsistencies regarding the contents of this paper.
- 1 cf. Kent Bach, Routledge Encyclopedia of Philosophy, ‘ambiguity’ entry.
- 2 Broadly speaking, such ambiguity can be accounted for by the lexicon (for instance, whole sentence (...)
- 3 See Guillaume 2006 and 2009 for a study of other ambiguous or hybrid subordinates in English, name (...)
- 4 From the point of view of etymology, since comes from sithen and has a temporal origin (cf. Molenck (...)
1Ambiguity is a well-known concept in the field of linguistics, and, at least at first glance, it appears to be well-defined and well-studied. I shall start from a simple definition of it as a word, a phrase, or a sentence which has more than one meaning1. Such a definition makes it clear that the conditions for ambiguity are situated both at a syntactic and at a semantic level. It should also be noted that ambiguity is more or less equated with polysemy here, in that one form has (at least) two different meanings. In a more strict acceptation of the term though, ambiguity is taken to be a special case of polysemy, restricted only to what happens at sentence level2 (Dubois et al. 1999: 31-2). In this paper, I shall concentrate more specifically on the since- subordinates in English, and on possible ambiguity regarding their interpretation. As with other cases of ambiguous subordinate clauses in English3, it is mostly because since as a subordinator is polysemic4 that since- clauses may turn out to be ambiguous, under circumstances which I shall attempt to clarify. Then, in turn, ambiguous clauses pave the way for ambiguous sentences. In a majority of cases, however, the ambiguity regarding the use of the subordinator since can be lifted thanks to a number of disambiguation criteria, which may be connected either with the context of the utterance or with its internal syntactic features. When recourse to such criteria fails to return consistent results, the original ambiguity is maintained, while the semantic content of the subordinate is unhampered: in such a case, the since- subordinate is deemed to be hybrid.
- 5 On the different values of the operator of location, see Gilbert 1993: 68.
- 6 See De Cola-Sekali 1991: 68-70 for a comprehensive account of the enunciative operations set in mo (...)
2When since is used as a subordinator,it can introduce either a temporal adverbial clause or a causal one. The nature of the clause is the same in both cases: since as a subordinator will always introduce an adverbialclause. The difference between a temporal and a causal adverbial clause lies at the semantic level. In an enunciative approach, the relationship between the main clause and the subordinate is described in terms of location: the subordinate provides a locator for the main clause. Since as a subordinate marker consistently signals a relationship of identification5 between the predicative relation enacted in the subordinate and that of the main clause. In a temporal clause, the event described in the subordinate clause marks the beginning of what is described in the main clause; it is therefore mostly a quantitative identification (in A. Culioli’s terms, it is the T parameter, connected with existence, time and place, which is concerned). Conversely, the cause and effect relationship underlying the articulation ofa causal since- subordinate with its main clause corresponds with identification of the qualitative type (which is linked to valuation or assessment, as well as to the S parameter in Culioli’s theory, S being for subjective)6. In a minority of cases, this can pave the way for ambiguity:
(1) Since he got his PhD at Cambridge, John has been offered several interesting jobs. (Aarts 1979 : 603)
(2) She’s been walking to work since her car packed up. (Wyld 1993 : 37)
3These two examples are invented, and quoted in papers on the uses of since to demonstrate that such a subordinator can indeed be ambiguous; yet, the potential ambiguity of such a marker is not in fact the subject of the papers in question, and, as a result, the authors do not study the conditions for ambiguity which are met in these examples. Indeed, it would probably be easy to discriminate between cause and duration in the since- clauses of examples (1) and (2), should they be given a larger context. Indeed, A. Hamm makes the point that “many of the ‘classical’ examples of ambiguity are really ambiguous only in context-free sentences, and (…) in most cases, our real-world knowledge and the discourse context or situation offer, if not the interpretation, at least strong discursive orientations or clues” (2001: 29).
- 7 As far as the oral examples are concerned, example (9) is a BNC example, which means that I am not (...)
- 8 See Goodwin and Duranti’s comments on the option of restricting context at sentence scope: “what i (...)
