Polysemy and English THROUGH

Carey Benom

Here, I ask the question “How can we tell if linguists are employing in their analyses of polysemy the same senses that language users employ in their cognitive behavior?”. My primary purpose is to demonstrate an empirically responsible lexical semantic analysis of polysemy within the framework of cognitive linguistics, grounded in empiricism and explicit, clear definitions, with the purpose of ensuring that languages users are actually employing the senses described in the analysis. To do this, I take an approach to polysemy that resolves the age-old issues plaguing polysemy research, employing it to good effect in the analysis of THROUGH. I take polysemy to refer to a lexeme or construction with multiple senses which are differentiable syntactically as well as semantically, with a constrained number and variety of potential relationships holding between the senses. Because my definition of polysemy is grounded in syntax, the senses described in the analysis can be assumed to be cognitively ‘real’ to language users, as they actually employ these senses. Furthermore, my analysis is grounded in corpus data. The analysis involves providing both syntactic and semantic evidence for the independence of each sense, as well as describing a plausible motivation for the synchronic relations between the senses. The success of the approach can be seen in part as an argument against overly simplistic analyses of polysemy. Even syntactically differentiable senses, which are presumably more deeply incorporated into speakers’ linguistic routines than those that show no syntactic distinctions, exhibited a high degree of syntactic and semantic overlap and ambiguity. This monograph contributes to the study of polysemy by resolving some long-running methodological issues hindering polysemy research, including ensuring that the linguist and the language-user are working with the same sense categories.

Polysemy and the structure of English THROUGH I. Introduction and purpose Most words have more than one meaning. Essentially any dictionary one picks up will support this statement. But how can one establish the number of meanings of a word, and their distinctions? One primary purpose of this research is to provide a satisfactory answer to this question by modeling a lexical semantic analysis of sense differentiation grounded in empiricism and explicit, clear definitions. But this central question remains: How can we tell if language users are employing the same senses that linguists describe? Without an answer to this question, we are at an impasse, as it is not clear what linguistic analysis means if it does not reflect what speakers actually do. As linguists, are our analyses theoretically pleasing, or even beautiful, but with little bearing on reality? Or do they reflect speakers’ conceptual and behavioral processes? Of course, nobody objects to the elegance of an analysis, but without relevance to what language users are actually doing and thinking, analysis quickly becomes an exercise in futility. The main purpose of this article is to provide an answer to the question in bold font above, or at least a solution to the problem it suggests, by analyzing the semantics of the English preposition and particle through within the context of the discussion of polysemy, and to do so by providing an empirically rigorous definition of polysemy, heretofore lacking in the literature. I take polysemy to require both morphosyntactic and semantic differentiability of senses, and in doing so, I provide an empirical method of sense differentiation based on speakers’ behavior. In other words, because my definition of polysemy is grounded in syntax, the senses described in the analysis can be assumed to be cognitively ‘real’ to language users, because they are based on their (linguistic) behavior. Lexicographers, linguists, psycholinguists, computational linguists, and artificial intelligence researchers have been enthralled (and, at times, troubled) by polysemy since Bréal’s (1897) pioneering discussion. However, within generative linguistics, polysemy has received little attention – indeed, Cann (1993) suggests that linguists can effectively ignore polysemy for ‘expository convenience’ (1993:8). This is because, in the generative tradition, the lexicon is seen as a list of exceptions from general principles (Chomsky 1995:235), and hence different word senses are often simply represented as different lexical items, though this approach has been criticized for failing to take into account the systematic relationships between forms (Jackendoff 1997, Langacker 1991a, Levin 1993, Pustejovsky 1998, Tyler and Evans 2003:96). In contrast to this situation, polysemy can be said to be one of the early, core, defining issues of cognitive linguistics, and has been seen as one of its most successful early endeavors as well (Brugman 1988, Brugman and Lakoff 1988, Cuyckens 1993, 1995, Deane 1992, Evans and Tyler 2004, Geeraerts 1993, 1994, Lakoff 1987, Rice 1993, Sandra and Rice 1995, Tyler and Evans 2001, 2003, Vandeloise 1990, inter alia). Cognitive linguistic approaches to polysemy rejected formal semantic views of compositional meaning and instead viewed meaning as embodied, which means that it is not logical or semantic features or abstract propositions that speakers employ, but rather image schemas, or imagistic, schematic representations which arise from perceptual analysis of patterns that recur, or, in other words, from existing as a human, with a body, in the spatiotemporal world (Gibbs and Matlock 2001, Lakoff and Johnson 1999, Johnson 1987). Therefore, the meaning expressed in language is seen as mediated through human perception and conceptualization. This means that cognitive linguists assume a model of encyclopedic semantics, in which lexical items are not given “dictionary definitions” in speakers’ minds but are seen as access points to encyclopedic knowledge (e.g. Fillmore 1985 and Lakoff 1987), which replaces truth-conditional semantics with actual world knowledge, based on sensory-motor experience, perception, and conceptualization. Additionally, they assume a model of prototype categorization (Lakoff 1987, Rosch 1978), substituting necessary and sufficient conditions with prototype structure. Eventually, because of their adherence to the principles of construction grammar – and especially the idea that lexicon and syntax are not essentially different in kind, but belong on the same continuum, as both lexemes and syntactic constructions are forms with functions – cognitive linguists’ analyses of polysemy began to be applied to forms larger and smaller than that of the word (e.g. from morphemes, such as Panther and Thornburg’s (2002) analysis of the polysemy of –er, to syntax, such as Goldberg’s (1996) treatment of the polysemous meanings of argument structure constructions). However, within this apparently successful research paradigm, serious problems arose. In more than a dozen distinct accounts of the polysemy of over (e.g. Brugman 1981, Brugman and Lakoff 1988, Deane 1992, 1996, Dewell 1994, Kreitzer 1997, Lakoff 1987, Queller 2001, Rice 1993, Sandra and Rice 1995, Tyler and Evans 2001, 2003, Vandeloise 1990, and many others), cognitive linguists invariably reached different conclusions, and it was not clear how to compare and test their analyses. Importantly, the lexical network approach employed in most early studies of polysemy within cognitive linguistics was criticized on various grounds. Sandra and Rice (1995, see also Rice 1996, Sandra 1998) provided an early and apt criticism of the approach, using psycholinguistic evidence to cast doubt on the cognitive reality of sense distinctions resulting from such analysis. Their primary criticisms – that definitions of categories and theoretical apparatus (such as prototype) and explicit protocol for distinguishing senses were lacking, and that linguists’ analyses may bear little relation to cognitive reality – can be said to be have inspired the approach taken here. Taylor (2003a, b) calls attention to our cognitive models of language and how they can affect polysemy analysis, and later (2012) warns that purported polysemy may be due to context, rather than in the semantic specifications of the item itself. Tyler and Evans (2001, 2003, Evans 2004) advance the argument by clearly demonstrating some of the theoretical weaknesses in the cognitive linguistic treatment of polysemy. Their approach takes steps toward clarity and empiricism in sense differentiation, distinguishing polysemy from vagueness, and explicit definitions of key concepts, though their reliance on intuited 2 Polysemy and the structure of English THROUGH data and primarily semantic analysis is ultimately insufficient in resolving the problems with polysemy research, or in answering the questions raised in the first few paragraphs of this paper. Unlike many discussions of polysemy within cognitive linguistics, the approach I take here looks to the language user to provide evidence for the differences that are real. That is to say that an inductive approach (Glaser and Strauss 1967) to data analysis like this one begins with observations based on the data before looking for patterns that can be used to make and test tentative hypotheses, ultimately leading to the creation of theory. This type of data-driven approach is in opposition to deductive approaches, which begin with theory and attempt to find patterns in the data that support (predictions made by) that theory. One cognitive approach to polysemy developed more recently that is also data-based and inductive can be seen in the work of Gries and Stefanowitsch (e.g. Gries 2006, 2010, Gries, Hampe, & Schönefeld 2005, 2010, Gries and Stefanowitsch 2004, Stefanowitsch 2003, Stefanowitsch and Gries 2005). Their corpus-based methods take great strides towards empiricism, and their work represents a significant advance in the analysis of lexical meaning. However, some specific details of the methods – most importantly, implicit acceptance of what I refer to as “the correctness fallacy” (Author 2007, in press), or the myth that, for every question, there is exactly one correct answer – may cast some of their results into doubt. I will refer the reader to Author (2007:169-180) and Author (in press) for a detailed discussion of their methodological shortcomings, which I cannot discuss at length here due to lack of space. Here I will simply state that, by assigning every token in their data to a category, including categories that may overlap or have unclear boundaries (such as degree of involvement, extent of controllability, and abstract vs. concrete), without an explicit definition of the category or discussion of borderline cases, they fall prey to an implicit acceptance of the correctness fallacy, and the reliability of the results made based upon the assignment of data to these categories is called into question. Note that it is not the fact that they are using intuition to assign tokens to categories while coding the data to which I object, but it is rather the absence of explicit definitions of the categories to which the tokens are assigned, in addition to the lack of any mention of unclear or borderline cases, or description of coding protocol in such cases. It is not my intention to disparage the entirety of the approach sometimes known as collostructional analysis, created and practiced by Gries and Stefanowitsch, which I have also employed (and with good results) in my research (Author 2007, 2014), but simply to point out that even the most sophisticated quantitative methodology can be compromised by implicit acceptance of the correctness fallacy. Here, rather than attempting to assign every token in the data to a category, I will observe the range of syntactic and semantic behavior in the data and make generalizations based on these observations. I will also attempt to leave no relevant category ill-defined or undefined. Importantly, while Gries (to appear) may not believe in or have given up on word senses (“the notion of distinct senses appears more of a descriptive device rather than a claim about psycholinguistic reality”, to appear §4, see also Kilgarriff 1997), I will describe senses that are behaviorally differentiated, assuring their psycholinguistic reality. 3 I will consider polysemy to refer to a situation in which a lexeme or other construction (e.g. Goldberg 2006, Langacker 1987) has multiple senses (categories of use / meaning) differentiable both syntactically and semantically and a constrained number and variety of potential relationships holding between the senses, as detailed in Section 2 below. By requiring that senses of a polysemous term be syntactically as well as semantically differentiable, the burden of evidence – as well as its reliability – is strongly increased. This syntactic requirement is part of an explicit definition that will be applied in the analysis of corpus-derived data to construct a detailed picture of the structure of through, attesting to the utility of the approach. Grammatical polysemy may be particularly relevant to the study of cognition, given the claims of the primacy of grammatical operators in much of cognition (e.g. Talmy 2000). Research in polysemy is one avenue of broadening our understanding of lexical structure, and therefore, like other research in lexical semantics, facilitates our understanding of human conceptualization and knowledge in general. It is fair to say that this article is long overdue. I first wrote on this topic in my thesis in 2003, and later revised my analysis in my dissertation (2007). I then published several articles that invoked, or were based on, the analysis presented here (Author 2009, 2010a, 2010b, 2014, in press). The line of argumentation I take here does not differ substantially from that I put forth in 2007, but some implicit reasoning has been made explicit, many details have been refined, and I have edited what I presented there for clarity, consistency, and brevity. In this article, the term ‘sense’ will be used, with modification, in two ways. The first, primary sense, refers to a differentiable category of meaning and usage of a polysemous term, as in ‘there are four primary senses of through.’ The use of this term implies that there is morphosyntactic argumentation that reveals that speakers treat each primary sense differently. The other term, sub–sense, will refer to uses with a pragmatic or semantic distinction that is not represented in the morphosyntax of English1. Thus, I may discuss a sub–sense of a particular primary sense of through. In addition, I will use the term usage-type, as in Kilgarriff (1993), in a general way, subsuming what I call primary senses and sub–senses (e.g. for as–of–yet unanalyzed uses that appear to have different meanings, or where specification is not crucial or relevant to the argument); this is identical to the general use of sense often found in the literature on lexical semantics. The product of a usage-event will be referred to as a use, and each use in a corpus that is counted as data, a token. Below, I will describe the four primary senses of through, but I will first provide a preview here. I refer to the primary senses as Through1 (which I will argue is the ‘central’ sense), ThroughEB (in which there is a focus on the crossing of the final or ‘End Boundary’ of the Ground2), ThroughNB (in which there is a focus on the motion within the Ground, to the exclusion of the conceptualization of 1 I will use the term sub-sense to subsume Croft and Cruise’s (2004) sub-sense and facet, as I do not find it useful to distinguish them here. 2 I use the terms Figure, Ground, and Path as in Talmy (2000). 4 Polysemy and the structure of English THROUGH boundaries, and hence invokes ‘No Boundaries’), and ThroughALL (in which the conceptualization of filling of the boundaries of the Ground is strongly entrenched, commonly expressed as all through). For the sake of readability, other than for stylistic clarity, I will refer to these primary senses as 1, EB, NB, and ALL, respectively.3 These are the four primary senses represented in Figure 1 below, in which the representation of each primary sense includes, from top to bottom, its name, a representation of its geometry, a paraphrase, and an example. Figure 1 – Representations of the four primary senses of through (taken from Author 2007, 2010a, in press; slightly edited) In Figure 1, the Ground is represented by a rectangle, the Path by solid black arrows with tails; a dotted tail means that motion is required, while a solid line for a tail implies that motion is not necessary. The Figure’s scans, used to build up a mental representation of the entire Ground, are represented with slimmer arrows. Arguments for the analysis represented here will be provided below in Section 3. 1.1 On the organization of this article The article will proceed as follows. In Section 2, I will address several facets that can be considered to be the background to the study. First, I discuss the methodology I have used, including my source of data, and then I cover the background as well as my approach to polysemy, beginning with the traditional semantic tests and describing various problematic issues relating to these tests before moving on to my own theoretical and operational definitions of polysemy. I will then undertake a brief survey of the scant attention through has received before 3 For examples discussed in the text by number, the number will invariably be enclosed in parentheses, and therefore a bare numeral 1, lacking a nominal referent (such as “Figure 1” or “Section 1”), will always refer to the primary sense of through, as in EB and 1 overlap to a considerable extent. 5 moving to Section 3, in which I discuss each primary sense in turn, addressing in each case the syntactic and semantic latitude of the primary sense, and the syntax and semantics differentiating each primary sense from the others. In Section 4, I will describe some frequently occurring sub-senses of through before the conclusion in Section 5. II. Background 2.1. Methodology The analysis is based almost entirely on data derived from the 100-million-word BNC, even where I have not provided examples from that source for the sake of brevity. I say “almost” because, in addition to examining data from the BNC, I read the OED entry for through, and my intuition as a native speaker of North American English and observations of uses in my environment led to the description of several usage-types not represented in the BNC. In such cases, or when the BNC data led to a question or problem, data from the Corpus of Contemporary American English (COCA; Davies 2008-), which currently contains 450 million words, was considered. If no examples were found, I employed the enTenTen12 – an 11 billion-word corpus of English created by web crawling (Jakubíček et al. 2013). When even queries in this huge corpus did not suffice, as a last resort, I occasionally used data from Google. 4 Barring a specific reason to make use of such data, however, I have invariably relied on the BNC data. All corpus data have their source clearly labeled, and any data without a source listed should be considered to be intuited. My method of extracting data from the BNC was as follows. First, 300 tokens were pseudo-randomly derived from each of the seven genres; spoken, fiction, magazine, newspaper, non-academic, academic, and miscellaneous, and this total of 2100 tokens was carefully examined. After reading through all of these, I formed the basis of my analysis, and then I tested this analysis by examining more than 700 additional tokens (at least one hundred more from each genre, again, pseudo-randomly extracted). Following this, I extracted 250 tokens from the spoken portion of the corpus and 250 tokens from the written section of the BNC, and analyzed these 500 pseudo-randomly selected tokens in careful detail, as described in Author (2007). In total, I examined more than 3000 tokens of through from among the 81,184 in the corpus. Acceptability judgments are based on the author’s intuition, and marked as follows: For syntactic ungrammaticality, an asterisk (*) will be used. For semantic or pragmatic ill–formedness, a pound sign (#) will be used. For questionable cases, a question mark will be added to the asterisk or pound sign (?* for syntactically questionable data, ?# for semantically or pragmatically questionable data). Cases that are unclear, or for which it is not clear if the oddity is due to syntax, semantics, or pragmatics, will simply get a question mark. 2.2. On polysemy 4 All Google queries were performed (or re-performed) between November 20 and 24, 2014. 6 Polysemy and the structure of English THROUGH 2.2.1. Background on polysemy There is widespread belief in cognitive linguistics (as well as lexicography, natural language processing, and translation studies) that polysemy is endemic in language (Taylor 2003a, b). Studies of lexical semantics done by philologists, prior to the establishment of cognitive linguistics, tended to define polysemy based on the etymology of the lexeme in question. This reveals an interesting but non-equivalent phenomenon, and is ineffective in assessing synchronic state of speakers’ understanding (see, for example, Rudskoger 1952). The fact the native speakers are not etymologists was also argued to be significant (Menner 1936, Lyons 1977), as polysemy has typically been considered a description of the synchronic state of a lexeme or morpheme since the first use of the term (Bréal 1897, who describes polysemy as “the synchronic side of language change”, as mentioned in Blank 2003). Researchers within the framework of cognitive linguistics are, of course, concerned primarily with the state of understanding and conceptual categories of speakers, and so their definitions attempt to isolate a synchronically active network of links between senses of a single form. Traditionally there is a distinction between one word with multiple meanings, or senses (polysemy), and multiple words that happen to share a single form (homonymy), such as fluke (a fish) and fluke (a lucky coincidence). The crux of the distinction lies in whether the multiple meanings ‘are related’, though it is extremely difficult to define precisely what this means. Multiple scholars have situated polysemy in the center of a continuum with homonymy and vagueness (also called “contextual variation”, or “allosemy”) on the opposing ends (e.g. Croft and Cruse 2004, Cruse 2000, 2002, Geeraerts 1993, Tuggy 1993). Using this perspective, homonymous items possess multiple unrelated meanings, a polysemous item has multiple related senses, and a vague item has one meaning with limited variation or breadth possible depending on the context. As an example of a vague item, Croft and Cruse cite friend, and how it can imply a male friend or a female friend in certain contexts (they cite My best friend married my brother/sister)5. They argue that, “In the absence of contextual pressure, the neutral reading of friend will be construed.” (2004:127). Therefore, such differences are not taken to be evidence for separate senses of a word. Ambiguity is a term used in Lyons (1977) to encompass both homonymy and polysemy, but used as an equivalent to homonymy by those such as Blank (2003). I will avoid using it here except in a general sense, to describe some linguistic utterance with two or more potential interpretations. One defining factor in studies of polysemy is the posited structure of the senses, including the relationships between the senses and the presence (typically assumed) of a central, basic sense that motivates the extensions to the other senses (such as the ‘above-across’ sense of over seen in The plane flew over the yard, according to Lakoff 1987). Some semantic network theorists invoke extension processes recursively, such that a chain of related senses extending out from the central sense results (e.g. Fillmore and Atkins 2000). However, there is also the possibility that the different senses of a polysemous word are not represented as 5 This example of theirs depends on the impossibility of same sex marriage. 7 chains or lexical networks, but rather that speakers detect (or, possibly more appropriately, create) similarity among the senses during online processing. This view states that the relationships between different senses of a polyseme are created in the context of processing, instead of being determined based on a pre-existing lexical representation (Barsalou 1987, Gibbs and Matlock 1999). I will not address this question in detail, but by using a syntactic criterion for sense differentiation, I focus this analysis on those senses that are conventionalized, rather than created online (see Jackendoff 2002 for a discussion of related problems with regular and irregular morphological processes). Therefore, the primary senses I describe here are clearly not created (solely) during online processing, because their syntactic behavior shows that they are routinely behaviorally distinguished. 2.2.2. Semantic tests for polysemy The literature on polysemy contains many semantic tests claimed to diagnose polysemy. These are discussed in detail and exemplified by Croft and Cruse (2004), Cruse (1986, 2000), Geeraerts (1993, 1994), Kilgarriff (1992, 1997), Quine (1960), Ravin and Leacock (2000), and Zwicky and Sadock (1975). Most are based on posited linguistic constraints against referring to two usage-types of a word with a single use, including the ‘crossed readings’ test, which predicts that, if through is polysemous, for sentences such as Alice went through the tunnel, and so did the Mad Hatter, the usage-type of through must be the same for both Alice and the Mad Hatter. My intuition tells me that this is probably the case. However, it may not be in instances such as the created example below, in which the first use of through is an instance of ThroughNB, but the ellipted use should be an example of Through1 (described below). 1. Jack swam through the water; so did Jill, but she did it to cross the river, whereas he was just having fun swimming around. However, the clause with the ellipted through (realized as so did Jill) could be interpreted as NB initially, and later adjusted to reflect new knowledge. It is difficult to determine if this is the case, and thus we have already found the main problem with reliance on such semantic tests. ‘Zeugma’ (also called punning or syllepsis) refers to the situation of two antagonistic usage-types simultaneously being evoked by a single use of a word. The classic example is John and his driver’s license expired last Thursday. The prediction as it applies here is that sentences such as Jeremy went through the rain and the tunnel are only interpretable without a sense of zeugma if there is a single usage-type of through that applies for each Ground within the scope of through in the sentence. My intuition suggests that this may be the case. However, even a single primary sense (ALL) applied to different domains (space and time) may give a sense of zeugma here, as in the example below. 2. Joshua walked all through the afternoon and the tunnel. 8 Polysemy and the structure of English THROUGH One test called ‘compositional autonomy’ is described by Croft and Cruse (2004). It is based on a compositional process (such as the modification of a noun by an adjective) in which one of the elements will take a sub-set of the meaning of the other as its ‘scope’. The facet of meaning is described as showing compositional autonomy, and Croft and Cruse assume that a boundary has been drawn between it and the rest of the meaning (potential). They give the example of a steep bank, which causes the language user to ignore the ‘financial institution’ meaning of bank. However, this may still leave multiple possible senses of bank (“of a river” and “of a turn”). They are also ignoring the possibility of using the term steep bank to refer to a (e.g. cartoon) (financial) bank built on a steep hill with floors that slope steeply. There are many ways that compositional autonomy seems applicable to through. For instance, since NB requires motion (as described below), combining a verb that does not refer to motion with through, such as be through or be located through, prevents an interpretation as NB. However, this many not reveal an autonomous primary sense of through, or even a sub-sense, but may simply select for motion. Such V+ prep combinations are incompatible with NB, but do not reveal anything more about the structure of through. They could go with multiple other primary senses. Similarly, presenting the event as instantaneous forbids the interpretation as NB, such as by combining an adverb such as suddenly, twice, etc., but this still permits multiple primary senses of through. Additionally, contexts such as suddenly go/set off through the water can be interpreted as the Figure suddenly beginning to go through the water, permitting conceptualization as NB. Adverbs (or other linguistic cues) referring to a process do not typically co-occur with EB, such that get through smoothly and slowly go through do not easily allow an interpretation of EB, but they could reflect a use of either 1 or NB. In this way, one can see similarities with the steep bank example discussed above. The compositional processes studied do not select a single primary sense of through, though they do eliminate one or two senses as potential candidates. In many cases, if we increase the number of compositional processes, such as by adding an NP for the Ground, an adverb describing manner of motion, etc. this could select a single primary sense, but as Croft and Cruse’s discussion of the test does not mention a way to evaluate the resulting data other than by intuition, it is not clear how this differs from simply ‘examining the context’, other than the fact that all contexts are created by the linguist using introspection (as is true for all the semantic tests discussed here). Therefore, while I will describe tests of compositional autonomy (or ‘examine the context’) for descriptive purposes, I will make an effort to analyze corpus-derived data whenever possible, and use such data as the foundation of the analysis. The next semantic test I will mention suggests that evidence for the polysemy of through could be found if a sentence such as 3. Samantha walked through the park 9 could possibly be both true (e.g. she walked around within the park for a while) and false (e.g. she never made it through to the other side of the park) simultaneously (see Quine 1960). This is very similar to the test Cruse (1986) discusses which would state that evidence for the polysemy of through can be found if multiple usage-types could be referred to by a single token in some of the examples above, and in such a case they would be true but not redundant ( e.g. Samantha walked both Through1 (‘from one end to the other’) and ThroughNB (‘within’) the park). However, consistent with what we have already seen, it has been noted that such tests are problematic, not least for the fact that they are not reliable (i.e. different analysts will have different intuitions) and are inconsistent (i.e. different tests may lead to different results for a single analyst, and slightly manipulating the tests will lead to different results; see Cruse 1986, 2000, Geeraerts 1993, 1994, Goddard 2000, Lyons 1977). Furthermore, if meaning is seen as being a flexible and open-ended process of creation, rather than a static, reified collection of entities (e.g. Croft and Cruse 2004, Geeraerts 1993, 1994, Fillmore and Atkins 2000, Raukko 2003), it is not clear what such tests are really testing. Due to these methodological problems with the tests, as well as this theoretical issue that suggests that these tests are fundamentally misguided in their assessment of what meaning is, I will not rely on them for my analysis, and though I will mention them occasionally for expository or clarification purposes (other than ‘examining the context’, which I will do consistently), the foundation of the analysis lies in the syntactic and semantic differences between primary senses, as well as the type of motivation of their relationships. Finally, I will note that it will become apparent as we proceed that a monosemist position cannot account for the full range of uses of through. For instance, two senses will be shown to be inherently contradictory (one, ThroughEB, will be shown to involve the crossing of a boundary, and another, ThroughNB, will be shown to explicitly lack any notion of boundaries). Immediately below, other types of problems in studies of polysemy will be discussed. In the following section, I will respond to the question is Is polysemy real? Once I have briefly discussed some behavioral and brain-imaging evidence that polysemy is real, and should be considered a status with respect to the organization and processing of lexical items, I will consider some difficulties and problems in cognitive linguistic studies of polysemy. 2.2.3. Evidence that polysemy is real There is various behavioral and neuroimaging evidence that polysemy is not just simply created by linguists and lexicographers, but should be seen as a status of the cognitive organization of the lexicon. Several studies focus on testing for differences between homonymous and polysemous words, and show that they are processed differently. For instance, in eye-tracking studies, Frazier & Rayner (1990) show that polysemous words require shorter fixation times than homonymous words in reading tasks. Azuma & Van Orden (1997), who avoid labeling words homonyms and polysemes by referring to words with multiple meanings of high or low relatedness, show that those with unrelated and less related meanings produced the slowest 10 Polysemy and the structure of English THROUGH response times in lexical decision tasks, and that the number of meanings of an item has no effect on lexical decision performance. A group of scholars from the Centre for Research on Language, Mind and Brain at McGill University has run several experiments on the structure of polysemy using brain imaging techniques – primarily measuring event related potentials, or ERPs. One fundamental notion in ERP research is the N400, or a negative voltage deflection that peaks at approximately 400 milliseconds after presentation of the stimulus. For instance, Klepousniotou et al. (2005) measure the time-course of meaning activation of polysemous and homonymous words using both behavioral measures and ERP measures and tracking subjects’ responses to visual presentations of stimuli. Their findings show that homonyms and polysemes are processed differently. Tokens of homonyms show ‘dominance effects’: more frequent meanings result in a reduced N400 relative to those resulting from target words that are unrelated to the prime. Tokens of polysemous words, however, showed decreased N400 effects relative to unrelated targets regardless of whether a dominant (frequently occurring) meaning or a subordinate (infrequent) meaning was involved. This indicates differences in processing patterns between homonymous and polysemous words, suggesting that homonyms are indeed stored separately in a way that polysemes are not. This confirms earlier research by Klepousniotou (2002) that showed faster reaction times and greater priming effects for metonymically-related senses of words (such as turkey as animal and food) than for homonyms. Klepousniotou et al. (2008) show that comprehension of polysemous and homonymous words depends on the semantic overlap of word meanings, with metonymically-related meanings showing reduced effects of context, and metaphorically-related (and unrelated meanings) showing stronger context effects. More recently, Klepousniotou et al. (2012) measure ERPs in response to a lexical decision task, investigating the processing of polysemy vs. homonymy, and their results show a clear differentiation in processing of homonymy and of polysemy based on N400 content. While more research is needed on the representation and processing of polysemy, the work discussed here unanimously concludes that polysemy is a real, cognitive phenomenon. 2.2.4. Some thorny questions in studies of polysemy Early studies of polysemy provided the major source of momentum for cognitive linguistics during its inception and initial stages (e.g. Brugman 1988, Lakoff 1987). The questions of what polysemy is and how to study it empirically reverberate at the core of lexical semantics; though this has been the case in cognitive linguistics at least since Brugman’s study of over a quarter century ago, there are no clear, accepted solutions to these thorny problems. For instance, the English word run is claimed by linguists and lexicographers to be polysemous, as there is one way of using it that is paraphrasable as ‘fast pedestrian motion’ (I ran a marathon last June), another as ‘flow’ (The tears ran down his cheeks) and another as ‘manage’ (You sure run the business cautiously), among others (e.g. Gries 2006). However, the question of how the linguist 11 determines that the senses in his or her analysis are distinguished by speakers is typically not addressed (Gries 2006 being a notable exception). For instance, should the ‘flow’ sense of run be considered a special case of the ‘fast pedestrian motion’ sense, in which liquid, rather than a human being, is moving? Furthermore, in order to make such a claim, what kind of evidence would be required? Without a principled way of determining how to differentiate senses in such cases, an analysis is unconstrained, but any method of differentiation must reflect the (potentially covert) knowledge of the language-user, or else there is a risk of the analysis being true to the theory, but not to actual behavior. Additionally, it is conceivable that two English speakers may organize the senses of run differently, or that one speaker may use different semantic structures at different times, depending on the purpose, but these possibilities are rarely considered in studies of polysemy. The most fundamental problem in research on polysemy is this lack of an empirical operational definition of polysemy, as Geeraerts (1993) describes. For instance, Rice (1996) appropriately points out that “network models are especially dependent on the particular analytic skills or subjective aesthetic of the individual researchers” (1996:137-138). Without some constraints on one’s definition, the number of senses one may posit is essentially infinite. As noted by Clark (1983), “the meanings of most words are innumerable, in that they can express an unlimited number of meanings given the wide range of discourse contexts in which words appear” (Gibbs and Matlock, 2001:214-215, discussing Clark 1983). However, there is, to my knowledge, as of yet no satisfactory operational definition of polysemy that circumvents this problem. The general trend in studies of polysemy is to provide an argument by assertion for polysemy (e.g. Brugman 1988, Lakoff 1987, Talmy 2000). Some researchers, such as Hallan (2001), ignore the issue altogether, preferring to discuss “different functions” of prepositions without addressing the relationship between, or establishing satisfactory definitions of, multi-functional lexemes, polysemy, homonymy, and vagueness, though, of course, if polysemy is a real cognitive phenomenon, as I have argued, this does not do justice to it. The general treatment of polysemy may be epitomized in Lyons’ (1977) statement that “The difference between homonymy and polysemy is easier to explain in general terms than it is to define in terms of objective and operationally satisfactory criteria” (1977:550). In fact, the extent of the problem is sufficiently great such that Lyons wonders if distinct senses exist independently of the linguist’s mind: “It may well be that the whole notion of discrete lexical senses is ill-founded; and, if it is, there is no hope of defining lexemes on this basis.” (Lyons 1977:554). Brugman’s (1988) study of over is an example of an argument by assertion for polysemy. The representations she uses to represent each sense appear, intuitively, to give her analysis a degree of validity, although the sticky question of how she arrived at those particular schemas is not addressed in her work. If one can posit distinct image schemas (or draw differing representations) for two uses, is this sufficient evidence to argue that they belong to different senses? It hardly seems convincing. According to most analysts (e.g. Hampe 2005, Johnson 1987, Lakoff and Johnson 1999, Mandler 2004) there are several distinct types of image schemas, but it appears that a rigorous definition of any of them is difficult at this point. 12 Polysemy and the structure of English THROUGH Similarly, Herskovits (1986) seems to base her definition of polysemy on the existence of different spatial configurations to which a term may apply, although the question of whether untrained native speakers would consider them different is not addressed. This leads to a major problem, because her divisions of senses are based on real-world scenarios, rather than speakers’ understanding and categorization. Because language is about speakers’ conceptualization of the world, and not directly about the real world, it is clear that the study of language must take into account speakers’ cognitive organization as an intermediary level between the real world and language. Note that I fully accept the usefulness of collecting data on real-world situations for which a term is used, as this anchors two ends of the triangle of world, thought, and language to an observable and manipulable reality (i.e. both the linguistic and the real-world situations are observable and manipulable). However, there is a difference between studying speakers’ conceptualizations of real-world situations, on the one hand, and, on the other, studying the real-world situations themselves, and claiming to understand cognition based on these situations. For example, the range of real-world situations to which English in can apply includes situations of “partial inclusion” (The cigarette is in his mouth) and those of “full inclusion” (The brain is housed in the skull) (Bowerman and Pederson 1992, ms). In languages such as Guugu Yimidhirr (Pama-Nyungan, Australia), the respective translational equivalent wawu cannot be used in situations of partial inclusion, and many languages have a term for total, but not partial inclusion (such as dentro in Spanish, Portuguese, and Italian, and even English inside). However, one would be mistaken to claim that this is evidence for different senses of in in English, and therefore that in is polysemous. Yet this is not entirely different in principle than the arguments for polysemy of Brugman and Herskovits, who claim that real-world situations to which a term may apply are evidence of polysemy. Likewise, Talmy (2000) gives no evidence for his claims of polysemy, other than the fact that intuitively different situations can be described by a lexeme. The question of how to recognize polysemy is one that should reverberate in the core of the field of lexical semantics, and yet it is largely ignored or taken for granted, and argument by assertion has been extremely common in cognitive linguistics. Lakoff asserts that “polysemy occurs when a single word has more than one meaning - and when those meanings are systematically related” (1987:316). Here, his definition fails in several ways. For example, he neglects to define meaning, and thus we are left with a nebulous understanding of polysemy – we have neither a successful operational definition – we do not know how to recognize polysemy based on this definition – nor a solid theoretical one, as this definition fails to give us the ability to clearly differentiate polysemy from other possible analyses, such as vagueness. For example, based on the following sentences, are we to understand the English verb eat as being polysemous? 4. 5. I ate all the pizza. I ate all the spaghetti. 13 I eat pizza and spaghetti very differently; to eat pizza, I often roll up my sleeves and use my hands, whereas with spaghetti, I use a spoon and fork to twirl the pasta into a compact mass before putting it in my mouth with the fork. Here, the real-world events to which I’m referring are different, although there are many identical aspects (the use of the interior of my mouth, tasting, chewing, swallowing, getting nutrition through the process of digestion, and most, but not all, of the social, emotional, psychological associations with eating food, etc.). There is no way to know, based on Lakoff’s definition, which aspects are relevant to the meaning of eat. Assuming that he would not fall victim to the “real world fallacy” when assessing the situation described above, we could surmise that Lakoff would say that eat is either monosemous or polysemous based on how the speaker conceptualizes the meaning of the term. This appears to be his actual position on the issue as evinced by his later statement that “Polysemy involves cognitive organization in a lexicon” (1987:334). Any claim about the cognitive organization underlying the use of a lexeme, however, would require substantial evidence, and there is no discussion in Lakoff (1987)6 of what kinds of evidence would be appropriate or sufficient to support such a claim. The only argument he provides for his claim of a polysemous structure of English over (1987:416-461) is a case of argument by assertion, revealing his lack of an operational definition of polysemy. The point here is that Lakoff fails to explicate precisely how one can know which aspects of meaning are relevant to polysemy, and therefore how researchers can isolate or define such situations, as opposed to situations of vagueness, for example. His lack of precise definition of ‘systematically related’ raises similar problems. The inclusion of the term related is necessary to prevent a situation of considering words such as bank (“a financial institution”) and bank (“the side of a river”) to be (synchronically) polysemous, and yet once again, an empirical definition of such a concept is difficult, and most linguists ignore this issue. While Lakoff’s definition of polysemy has been shown to lack rigor, it is not my intention to imply that Lakoff is singularly irresponsible with respect to these matters. Rather, the entire field suffers from similar problems, and I have chosen to discuss the work of a leader in the field for expository purposes. As a whole, lexical semantics is prone to hand-waving where there should be empiricism, and this is but one example of a prevalent phenomenon. Careful study leads one to the conclusion that, in addition to Lakoff, researchers such as Brugman (1981, 1988), Talmy (2000), Evans (2009), and others who lead the field lack a satisfactory definition of polysemy. Taylor, for example, gives a definition extremely similar to that of Lakoff: “(polysemous words) have a range of distinct, through related senses…” (2002:98). Related to the lack of a definition of polysemy, there are other methodological inadequacies that commonly endanger analysis of polysemy. For instance, Geeraerts (1993) warns that the purported polysemy or indeterminacy of a word may reside in the polysemy or indeterminacy of the words used to define it. Murphy (2010) details the consequences of minimizing senses, which makes the job of 6 Or anywhere else in Lakoff’s work, to my knowledge. 14 Polysemy and the structure of English THROUGH processing more difficult, versus positing more senses, which makes the lexicon more complex but simplifies interpretation. Moving away from the lack of satisfactory definition, another common problem in research on polysemy is that most semanticists assume that each utterance of a polysemous word like over or run must ‘belong’ to (be motivated by) exactly one of the word’s senses. This fallacious assumption is discussed in more detail in Author (2007, in press), and is related to (though distinct from) the tricky question of which and what semantic variation is attributable to a given form, and how much semantic variation is attributable to semantic extension resulting from use in a given context. For instance, Evans (2009) argues that meaning is “not a property of words…(but) arises as a function of the way words…are deployed” (2009:22). However, he does accept that words do have stable lexical semantic representations associated with them, which he terms “lexical concepts”. It seems safe to assume that syntactic differences represent maximally entrenched, and therefore stable, meaning differences, and map the geography of meaning potential reliably, though at a crude level. Unlike Evans (2009), who frequently gives examples that he claims illustrate different senses of a word without explaining or arguing why he believes that they are different senses, and who does not address the question of how to tell when word uses constitute different senses, I will explicitly describe my methods that make use of both syntax and semantics for sense differentiation. This is insufficient, however, to deal with a particularly problematic issue that Langacker, Evans and others have raised, which is the problem of ascertaining how much and which part of the meaning of a use of a (what appears to be polysemous) lexeme comes from the lexical semantics vs. the syntactic construction in which the form is included vs. the larger context. By requiring syntactic differentiability of primary senses, I can address this issue by simply noting generalizations in the syntactic and semantic environments in which a lexeme can appear. Whether the distinct meaning comes from the lexeme itself or the syntactic construction in which it is embedded (or partially from each) will be investigated by analysis of the constructional semantics, and observing if these semantics account for the meaning difference between two uses of through. I will assume that it is the lexical semantics of through (with, possibly, some coercion – see e.g. Suttle and Goldberg 2011 and other papers in the same issue of Language) that permits it to be used (or prevents it from being used) in particular constructions, and therefore I study the constructions in which it can appear and the semantics it contributes to the larger utterance in each case. Thus, I will be aware of the contribution of the larger construction to the semantics of the utterance (e.g. Croft 2001, Goldberg 1995, 2006), but I will filter out this contribution in search of the semantics common to the use of through in various constructions. What, then, of the larger context, and its contribution to the meaning of the utterance (e.g. Taylor 2012:220ff.)? Here, I will adopt Langacker’s approach to the flexibility of word meaning in context. His discussion of the contextual modification of word meaning, which he terms “accommodation” (1987:76), was later employed successfully by cognitive linguists such as Nerlich (2003, Nerlich et al 2003). While 15 I generally embrace their vision of the flexibility of polysemy, I will focus my efforts on describing entrenched meanings in a quest for an empirical definition in my treatment of polysemy. The final problem in research on polysemy I will address is the difference between synchronic and diachronic polysemy. This is discussed immediately below. 