4As far as the notion of context is concerned, C. Goodwin and A. Duranti define it as the necessity to “[look] beyond the event itself to other phenomena (for example cultural setting, speech situation, shared background assumptions) within which the event is embedded”(1992: 3). In this paper however, the use which I make of ‘context’ and ‘contextualisation’ is more limited, as most of the examples given come from written material, namely from the British National Corpus7. As a result, the contextualisation which is performed is most of the time of an endophoric nature, in that reference is made to elements within the text itself; yet, I try, when possible, to go beyond the scope of the sentence, as ambiguous sentences cannot by definition receive a clear-cut interpretation within such a restricted scope8.
5The first two examples have no context whatsoever because they are invented. They are ambiguous because the completion of the PhD in (1) marks the starting-point of the job hunt, while at the same time being a necessary condition for it, in exactly the same manner as the breakdown of the car in (2) is both the temporal starting point and the reason for which the person has to go on foot.
- 9 As explained by J.J. Gumperz: “I use the term “contextualization” to refer to speakers’ and listen (...)
6And yet, even in the absence of a proper context, one tends to interpret the first instance as a causal clause, and the second one as a temporal one. Why is that so? In (1), it is for semantic reasons that the causal interpretation is prevalent. The reference to a Cambridge PhD can hardly be innocent: it triggers off an implicit contextualisation9, which plays on the cultural background shared between the speakers. There is no such clue in example (2); yet, despite the obvious cause and effect type of relation between the subordinate and the main clause, the choice of tenses and aspects, and more particularly the use of the progressive present perfect in the main clause, is much more typical of the temporal since- clause, as we shall see.
7The order in which the clauses occur is thought to be relevant as well (cf. Lattes 2005: 240); typically, the subordinate is fronted in (1) while it is postponed in (2). Finally, the presence of a comma separating the subordinate from the main clause is worth mentioning in (1), as well as its absence in (2). In the light of such criteria, example (1), with its fronted subordinate and the comma which coordinates the two clauses, calls to mind the canonical syntactic layout of a causal since- clause, whereas example (2), which exhibits neither feature, is more likely to be temporal.
8The examination of these examples, which are fairly uncomplicated because they are invented, has enabled me to underline several disambiguation criteria which can be resorted to in the few cases in which a since- subordinate can prove to be ambiguous. The indications given by context, whether endophoric or exophoric (shared knowledge, for instance) are essential, but when such a contextualisation is not possible or fails to return relevant information, there remain several purely syntactic features which can help disambiguate between the two types of since- clauses.
- 10 The other findings from this corpus are currently being processed. As far as the number of occurre (...)
9I have compiled a corpus of more than five-hundred examples of since- subordinates in order to study such criteria, among other things10. In my 513-example corpus selected from the BNC by means of a random search, 64% are causal while only 24% are temporal. A few examples are hybrid, another handful are impossible to classify, while about one tenth of the corpus is accounted for by parsing or tagging errors in the SARA software, which returned some irrelevant examples of since used as an adverb or as a preposition, whereas my request was for since- subordinates only:
10Let’s now turn to more specific results gleaned from the analysis of this corpus. As far as the placement of the subordinate is concerned, the hypothesis according to which causal since- clauses are more often fronted than temporal ones is accurate but it must also be qualified:
- 11 I wish to thank Sylvie Mellet (UMR BCL) for suggesting the use of this test and for performing the (...)
- 12 See in particular the analysis of examples 3 and 6.
11According to the chi-square test, the aim of which is to assess the significance of such statistical results, this distribution is not random and it does reveal a strong tendency for the causal since- clauses to be fronted while the temporal ones will more often be placed after the main clause11. However, these data also show that one temporal since- clause in four in the corpus happens to be fronted, which makes fronting a fairly common feature even in the temporals. In the same manner, according to the absolute values in my sample, a causal since- clause is even more likely to be placed after than before its related main clause. As a result, the placement of the since- subordinate by itself is not a reliable enough criterion for the case in point, as it cannot be used systematically in order to disambiguate between the two types of subordinates. It can, however, be confronted with other criteria, and if they corroborate one another by pointing towards the same interpretation of since, then the placement of the subordinate can be taken as a good indicator regarding the nature of the subordinate12.
- 13 The presence of a punctuation mark corresponds orally to a break.