2.2.5. Polysemy: synchronic or diachronic? Is polysemy a synchronic or a diachronic phenomenon? That is to say, is it a question of the motivated extension of meaning via processes such as metaphor and metonymy over time, which may be quite irrelevant to speakers synchronically, or is it a synchronic question of speakers’ conscious belief in relatedness of senses, or should it be based on their unconscious treatment of senses? It is essential to address this question, as it has been argued that many analyses of polysemy reflect the historical extension of the senses rather than a synchronic status: “In the last analysis, the radial model may be little more than a hypothesis concerning the plausible historical developments that resulted in the array of conventionalized uses in the modern language.” (Taylor 2002:479). There are different answers to this question depending on how polysemy is defined, including the researcher’s method of differentiating senses. Over its history, cognitive linguistics has tended to slowly become more and more concerned with methodological considerations, particularly motivated by severe but appropriate chastising such as that by Sandra and Rice (1995). They point out that a large portion of the debate in cognitive linguistics over polysemy arises from methodological inexplicitness, and particularly the lack of transparent and logical criteria for differentiating between uses. They describe research suggesting that there may not always be a central sense of a polysemous item and questioning the view that polysemous networks are strictly synchronic. This is an empirical question, for which we are still lacking evidence, but it should become easier to answer as experimental techniques become more sophisticated and more experimental evidence becomes available. Even if experimental evidence such as grouping and similarity studies show that speakers treat two putative senses of a word differently, however, the danger that this is due to pragmatic factors, rather than the structure of the linguistic item in their minds (Gibbs and Matlock 1999), so the ‘magic bullet’ of experimental evidence may be somewhat further away from providing a definitive answer than many cognitive linguists imagine (though empirical work such as that in Bietel et al. 1997, Cuyckens et al. 1997, Gibbs & Matlock 2001 and Sandra and Rice 1995 shows the potential of this route of investigation). It seems very likely that many semantic networks posited by linguists represent diachronic relationships between senses, but the extent of the synchronic relationships between senses, based on the methodology typically employed, is more dubious, as noted by multiple linguists. Vandeloise (1990) and Ruhl (1989), cautioning against the unconstrained positing of senses for polysemous lexical items, take monosemist positions. Tuggy (1993), Geeraerts (1993), Croft (1998), Sandra (1998) and others also scold cognitive linguists for the intuition-based methodology typically used in polysemy studies, by arguing, for instance, that it may not always 16 Polysemy and the structure of English THROUGH be possible to determine how many senses a word has, that traditional semantic tests for polysemy are not reliable (as discussed above), and that intuition can not serve to differentiate between various linguistic theories of polysemy. One basic question raised is whether language users are actually differentiating the posited senses (or whether the analysis makes sense to linguists, but fails to accurately describe actual linguistic behavior). 2.2.6. Towards an empirical definition of polysemy: Blank’s contribution One apparent exception to the trend of methodological inadequacies in work on polysemy is seen in the work of Blank (2003), who defines polysemy as a lexeme with multiple lexicalized senses which are related by one of the seven synchronic semantic relations in his taxonomy. His taxonomy is described below. In order to begin, Blank (1999) created a corpus of more than 600 documented historical changes, the vast majority in Romance and Germanic, and described a taxonomy of the types of relationships attested. While the bias of his sample is unquestionably troubling, I will assume that at least what he has described can occur. Furthermore, in studies of polysemous items in other language families (e.g. Del Corro 1990, Kristoffersen 2001, Nightingale 1999), the same relationships between senses are described, suggesting that few, if any, other relationships exist. Based on this taxonomy of diachronic semantic change, Blank later (2003) argues for a taxonomy of synchronic relationships among senses of polysemous words. Because his work on synchronic polysemy is founded in the diachronic creation of polysemy, as is the case in natural language, Blank’s taxonomy is both principled and plausible, and I will employ it to constrain the types of possible motivated relationships between polysemous senses. Blank’s taxonomy of 11 types of lexical semantic change, and the resulting seven synchronic relationships between senses, are listed and exemplified below in Table 1. Table 1 – Blank’s taxonomy of sense relations (from Blank 2003:270-271) type of lexical semantic change resulting synchronic relation between senses 1. metaphor E mouse ‘small rodent’ > ‘computer device’ It afferrare ‘to grasp’ > ‘to understand’ L brevis ‘short (spatial)’ > ‘short (temporal)’ A. metaphoric polysemy E mouse ‘small rodent’, ‘computer device’ It afferrare ‘to grasp’, ‘to understand’ L brevis ‘short (spatial)’, ‘short(temporal)’ 2. co-hyponymous transfer ? ratt- ‘rat’ > F (reg.), It (reg.) ‘mouse’ Pt aborrecer ‘to annoy s.o.’ > ‘to bore s.o.’ B. co-hyponymous polysemy F (reg.) rat, It (reg.) rat, ratta, ratto ‘rat’, ‘mouse’ Pt aborrecer ‘to annoy s.o.’, ‘to bore s.o.’ 3. semantic extension MF pigeon ‘pigeon raised for eating’ > ‘any kind of pigeon’ Sp tener ‘to hold’ > ‘to have’ C. taxonomic polysemy F pigeon ‘pigeon raised for eating’, ‘any kind of pigeon’ Sp tener ‘to hold’, ‘to have’ 17 type of lexical semantic change resulting synchronic relation between senses 4. semantic restriction VulgL homo ‘human being’ > ‘man’ F homme, It uomo, Sp hombre etc. ‘human being’, ‘man’ F blé ‘corn’, ‘wheat’ F blé ‘corn’ > ‘wheat’ 5. lexical ellipsis (absorption) a) absorption into the determinatum Sp coche ‘coach’ > ‘car’ (< coche automóvil) G Schirm ‘shelter’ > ‘umbrella’ (< Regenschirm) Sp coche ‘coach’, ‘car’ G Schirm ‘shelter’, ‘umbrella’ b) absorption into the determinans F diligence ‘velocity’ > ‘stage-coach’ (< carosse de diligence) G Weizen ‘wheat’ > ‘beer made of wheat’ (< Weizenbier) D. metonymic polysemy F diligence ‘velocity’, ‘stage-coach’ G (der) Weizen ‘wheat’, (das) Weizen ‘beer made of wheat’ 6. metonymy L lingua ‘tongue’ > ‘language’ L defendere ‘to defend’ > F défendre ‘to forbid’ G während ‘while (temp.)’ > ‘whereas (advers.) L lingua ‘tongue’, ‘language’ F défendre ‘to defend’, ‘to forbid’ G während ‘while (temp.)’, ‘whereas (advers.)’ 7. popular etymology F forain ‘non-resident’ > ‘belonging to the fair’ F forain ‘non-resident’, ‘belonging to the fair’ (< foire) Lat somnium ‘dream’ > Sp ‘sleep’ (< somnus) Sp sueño ‘dream’, ‘sleep’ 8. auto-converse change It noleggiare ‘to lend’ > ‘to borrow’ L hospes ‘host’ > ‘guest’ E. auto-converse polysemy It noleggiare ‘to lend’, ‘to borrow’ F hôte, It ospite, Cat hoste, ‘host’, ‘guest’ 9. antiphrasis F villa ‘country house’ > F (argot) ‘prison’ It brava donna ‘honorable lady’ > It (gergo) ‘prostitute’ F. antiphrastic polysemy F villa ‘country house’, F (argot) ‘prison’ It brava donna ‘honorable lady’, It (gergo) ‘prostitute’ 10. auto-antonymy E bad ‘not good’ > E (slang) ‘excellent’ Sard. masetu ‘gentle’ > ‘irascible’ G. auto-antonymic polysemy E bad ‘not good’, (slang) ‘excellent’ Sard. masetu ‘gentle’, ‘irascible’ 11. analogous semantic change F polir ‘to polish’, ‘to steal’, fourbir ‘to polish’ > ‘to steal’, nettoyer ‘to clean’ > ‘to steal’ etc. L levare ‘to lift up’, ‘to erect’, Sp alzar, It alzare ‘to lift up’ > ‘to erect’ all relations possible, e.g.: metaphoric polysemy F fourbir ‘to polish’, ‘to steal’ metonymic polysemy Sp alzar, It alzare ‘to lift up’, ‘to erect’ 18 Polysemy and the structure of English THROUGH At this point, a few comments on the taxonomy are in order. In the first type of semantic change, a metaphoric change leads to the creation of multiple senses that are then related synchronically by metaphor. In the second, co-hyponymous transfer, similarity within one conceptual domain or folk taxonomy is the basis for the change, and it gives rise to a synchronic situation of co-hyponymy as well. The third and fourth types are commonly called ‘widening’ and ‘narrowing’ respectively, and are also based on the similarity of concepts within a single domain. The fifth type, what Blank alternatively calls ‘lexical ellipsis’ and ‘absorption’, has two sub-types, defined based on what part of a complex word ‘absorbs’ the sense of the complex word. These sub-types do not form a one-to-one relationship with synchronic relationships of polysemous items, as seen in the table above, as absorption into the determinatum (or head) results in a synchronic situation of taxonomic polysemy, whereas absorption into the determinans (or modifier) results in a synchronic situation that is motivated by metonymic polysemy. Predictably, diachronic metonymy (the sixth type) results in synchronic metonymy as well, as Blank claims that both are based on conceptual contiguity. The seventh type of diachronic change, which Blank calls ‘popular etymology’, is seen in to result in synchronic metonymy, though Blank notes that both conceptual contiguity and formal similarity are motivating factors in such cases. It does not seem clear that such cases must be limited to resulting in a synchronic situation of metonymy, since popular etymology should, in principal, be unconstrained. The eighth type, auto-converse change, relates to the reciprocal relationship of participants in a frame, and results in an atypical case of contiguity that nevertheless can be defined as a type of metonymy. The ninth and tenth types involve oppositeness as well, with antiphrasis being an underlying association on the connotational level, and auto-antonymy being an underlying association at the ‘core content’ level (Blank 2003:269). Each results in a corresponding synchronic situation of polysemy. Finally, the eleventh type of semantic change is based on analogy with some already-existing polysemy of any kind (metaphoric, metonymic, etc.), and therefore may result in any type of synchronic polysemy. While this approach is a laudable attempt to avoid the type of problem we have discussed, it ultimately fails to define polysemy, as it doesn’t differentiate entrenched and novel uses. In other words, following this approach, if we encountered the sentence The ham sandwich in the corner just spilled beer all over himself (Lakoff and Johnson 1980), we would be forced to claim that ham sandwich is polysemous, as there is a metonymic relationship between the ‘default’ use of this term (referring to a food) and the use in this data (referring to a person). If we determine that any metonymic or metaphoric (etc.) use of a lexeme reveals polysemy, our analysis is essentially unconstrained. Essentially, Blank solves the problem of defining ‘related’ senses, but the problem of how we can tell if the posited senses are established or entrenched in the minds of language users is not addressed. Therefore, if we see Blank’s contribution as a taxonomy of the types of relationships that hold between the senses of a polysemous item, we then need to additionally provide evidence that speakers consider these senses distinct. I will provide a definition that assures us that speakers distinguish the various senses in the following pages. 19 2.2.7. My definition of polysemy At this point, I will discuss the definition of polysemy I adopt for this research, beginning immediately below with my theoretical definition before moving to my operational definition. 2.2.7.1. My theoretical definition of polysemy Within cognitive linguistics, Langacker (1991b) successfully provides a theoretical definition of polysemy. He refers to a situation in which there are multiple uses of a single term for which a single abstract meaning accounting for all the uses, and only the possible uses, of the given term cannot be found (1991:55). According to Ravin and Leacock (2000), this definition can be traced back all the way to Aristotle. I will generally embrace this theoretical definition. However, I will state that I will assume that the multiple paraphrases necessary to define the item must be related in order to speak of a single term, rather than multiple items which happen to be phonologically identical. As an example, consider the uses of through in the following sentences (with my analysis of the primary sense, or sub-sense, given in parentheses). (1): (NB): (EB): (ALL): (1 with time): (‘via’ sub-sense of 1): (ALL): The ball went through the window. The fish swam through the water. When will you be through? Mold spores sprung up all through the town. She doesn’t want to work Monday through Saturday. I got the tickets through a friend. The children got soaked through. Even attempting to come up with a single characterization of through that covers the first two of these examples is difficult. Such a characterization would necessarily be something like “movement within the boundaries defined by the Ground”. With this definition, the window in the first example is seen as the Ground through which the Path goes – the fact that the boundaries are crossed is irrelevant. The problem with such a definition is that it is overly general – it permits situations to be described by through that would never be so described in actual discourse. For example, someone who took a few steps into a two-mile- (three-kilometer-) long tunnel and immediately turned back could say, based on this characterization of through, “I walked through the tunnel.” I could move my hand up and down in the vertical plane of the (open) window, and say that my hand had moved through the window. This is not how the word is used, as shown by the fact that any native English speaker would view these evaluations of the events as inaccurate. Therefore, to claim that all of these uses can be encompassed by a single definition of through is clearly problematic Of course, to account for the third example, the element of motion must be discarded. As more uses are considered, any single definition must become more general, to the point where it becomes ridiculous to proffer that one is describing only the word through, and more and more uses not attested in speech are predicted. Therefore, based on this definition, it can be seen that through is polysemous. 20 Polysemy and the structure of English THROUGH 2.2.7.2. My operational definition of polysemy Even with an established theoretical definition of polysemy, the more difficult task of providing an operational definition remains. As discussed in Section 2 above, the semantic tests for polysemy in the literature all involve methodological problems that warrant their disclusion from this analysis, or at least prevent me from using them as salient evidence for my claims. Here, I will offer the following as my operational definition of polysemy: a construction with multiple, related primary senses. Of course, this looks extremely similar to the definitions criticized above, but I will take pains to explicate precisely what I mean by each of the terms used. To begin, by construction, I refer to a pairing of form and meaning, following Goldberg (1995, 2006) and Langacker (1987, 1999). This includes lexemes and other form-meaning pairings (e.g. morphemes and argument structure constructions). This leaves the problems of differentiating senses (or, in this case, defining primary senses) and identifying whether they are related – the same problems found in Lakoff (1987) and elsewhere, described above. For me, the clearest and most reliable way to differentiate senses is based on syntax, which is one easily observable way that native speakers behaviorally differentiate senses of lexical items, and hence I focus on primary senses here. I avoid referring to polysemy as “the association of two or more related meanings with a single phonological form” (Taylor 2003b:32, see also Cuyckens and Zawada 2001:ix), in order to remove the ambiguity of the term meaning and replace it with primary sense, which I will define as “a sub-section of the full meaning potential of a construction that is sufficiently strongly entrenched (i.e. stored in semantic memory, e.g. Croft & Cruse 2004, Evans 2005, Langacker 1999) such that it is reflected in syntax”. The use of syntax in definitions of polysemy is almost unknown (but see Author 2003, where I include a definition similar to the present one, and Evans 2006, who uses syntax as a criterion for sense differentiation, albeit an optional one. Additionally, most work by Gries and Stefanowitsch (as well as work in their tradition) uses syntax to isolate semantic tendencies when differentiating senses, but it does not appear to be part of their definition of polysemy or a necessary requirement in differentiation, e.g. Gries 2006). Because for this paper I have focused on a synchronic description of through, the use of syntax as a defining factor is essential, as we can be certain that speakers’ (synchronic) behavior differentiates senses, solving the problem of how to tell if posited senses are actually relevant to speakers’ conceptualization and behavior. This methodology explicitly identifies conventional senses, not meanings derived from context, as stated above. However, two senses having divergent syntax doesn’t mean that they are necessarily considered related by speakers. Therefore, this description relies on the kinds of connections between senses for evidence of relatedness. If we can find some general, well-supported cognitive principle of meaning extension applied to relate one sense to another, we can determine that speakers are much more likely to consider the senses related, and that their relationship will be motivated. If, on the other hand, the senses of a lexeme, as determined by syntactic and semantic differences, can not be described as being related in a straightforward manner, we can consider this evidence that speakers are probably not treating the lexeme as 21 polysemous, but homonymous. Therefore, in order to define related, I will require there to be a ‘natural’ or motivated relationship between the (primary) senses, based on Blank’s taxonomy. By employing Blank’s taxonomy of synchronic relations between senses of polysemous items in order to define what I will consider a motivated relationship between primary senses of a word, in combination with my reliance on syntactic factors to differentiate primary senses, I believe that I have created an operational definition of polysemy that is principled, replicable, and reliable. Significantly, the linguist need not wonder if speakers actually differentiate the various primary senses of a polysemous item or if the results are an artifact of the linguist’s analysis. In this way, I will claim to have solved the problem of an empirical operational definition of polysemy, but more explication of these topics is needed. I will discuss my use of syntax to identify polysemy in more detail immediately below. 2.2.8. Why use syntax to diagnose polysemy? I agree with Sandra and Rice (1995) that the real issues in studies of polysemy are whether native speakers, rather than linguists, treat the different senses of a polysemous item differently, and whether they consider there to be a connection between them. Without some kind of evidence that speakers treat the purported senses of a lexeme differently, it is not clear if the analysis mirrors the language user’s mind, or that of the linguist, as Sandra and Rice point out. In order to create an operational definition of polysemy that is reliable, and reflects the way that native speakers treat the different senses of a lexeme or morpheme, I have used syntax to identify when senses are different. As an example, the English word cup can be seen to be polysemous based on the following data: 6a. 6b. I only drink from my favorite cup. You’d better cup your hands over your ears to keep the sound out. Because cup is a noun in the first example above and a verb in the second, and because the meanings are related via metaphor (based on a resemblance of shape), we can define cup as polysemous. This definition permits polysemy within a word class, it should be noted, if there is some syntactic difference between (primary) senses of a word. Without such a syntactic difference, data such as those below will be considered instances of vagueness. 7a. 7b. I need a cup of sugar. I need a cup of coffee In (7a), the referent of cup is the amount measured by a standard measuring cup (250 ml., or 8 oz.), whereas in (7b), there is less precision – any reasonably sized cup will suffice. Because these usage-types of cup are not differentiated in the syntax of English, they cannot be considered different (i.e. as representative of multiple primary senses of cup) based on my definition. 22 Polysemy and the structure of English THROUGH In this research, I have adopted what must certainly be seen as an extremely conservative view of what constitutes evidence for polysemy. I believe that the use of the strict condition of syntactic differentiability is necessary to ensure the methodological validity surrounding claims of polysemy. This definition works because it is grounded in syntax, and the methodology is replicable. It also differentiates polysemy from the neighboring categories of vagueness and homonymy, as the first shows no syntactic differentiation between senses, and the latter lacks motivated relationships. 2.2.9. Potential criticisms of my definitions The major potential drawback to the use of this operational definition is that is insensitive to semantic shifts which have already occurred in the minds of speakers, but which have not yet resulted in differentiated syntax. Semantic changes occur historically prior to syntactic ones; that is to say that the semantics of a morpheme, lexeme, or other construction evolves first, and only then is the related syntax altered to reflect the new conceptualization of the morpheme or lexeme in question. For instance, the history of the Modern English progressive construction can be seen to have evolved through a pivot or bridge construction, which originally had a single interpretation as a type of locative expression, and later changed to approximate the aspectual meaning of imperfective. Following this shift, the grammar of the construction changed, presumably to reflect speakers’ newfound understanding. Thus, what was originally a locative construction, exemplified as 8. He is at hunting of bear.7 became reinterpreted as an imperfective, resulting in major changes in diachronic syntax: the deletion of the locational preposition at, as well as the partitive of, in Modern English (Traugott 1972). Before the syntactic adjustment, however, speakers must have already reanalyzed the construction, or else the deletion of at would not have made sense. Therefore, we can assume that the analysis presented here reflects better–defined, longer–standing, and more entrenched contrasts between senses than would an analysis without a syntactic basis. This definition will not permit the researcher to discriminate a recently developed sense, for which the syntax does not yet reflect speakers’ cognitive organization. In the above example, the original locational construction was first reinterpreted, and only then did speakers have reason to change the syntax in order to better reflect their new understanding of the semantics of the construction. To contrast with this inability to pick up recently-developed categories of meaning, the benefits of using such a methodology are many: it relates the concept of polysemy to syntax, which allows the researcher to test for polysemy and obtain empirically reliable results. In doing so, it leaves open the possibility of underrepresentation of the number of entrenched senses, but ensures no 7 Modern English spelling has been used for convenience. 23 overrepresentation. At the least, it identifies a secure subset of all senses that are behaviorally distinguished by speakers. Another problem with my definition, as well as all others, is the inescapable leap of faith necessary to accept that, given my best efforts, the theoretical and operational definitions are identifying precisely the same phenomenon. However, I choose to employ these definitions, despite their potential problems, as I see the first situation a necessary sacrifice of sensitivity for the purpose of approaching semantics as empirically as possible, and the second, which I have attempted to address to the greatest extent possible in the creation of the definitions, is nonetheless inevitable. Note that analyses of polysemy are always open to various alternative hypotheses, particularly that of vagueness. Vagueness could look like polysemy because of pragmatics, and critics could protest that there is no such conceptual system as that claimed by analyses of polysemy in speakers’ minds. In their view, there is one general meaning, differentiated by pragmatics. In the view of these critics of polysemy, what some would call polysemous forms are simply distinct lexical items that happen to share a phonological form. However, as I mentioned in the introduction, this critique of polysemy has been criticized for not taking into account the systematicity of the relationships between forms. Essentially, such a perspective fails to deal with the difficulties of research on polysemy, and merely pushes them into another realm (that of pragmatics). 2.3. Previous studies of through Through is not given more than the most peripheral treatment in most of the literature that I have been able to find, with the exception of Author (2003, 2007), Martín (2000), Evans and Tyler (2004), and, less directly, in Lakoff (1987), Lakoff and Johnson (1999), and Talmy (1983, 2000). What has been claimed is summarized below: Through is ‘neutral’ with respect to the shape of the Ground and the shape of the Path (see Talmy 1983:39): Shape of Ground: The ant walked through the circle/ square/ triangle I had drawn. Shape of Path: The ant weaved/ zig–zagged/ spiraled/ arced through the tunnel. Through is also neutral with respect to the make–up of the Ground (Talmy, 1983:39): continuous mass: aggregate: liquid: The meteorite fell through the roof. Randy ran through the forest, singing. Carol swam backstroke through the lake. Lakoff (1987:441) notes that through is unspecified as to whether the Figure is a multiplex or a mass: multiplex: The fans poured through the gate. 24 Polysemy and the structure of English THROUGH mass: He poured the juice through the sieve. Lakoff also notes that through permits situations in which the Path as multiplex is a sequence of points and the Path as mass is a one–dimensional trajector: sequence of points: I coughed through the night. one–dimensional trajector: I slept through the night. Some uses of through can refer to a Figure that is a moving trajector, one that is a static trajectory (Talmy’s ‘fictive motion’, 2000), or a location at the end of the Path (Talmy’s ‘localizing Figure by means of Figure–encountering Path’, 2000), as in: moving trajector: static trajectory: location at end of Path: Ronald skipped through the tunnel. The road goes through the tunnel. The room is through that hallway. If the Ground is an aggregate, this works as well: moving trajector: static trajectory: location at end of Path: Wendy hiked carefully through the forest. The road goes through the forest. The lake is through the forest. This also works with a Ground with collapsed boundaries: moving trajector: static trajectory: location at end of Path: Jordan ran through the door. The path ran through the door. The office is through that door. and with an aggregate medium: moving trajector: static trajectory: location at end of Path: Harry went through the wall of trees. The road ran through the wall of trees. The kitchen is through those curtains. Importantly, because Talmy and Lakoff do not differentiate the four different primary senses of through, some of the above claims are valid for some but not all the primary senses of through. However, to these I will add: through can be unspecified as to contact between Figure and Ground. contact: no contact: The snake slithered through the tube. The bee flew through the tube. In the tradition of Brugman (1981) and Lakoff (1987), Martín (2000) analyzes the semantics of through, using the “full specification method” (extreme splitting) in 25 which all semantic differences are taken to represent sense differences. Importantly, some of her data casts a doubt on her research methods and conclusions.8 For instance, after stating that “the first example/s presented within each category is considered to be the most basic one” (2000:20), for the first two examples of what she calls “the most basic schema for the entire category through” (2000:23; emphasis in original), she provides the following, without further elucidation9 (her (26-27)): 9. 10. The ball came through a bit quickish. (sic) The ball screamed through away on a very high orbit through into the crowd. (sic) Neither example seems plausible, with the second being particularly egregious. Since her data is derived from “a computerized corpus of spoken English” (2000:12; she cites Svartvik and Quirk, 1980), it should be possible to check the context from which it was drawn, but neither of these examples was found by this author in searches of the London-Lund corpus of spoken English. Furthermore, her argument is an instance of argument by assertion, and it is unclear on what basis she creates “tentative image schemas”, including for the two examples given above. In any case, her focus is on semantic, not syntactic, analysis, and her theoretical stance that polysemy can somehow be accounted for simply by structuring lexical items into semantic networks (2000:37) is sharply at odds with the position taken here. Evans and Tyler (2004) treat through directly, but primarily in the context of their argument that ‘prepositions of movement’ such as English through and to do not, by themselves, refer to motion as part of their lexically specified meaning. In Author (2003, 2007), I show that the primary senses 1, EB, and ALL need not involve motion, but that this is not the case with NB (this is also discussed below). As in the briefer treatments of through in Lakoff (1987) and Lakoff and Johnson (1999), Evans and Tyler make no distinction between the four primary senses, discussing through as if it were a unitary entity (which looks identical to Through1). We will see below that there are clearly four distinct but related and sometimes overlapping topographies of meaning sharing a single name. Talmy (1983:14, p.c.) claims that there are two senses of through; one in which the Ground is a medium (‘anything from aggregate to continuous mass’) and another in which the Ground is a linear enclosure (e.g. a cylindrical form), though he cites no evidence. As I have analyzed it, two primary senses (NB and 1) appear to correspond to the two that Talmy mentions. I will be systematic in describing and providing evidence for the existence of each primary sense, as well as describing and motivating sub–senses arising from each use. 8 I only became aware of this paper recently, and therefore my previous work does not make reference to it. 9 Other than to note that the landmark (Langacker 1987; cf. Talmy 2000’s Ground) is unspecified. 26 Polysemy and the structure of English THROUGH Finally, I will mention that, in Lakoff and Johnson’s (1999) discussion of the Source-Path-Goal schema, it is mentioned that through and other spatial relations terms (they cite toward, away, and along) are defined using this schema. They say that the Source-Path-Goal schema characterizes our most fundamental knowledge of motion (1999:34). I will describe the central sense of Through1 as being motivated by this schema in combination with the Container schema applied to a salient Ground. As a complex10 spatial relations term invoking the notion of Path, this primary sense of through is amenable to motion, though it doesn’t require motion. Because it is claimed to be fundamentally based on in, the relevance of the Container schema is predicted. However, the four primary senses of through make use of the Source-Path-Goal schema to differing extents. III. The primary senses of through Here, I will describe the primary senses of through, including their semantic characterization and the syntactic and semantic evidence distinguishing them. I will begin with what I will call the central sense of Through1. 3.1. Through1 In a prototypical use of Through1, a Path crosses boundaries defined by the Ground. The Path traverses the center of the Ground (i.e. with the same amount of Ground on each side / all sides). By crossing the first boundary, the Figure is brought into an ‘in’ relationship with the Ground. By crossing the second boundary, the Figure is brought into an ‘out’ relationship with the Ground. The boundaries crossed are opposite one another. Through1 defines the Source-Path-Goal and Container image schemas maximally relating (the Path and the Container relate most directly). Moreover, this relationship must be defined in a functional way. A functional relationship is defined by i) object, object interaction, and situation knowledge and ii) knowledge of potential and desirable interaction, including force dynamics (or what Coventry and Garrod 2004 call ‘object knowledge’ and ‘dynamic-kinematic routines’ respectively; I follow their classification in constructing the definition in this way). This is to say that what it means for the particular Path and Container to relate maximally is based on what is relevant to real world activities of language users, or “makes sense”. I will define ‘opposite’ in functional terms as well. Therefore, if we attend an art exhibition staged in the entirety of an L-shaped building, a default understanding of an unmarked utterance such as we already went through, so we’ll meet you at the car may construe the boundaries as the doors on perpendicular facades of the building. In fact, if one were to take a direct path from the entrance to the parallel and opposite door, or exit at the ‘bottom of the L’, one could be accused of having ‘only gone through half of it’. I walked through the maze in 11 minutes is another case showing the importance of defining oppositeness functionally, rather than geometrically, as the 10 See my dissertation (2007:13) for a description of what I call “basic” and “complex” spatial relations terms in English. 27 functionally opposite exit may be immediately adjacent to the entrance. Additionally, to be through an exam means to have passed it, rather than just to have finished it, according to the OED. This is motivated because not passing could possibly mean having to take the test again, so finishing but failing doesn’t lead to the functionally opposite boundary. Finally, if someone says ‘did you get the knife through that salami?’, they are probably not referring to stabbing it lengthwise, or the way one would go through a tunnel. These examples show that knowledge of functionality plays no small role in determining what is to be considered opposite. A representation of Through1 is presented below. Figure 2 – Schematized representation of Through1 In Figure 2, the arrow represents the Figure’s Path, and the rectangle, the Ground. It can be seen that the Figure was initially outside the Ground, then crossed a boundary to enter into an ‘in’ relationship with the Ground, finally crossing an opposite boundary, thereby terminating the ‘in’ relationship. 3.1.1. Through1: prototypical uses Fairly prototypical uses of Through1 can be seen in the following examples: 11. drink any water straight through the straw (BNC) 12. The older sister, Manyara, left her village and went all alone through the forest to the city, all by herself. (BNC) 13. The marchers did not walk through the Tunnel. (BNC) 14. They replaced the curtains with metallic finish Venetian blinds, punched through with tiny holes to let the light filter through. (BNC) I consider this to be the prototypical usage-type of Through1 because it is the least marked of the usage-types of this primary sense. I am speaking in very general terms, and ignoring certain elements that deviate from a prototypal scenario, such as the negative in the penultimate example above, or the abstract nature of the Figure in the last example.11 Note that while it is not necessary that the Ground completely enclose the area surrounding the Path, it appears that the Ground is typically conceptualized as 11 I am also ignoring the fact that it is entirely conceivable that the speaker of (15), could intend it to be an instance of ThroughNB. My claim of prototypicality assumes that the language user conceptualizes a particular geometric configuration in which the forest has a salient entry and exit, ideally on opposite sides. 28 Polysemy and the structure of English THROUGH enclosing the area surrounding the Path. Thus, even a prototypically 2D Ground such as a square can be used as if it were 3D, as in the example below. 15. The car slowed only slightly as Julie drove through a deserted Berkeley square. (BNC) Author (2003) discusses the results of a study showing that through is selected by subjects relatively more often in a forced-choice paradigm (with across and over as the other choices) when an arrow representing Path is represented as entering and emerging from the Ground. Subjects who saw the stimulus in Figure (3a) were less likely to choose through than those who saw Figure (3b). Figure (3a) Figure (3b) The results are presented in Table 2. Table 2 – Conceptualization of containment/support (Fig. 3a) Arrow above grass (Fig. 3b) Arrow within grass ‘through’ responses 7 of 17 12 of 16 as percentage 41.2% 75% This difference is very nearly statistically significant (a t-test returns a p value of .0514, t=2.0266, df=31, standard error of difference 0.167), but misses the benchmark, which situations such as these help to remind us is arbitrary. Additionally, the greater the extent to which the Ground encloses the Path, the more subjects chose through in their responses to forced-choice questions (again, with across and over) when choosing whether the mouse in (4a-c) below wants to go through, across, or over the bamboo. 29 Figure (4a) Figure (4b) Figure (4c) This was predicted to be significant based on the assessment that the geometric speifications of through1 are argued to be based on those of in (Author 2003, 2007), and the concavity of the Ground has been shown to be relevant in speakers’ use of in (Feist 2000, which, along with Mark and Egenhofer 1994, inspired the experimental approach I am mentioning here). Subjects chose through most often with Figure (4a), less often with (4b), and least often with (4c), across all conditions. The conditions manipulated were the animacy and species of the Figure (mouse, spider, worm, coin, ball, dice; no effects were seen) and the name given to the Ground (bamboo, tube, wood; subjects disliked using through if the Ground in Figure 3 is referred to as wood, and preferred to use through when the Ground was referred to as bamboo or a tube). A summary of the results across all conditions is presented below in Table 3. Table 3 – Effects of geometry on Ground (Fig. 4a) High boundaries (Fig. 4b) Medium boundaries (Fig. 4c) Low boundaries ‘through’ responses 46 of 66 26 of 66 13 of 66 as percentage 69.7% 39.4% 19.7% This is the same type of categorization that allows the use of in with both two– and three–dimensional Grounds, such as in 22) The ball is in the circle / box. 30 Polysemy and the structure of English THROUGH 3.1.2. Collapsed boundaries In case the boundaries of the Ground are collapsed to form essentially a single plane, as with a door or window, Through1 can still be used. 16. Dashing his pint mug down on the bar, Greg Hocking darted through the door before any of the idly curious could bar his way. (BNC) 17. That's it now put your foot right through the window. (BNC) There is no grammatical or other evidence, however, that speakers treat this type of example as different in any way than the central sub–sense of 1. As such, it will be subsumed under the same primary sense for this analysis. It does diverge from the prototype in the sense that motion within the boundaries of the Ground is (nearly) instantaneous; with windows and doorways there may be no single instant during which an observer can say ‘the ball is part–way through but not all the way through yet.’ 3.1.3. Without motion Through1 can be used in cases without any actual motion: 18. fortunately she looked through the window and er found him (BNC) 19. Farrell saw the smashed bone sticking through skin and material. (BNC) 20. Her hair was tied in a bun and, like Topknot, it had a long grey feather stuck through it. (BNC) The first example shows virtual motion (Talmy’s ‘fictive motion’), as well as a collapsed Ground, as discussed above. The second and third examples, lacking motion entirely, make use of stick as verb (or past participle form of the verb, used as a modifier), with a stative, rather than eventive, reading. There is no evidence that this usage-type of through differs from the central meaning of 1, and so I will consider these data to reveal the broad range of real world events to which this single primary sense of through applies. 3.1.4. Metaphorical extension with target domain of time Through1 can be metaphorically extended to refer to an event taking place over a period of time, with the boundaries of the event as the boundaries of the Ground: 21. Then, half–way through the coffee, the gardener brought in Tobermory's blood-stained body. (BNC) 22. Half-way through the last act we'll call a doctor (BNC) 23. Half-way through the movie Ms Dash switches on the mental light bulb and asks her suitor…(BNC) 24. Injecting a different perspective of this kind part way through a project can also have its uses. (BNC) 25. Part way through the reconstructed Violin Concerto…(BNC) 26. his companion was half-way through a twelve-year stretch for a similar crime (BNC) 31 27. He doesn’t want to work Monday through Saturday. It should be noted that the use of Through1 exemplified in the final sentence, with an explicit mention of time words as Ground (as opposed to a Ground that can be construed as occurring over a time period, such as a movie, or even a twelve-year stretch, seen in the examples above), was non–existent in the BNC, and the example given above was considered unacceptable to the several British English speakers I asked. According to an informal survey of these speakers, this type of expression is rarely, if ever, used in British English, though it seems common in North American English (e.g. millions of hits are returned for Web queries for “Monday through Friday”). If the temporal use of Through1 were as basic as the spatial use, one would expect this type of use to exist uniformly in all dialects. The fact that it doesn’t suggests that the temporal use of 1 is an extension of the more basic, spatial use. I will return briefly to this usage type in Section 4 below. Importantly, searches of the BNC for the pattern N1 through N2 reveal that there is no other instance of a usage-type of through occurring in environment N1 through N2 in which the N1 is the entrance to the Ground, and N2 the exit. The most frequent results include passage through parliament, bill through parliament, and knife through butter. In most results, N1 refers to the Figure, (bill, knife) or the Path (passage). In a minority of results, N1 refers to the entire Ground (season through injury, game through injury) and the “via” usage-type of Through1 is employed (see Section 4 below for a description of this sub-sense). In no cases is N1 used to express the entrance to the Ground, and so we find (missing a) day through sickness/illness, but not *day through night, and death through drugs but not *birth through death. It appears that this usage-type does not exist in the BNC, and may be a (North) Americanism. However, the range of semantic categories to which N1 can apply is clearly a semantic distinction, not a syntactic one, and therefore I will not posit a distinct primary sense to account for this usage-type, instead relegating it the status of a sub-sense. In Section 4 below, I will describe this in more detail as an instance of the ‘list’ sub-sense of Through1. 3.1.5. Through1 as an adjective In the BNC, there are adjectival uses that are clearly expressing the semantics of Through1, such as 28. Chris Whyte and Gary Speed watched a through ball sail over their heads (BNC) 29. …it must be the best quality you can afford because hallways take a lot of through traffic. (BNC) 30. And where else on earth would a through train from Glasgow to Mallaig unhitch the guard's van… (BNC) In all of these cases, through is expressing the semantics described for Through1, with a Figure moving via a Path from one side of the Ground to its functional opposite. Therefore, we can say that Through1 permits expression both as a preposition and as an adjective. 32 Polysemy and the structure of English THROUGH 3.1.6. Through1 as a verb? There is a use of through, which should be considered extinct, as what is most likely a verb. It is limited to the specific context of ellipsis after an auxiliary verb, as in the OED example from 1906 We must pass, we must through this moment. Searches of the BNC reveal no such uses, and therefore I will consider this to be sufficiently rare such that it is essentially not a current phenomenon. However, it is worth noting that the semantics expressed by such examples coincides closely with the basic semantics of 1, in that a Path that exists within the Ground and crosses the final boundary is described, as will be detailed immediately below. 3.1.7. Through1 describing a Path that is in a Ground Based on what the above data reveal, Through1 can be seen to prototypically describe a Path that twice crosses the boundaries of a Ground with which it is temporarily in an in relationship, such that the two boundary crossings define the limits in space and time of the existence of the relevant in relationship. This describes the meaning of a complex spatial relationship in terms of a basic one (that of in) and several other fundamental primitives, such as: a Path, a Container, the crossing of boundaries, functional oppositeness, and a temporal span including a specific sequentiality of sub-events. A nearly identical definition can be offered very preliminarily for the central sense of the lexical alternative across, as it describes a Path crossing the boundaries of a Ground with which it is temporarily in an on relationship (c.f. Talmy 2000: II: 324). It may be that complex spatial terms are typically based on certain universal primitives, such as those mentioned above (a Path, a Container, temporal sequentiality, etc.), in combination with language–specific coding of concepts for (again, language–specific) basic spatial relations terms (in this case, in). While it is not my intention to do so at this time, I believe that investigation into the cross–linguistic applicability of this hypothesis warrants further research. See Author (2003, 2007, 2010b) for an argument that the range of spatial relationships to which through applies is fundamentally based on that of in. 3.1.8. Monosemy of Through1 As I have argued above that a monosemous account of through cannot account for all of its varied uses, an argument that Through1 is, indeed, monosemous, is in order. The methodology used above to claim that through is polysemous can be applied to the primary sense labeled Through1, with the result that a definition that links all the examples of Through1 given in this paper is not overly general, and therefore will not apply to words other than through or primary senses other than Through1. This type of argument will serve as the evidence for the monosemy of all primary senses of through claimed in this paper. 3.2. ThroughEB 3.2.1. Characterization of ThroughEB The first extension from Through1 to a distinct primary sense that I will discuss is a case of profiling (Langacker 1987) or metonymy, taking motion across two boundaries and focusing on the crossing of (or state of having crossed) the second, 33 criterial boundary. This use of through focuses on (the crossing of) the end boundary, and begins to function like a perfective marker. Through occurs as an adverbial particle and takes on the meaning ‘finished, in the state of having/being finished’. I will refer to this primary sense of through as EB (“end boundary”). With this primary sense, in what is clearly a case of metonymy or profiling, the crossing of the second boundary comes to be what people are talking about, and takes the label, through, which once only applied to the entire (Through1) event, including crossing two boundaries and moving within a medium. A schematic representation of EB is given in Figure 5 below. Figure 5 – Schematized representation of ThroughEB In Figure 5, the arrow represents the Figure’s Path, and the rectangle, the Ground. The scene begins with the Figure in an ‘in’ relationship with the Ground, and then it crosses a boundary to terminate the ‘in’ relationship. The boundary of the Ground is darkened in suggest its salience, and the arrow representing the Path is considerably shortened, as the only facet of the Path that is relevant is in the fact that it crosses the end boundary. By comparing the representation of Through1 in Figure 2 above with that of EB in Figure 5, it becomes clear that EB is profiling the end boundary that is relevant in, and part of the schema of, Through1. EB often co–occurs with a ‘be’ verb and in idiomatic expressions such as fall through, pull through, think it through, etc. It can be seen below, with examples that are clearly referring to the state of having crossed or crossing the end boundary, while the first boundary and motion in–between boundaries are not relevant: 31. I was through with thinking about Ardakkeans. (BNC) 32. When the plan fell through, he put them back into the larder till the next time. (BNC) 33. Finally, I was given an experimental drug and I managed to pull through. (BNC) 34. …he had no intention of going through with it. (BNC) 35. …buyer, who has thought this through before the supplier appears (BNC) A possible motivation for this extension is the fact that people often say things such as We’re almost Through1 and We’ll be Through1 in just a few minutes. In both of these situations, it is extremely salient that the passing of the end boundary is equivalent to achieving the state of being ‘through’. People may consciously experience the crossing of the end boundary, which means that they are (finally) ‘through’, whereas the crossing of the initial boundary means that they are in some location. This naturally leads to a focus on the end boundary as demarcating when one is finally ‘through’. 