12The factor of the placement of the subordinate, before or after the main clause, is also related to its integration within the sentence, that is to say whether or not the since- clause is separated from the rest of the utterance by a comma (or more rarely by another punctuation mark)13. According to the analysis of my corpus, both types of since- subordinates can be either preceded or followed by such a punctuation mark; it is true, however, that the recourse to punctuation in order to coordinate the clauses is significantly higher in clauses of reason. Therefore, this is a further criterion which can be confronted with other such indicators in order to determine the nature of the subordinate.
- 14 See Groussier and Rivière 1996: 178, as well as Chuquet, 2001: 166.
13The placement of the subordinate and the punctuation are in fact indicative of whether or not the subordinate serves as an initial locator for the utterance as a whole, that is to say whether it is the main locator in the utterance14. Given the distribution of the placement of the subordinates as well as that of the punctuation, it seems that the two types of since- clauses can work as initial locators (even if this is more commonly the case for the clauses of reason), which again disqualifies this criterion as a reliable way to disambiguate between them, if it is used in isolation.
- 15 T. Lattes’s experiment on just one manipulated sentence (the two predicative relations <he / not us (...)
14The other syntactic criteria which can turn out to be more specific for one type of since- clause than for the other are the tenses, aspects and modality used in the subordinate and in the main clause respectively, and also their combination. According to M. De Cola-Sekali, however, it is mostly the choices that appear in the main clause which are relevant: the use of the perfect (whether present or past, simple or progressive) is characteristic of the temporal use of since, whereas simple tenses or modals tend to signal a causal relationship between the clauses (1992: 131-4)15.
- 16 By convention and in order to make the comparison easier, the combination of markers which is give (...)
15The study of my own corpus confirms these tendencies: the choice of tense, aspect and modality at clause level as well as at sentence level16 is clearly not the same as far as the two types of since- subordinates are concerned. And yet, while being based only on the cases for which there was no ambiguity whatsoever regarding the nature of the since- subordinate, the diagram shows that there are a few possibilities for overlap, which is enough to create the syntactic conditions for ambiguity:
16Let’s now turn to attested examples to see whether the semantic (contextual) or syntactic disambiguation criteria which have just been listed are really operative. For instance, the nature of the following since- clause may prove somewhat ambiguous if it is taken completely out of context, that is to say in the same manner as examples 1 and 2:
(3) CEM 1422 The multi-millionaire actor has maintained a round-the-clock vigil at the bedside of TV reporter Maria Shriver since she was admitted to hospital with a soaring temperature.
17From a semantic point of view first of all, the subordinate is compatible with the two distinct interpretations: Maria Shriver’s serious condition is undoubtedly the cause for the display of unfailing devotion on the part of her husband, while her admission to hospital also marks the starting-point of the vigil in question.
- 17 These examples again support the view expressed in Goodwin and Duranti (1992: 12-3) that it is nece (...)
18As for syntax, it too proves somewhat ‘deceptive’, mostly because the combination of tenses and aspects, namely preterite + simple present perfect (cf diagram C, 5th line from the top), belongs to the list of cases which are documented in my corpus for both temporal and causal since- clauses. But if more textual context is retrieved from the BNC17, it becomes obvious that the temporal interpretation must prevail:
(3’) CEM 1422 Arnie's wife struck by killer brain bug - ARNOLD Schwarzenegger's wife is being treated for the killer virus meningitis. The multi-millionaire actor has maintained a round-the-clock vigil at the bedside of TV reporter Maria Shriver since she was admitted to hospital with a soaring temperature. The disease causes inflammation of the brain but doctors are optimistic Maria, 37 will make a full recovery.
19The larger context given in (3’) shows that the journalist’s serious illness has already been mentioned twice in the span of a few lines (Arnie's wife struck by killer brain bug / Schwarzenegger's wife is being treated for the killer virus meningitis). As a result, the cause for the actor’s presence in the hospital has been made clear to the reader, which renders the causal interpretation very unlikely. This semantic disambiguation criterion supports and strengthens the syntactic evidence found within the sentence, namely the placement of the subordinate after the main clause without a punctuation mark in between, while the combination of preterite in the subordinate with simple present perfect in the main clause (cf. diagram C) is overwhelmingly associated with a temporal interpretation according to my corpus. As for the following examples, the syntactic evidence regarding the choice of tenses and aspects turns out to be even more decisive than it was in the case of example (3):
(4) HH3 9068 Since my wife was diagnosed with the illness, I have been researching alternative cures during which time I have come into contact with the ‘Association stop au cancer’ (…).
(5) A9J 130 It is not as if the Administration really believes that any more. Weary of Israel's ceaseless importuning, it recently produced a long and scholarly ‘confidential report’ which concludes that, since Arafat told the UN General Assembly last December that he recognised Israel's right to exist and ‘renounced’ terrorism, the PLO has not been speaking with ‘forked tongues’ and its declarations have ‘for the most part been consistent, regardless of the media in which they appear.’
20Example (4) could well be given a causal interpretation; besides, the fronting of the subordinate as well as the presence of a comma between the clauses would be in keeping with this. And yet, the use of progressive present perfect in the main clause is a syntactic feature which is in fact incompatible with causal subordinates, according to the findings in my corpus: the subordinate is temporal.
- 18 The fact that the excerpt contains reported speech (or rather reported writing for that matter, as (...)
21The main clause in example (5) exhibits that same syntactic characteristic: it can only be interpreted as temporal, even if a cause and effect relationship can be felt, in this case wrongly, between the clauses. Besides, a closer look at the textual context confirms the temporal interpretation, with phrases such as any more,or again recently,which put an emphasis on the time frame of the events described18.
- 19 Both the combinations present perfect + modal and present perfect + simple present occur exclusive (...)
22As for example (6), it was not taken into account in my statistics: as it is not a BNC example, it does not belong to my original corpus. It displays a combination of simple present perfect in the subordinate associated with a modal in the present tense in the main clause; such a combination never takes place with a temporal use of since in my corpus, whereas it does with causal subordinates19. And yet, from a semantic point of view, the temporal interpretation is not completely unlikely, as the king might be referring to specific occasions on which he had the opportunity to find out more about Madame de Longueville, even if the use of the preterite would have afforded greater clarity in that case. However, the French ambassador’s reply unquestionably supports the causal interpretation, since he does not inquire about the circumstances in which the king gained his information, but rather seeks qualitative clarification about what it was that made the monarch think that Madame de Longueville was a more suitable potential bride, which justifies the cause and effect interpretation of the previous sentence over the temporal one:
(6) (After the death of Jane Seymour, Henry VIII starts planning on his 4th marriage.)
Henry VIII: I wanted to talk to you in private Monsieur l’Ambassadeur, because I’m inclined towards a French bride. (…) Milord Privy Seal spoke to me of two potential French brides: Maria, the King’s daughter, and Marie de Guise, Madame de Longueville. Since I have heard further of Madame de Longueville, I cannot refrain from considering her as a wife.
French ambassador: (…) Hum… What have you heard?
Henry VIII: The King’s daughter’s too young for me. But Madame de Longueville is much more suitable: being herself a widow, and having already borne two sons, and being as they say very… voluptuous. (The Tudors, Showtime; 306: 4’)
23But what happens when syntax is deceptive, and semantics does not come to the rescue? In previous work, I have labelled such cases as examples of hybridism (Guillaume 2009: 199). Generally speaking, hybridism is often understood as the product of the amalgamation of two components of different origins, but there is a tendency to conceive of such an amalgamation in idealised terms – that is to say, to assume that the components necessarily have equal weight. In reality, the examples of this phenomenon in linguistics are not situated precisely halfway between two syntactical categories, such as those of interrogatives and relatives as far as certain wh- clauses are concerned (see Guillaume 2009), or between two types of adverbial clauses when taking the since- subordinates into consideration. Indeed, most of the time, the subordinates in question do not exhibit as many typical properties of one type as they do of another type. Still, the concept of hybridism proves helpful in studying several cases which resist the traditional categorisation of subordinates in English, as they happen to possess at least one property which is in fact not in keeping with the type to which they ought to belong according to most of their other characteristics. Huddleston and Pullum, for their part, make a somewhat more blunt comment in relation to examples which involve the use of the –ing suffix and which display both nominal and verbal characteristics, a comment which, after all, might apply rather well to the concept of hybridism: “such examples resist elegant description” (2002: 1189).