34 Polysemy and the structure of English THROUGH Note that the end boundary of EB is not always necessarily an end boundary that is defined by the natural or inherent limits of the Ground. In situations in which EB functions to express completion, as a perfective marker, the ceasing of the activity creates an end boundary, such as in: 36. Donna reckoned they should be through with all that kind of stupidity (BNC) 37. ' I 'm through with Harvey,' she said.' (BNC) 38. He’s through with the cherries (so you can have what’s left). Here, the total number of cherries or total potential amount of stupidity (or of “Harvey”) is irrelevant, and what’s relevant is the boundary created by the cessation of the action of eating, interacting with Harvey, or thinking in a “stupid” way. With the addition of EB to the posited senses of through, the following schema of the senses of through posited until this point become relevant: Figure 6 – Primary senses of through discussed so far EB 1 Oval = primary sense of through Arrow = extension to alternate primary sense 3.2.2. Evidence for status as a primary sense The fact that EB occurs in different syntactic constructions than Through1 (EB is a particle, can occur immediately preceding a preposition such as with) is sufficient to show that speakers treat the two differently. The metonymic relationship between the primary senses EB and 1 (discussed in more detail below) is included in Blank’s taxonomy, which qualifies it as a motivated relationship by our standards. Some additional semantic evidence is provided below. 3.2.3. Semantic evidence – minimal pairs Minimal pairs such as those below show the semantic differences between the two primary senses (which are given in parentheses before the examples here, and below throughout the paper, either in parentheses or as subscript, when appropriate): 39. (1) I’m through school. = I graduated. 40. (EB) I’m through with school. = I stopped, but we don’t know if I graduated. 41. (1) He’s through the cherries (and onto the peaches). =He ate all of them (or otherwise made use of them, such as by picking or canning them). 42. (EB) He’s through with the cherries (so you can have what’s left). = He stopped eating them, but some cherries may not be eaten. 35 43. (1) I’ve gone/I am through the book (but I may need to study it some more before the test). 44. (EB) I’m through with the book (so you can have it). 45. (EB) ?#*I’m through with the book, but I may need to study it some more before the test. In the first example of each group above, through is a preposition, while in the second, it is a particle. It is also significant that the implication that some initial boundary was crossed exists for each use of Through1, whereas with ThroughEB, this is not the case. (55), for example, is unacceptable or less acceptable if the subject started eating the cherries after someone else had already eaten many or most of them. (56), however, I find perfectly acceptable even if he only ate one or two cherries, and he started after others had had their fill. In fact, even if he had just been using the basket of cherries to impress someone or as a doorstop, this is a possible utterance, whereas the first sentence is not. The same holds for the examples in (57–59) in which, in (57), we can assume that the speaker started reading the book at some point in time. In (58), the speaker probably made use of the book, but it could have been as a bookend or to kill an insect, and the lack or presence of any starting boundary is completely irrelevant. In fact, the addition of ‘Actually, I never even started it.’ clearly reveals the absence of a Path within the Ground and the crossing of a starting boundary: 46. ?# I’ve gone/ ?# I am through the book. Actually, I never even started it. 54. (EB) I’m through with the book. Actually, I never even started it. The crossing of the initial boundary is implied with the use of Through1, despite potentially lacking relevance to the particular situation, whereas with EB, whether the initial boundary was crossed or not is not implied. Additionally, the Path in EB may simply cross a boundary (i.e. mark termination of a state or event), but it need not be seen as traversing the Ground. 3.2.4. ThroughEB and time EB often describes instantaneous situations (i.e. those without any process or duration), such as in all of the examples of EB given above. However, there is an infrequent usage-type in which EB is used with adverbials of duration, as seen below. 55. …wondering if you will have enough willpower to see this through for the whole 28 days (BNC) 49. I think it would be better if I worked through for the rest of today (BNC) While this usage-type appears to represent a very low fraction of all uses of EB, it forbids the simple assignment of clauses with through and adverbials of duration to another primary sense. While they are apparently absent from the BNC, intuition provides apparent cases of metaphorical extension of ThroughEB to time, such as the examples below: 36 Polysemy and the structure of English THROUGH 50. 51. We’re almost through with the president’s first term. We’re finally through the coldest months of the winter. However, the essential meaning of this primary sense of through is that an event of some type is terminated. Thus, it is likely impossible to differentiate between events defined by time periods, such as winter, and those defined by less obviously temporal abstractions, such as a human relationship. This is because EB selects a Ground that is durational (i.e. exists over a span of time, such as school, or requires a span of time to interact with or process, such as a book). 3.2.5. ThroughEB with the Present Tense Many of the above examples of EB contain verbs in the present tense. This is not the case for 1, which requires a particular context to appear with a simple present tense verb (i.e. not one marked for progressive aspect), such as what Talmy (1983:32) describes as the ‘freeze–frame’ talk of, for example, sports announcers, as in: 52. …he’s around the corner...now he’s through(1) the tunnel, and into the home stretch… Through1 can appear with a present tense verb when the element of motion is lacking: 53. Rebecca’s arm is through(1) the window. Through1 can also occur with a verb that is marked for the present tense, but is only interpretable in a habitual sense, as in: 54. 55. He doesn’t want to work Monday through(1) Saturday. I go through(1) the Lincoln Tunnel on my way to work. Thus, we have discussed four uses of through that take present tense verbs: Through1 with the ‘freeze–frame’ phenomenon, Through1 without motion, Through1 with a present tense verb construed as habitual, and, without restrictions, ThroughEB. 3.2.6. Monosemy of ThroughEB Again, the syntactic evidence differentiates the primary sense of EB from that of 1. The definition, including the syntactic behavior, of EB is sufficiently broad to capture its uses without applying to another primary sense of through, or any other morpheme. Therefore, I will consider EB to be monosemous. 3.3. ThroughNB 3.3.1. Characterization of ThroughNB With this primary sense, the boundaries of the Ground so integral to an understanding of 1 and EB are irrelevant, and only the Path within the medium of 37 the Ground is pertinent. Motion is essential, unlike with 1 and EB. The motivation for the existence of NB can be found in a metonymy in which the motion within the medium of the Ground is seen as sufficient to stand for the entire situation of through, even though the Path does not cross any boundaries. With ThroughNB, - the Ground is conceptualized as a medium, - the Figure moves within the medium, - there is no sense of boundaries being crossed, - there is no sense of completion. A schematic representation of ThroughNB is given below. Figure 7 – Schematized representation of ThroughNB Here, the Figure’s Path is not relevant, but the fact that the Figure is in motion while being in an in relationship with the Ground is critical (and is represented by the dotted line). Though I have represented the Path as ‘wandering’ slightly, there is no reason I couldn’t have drawn a spiral or circular Path (or a shorter Path, or only the arrow head, or one that repeatedly traverses the Ground, etc.) instead. 3.3.2. Prototypical uses of ThroughNB Some fairly prototypical examples of ThroughNB are presented below. 56. However, in many species, the larvae travel considerable distances through the body before settling in their final (predilection) site and this is the migratory form of life cycle. (BNC) 57. There was no struggle, like trying to walk through water; but she had a sense of floating, as if her body had become insubstantial, like an air bubble. (BNC) 58. River rafting is definitely the thing to try if you have the spirit of adventure coursing through your veins. (BNC) 59. The maximum speed through the tunnel is 75 mph. (BNC) 60. The train rumbled through the dark. (BNC) 61. … he treks off through the dubs of a drizzle… (BNC) 62. I screamed at him as I ran off through the woods… (BNC) Note that, unlike what we saw with Through1, the final two examples above involve an inchoative reading, due to the fact that motion within the Ground, but not the crossing of any boundaries, is relevant to the use of NB. Therefore, it can (and relatively often does) co-occur with inchoatives, whereas Through1 prototypically requires the crossing of a final boundary, making it less compatible with inchoatives (though examples such as I began to go through the tunnel every day do exist). 38 Polysemy and the structure of English THROUGH The primary senses of through that have been discussed are represented in Figure 8, presented below. Figure 8 – Primary senses of through discussed so far 1 EB NB Oval = primary sense of through Arrow = extension to alternate primary sense This primary sense of through can be shown to differ from 1 and EB with the following evidence. 3.3.3. Addition of time adverbials If, for the Ground, we choose the word crowd, which could be ambiguously interpreted as a spatial region with two boundaries or as a medium, the following form a minimal pair: 63. 64. (NB) He walked through the crowd for hours, singing. (1) He walked through the crowd in ten minutes, singing. Note that it is not possible to get a reading of 1 for the first example, or of NB for the second. Similarly, the following examples, which vary in the explicit presence or absence of a Goal and in the use of a durational (for-TIME) or an instantaneous (in-TIME) adverbial, highlight the difference between 1 and NB: 65. 66. 67. 68. (NB) He walked through the forest for hours, without a destination. (NB) ?#*He walked through the forest in two hours, without a destination. (1) ?#*He walked to the city through the forest for hours. (1) He walked to the city through the forest in two hours. Through1 can sometimes co-occur with a period of time, however, as in: 69. …. huge lorries are passing through what were envisaged as quiet residential neighbourhoods (BNC) 70. As I looked through the viewer I had the feeling, momentarily… (BNC) 39 This is motivated by the repetition or temporal extension (in the first and second examples above, respectively) of the situation described by 1. Unlike instances of NB that occur for durations of time, which are internally complex events with some type of consistent change necessary, such examples of 1 require either iterativity (as in (89), and in I went to work through that tunnel for ten years, every day of my life) or a lack of change in the situation described by through (as in (90), and in I was looking through the telescope for hours). 3.3.4. Addition of other adverbials Because with NB the boundaries are not relevant, only 1 can co-occur with adverbial clauses such as ‘all the way’: 71. 72. (1) – George walked all the way through the tunnel, singing to himself. (NB) – #George walked all the way through the streets, singing to himself. 3.3.5. With progressives Since 1 implies completion, it cannot easily appear in a clause with a progressive. NB, however, can (and often does): 73a. (NB) I was going through the tunnel when I heard the noise. 73b. (1) #I was going (all the way) through the tunnel when I heard the noise. The use of iterative or habitual aspect makes possible something apparently similar to the above examples, as in 74. This saves the chore of going through large media directories every time you want to contact the media. (BNC) Note that an event expressed in the simple past or present can be related to the event involving through even in such cases, as the end of the sentence shows. 3.3.6. With be NB cannot appear in a clause in which the main verb is any form of be: 75. 76. 77. 78. 79. (1) My arm was through the window. (without motion) (1) The only way into his protective case was through a little trap door he kept locked night and day. (BNC) (fictive motion) (1) He’s through the tunnel. (‘freeze–frame’) (EB) I am through with finals. (NB) *I am through the streets. This is because be is a stative verb, and NB requires an eventive reading. Thus, more than just being incompatible with the verb be, NB cannot describe scenarios lacking motion in general, as in 40 Polysemy and the structure of English THROUGH 80. 81. (NB) *The blood lay through the streets. (NB) *The soldiers stood/sat through the crowd.12 The fact that it cannot appear in clauses with be, or any other verb entailing a lack of motion, or in any stative clauses, effectively distinguishes NB from both 1 and EB. 3.3.7. Contrasting uses Compare the following examples: 82. 83. I went through(NB) the tunnel, thinking, but I couldn’t go all the way through(1). He staggered through(NB) the room clumsily, too drunk to make it through(1) before the bomb went off. Notice that the first use of through in each of the sentences above is an example of NB, while the second is an example of 1. The fact that the first through is affirmed while the second is negated shows that they are used differently. The crossed readings test, discussed above, is designed to reveal contrasts in two senses of a word. It results in a firm distinction being drawn between 1 and NB, as exemplified in the following: 84. 85. 86. #He walks through(NB) the rain, but not the door (1). He goes through(1) the door, but not the window (1). He will walk through(NB) the rain, but not the snow (NB). This semantic evidence is relevant, but will not be relied on for the purpose of sense differentiation. More evidence for the distinction of primary senses NB and EB will be presented below. 3.3.8. NB vs. EB The thrust of the evidence above distinguishes NB from 1, though one piece of evidence – that involve stativity and ‘be’, does successfully differentiate NB from EB as well. Here, I will focus on additional ways that English speakers distinguish NB from EB. First, and most fundamentally, there is the obvious fact that EB is a particle, and NB is a preposition. In itself, this is sufficient, but we will see that, in addition, there are several other syntactic discrepancies arising from their semantic differences. 3.3.8.1 Point in time With both 1 and EB, a point in time may be specified, whereas this is impossible for NB, which requires a time period with extended boundaries: 12 Note that, with the addition of all, these two examples become possible instances of ThroughALL, discussed below (e.g. The soldiers were stationed all through the crowd). 41 87. 88. #*I go through(NB) school on Monday.13 ?#I went through(NB) a lot of anguish at 7:03. The only possible way to interpret the second example is to increase the granularity with which the event in the adverbial clause is viewed, so that it takes extended boundaries rather than being a point in time. If we change the point in time to a time period, such as in 2012, the sentence is immediately completely normal. 3.3.8.2. With a present tense verb The use of a present tense verb appears to be fairly uncommon with NB. are given below: 89. 90. 91. Examples I go through(NB) work with a smile on my face, despite my true feelings. I go through(NB) a lot every day, so don’t give me any extra trouble. Our stress levels suddenly rise markedly when we go through(NB) a major life event such as perhaps a divorce, serious accident, or death in the family. (BNC) EB is often used with a present tense verb. Unlike the above examples of NB, no difficulty is implied, but termination is the essence of the meaning (and this termination is already completed). Two examples are given below: 92. 93. (EB) I’m through with school now, so I should look for a job. (EB) I am through with all of those papers, so I can rest now. Compare, as well, the following examples: 94. 95. (EB) I am through with school. (NB) I go through a lot in school every day. Notice that with the first sentence, the speaker is in the state of being finished with school, and in the second, the speaker is describing a habitual process. 3.3.8.3 With manner adverbs Because EB refers to a state, it cannot easily occur in a clause with a manner adverb such as happily, well, etc. NB, however, focuses on the process of motion through the medium, and so it can easily take a manner adverb, as seen in the following examples. 96. 97. 13 (EB) *I am through with my finals happily. (NB) I go through school happily every day. Note that, if the school in question only involves a few hours of time, and therefore matriculation and graduation could both occur on the same day, this sentence, with Through1, is entirely possible. 42 Polysemy and the structure of English THROUGH 98. (NB) The staff thought it better to let her go through her labour cheerfully. (BNC) 3.3.9. ThroughNB: metaphorical extension to time NB can be metaphorically extended to refer to time, as in: 99. … it provides much–needed living ground for animals through the winter. (BNC) 100. Can't say I'd like to go through life with a name like that. (BNC) 101. In truth through the years Tiller Girls danced in ill–fitting shoes and had bunions to prove it for the rest of their lives. (BNC) 102. the information needed to optimise the use of the information through its life cycle is captured as part of the creation process. (BNC) As was the case with Through1, the use of NB in the domain of time is not differentiated syntactically, and so it will be subsumed as a sub–sense under the primary sense NB in this analysis. 3.3.10. ThroughNB with time vs. ThroughEB The examples of NB with time may look superficially similar to those of EB in (99-102). There are clear differences, however. For EB, the focus is on the termination of the event or the resulting state, while with NB, the focus is on the process itself. Certainly, the examples of NB above clearly refer to the process, rather than the end boundary or resulting state. All of the evidence differentiating NB and EB is relevant here, but certain evidence will be highlighted below. 3.3.10.1. With be We can see differences arise in the following sentences: as shown above, EB can co–occur with a form of be as a main verb; NB can’t. This holds for this extension of NB as well, as in the following example. Again, the stative verb be conflicts with the eventive reading required for NB. 103. (NB w/time) *I am through life with a name like that. (cf. (100) above) 3.3.11. Different uses of through with go We have seen forms of the verb go used with all of the primary senses of through that we have observed so far. Minimal pairs abound, but in each case, the interpretation is significantly different. This is more evidence that it is, in fact, through (or the V + P construction) that is determining the differences in meaning, rather than the verb. 104. (1) I can’t go through it (because I could never walk that far). (= enter a tunnel, etc. and exit from the opposite side) 105. (EB) I can’t go through with it (because it is against my moral code). (= bring the plan to completion) 106. (NB) I can’t go through it (because it is too dense or cold, etc. for my body to 43 be able to move in) (= move within the boundaries of water, space, etc.) 107. (NB) I can’t go through it (because I still have my dignity) (= endure the torture, hardship, etc.) 108. (NB w/time) I can’t go through another winter here (because I am not physically or emotionally capable) (= endure the difficulties of the time period) 109. 110. 111. (1) Phil will go / be14 through the tunnel at 3:00. (EB) Phil will #go / be through the tunnel at 3:00. (NB) Phil will #go / #be /be going through the tunnel at 3:00. 3.3.12. Monosemy of ThroughNB The syntactic evidence links all sub–senses of NB, including the metaphorical extension of NB used with time. Here, I will once again call on the argument that a single definition linking all uses of NB, as provided above, is not overly broad, and does not permit other morphemes to fall within its range. This is certainly true for other primary senses of through, but what about something like in, which overlaps in usage with NB? It can be seen that the requirement of motion differentiates the two, as in with motion – I swam in the water. – I swam through(NB) the water. without motion – I was in the water. – #I was through(NB) the water. Based on this, I will consider NB to constitute a single primary sense. 3.4. The final primary sense: ThroughALL Through1 can be extended to refer to a Path that penetrates or traverses (approximately) all parts of a Ground. A modifier such as all can be (and is often) added, but is not necessary. I will refer to this primary sense as ThroughALL. Some examples are provided below: 112. He'd washed his hair and piled all his plastic carrier bags into a corner instead of strewing them through the house like the clues of a treasure–hunt. (BNC) 113. Soon there are delicious smells wafting through the house. (BNC) 114. …but seeing the electrifying force of spontaneous, hate–inspired rebellion sweeping through the crowd of five hundred coolies, they froze where they stood…(BNC) 14 This use of be requires a “freeze-frame” reading. It is interesting how similar it is in meaning to the following example, of EB, and how the syntactic environment is identical. This type of “bridge context” may have been one impetus for the development of extension EB. 44 Polysemy and the structure of English THROUGH 115. Pass all through the mill of your mind and don't use forms unmeaningly… (BNC) 116. Years later my mother could hardly wait for me to pass it on to her. It went all through the family. (BNC) 117. woven all through like in a tapestry we find continuity, of faith, of worship. (BNC) 118. the sharp crack of the ball on the end-wall carries all through or even outside the villages. (BNC) 119. it spread all through Europe . (BNC) 120. they were chasing, were chasing all through the common. (BNC) 121. you used to carry them all through all round the village. (sic) (BNC) 122. All through the long, twisting lanes the route was marked by twigs (BNC) In the first three examples, the semantics of ThroughAll is expressed without use of the lexeme all, but many or most uses of this primary sense occur with all. Note that the penultimate example above – apparently a performance error found in the spoken portion of the corpus – the token of through could be motivated by Through1 or by ALL, but the speaker’s self-correction (all round) suggests that it was ALL that was intended. Note also that the last example above involves fictive motion. 3.4.1. Without motion – all parts of / everywhere A variation of ALL in which motion is unnecessary exists, and the focus is on the medium itself. It expresses the concept of all parts of, or everywhere in the medium, in the sense of throughout.15, or ‘chaotic dispersal’, as Queller (2001) offers in an analysis of all over. Examples are given below. 123. ‘Twas the night before Christmas, and all through the house, not a creature was stirring, not even a mouse’. (poem attributed to Clement Moore) 124. Such ambiguities are not unique to Tanzania but run right through relationships between development agencies and African governments. (BNC) 125. The sirens ecoed (sic) through the street. (BNC) 126. When you've been through death and live again you're changed all through. (BNC) 127. Ruth stretched out her hands to the red glow, wondering if she would ever again be warm all through her. (BNC) 128. All through the pages of Ramakrishna's biography appears his gentle humour (BNC) 129. I imagined it was all through my body and that I was going to die. (BNC) 130. Mold spores sprung up (all) through the town. 131. Soldiers were stationed (all) through the town and the fields. 15 It is certainly not a coincidence that these words are closely related. 45 I will not differentiate between the ALL that includes motion and the one that lacks motion, as I have not found any syntactic evidence warranting such differentiation. 3.4.2 On forming a phonological unit with all Note that the lexeme all that commonly precedes ThroughALL is not necessary, and it can occur with primary senses other than ALL. It is possible, for instance, to get all through with clear instances of EB, as in: 132. He was feeling smug after his win from Rab; his hunch, late in play, that Rab was shot, his concentration all through. (BNC) 133. When his tests were all through, he decided to ask her out. (BNC) However, it is noteworthy that a use of the primary sense of ThroughALL can form a phonological unit with the lexeme all, as seen by contrasting the phonological realization of the following: a) You carried [them all] through the village. (1 or NB) b) You carried [them all] through the village. (1 or NB with stress on all) c) You carried them [all through] the village. (ALL) In (a), which refers to a Path that traverses the village from one side to another (1) or moves within the village (NB), all is unstressed and the final intonation unit begins with through, which may be produced with slight stress. In (b), all is stressed, and through cannot be stressed, but it again begins the new intonation unit. In (c), however, which refers to a Path that (via summary scanning) traverses (roughly) all parts of a village, both all and through take stress as part of the same intonation unit. 3.4.3. On representing ThroughALL It seems that one could represent an idealized schema of ThroughALL in multiple ways; one type would show an image of a Path extending through essentially all of a Ground without crossing the boundaries, but with the addition of several glances around the Ground, as in Figure (9a), below, in which thinner arrow heads are meant to represent scanning; a second type could represent a multiplex in a state of chaotic dispersal, with the outline of the entire group covering approximately all of the Ground, as in Figure (9b); a third simply shows a Ground (container) filled entirely with a Figure, as in Figure (9c). Figure 9 – Possible schematized representations of ThroughALL Figure (9a) 46 Polysemy and the structure of English THROUGH Figure (9b) Figure (9c) In some sense, (9c) best captures the meaning of ALL, as it shows the entire Ground completely filled, but in another sense, (9a) is to be preferred, as it includes the Path and the summary scanning (Langacker 1991a) that will be argued below to help motivate the extension to ALL from 1. Finally, (9b) portrays the element of ‘chaotic dispersal’, present in some uses of ALL (such as those in (124 and 128) above). Therefore, no single schema completely captures the essence of ALL, and thus I have provided multiple schemas in Figure 9 to capture different aspects of the full meaning potential.16 3.4.4. ThroughALL as an adverb In some uses, through can also serve as an adverb, both as a particle and as a post-adjectival modifier. Above, we saw that EB is expressed solely as an adverb, but here I will first present a series of examples that represent a semantic bridge between EB and ALL, and then I will show some clear instances of ALL as an adverb. To begin, observe the following: 134. …4–6 minutes on each side, until opaque and cooked through. (BNC) In this example, it seems possible to conceptualize the progression of the heat or “line that is cooked” as a Path through the food that finally reaches the limit (i.e. where it meets an opposing Path). Such a conceptualization could be structured in terms of EB or of 1. It is also conceivable that one could conceptualize this as “cooked in all places within”, motivated by ALL. The next example is similar: 135. You're soaked through with sweat, right enough. (BNC) However, in this case, it seems unlikely that the speaker is referring to one single location (e.g. the upper back). Instead, I would suggest that the speaker is likely 16 It may also be true that the “filling in” of the Ground is a conceptualization process that applies during processing of one of the two leftmost schemas, and therefore that two (or even all) of these schemas may be relevant to the processing of a single use. 47 expressing the semantics of EB (through to the outside of your clothes) in addition to “everywhere” (or chaotic dispersal). In the next example, “everywhere” is the most salient meaning, while the semantics of EB may be absent. 136. At least the cameras had survived. Everything else was wet through. (BNC) While I do see the above examples as a type of bridge context, I am not claiming that ALL is derived from EB – merely that potentially ambiguous examples exist, suggesting that ALL’s (presumed) extension from preposition to (preposition as well as) adverbial particle is a motivated one. The semantics of EB and ALL are insufficiently similar to warrant deriving ALL from EB (or vice-versa), and instead I will claim below that ALL is more closely related to (and motivated by) Through1. However, at this point, I will first present some data I judge to represent adverbials expressing of the semantics of ALL less ambiguously. 137. …warmed by rich fabric wallcoverings and dotted through with Murano glass fittings. (BNC) 138. This classic bob has been softly layered and highlighted through with delicate copper tones to add warmth, shine and movement. (BNC) 139. This is in another class altogether: literate, thoughtful, at times profound, and shot through with wit. (BNC) 140. Unfortunately the timber industry is shot through with economic inefficiency. (BNC) 141. The air was close, soured through with the smell of size, canvas and stewed tea (BNC)17 In these examples, through is clearly serving as a particle with the semantics of ALL (‘(approximately) everywhere within’). It is not clear if all of them involve a Path, however. For instance, does shot through include conceptualization of a Path (e.g. that of the metaphorical “bullets”), or simply the notion that something exists at approximately all (or many, chaotically dispersed) locations within? Is something that is dotted through with certain glass fittings conceptualized as having the glass fittings in many places throughout, or is the Path (of the architect or designer, or possibly the viewer) relevant? I have no empirical basis from which to answer this. In any case, we do not need to alter our analysis to reflect this syntactic differentiation, because of a lack of corresponding semantic differentiation. Therefore, we have the following situation: ThroughALL can express two related semantic possibilities: a Path that penetrates or traverses (approximately) all parts of a Ground, or a notion of the entirety of the Ground, lacking any element of Path. The syntactic possibilities associated with these meanings include realization as a preposition and as an adverbial particle. 17 The OED claims that the use of through as an adverb is only possible while referring to a physical condition, but these final three examples suggest otherwise. 48 Polysemy and the structure of English THROUGH This brings us back to the question of whether examples (134-136) are instances of the expression of the semantics of ALL or 1. There is no easy way to resolve this question, which involves the language user’s conceptualization of the scenario. 3.4.5. Summary scanning and ThroughALL The extension to ALL from 1 (described in detail below) involves the cognitive process of summary scanning (Langacker 1991b, 1999). Langacker describes summary scanning as a process in which ‘the various facets of a situation are examined in cumulative fashion, so that progressively a more and more complex conceptualization is built up; once the entire scene has been scanned, all facets of it are simultaneously available and cohere as a single gestalt.’ (1991b:78) It is the mode of processing that allows us to say 142. There are houses all through the valley. This could be paraphrased by saying that, as one goes through the valley, one sees houses everywhere. Through summary scanning, and aided by real world knowledge, the houses can be conceptualized as existing simultaneously in the valley. Notice that there is no need for the speaker of this sentence to have crossed the end boundary of (i.e. to have gone Through1) the valley; it is perfectly acceptable if s/he walked most of the way through(1) and then turned back, as long as s/he had reason to believe (such as having seen) that it was not the case that there were no houses in the final part of the valley. S/he could also have flown over the valley. The fact that the Ground is filled is criterial, whereas whether the speaker actually moved all the way through(1) the Ground is irrelevant. At times, the performance of a summary scan on a situation involving chaotic dispersal of the Figure permits ALL to refer to the entirety of a Ground. Using all with what could otherwise be 1 or NB typically changes the interpretation to that of ALL. For instance the following would be categorized as different primary senses in my analysis: 143. I walked through the forest. 144. I walked all through the forest. Note that the first example could be an instance of either 1 or NB (depending on the geometry of the forest and the Path taken, and, most importantly, the language users’ conceptualizations). If we make the first example above a clear instance of NB by adding an adverbial expressing a duration of time, it becomes more difficult to add all, as in the first example below; with a clear example of 1, the addition of all sounds more natural, as in the second example. 145. I walked (?all) through the forest for an hour. 49 146. I walked (all) through the forest in an hour. However, compare the following example, in which a clear instance of NB can be modified by all, resulting in an instance of ALL: 147. I enjoyed walking (all) through the forest, but I ended up leaving via my entry route, never having made it to the opposite side. This is due to the fact that ALL and NB can share a focus on the Path within the Ground (as seen in their paraphrases ‘everywhere within’ and ‘motion within’) unlike 1, which inevitably involves the crossing of boundaries. In fact, some tokens are ambiguous between NB and ALL, as in the examples below, in which it is not clear if we are to understand the Path as existing in all (or sufficiently many) parts of the Ground, or simply as moving within the Ground. 148. A murmur of approval passed through the crowd (BNC) 149. The second SAVE publication sent shock waves through the country. (BNC) 3.4.6. Through and through and throughout It should be noted that ALL is extremely close, in terms of semantics, to ‘through and through’, as in: 150. …Comb the fibres through and through, Twist them till they're strong and true…(BNC) 151. Through and through the inspired leaves Ye maggots make your windings... (BNC) 152. Einstein, of course, was a realist through and through. (BNC) 153. The childish words went through and through his mind, blotting out all coherent thought. (BNC) 154. … he was the pure, 100 per cent archetypal don, a don through and through, albeit a don whose special subject was politics in practice. (BNC) It should also be noted that ALL is analogous to ‘throughout’, as in: 155. Open racism in court may continue throughout the trial…(BNC) 156. Directors should keep their finger on the pulse of a production throughout it's run. (BNC) 157. Using wood throughout ties the design together. (BNC) 158. She described her itinerary over gazpacho, her beach clothes throughout a pasta salad…(BNC) 3.4.7. ThroughALL with time In addition to its uses both with and without motion, ALL can be extended to apply to the domain of time. This is by far the most common use of ALL in my data. Some examples are provided below. 50 Polysemy and the structure of English THROUGH 159. All through the night he lay awake, twisting and writhing, and moaning in low desperate tones. (BNC) 160. The 12 jurors who eventually emerged, acquitting themselves both responsibly and wisely, were mocked all through the trial for their necessary ignorance. (BNC) 161. He worked all through the night and into the next day. (BNC) 162. All through the 1960s the package holiday market had grown and by the early 1970s they accounted for nearly half of all overseas travel. (BNC) 163. We had to keep quiet all through the evening. (BNC) There is no morphosyntactic evidence differentiating the use of ALL with time from that with other mediums, and thus I will not treat them as distinct usage-types. A representation of all of the primary senses of through is provided below in Figure 10. Figure 10 – Representation of all primary senses of through 1 EB ALL NB Oval = primary sense of through Arrow = extension to alternate primary sense 3.4.8. ThroughALL vs. the other primary senses It is true that some examples of ALL look very much like other primary senses, as discussed above, and I will continue treating this topic below. However, there are distinctions between all of these that warrant an analysis that treats them as distinct primary senses. It is true that many, and maybe even most examples of ALL do take the modifier all, but there are certain contexts, presumably where there is little danger of ambiguity, in which ALL does not – or cannot – occur with the lexeme all. For instance, some of the examples above could not take the quantifier all, such as this, brought down from above: 164. This is in another class altogether: literate, thoughtful, at times profound, and shot through with wit. (BNC) Others take different intensifying morphemes, such as right, that may convey that the full extent of the Ground is relevant or affected: 51 165. Such ambiguities are not unique to Tanzania but run right through relationships between development agencies and African governments. (BNC) It should be noted, however, that right can modify Through1 as well (he ran right through(1) the tunnel). By expressing intensification, it appears that (this sense of) right only can be used with primary senses of through for which multiple boundaries are relevant; the addition of right refers to the speed, the ease, or the fullness of the extent to which the Figure filled (ThroughALL) or traversed (Through1) the boundaries of the Ground. Some examples are given below. With ThroughALL: 166. Stage nerves are unlike anything else in the whole world, and many good actors will confess to them right through their careers. (BNC) 167. Intertwined with this line of attack on Labour, and running right through yesterday's speech, was another theme, that of freedom. (BNC) 168. … the social repercussions of external and internal war and revolution flowed on right through the 1920s. (BNC) With Through1: 169. They were soaked right through to the skin and shivering in the cold. (BNC) 170. ‘Did the blade pass right through the person?’ he asked. (BNC) 171. The hole goes right through to the inner mechanism…(BNC) Right is not normally used with NB or EB, however. This is presumably because the boundaries are not relevant for the former, and only a single boundary is relevant for the latter, while right implies smooth passage for the full extent of the Path that is internal to the Ground. 172. ?*The maximum speed right through(NB) the tunnel is 75 mph. 173. ?*The train rumbled right through(NB) the dark. 174. ?* I went right through(NB) the tunnel, thinking, but I couldn’t go all the way through(1). 175. ?*She had to go right through(EB) with it now that she had signed the contract. 176. ?* John was really sick for a while there, but after starting this new medicine, he pulled right through(EB). 177. ?*…the project fell right through(EB) with the arrest of his opponent. 3.4.8.1. With except for Because NB does not invoke any boundaries, unlike the other primary senses, it cannot occur with the adverbial except for easily, unless all is added, changing the semantics to those of ALL: 178. ?#He ran through(NB) the tunnel for about ten minutes, except for the west end. 179. He ran all through the tunnel ?for/in about ten minutes, except for the west end. 52 Polysemy and the structure of English THROUGH 180. ?#He swam through(NB) the water, except for the deep end. 181. He swam all through the water, except for the deep end. 182. ?#She ran through(NB) the crowd, except for the group in back. 183. She ran all through the crowd, except for the group in back. Note that, for the second example of each pair, the addition of all changes the interpretation of through from that of NB to that of ALL. Unlike this situation, with ThroughALL, except for is easily added, even without all: 184. It echoed through(ALL) the house, except for in the sound–proof room. 185. The mold sprung up through(ALL) the house, except for in the bedroom. 186. The news spread quickly through(ALL) the crowd, except for the people dancing, oblivious, in the back. Here, the modifier except for presupposes that the Ground is entirely filled by the Figure. This is not presupposed with uses of NB, and hence a modifier (all) that alters the situation expressed so that it fulfills the requirements of the presupposition is necessary to make the sentence felicitous. No such modification is necessary with ALL. This is predicted by the characterization I have given ALL. 3.4.8.2 Discrete entities as Figure While NB can co–occur in a sentence with a singular figure, ALL cannot do so (if the subject is a count noun) unless it is describing a motion event, even if the speaker’s construal of the boundaries of the Ground permit a single figure to ‘fill’ the Ground. If an ALL sentence describing a motion event is given a singular subject, it will take the meaning of a Figure in motion defining a Path as a sequence of points. These points act much like the mold in example (130) or the soldiers in (131), conceptually filling the Ground. Talmy (1983:31) notes that through can be used to cognize a scene with a temporal sequence of close–up perspectives. Since a singular figure with a stative verb doesn’t allow for this interpretation, it is an ungrammatical combination, as seen in the following examples of ALL: 187. the bugs landed all through the house (possibly simultaneous) 188. the bug landed all through the house (= a single bug landing multiple times) 189. #all through the house, I just sat there, exhausted. 190. all through the house, my friends just sat there, exhausted. This is not a requirement on the relative size of the Figure and Ground; as shown by the following examples: 191. #all through the hangar, I just laid there, exhausted. 192. #all through the coffin, I just laid there, exhausted. Note that ALL accepts singular mass nouns as Figure: 53 193. the trash stunk/rotted/decomposed all through the house 194. the trash was (strewn) all through the house 195. the water flowed /spilled all through the house 196. the water was all through the house This is because mass nouns, as is the case with plural nouns, can be summary scanned whereas count nouns cannot. This reveals a very interesting situation: mass nouns are normally treated as singular by the grammar of English, as shown by the use of singular verb forms, and yet they are permitted in a context, heavily influenced by pragmatics, where a singular noun is not permitted, but plural nouns are. This reveals that mass nouns have properties of both singular and plural nouns in the grammar of English, which is not surprising, as they have properties that make them seem like they are both singular and plural in the ‘real world’, as well. 18 Is it ever possible to get a clause with ALL with a singular count Figure that does not overtly express any motion? Consider the following: 197. He worked all through the night and into the next day. (BNC) 198. All through the winter the nightmare went on, through the worst of the weather… (BNC) Here, the Figure is unquestionably a singular, countable entity. I would claim that, for this type of data, time is being construed as a location (an extremely well–attested process; see Lakoff and Johnson 1980, 1999, inter alia), and the Path is conceptualized as passing ‘through’ the medium of time. Importantly, we see that verbs of motion can be used with these Figures, such as went in the second example above. This is the same phenomenon that allows for the acceptability of the second example, but not the first, in the following data: 199. #All through(ALL) the hangar, I just sat there, exhausted. 200. All through(ALL) the play, I just sat there, exhausted. This requirement is not one specific to singularity, but rather is a demand that the Figure be capable of conceptually filling the Ground. Therefore, it is odd to get a small number of Figures with ALL, but if the number is increased, the sentence becomes more acceptable. 201. When I returned, his ?#two/twenty–two friends were all through(ALL) my house. 202. ?#There were several descriptions of her emotional state all through(ALL) the book. 203. There were dozens of descriptions of her emotional state all through(ALL) the book. 18 A discussion of the commonalities of mass and plural nouns is beyond the scope of this paper. The interested reader is referred to the works of Ronald Langacker (1987, 1991a,b, 1999). 54 Polysemy and the structure of English THROUGH Notice, however, that the following, with 1, is fine: 204. There were several descriptions of her emotional state through the book. This can be paraphrased as ‘…as I read through(1) the book’. This restriction on ALL effectively differentiates it from all of the other primary senses of through, as seen in the following data. 205. (EB) They’re all through with the book. 206. (1 with ‘Freeze–frame’)They’re all through the tunnel. 207. (ALL) They’re all through the house. 208. (EB) He’s (already) through with the book. 209. (1 with ‘Freeze–frame’) He’s through the tunnel. 210. (ALL) #* He’s (all) through the house. Again, the restraint on ALL taking a singular, count noun Figure in clauses that lack motion is the determining factor in the grammaticality of these examples. This is evidence that clarifies both the definition and origin of ALL. The use of this primary sense does not simply describe a Figure filling the boundaries of a Ground, but summary scanning is part of what defines it. Based on this definition, we can predict that a Ground that is small enough to be ‘taken in’ with a single glance should not permit the use of ALL, whereas one that is larger and requires the speaker to build up a conceptualization of one portion at a time should allow the use of ALL. Therefore, we would not expect to get sentences such as ?#There is dirt all through the cup, whereas we would expect speakers to produce sentences such as There is dirt all through the house. Indeed, searches for collocates coming immediately after all through the in the BNC support this prediction, as seen in the list of all collocates with MI of 3+ coming one place after all through the in the BNC, in Table 4, below. Table 4 – Collocates of ThroughALL in the BNC Collocate WINTER NIGHT WAR SUMMER MORNING WEEK DAY YEAR Freq 9 23 15 6 5 6 9 9 Total Freq in Corpus 7046 34119 26881 10955 19846 31230 59298 72143 % 0.13 0.07 0.06 0.05 0.03 0.02 0.02 0.01 MI 7.23 6.31 6.03 6.01 4.89 4.50 4.16 3.87 All collocates are sufficiently complex such that a single glance is potentially insufficient to “take them in”, and a conceptualization must be built up over time. 55 3.4.8.3. With be Notice that ALL is used with be in some examples immediately above. We can add ALL to our list of uses of through that can occur with be, differentiating it from NB, though this doesn’t hold with singular count noun Figures, as (217) below shows. 213. (1 without motion)My arm was through the window. 214. (1 with ‘freeze–frame’) He’s through the tunnel. 215. (EB) I am through with finals. 216. (NB) #The water is through the streets. 217. (NB) #I am through the streets. 218. (ALL) The water was all through the house. 219. (ALL) #I was all through the house. 3.4.8.4. Different uses of through with go and be We have seen go used with all of the primary senses of through that we have observed so far. Minimal pairs (actually many quadruplets) abound, but in each case, the interpretation is significantly different. This is more evidence that it is, in fact, through that is determining the differences in meaning, rather than the verb. 1 – I can’t go through it (because I could never walk that far). (= enter a tunnel, etc. and exit from the opposite side) EB – I can’t go through with it (because it is against my moral code). (= bring the plan to completion) NB – I can’t go through it (because it is too dense, cold, etc. for my body to be able to move in) (= move within the boundaries of water, space, etc.) ALL – I can’t go all through the house (because I lack energy, and the house is so large). (= physically go to most or all locations in the house) Several near minimal pairs exist with go and be, such as the following, in which it may be noticed that aspect serves to differentiate 1 and NB: 1 EB NB ALL – Philip will go / be through the tunnel at 3:00. – Philip will go / be through with his walk in the tunnel at 3:00. – Philip will ?#go / #be /be going through the tunnel at 3:00. – Philip will go/*be all through the tunnel (starting) at 3:00. 3.4.9. Monosemy of ThroughALL Once again, the grammar of English differentiates the use that I have labeled ALL from all other primary senses of through, and the semantic range of uses of ALL does not require an overly general definition which would apply equally well to other words. 