24As with ambiguity, hybridism implies that one subordinate turns out to have more than one possible meaning; nevertheless, the syntactic clues are either not telling enough, or they yield contradictory results, while at the same time the recourse to context is not sufficient to eliminate the ambiguity and opt for one interpretation while excluding the other completely. Besides, these two meanings are complementary, not contradictory, a condition which is easy to meet in the case of the since- subordinates as the same event can often be considered both as the cause and as the starting point of what is described in the main clause, as has already been shown for some of the examples studied here. Therefore, in the case of hybridism, the type of disambiguation which I have attempted to put into practice so far will fail. As a result, syntax is not just ambiguous, it becomes deceptive, for both purely syntactic and also for semantic reasons.
25From a more technical point of view, it can be said that the type of location which is at play between the main clause and its subordinate in the case of hybrid since- clauses displays both quantitative and qualitative features, which in turn makes room for two possible interpretations of the clause at the semantic level. Let us now study several such examples:
(7) EVS 1821 The Catholic religion has marginalized women, in part, one might say, since it was decided that God was male. In the nuclear family, women are either seen as 'mother' or 'lover', the root cause of the unhealthy machismo of Salvadorean men. Now women participating in the liberation struggle are finding new roles. Obviously we are not going to cure the blight of centuries in a few years.
26According to the statistical analysiscarried out on my corpus, the combination of the preterite in the subordinate and the simple perfect in the main clause (see 5th line from the top on diagram C) accounts in overwhelming numbers for the temporal use of since, even though it is attested in a few cases with the causal use as well. The temporal interpretation of example (7) is also supported by the fact that the reader may be prompted to wonder when exactly it was decided that God was male.
27In spite of all this, the interpolation of the adverbial phrase in part explicitly signals the fact that the subordinate which follows will only give a partial answer regarding the problem at stake. Such a description could hardly apply to a temporal starting-point: it makes a lot more sense to say that the decision that God was male is only a partial cause of the marginalisation of women in the Catholic religion. The case would have been different with the use of in particular,or particularly,for instance, instead of in part, as this would have been compatible both with a temporal and a causal subordinate. But in its given form, there is some dissonance in this example between the syntactic arrangement of the utterance (namely, the choice of tenses and aspects) and the meaning of the adverbial phrase in part which bears on it. The fact that the subordinate comes after the main clause is not very telling in itself, while the use of commas is due to the interpolation of in part and one might say respectively. As a result, contrary to all the other examples of since- clauses which have been examined so far, the two interpretations of the subordinate really do seem to co-exist as they may be different in theory, but hardly so in practice. In such a case, it seems more productive to admit that the subordinate displays a blend of characteristics from the two types of adverbial clauses on the syntactic level, while its semantics remain unchanged.
28In the next example, the syntactic analysis coupled with the taking into account of the context again yields mixed results, which is consistent with a type of location of a dual nature (partly quantitative, partly qualitative) between the subordinate and the main clause:
(8) FDC 118 Before the coming into force of the Children Act 1989, the appropriate step would have been an application to make W. a ward of court. Since that Act came into force, a child who is the subject of a care order, as W. was and is, cannot be made a ward of court: see section 100(2) ( c ) of that Act.
- 20 According to the chi-square test which was carried out on the results given in diagram B.
29If one first takes into consideration the syntactic arrangement of the sentence, all the indicators point towards a causal interpretation, without formally ruling out a temporal interpretation. Even if the combination of preterite in the subordinate with a modal in the main clause is more typical of the clauses of reason (see 12th line from the bottom on diagram C), it is also attested for temporal clauses. In the same manner, the fact that the subordinate is fronted and followed by a comma indicates a statistically higher chance of its being causal20, but again temporal subordinates are also compatible with such an arrangement. From a semantic point of view, however, there is a striking temporal contrast between the two temporal points of reference, namely before the coming into force of the Children Act 1989 vs. since that Act came into force.
30Finally, the following example displays a combination of simple present perfect + simple present perfect (see 6th line from the top on diagram C) in the subordinate as well as in the main clause, a combination which is rather rare in my corpus for both causal and temporal clauses, even if it is a little more common with temporal clauses:
- 21 It should of course be noted concerning this example that it is a transcription from the oral, whic (...)