56 Polysemy and the structure of English THROUGH 3.5. The relationships between the primary senses From the evidence presented above, it is clear that 1, EB, NB and ALL are each treated differently by English speakers. What has not yet been described in much detail is the motivation for the existence of these different uses of through, and in particular, the relationship between the senses. It can be seen that Through1 is concerned with both the crossing of boundaries and the temporary in relationship of the Figure and the Ground. As described above, the Figure enters the Ground, crossing a boundary, moves to the (functionally) opposite side of the Ground19, and then exits. NB is an extension from 1. With NB, the boundaries become insignificant, and the mere fact of motion while in an in relationship defines the primary sense. Thus, the temporally and spatially central portion of Through1 is taken as the core sense of NB. This extension from 1 to NB is an example of metonymy in which one part of 1 comes to stand for the entire primary sense of NB. By comparing the descriptions of Through1 with that of EB, as given above, it can be seen that EB is profiling the end boundary that is relevant in, and part of the schema of, Through1. In other words, this is also a case of metonymy. This motivation for the derivation of NB from 1, as well as that of EB from 1, can be seen in other domains of grammar. The same process also motivates some languages, such as English, to permit the use of instrumental subjects; rather than view the entire event sequence, the speaker begins with a point further to the right on the chain of causation (DeLancey 1984), so that in a sentence like: AGT INST - The janitor opened the door with the key. the starting point of the entire event (the janitor) can be excluded if the speaker cognitively ‘takes a step’ nearer to the event line20, and one gets: - The key opened the door. As discussed by DeLancey (1984), this is the result of conceptualizing or viewing a restricted portion of the chain of causation, rather than the entire sequence of sub–events. The conceptualized portion starts with the key, which is not the ultimate cause of the event, and continues to the point where the door is open. Although an agent still is necessary21 to cause the key to turn, mention of the agent is purposely avoided by the speaker for pragmatic reasons. This same type of reasoning applies to forms expressing e.g. imperfective aspect, which involves viewing an event from “up close”, and considering only a certain central portion of 19 This could be fictive motion (He looked through the trees), or the Figure could not move at all (His arm was sticking through the head–hole), but if its shape were drawn, it would trace a path that, if followed, would carry it through the Ground. 20 Of course, certain pragmatics are required for the felicitous utterance of such a sentence. 21 With the exception of animate keys in cartoons, for example. 57 the event22, vs. perfective aspect, which involves taking a “step back” to see that an event was both started and completed. Thus, phenomena that reflect zooming in, such as the existence of instrumental subjects and aspectual systems, provide examples of the part-for-the-whole metonymy that is fundamental to the extension of Through1 to ThroughNB, as well as that of Through1 to ThroughEB. The extension to ALL from Through1 involves a metonymy, in that part of Through1 (a Path crossing the full extent of a Ground; the crossing of boundaries is not relevant) is sufficient to motivate this use of the form through (referring to (a Path covering) the full extent of a Ground). Other aspects of 1, including the crossing of boundaries and the suggestion of force dynamics (see Author 2014; I cannot discuss this here due to space limitations), are not relevant to ALL. When it is used in the domain of time, 23 a Path crossing all of the Ground is indistinguishable (in many respects) from one covering all of the Ground, so that some uses are potentially motivated by either ALL or 1, such as all through the night, I sat there. Crucially, however, with ALL, summary scanning is also employed, as described above, resulting in uses such as I coughed all through the night. From here, the metonymy of a person’s attention for the person (or the metaphor that a person is located where their attention is focused) may be applied, such that someone taking a Path prototypical of Through1 may say 220. I walked all through the museum, but I never found the painting you described. That is to say, the speaker walked Through1 the museum while paying attention all ThroughALL (throughout) the time covered by the event. Thus, even in such cases, we find the basic metonymy deriving ALL from 1 to be required in order to motivate the usage-type. More complex cases such as the famous 221. Twas the night before Christmas and all through the house, not a creature was stirring, not even a mouse (repeated from (123) above) involve virtual presupposition (if someone had walked Through1 the house, all ThroughALL the time s/he would have been walking and observing the condition of the house, s/he would not have seen even one creature stirring). They also invoke the (assumption of the) metonymy that motivates the extension of ALL. The extensions of the three primary senses EB, NB, and ALL from 1 fit the definition of metonymy most commonly used in cognitive linguistics, based on a semantic shift within a domain, or frame (see Author 2007:135, Croft 2001, 2006, Croft and Cruse 2004, 22 While real-world knowledge tells us that all events must have a beginning and an ending, the use of an imperfective suggests that the speaker is describing one portion of the event (the middle) as the whole event, in a case of metonymy. 23 Time is often represented one-dimensionally. 58 Polysemy and the structure of English THROUGH Fillmore 1982, Kövecses 2002, and Langacker 1987, 1999 for discussion of metonymy, frames, and domains in cognitive linguistics). 3.6. Image schemas and the four primary senses of through The meaning of through relies most fundamentally on two spatial relations concepts: the Container schema and the Source-Path-Goal Schema (Johnson 1987, Lakoff and Johnson 1999, Lakoff and Turner 1989). Phrased in the terms of Croft and Cruse 2004 except where noted, we have the following mapping of image schemas to prototypes of the primary senses of through. Figure 8 - Image schemas invoked by the primary senses of through 1: Path, Bounded Space, In, Out NB: Path, In EB: Path, Bounded Space, In, Out ALL: Bounded Space, In, Full While 1 and EB are mapped to identical image schemas, they are applied differently, as 1 prototypically requires the sequence Out-In-Out, whereas EB requires In-Out (i.e. the Source of the Source-Path-Goal schema is not relevant to EB). EB and 1 may also invoke the Path-to-endpoint transformational schema (Lakoff 1987; She is through the tunnel/with her homework). The Center-Periphery schema (Johnson 1987) is relevant to the application of Through1, but not EB or the other primary senses. Many other image schemas can contribute their ‘logic’ to a conceptualization of a use of through. For instance, contexts with NB or 1 can reflect the Multiplex Path schema (Laughter spread through the crowd), as can uses of ALL (There are bugs all through the house). Uses of NB and 1 can invoke the Multiplex Mass schema (Johnson 1987 refers to the Undefined Mass-count schema; I fought my way through the crowd). Further discussion of some of the image schemas associated with through can be found in Ochoa (1998) and Martín (2000). 3.7, Event structure associated with the primary senses of through Due to many different factors, including the use of one primary sense (EB) as a perfective marker and the unboundedness of another (NB), the primary senses of through are associated with a limited range of possible types of event structure. In addition to the strict requirements of EB and NB, 1 and ALL each have a preferred event structure. ThroughALL prefers durational events, based on the noun collocates of all through in the BNC, as discussed above. Through1 typically involves the crossing of a final boundary, and so it tends to be used in telic events. This is not necessary in the right circumstances, however, such as when the geometric relationship defined by 1 continuously applies throughout the event described, such as in 220. Whose face would it be, looking through the small square window behind us? (BNC) 59 3.8. Summary of the primary senses of through Below in Table 5, a summary of the characterization of each of the primary senses of through is given. Most information given in the table has been discussed here, but for certain aspects, such as force dynamics, the reader is referred to Author (2007, 2014). 60 Polysemy and the structure of English THROUGH Table 5 – Prototypes of the four primary senses of through (adapted from Author in press, see also Author 2007:139, 2010a) Primary sense: Characterization of prototype semantics Path Ground Motion Lexical class Collocates Duration Telicity Category of event structure (Vendler 1967) Through1 Enter, cross, exit EB Finished/ terminative, exit Specifications Crosses two Crosses boundaries final boundary NB Motion within ALL Everywhere within Stays within Ground Fills Ground (or not relevant)24 Relevant, filled May be Relevant; as irrelevant25 ; Medium may be focused on26 Possible Possible Yes Morphosyntactic Realization Particle Prep Prep27 Precedes Mass Physical entities28 with, often nouns occurs with copula Associated event structure Possible No Yes Usually telic Telic Atelic Usually States Activities Accomplishments, Achievements Relevant Possible Prep, Particle Follows all Usually Either Usually Activities, Accomplishments Based on the information seen in Table 5, it is clear that the union of the four primary senses of through is internally inconsistent. For example, while ThroughEB is telic, ThroughNB is atelic, and while the latter requires motion, the other primary 24 The Path for ALL may (conceptually) fill the Ground, as in I walked all through the building, or may not exist in the real world, as in There are flies all through the building, which seems to involve fictive motion as well as summary scanning. 25 Examples include Thanks for coming through in a pinch. 26 Examples include The night is almost through. 27 It is possible that the semantics of Through1 can be expressed in a verb particle, though the discussion in Author (2007 §4) was inconclusive. 28 Among the strongest collocates of through in the BNC (within 5 words to the right) are archway, alleyways, sieve, channels, undergrowth, turnstiles, keyhole, megaphone… 61 senses can be used without motion. For this reason, a monosemist approach cannot explain the breadth of meaning of the four primary senses.29 At this point, I will present a schematic representation of the basic semantics of primary senses of through that is organized based on semantics. It represents the organization of through and its relationship to various lexical alternatives, and is given in Figure 11, with definitions/ explanations given in the notes below the Figure. 29 This is also why I argue for the hypothesis of ‘unspecification’ in Author (2007, 2008, 2010c, in press; the terminology used in these sources varies), which states that some semantic features associated with a lexeme are undetermined or unrepresented in the semantic representation created by processing the utterance including the lexeme. In other words, I argue that a language user can create a mental representation of the utterance in which a lexeme occurs without fully utilizing its lexical semantic meaning potential. Logically, if a speaker were to specify a unique primary sense, the presence or absence of a boundary, the telicity, etc. would need to be selected, but I show that the linguistic environment of many uses of through does not specify these (seemingly essential) semantic facets. In Author (2007, in press), I show the results of a quantitative study of corpus data revealing that 40% of the tokens examined are not unambiguously motivated by a single primary sense (I will describe this study again below); I also describe a qualitative study that shows the wide variety of usage-types that can serve as candidates for motivating an ambiguous use. 62 Polysemy and the structure of English THROUGH Figure 11 – Schematized representation of the relationships of the primary senses of through, including several lexical alternatives. (Author 2007, 2010a) Notes on Figure 11: -Primary senses of through are represented as ovals. Those derived from the central sense of Through1 are represented as dashed ovals. -Related lexical items are represented as rectangles. The ‘basic’ spatial relations of the IN/ON continuum in English (in, on, and above) have been represented with dashed rectangles. -Text in ovals characterizes basic semantics of primary senses of through. -Regarding the text in parentheses: both motion and resistance are optionally (but relatively often) found with uses of Through1. The latter is also a tendency with uses of NB, but movement is required with NB. -ALL takes the ‘within’ and the ‘to the boundaries’ from 1. In order to show the relationship of the senses, lines have been included linking both the initial and final boundaries of 1 with ALL. -Through, across, and over, being complex prepositions, are represented by longer horizontal shapes, iconically invoking a Path (which is present in prototypical instances of all three words). 63 -Through and across are based on in and on, respectively. Over, however, is related to above as well as on, suggesting a more distant relationship with through (due to the proximity of above and on; see Bowerman and Pederson 1992, ms and Author 2007). -The relative size of the objects in Figure 11 was determined based on the size needed for text, the focus of this study, and aesthetics, and is not meant to represent anything else. 3.9. The centrality of Through1 This issue of which of the primary senses of a through (if any) is central will be discussed here in light of the analysis presented in this chapter. Centrality is a concept often invoked in cognitive linguistic studies of polysemy, though it may be more appropriately divided into several different components that may or may not overlap (c.f. Jesperson 1922). Because several different components of centrality come together unanimously in the case of through (as described below), I will refer to ‘centrality’, though in general it may be more appropriate to discuss the individual components (frequency, motivation of the extensions to other primary senses, and others). A discussion of the relative frequency of each of the primary senses is located in Author (2007, in press), so I will only briefly summarize here: in an in-depth study of 500 tokens extracted from the BNC (with half coming from the spoken portion of the corpus), clear instances of Through1 dominated at 54.8% (or 274 of 500 tokens studied). In distant second place was Through NB, at 4.4% (or 22 of 500). (199 of the 500 tokens, or essentially 40%, were assigned to multiple potential primary senses – see fn. 27 in this paper as well as Author 2007, in press). We can now discuss which primary sense most naturally motivates the extensions to other senses. In the analysis above, it was argued that the ultimate source of the extension of all four primary senses is Through1. As we have seen, the relationship between Through1 and the other primary senses can be motivated by appealing to the process of metonymy; if one hypothesized that Through1 is an extension from the more central NB, a less satisfactory motivation for the extension would be necessary. One would have to claim that ‘from the sense describing motion within a medium (NB), the Path within the medium becomes somewhat less salient, although the fact of the Figure and Ground being in an in relationship retains its significance, and crossing the boundaries of the medium becomes the essential meaning of the extended sense (Through1).’ This is not a motivated series of extensions, however, as the question of how the crossing of boundaries arose is not easily resolved. Additionally, though the preference for simplicity of description in science may be an aesthetic bias, it too argues against such an analysis. This path of extension is less well motivated than if Through1 is considered the central sense. The same argument holds for the other primary senses. In other words, this analysis says that Through1 is more central than the other primary senses because of the way in which the links between the primary senses are motivated, in addition to the relative frequency of the primary senses. 64 Polysemy and the structure of English THROUGH IV. Some sub-senses of through 30 Here, I will describe the most common sub-senses of through. While they are not syntactically differentiable in any way that I can find, they are semantically distinct, warranting some description. However, I cannot guarantee that these meaning differences are arising due exclusively or primarily to lexical semantics, rather than contextual effects (particularly the effects of nearby lexemes and grammatical constructions, and, as always, real-world context). 4.1. The ‘via’ sub-sense of Through1 While the prototype use of Through1 includes information about each of the three aspects of Source, Path, and Goal, this is not necessary in all uses or usage-types. For instance, in some uses, the Path exists completely within the Ground (e.g. I work Monday through Friday; I read pages 125 through 170; this usage-type was discussed above), and reaching the Goal does not involve crossing the final boundary, but arriving at it, in such cases. But the ‘via’ sub-sense, which I will address here, is about conceiving of the Ground as an intermediary and a means by which the Figure reaches the resulting state of the Goal, which is increased in salience. Possible paraphrases include by way of, by means of, via, via the medium of, and through the use of. This sub-sense is extremely common; about half of the tokens labeled clear instances of Through1 in the corpus study in Author (2007, in press) were judged to be cases of the ‘via’ sub-sense. The following is a prototypical example. 221. They came through Portland, since direct flights cost more The ‘via’ sub-sense of 1 is not syntactically individuated in any way as far as I can tell. It is clear that it should be categorized as an instance of 1, since there is a coherent connection between the two usage types, and the breadth of meaning of 1 subsumes multiple specific ways of viewing a Through1 scene, one of which is as a ‘via’ event. A person may serve as the Ground with the ‘via’ sub-sense, typically in combination with the conduit metaphor (Reddy 1979) applied to communication or any other abstract or concrete entity, including a location, a state, a quality, or a medium (e.g. one can hear a story, get tickets, or be exposed to an illness or an idea through a friend, or through sheer luck, and one can have come through Tokyo or through foresight). Some instances from the BNC are presented below: 222. I have probably got a chance of getting tickets through Jack Charlton. (BNC) 223. No matter how long it takes, communication shouldn't be through a third person unless it's absolutely necessary. (BNC) 224. If we wanted to speak to him, we'd have to go through the headmaster. (BNC) 30 All sub-types described in this chapter are represented in the BNC, with the exception of ‘Ground-prominent EB’ and the ‘thoroughly’ sub-sense, but for the sake of brevity and simplicity, some examples given for sub-senses are intuited. 65 225. He had most likely become infected through a partner who had been receiving zidovudine. (BNC) This sub-sense of Through1 reflects a relative prominence of the Goal; the Goal is typically mentioned before the Ground (But not always; c.f. We’d have to go through the headmaster to get to him, in which him is the Goal), rather than the other order, which is much more common for other usage-types of 1 (and other primary senses, presumably since it iconically represents the order of the sub-events as experienced). The Goal must be either explicitly mentioned or clearly implied by context, or else it is difficult to justify calling a token an instance of the ‘via’ sub-sense. It includes uses with abstract nouns such as the following serving as the medium: 226. It lives through a wealth of images. (BNC) 227. For 36 years, he has survived through a sure–footed personal response to problems. (BNC) 228. Through our advanced technology MERYL and SETILA microfibres, Rhône Poulenc helps to bring about a world of greater comfort and performance. (BNC) 229. …reducing costs at the exchange through cutting down on the amount of paperwork. (BNC) 230. Her hankering for the outside world is shown through a solo where she is cruelly tangled in the legs of a chair (BNC) Though it may appear that these examples could be instances of NB, it is preferable to consider them to be instances of 1. In (226), for example, there exists some entity which can be seen as arriving at some metaphorical location (the state of ‘life’) via the process of becoming embodied, described as an abstract medium (‘a wealth of images’). The other examples can be analyzed in a similar manner. I will provide more evidence that ‘via’ is a sub-sense of 1. A certain subset of those tokens in the BNC that are considered instances of 1 based on this analysis is paraphrasable as ‘via’. A second subset is not paraphrasable this way, including the following: 231. ... to work my way through the manuscript … (BNC) 232. … hammers a tap through the wooden or plastic keystone… (BNC) This is because the focus in this example is on the Ground, rather than viewing the Ground as means to the Goal. A third, smaller subset of the BNC data is either ambiguous or not easily analyzable. This is because the prerequisite focus on the Ground as intermediary is not a binary choice but a gradation, such that the following examples may not be cases of ‘via’ at all, depending on how they are interpreted: 233. image of the Madonna is softened through the centuries into a more human and tender figure (BNC) 66 Polysemy and the structure of English THROUGH 234. a transaction entered through a conventional keyboard (BNC) The first example may be paraphrased as ‘via the centuries’ passing’ or ‘during the passing of time for centuries (up to now)’. The second could reflect a conceptualization of a message moving through the typist’s hands by making use of the conduit metaphor, or it could be a prototypical case of ‘via’. Tokens that don’t mention the Goal, or for which the Goal is not salient, such as the first example below, are unlike the prototype of ‘via’. This is also true to a lesser extent for tokens in which the Ground is overly salient or focused on (the last two examples below): 235. As I looked through the viewer I had the feeling, momentarily…(BNC500) 236. …my heart leapt when I saw, through the Bride and the Bachelors, through the Milky Way and…(BNC500) 237. We will meet again one day through the fog that we will clear (BNC500) This interpretation of the ‘via’ sub-primary sense could be represented as in Figure 10 below, not different in principal from the depiction of 1 given earlier (only the focus differs). Note that, in some cases, ‘via’ can completely lack reference to a Path, focusing on the aspect of enabling or causing: 238. the division between the living room and the dining room is through a pair of glass doors (BNC500) 239. democracy is at risk through our continued membership (BNC) In the first example, there is no Path through the doors, since what is being described is a division. In the second, the abstract idea of a cause is expressed. In these uses, ‘via’ can be interpreted in an analogous manner to by means of or because of, which include the notion of cause, in addition to the geometric relationships basic to Through1. Interestingly, the OED states that through has an extinct usage-type in which it indicated the agent of a passive clause (as Dutch door ‘through’ currently does), and another in which is was used to mean ‘in the name of’ in oaths and adjurations, suggesting a connection to the causal meaning aspects of the ‘via’ sub-sense. I considered all of these to be instances of Through1 both because there is no syntactic evidence differentiating the usage-types, and because semantically 1 could be said to subsume ‘via’. Therefore, based on the operational definition of polysemy employed here, I will not distinguish the ‘via’ sub-sense from the larger sense of 1 for other than descriptive purposes. 