(9) KB2 1937 he used to be right out here Frank didn't he? He used to be right out here Frank, he used to be a right paunch he's er, since he's had his heart attack he's slimmed down.21
31From a semantic point of view, the cause and effect relationship between the two clauses is undeniable: it is because Frank has had a heart attack that he has slimmed down. Nevertheless, the recourse to the aspectual phrase used to, which isrepeated no less than three times in the short space of this transcription, draws attention to how much bigger the person was, and underlines the contrast between now and then; as a result, the emphasis is on the time frame rather than the cause and effect relationship. Of course, even if the time frame is prevalent, the heart attack is still very clearly the cause of Frank’s weight loss, which makes example (9) a good candidate for hybridism, since the two possible interpretations are allowed to co-exist both from a semantic and from a syntactic point of view.
32As far as such examples are concerned, there remains the important question whether such occurrences (or some of them) are deliberate on the part of the speaker / writer. There is indeed no denying that the resonance of the subordinate and, as a result, that of the context in which it is used, becomes enriched by the fact that there is more than one interpretation, all the more so as these add up rather than replace each other. This creates some tension between the competing interpretations and is necessarily more complex from a semantic point of view. It can hardly be called a stylistic effect however, since it clearly is not voluntary. Also, from looking at the examples, it is obvious that such a phenomenon can be generated irrespective of the language register, as in example (9), for instance, which contains substandard English. Therefore, it really is the human factor which is at play in the use of language which prevails in the invention of such subordinates, not the stylistic factor. This is brought about by the felicitous coincidence between an ambiguous syntax and a propitious semantic context: several meanings are allowed to co-exist, thanks to a deceptive syntax and a favourable environment.
- 22 See Guillaume 2006 and Guillaume 2009 for problematic assignments of part-of-speech categories as (...)
33What is also made plain by such a phenomenon at the syntactic level is that the traditional categorisation of the subordinates (concerning both their nature, i.e. their part-of-speech22, and their function) sometimes turns out to be relatively inadequate. In such cases, one is tempted to consider the possibility of an overlap zone between certain categories of subordinates, while the categories themselves prove porous. Yet I am certainly not advocating the replacement of the traditional classification altogether by fuzzy categories, or by a merging of all types of subordinates into one. The traditional description and classification of subordinates is essential, but it can only be effective if it makes some room for a constitutive remainder inherent to language, a concept introduced into the field of linguistics by Jean-Jacques Lecercle. Using cartography as a metaphor for grammatical rules, Lecercle explains: “The details that any grammatical map necessarily leaves out constitute what I call the remainder” (1990: 19).
- 23 Lecercle in his turn was inspired by Judith Milner’s characterisation of the concept of frontier ( (...)
34Indeed, regarding some of the examples examined in this paper, the linguist is left at the end of the day with only two choices: either to take for granted the fact that such examples “resist elegant description”, in Huddleston and Pullum’s words, or to find in them an opportunity to acknowledge the limitations of grammatical description. Still building on J.-J. Lecercle’s ideas, such inadequacies in fact turn out to be frontiers23 rather than limitations, but frontiers which are “arbitrary and changeable” (1990: 25), and are apt as such to account for the complexity of human language, irrespective of style and register. It is indeed an illusion to expect grammar, even a very accurate, corpus-based, grammar, to mirror exactly the reality of language. As Lecercle puts it, “the only comprehensive grammar of a language would be coextensive with the language itself” (1990:18). There can and there must remain certain grey areas which do not necessarily call for a more precise description, beyond recognising that they exist and make sense somehow.
- 24 See Guillaume 2005 as far as the choice of tenses in a when- subordinate is concerned.
35To return to the question of a ‘deceptive’ syntax, I would also like to underline the fact that, in most cases, not only is syntax not deceptive, but it is a major disambiguation criterion, for instance through consideration ofthe choice of tenses and aspects in the case of since- or when- subordinates24. The corollary to this is that, in the rare cases in which syntax itself is a source of ambiguity, it opens no less than a Pandora’s box by creating the possibility for confusion, vagueness, or even misunderstanding.