4.2. Other sub-senses of Through1 Many sub-senses of Through1 can be motivated by appealing to metaphor and functional oppositeness. More than a few are based on the metaphors ‘time is space’ and ‘states are locations’ (conceptualized in such cases as containers, likely because states existing over a period of time can be seen as a one-dimensional version of a container encompassing an area of space). Therefore, the cessation of a state is 67 emerging from a container (prototypically via the functionally opposite boundary). Other metaphors often applied include ‘life is a journey’ (I’ve been through some hard times and emerged victorious) and ‘seeing is touching’ (or, perhaps more appropriately, ‘a person is located where their attention is focused’), alternatively analyzed as the metonymy of a person’s attention for a person (I’ve been through this list twice already, and I still can’t find the error; see Lakoff and Johnson 1980, 1999 for a more detailed description and analysis of ‘seeing is touching’ and ‘life is a journey’, which Ochoa 1998 first noted are commonly used with through). The impetus for this situation comes from both the full scope of meaning of the various usage-types of through and the process of blending (Fauconnier and Turner 2002) that results in the meaning of the V+P construction (and any larger constructions in which through participates). Such metaphors motivate usage-types paraphrasable as ‘success’ (They’re through to the next round) and ‘failure’ (The deal fell through). In the former case, the Ground (such as a round in a tournament) is seen as a risky situation or a gamble, and crossing the opposite boundary (completing the round successfully) is the Goal; the latter is used in situations for which success is defined by staying in the container, and to fall through brings to mind a crossing of a boundary that is not the functional opposite of the entrance (such as the ‘floor’, rather than the desired exit). Compare The deal went through, which, with the use of an unmarked motion verb, describes success based on the idea that the final boundary was crossed. If the situation being described is one with an inherent endpoint/boundary that is also the Goal, such as a race, then to be through (if it is elliptical for to be through a stage/the tape/the race) can be good. If the situation is one with no inherent boundary, and the goal is to remain active for as long as possible, such as a basketball tournament in which teams are eliminated until the one remaining undefeated team is considered victorious, then They are through means that they have been defeated (distinct from they are through the round and they are through to the next stage, which both include explicit boundaries). One usage-type is paraphrasable as ‘survival’ (we made it through alive) when the container is a dangerous state. If the container is a process, then cessation in this way is ‘finishing’ (we really want to get through the business as soon as possible). If the state/container is possession of goods, and they are meant to be used or sold, and in such a way the supply is exhausted, one can be said to have gone through the goods. I will refer to this as the ‘use-up’ sub-sense. An example of an abstract use of this sub-sense is given below. 245. Clearly Marty was a back number now; Lori seemed to go through men like a knife through butter. (BNC) Note that this sub-sense depends on functional opposition: goods that one has gone through have been used. If the goods are lost or stolen, however, even though they are gone, one did not go through the goods. Additionally, it is important to mention that this sub-sense might be motivated or sanctioned by ThroughEB, rather than (or in addition to) Through1. However, in the corpus example above, the fact that a 68 Polysemy and the structure of English THROUGH knife going through butter is an instance of Through1 may suggest that (this use of) this sub-sense is also motivated by Through1. Similarly, if the entry into, existence within, and exit from a state or situation are scanned over time, the ‘experience’ sub-sense (as in I went through a lot of difficulties to get to this point) can be seen to be motivated.31 If states are containers, then it follows that activities, which occur over time, can be seen as containers as well (i.e. both states and activities can be represented by extended contiguous locations on a timeline), and we find instances of this in the data as well, as what I will refer to as the ‘perform/recite’ sub-sense (Violins, go through the first two movements again). This sub-sense is often used with the metonymy of a person’s attention, which progresses within the (spatio)temporal span of the Ground, for the person (We’re almost through the movie). Such usage-types may be iconically motivated, since the real world event involves application over time of attention within the medium of the physical manifestation such as the text, from one end to the other, opposite end. This is true for reciting (and reading) prose or poetry, listening to and performing music, and other types of performances and recitations, which may be structured recursively, as one reads from one end of a line to the other, and from one end of a page to the other, and from one end of a book to the other. Many uses could belong to the ‘perform/recite’ sub-sense or the ‘finish’ sub-sense, such as I got through all of Plato/the play in one summer/week; I do not see a principled way of distinguishing. Another sub-sense means to ‘penetrate through understanding’, as in She saw through his façade. This may only occur with verbs of perception, but other contexts could possibly suggest such an interpretation (I had to break through my initial false impressions to arrive at a deeper understanding). The difference between the ‘penetrate’ and ‘via’ sub-senses is typically clear from the larger context – is the focus on the crossing of the final boundary in order to reach the Goal, as in ‘penetrate’, or does one go through the Ground in order to get to the Goal, as with ‘via’? This difference is not about the semantics of through, per se, but rather the available semantics of through utilized (sometimes imaginatively) in combination with other elements, such as verbs of perception, within the overall linguistic and non-linguistic context. The final sub-sense of Through1 I will discuss is what I call the ‘list’ use, which involves progression within and to the end of a sequence, as in We went from A through M this morning, so we may finish by the end of the day. I discussed the pattern Monday through Friday in Section 3, describing it as N1 through N2 in which N1 refers to the entrance to the Ground, and N2 the exit, and both N1 and N2 are words expressing time, noting that this sub-sense of Through1 does not exist in the BNC, though it seems frequent in North American English. This is a specific instance of the ‘list’ sub-sense, which is only extremely marginally represented in the BNC itself. For instance, “A through Z” does not exist in the corpus, and searches for “A through” return 57 results, of which nearly all are adjectival uses of 1, the lone exception being in a scientific paper. Searches for “1 through” return just 31 With the use of metaphors such as emotions are physical objects and time is space. 69 three tokens, of which two should be considered instances of the ‘list’ sub-sense, and both are from the scientific portion of the corpus. (For a comparison, searches for “1 through” in the COCA, currently 4.5 times the size of the BNC, return 341 tokens, of which 99 of the first 100 were judged to be instances of the ‘list’ sub-sense.) This sub-sense involves a fictive operation that Langacker (2005) calls ‘sequential examination’, referring to “individually looking at the members of a group, one by one, until they have all been examined (like checking a stack of letters to make sure they all have stamps).” (Langacker 2005:192). This reveals one connection between fictivity and metaphor – it is only possible to use through in such cases because a sequence can be conceptualized as a container (choose any number in the sequence; Stop me in the middle if you’ve heard this). Only if speakers are first able to use metaphor in order to view something as a container can it then become the Ground of a fictive scan. The ‘list’ use commonly occurs with mentions of time (from the sixties through (to) the eighties). There are cases in which it is not clear if the ‘list’ or ‘finish’ sub-sense is a more likely (or the sole) motivation for a use (The problems continued through the rest of the week). 4.3. Ground-prominent ThroughEB (GP.EB) There are uses of EB in which the NP expressing the Ground is prominent, as in the second of each (minimal) pair of examples: 240a) I’m most of the way through the movie, but you can watch the end with me. 240b) The movie is most of the way through, so let’s just watch the end, okay? 241a) We are almost through January of 2010…(enTenTen12) 241b) November is almost through and we are fast approaching Christmas… (enTenTen12) In the (a) examples, the Figure is expressed as subject, and the Ground in an NP immediately following through. In the (b) examples, the NP expressing the Ground is the subject of an intransitive clause, and the Figure is left unexpressed. In these examples, there is no real motion, but fictive scanning (Langacker 2005) and the metaphor ‘time is space’ play roles in creating a sense of motion. The ego-moving metaphor, in which the “ego” progresses along a timeline toward the future, motivates the (a) examples, and the time-moving metaphor, in which time moves past the (stationary) ego like a river motivates the (b) examples (Lakoff and Johnson 1980, Boroditsky 2000). More examples are given below: 242. Acquisitions are normally kept secret until the deal is almost through. (enTenTen12) 243. …currently working on my first novel, which is half way through. (enTenTen12) 244. When the day is through, prepare a new list… (enTenTen12) 70 Polysemy and the structure of English THROUGH In these cases, to ask ‘What is the deal/novel half-way through?’ can only be interpreted as wordplay. This is because, in some sense, this usage-type is reminiscent of a passive construction (the analogy is particularly striking if one notes that Figure is often mapped to the subject, and Ground to direct object), in which the Ground/direct object is highlighted at the expense of the Figure/subject, which is downplayed (see e.g. Givón 2001 on the function of passives). It is also reminiscent of English verb alternations in transitivity such as I broke the stick vs. the stick broke (e.g. Levin 1993, Fillmore 1970), in which the patient / object of the transitive is used as the subject of the corresponding intransitive. Rather than a patient being highlighted, however, it is the Ground which is emphasized (syntactically and semantically), and so I will call this sub-sense of EB ‘Ground-prominent EB’ (or GP.EB). This usage-type typically occurs as a particle with a ‘be’ verb, and the semantics express termination, and thus we must ask whether to treat this as a distinct primary sense altogether. After all, both syntax and semantics differ, as seen below: Table 6 – Contrasting EB and GP.EB Label Example, syntactic + semantic analysis Figure Path Ground Subj be P NP EB I’m most of the way through (with) the movie Ground Path/Goal Subj be P GP.EB The movie is most of the way through Constructional syntax + semantics Figure = subj. Ground = NP32 through = particle, verb = ‘be’ Ground= subj. through = particle verb = ‘be’ Observing the table above, it is clear that the syntax and the semantics of these two usage types differ.33 Therefore, we must ask if another primary sense should be posited. Here, I will choose not to do so, attributing the differences to the construction in which each is employed. My reasoning is as follows. The construction exemplified by the use of EB is a locative construction (cf. I’m in the house) with an overt expression of Path. Other examples of this locative + Path construction include he is not even through the first 32 I am not sure how to refer to this NP at the level of syntactic relations, to match subject. If through is considered a preposition that combines with the NP to form a PP, it is the object of a preposition. But through appears as a verb particle here, and with is optional (and not relevant to the Ground). Instead of the Ground, if we express the Goal, instead of an NP, we would get a PP, as in I am almost through to the surface. This should be considered a distinct construction, however, regardless of how we label the NP, as the one in which we are interested here specifies that the Ground is expressed as a post-particle NP. 33 An additional semantic distinction is that the example of EB requires either the metaphor ‘the person is located where their attention is focused’ or the metonymy ‘the person for their progress’, whereas this is not relevant to the interpretation of GP.EB. 71 chapter and she’s almost over her ex.34 The construction we see used to exemplify GP.EB, on the other hand, lacks a Figure entirely, and is a predicative modification construction (cf. the movie is quite boring) in which the subject is not the Figure (but is the Ground or Patient) (cf. dinner is nearly ready) and in which the modifier is a location on a Path (almost like your package is almost here, but instead of a Figure, the subject encodes the Ground, as in summer is mostly over). Thus, it seems that the semantics of the larger syntactic constructions are responsible for the differences in meaning that we had attributed to through. For these reasons, I will consider the construction to be the cause of the semantic differences, and treat GP.EB as a sub-sense of EB. Searching the BNC for tokens of this usage type, however, we find only cases like that exemplified below, which are probably not examples of this phenomenon, as whether the Figure or the Ground is “missing” from the sentence is ambiguous: 245. Just how black isn't fully apparent until about half-way through, when you discover that …(BNC) In this case, we don’t know if it can be paraphrased as “until we are about half-way through” or “until it is about half-way through” so this may be a case of ambiguity between multiple usage types. No clear cases of Ground-prominent EB were found in the BNC, either in the 500 tokens of through studied in-depth (Author 2007, in press) or in examinations of all results of searches for all combinations of is/are + already/almost/ part-way/Ø + through. The usage-type that does exist in the BNC, in which neither Figure nor Ground is overtly mentioned, also may be a bridge context creating the potential for the development of a unique sub-sense, which may have only been developed in some dialects of English. Another way that Ground-prominent EB is distinguished from EB is that it is possible to get a Goal with the former (c.f. the impossibility of including a Goal, Y, in prototypical examples of EB such as *I’m through with X to Y). This can be seen below. 246. The project is half way through completion (enTenTen12) 247. Check them regularly for wear. If the sole is nearly through to the rand (80% worn) send them to us. (Google) However, it is also true that such cases seem infrequent, and most instances of Ground-prominent EB can not take a Goal (for example, all BNC, COCA, enTenTen12, and Google queries for “summer/winter/night/day is almost/nearly/Ø through to” return no results). Therefore, I will consider this to show the semantic latitude of Ground-prominent EB, rather than giving evidence that it is a sub-sense of some other primary sense of through or a primary sense in its own right. 34 This sentence represents a specific instantiation of the larger locative + Path construction in which [Sub be over Person] expresses that the subject referent was, but is no longer, suffering from being emotionally tied to the Person. 72 Polysemy and the structure of English THROUGH 4.4. The ‘thoroughly’ sub-sense One usage-type can be paraphrased as ‘thoroughly’35 (I cooked it through, the fire warmed him through). Based on semantics alone, it is hard to determine if a ‘thoroughly’ usage is an instance of 1, or of ALL, or possibly of EB (though the lack of ‘be’ verb makes this doubtful). Is the process of “cooking something through” conceptualized as an instance of one of these primary senses, and if so, how can we tell? Semantically, going from one side to the other (i.e. 1) seems as possible to be relevant as progressing until the end (i.e. EB) or existing everywhere within (i.e. ALL). Significantly, all such uses are particles, suggesting that the ‘thorough’ sub-sense should be characterized as a sub-sense of either EB or ALL, but not of 1. It includes a usage type in which the Ground is in focus as the subject, as we saw with GP.EB above, exemplified below. 248. the liquid has evaporated and the liver is cooked through but still tender (enTenTen12) 249. I found myself warmed through with happiness (enTenTen12) 250. It was crooked on leather hinges that seemed about to either crack of (sic) rot through. (COCA) This is similar to GP.EB, but we don’t get a ‘be’ verb, and the semantics are probably closer to those of ALL than EB. I will tentatively consider this to be motivated by ALL. V. Conclusion In this article, I have sketched a picture of the four primary senses of through. I have given syntactic and semantic evidence for the independence of each primary sense, as well as describing a plausible motivation for the synchronic relations between these primary senses. This represents my effort to engage in a lexical semantic analysis of sense differentiation grounded in empiricism and explicit, clear definitions for the purpose of ensuring that language users are actually employing the senses described in the analysis. By defining polysemy in a replicable way and grounding my definition in syntax and Blank's (2003) taxonomy of motivated relationships, I have successfully resolved the problem of not knowing if linguists and speakers are making use of the same senses. By using my strict definition of polysemy, I have assured that the (primary) senses of a polysemous lexeme are differentiated by speakers, and therefore provided a way for linguists to be certain that a word is polysemous in a way that reflects linguistic behavior. By employing Blank's taxonomy, I have constrained the types of motivated relationships holding between senses of a polyseme. In doing so, I discovered four primary senses of through that are empirically differentiated. 35 Through and thorough share a single root, according to the OED. 73 By requiring both syntactic and semantic differentiability in order to diagnose polysemy, I provided an empirical method of sense differentiation based on speakers’ behavior. The clear and well-motivated results give support to the methods employed. I performed this analysis without implicit acceptance of the correctness fallacy – without requiring that all uses be motivated by exactly one primary sense, for example. This approach can be seen in part as an argument against overly simplistic analyses of polysemy. Even syntactically defined senses, which are presumably more deeply incorporated into speakers’ linguistic routines than those that show no syntactic distinctions, exhibited a high degree of syntactic and semantic overlap, ambiguity, and ‘underspecifiability’ (Author 2007). The four primary senses of through can be seen as extremes or prototypes of gradient phenomena that are reflected in the grammar of English. The structure of the semantic architecture of through and the relation of its semantics to the possibilities for syntactic realization leads naturally to cases of overlap or ambiguity between primary senses. Significantly, I have demonstrated an approach to polysemy that resolves the age-old issues plaguing polysemy research, and employed it to good effect in the analysis of through. Bibliography Author. 2003. On the Psychological and Linguistic Reality of Semantic Networks: A Cognitive and Semantic Analysis of English through including a Crosslinguistic Study of Japanese tooru, nukeru, and toorinukeru. MA Thesis, University of Oregon. Author. 2007. An Empirical Study of English through: Lexical Semantics, Polysemy, and the Correctness Fallacy. PhD Dissertation, University of Oregon. Author. 2008. “‘Unspecification’ and polysemy.” Presentation at the Language, Cognition, and Communication (LCC) Conference, University of Brighton, UK. Author. 2009. Cross-linguistic support for polysemy analyses: The case of English through and three Japanese verbs (tooru, nukeru, and toorinukeru). In Proceedings of the Ninth Annual Meeting of the Japanese Cognitive Linguistics Association, 2009, pp.34-38. Author. 2010a. On the (im)possibility of translation. Kyudai Eibungaku 52, pp. 77-97. Available on the web at http://catalog.lib.kyushu-u.ac.jp/handle/2324/20044/p077.pdf Author. 2010b. The IN/ON continuum and three Japanese verbs. Kyushu University Papers in Linguistics 31, pp. 89-108. Available on the web at http://www2.lit.kyushu-u.ac.jp/~linguist/kupl/doc/KUPL31_Benom.pdf 74 Polysemy and the structure of English THROUGH Author. 2010c. Polysemy, ‘unspecification’, and good-enough representations. Presentation at Conceptual Structure, Discourse, and Language. University of California, San Diego, USA. (Poster presentation.) Author. 2012. The semantics of some verbs of separation in Japanese. Kyushu University Papers in Linguistics 33, pp. 27-52. Available on the web at http://www2.lit.kyushu-u.ac.jp/~linguist/kupl/doc/KUPL33_benom.pdf Author. 2014. English through and the gradience of force dynamics. International Journal of Cognitive Linguistics, vol 5 no 1 pp. 29-51. Author. In press. The Correctness Fallacy and Lexical Semantics. Kyushu University Papers in Linguistics 35. Azuma, T., and G. C. Van Orden. 1997. Why SAFE is better than FAST: The relatedness of a word’s meanings affects lexical decision times. Journal of Memory and Language 36: 484– 504. Barsalou, L.W. 1987. The instability of graded structure in concepts. In U. Neisser, ed., Concepts and conceptual development: Ecological and intellectual factors in categorization, 101-140. New York: Cambridge University Press. , . Beitel, R., C. Schreiner , S. Cheung X. Wang , and M. Merzenich 2003. Reward- dependent plasticity in the primary auditory cortex of adult monkeys trained to discriminate temporally modulated signals. Proceedings of the National Academy of Science USA, 100(19): 11070-5 Blank, A. 1999.Why Do New Meanings Occur? A Cognitive Typology of the Motivations for Semantic Change. In Blank and Koch, eds., Historical semantics and cognition, 61-89. Berlin, New York: de Gryuter. Blank, A. 2003. Polysemy in the lexicon and in discourse. In B. Nerlich, Z. Todd, V. Herman, and D.C. Clarke, eds., Polysemy. Flexible Patterns of Meaning in Mind and Language, 276-93. Berlin, New York: Mouton de Gruyter. Boroditsky, L. 2000. Metaphoric Structuring: Understanding time through spatial metaphors. Cognition 751: 1-28. Bowerman, M. and E. Pederson. 1992. Crosslinguistic perspectives on topological spatial relations. Paper presented at the annual meeting of the American Anthropological Association, San Francisco, December. Bowerman, M. and E. Pederson. ms. INwards from ON and ONwards from IN: The crosslinguistic categorization of topological spatial relationships. Bréal, M. 1897. Essai de sémantique. Brionne: Monfort. Brugman, C. M. 1981. The story of OVER. MA thesis, University of California: Berkeley. Brugman, C. M. 1988. The story of over: polysemy, semantics, and the structure of the lexicon. New York: Garland. 75 Brugman, C. and G. Lakoff. 1988. Cognitive Topology and Lexical Networks. In S. I. Small, G. W. Cottrell and M. K. Tanenhaus, eds., Lexical Ambiguity: Perspectives from Psycholinguistics, Neuropsychology and Artificial Intelligence, 477-508. San Mateo, California: Morgan Kaufmann. Cann, R., 1993. Formal Semantics. Cambridge University Press, Cambridge. Chomsky, N. 1995. The minimalist program (Vol. 28). Cambridge, MA: MIT press. Clark, H. H. 1973. Space, time, semantics, and the child. In T. E. Moore, Cognitive development and the acquisition of language, New York: Academic Press. Coventry, K. and S. Garrod. 2004. Seeing, Saying, and Acting: The Psychological Semantics of Spatial Prepositions. Great Brittan, USA, and Canada: Psychology Press. Croft, W. 1998. Linguistic evidence and mental representations. Cognitive Linguistics 9: 151-173. Croft, W. 2001. Radical Construction Grammar: Syntactic Theory in Typological Perspective. Oxford: Oxford University Press. Croft, W. 2006. On explaining metonymy: comment on Geeraerts and Piersman, "Metonymy as a prototypical category". Cognitive Linguistics 17, 317-326. Croft, W. and D.A. Cruise. 2004. Cognitive Linguistics. Cambridge: Cambridge University Press. Cruse, D. A. 1986. Lexical semantics. Cambridge, England: University Press. Cruse, D. A. 2000. Meaning in language: An introduction to semantics and pragmatics. New York: Oxford University Press. Cruse, D.A. 2002. Paradigmatic Relations of Inclusion and Identity III: Synonymy. In D.A. Cruse, F. Hundsnurscher, M. Job and P.R. Lutzeier, eds., Lexicology: An International Handbook on the Nature and Structure of Words and Vocabularies, 485-497. Berlin: Walter de Gruyter. Cuyckens, H. 1993. Spatial prepositions in French revisited. Review article on C. Vandeloise (1991) Spatial prepositions: A case study from French (Chicago : The University of Chicago Press). Cognitive Linguistics 4 (3): 291-310. Cuyckens, H. 1995. Family resemblance in the Dutch spatial prepositions door and langs. Cognitive Linguistics 6 (2/3): 183-208. Cuyckens, H. , D. Sandra and S. Rice. 1997. Towards an empirical lexical semantics. In B. Smieja and M. Tasch eds., Human Contact Through Language and Linguistics, 35- 54. Frankfurt: Peter Lang. Davies, Mark. (2008-) The Corpus of Contemporary American English: 450 million words, 1990-present. Available online at http://corpus.byu.edu/coca/. 76 Polysemy and the structure of English THROUGH Deane, P. D. 1992. Polysemy as the Consequence of Internal Conceptual Complexity: The Case of Over. Proceedings - Eastern States Conference on Linguistics 9: 32-43. Deane, P. D. 1996. On Jackendoff’s Conceptual Semantics. Cognitive Linguistics 7-1: 35-91. Del Corro, A. 1990. An analysis of polysemy in Kapampangan and Japanese, using the systems correspondence theory. Studies in Philippine Linguistics 8-1: 52-76. DeLancey, S. 1984. Agentivity in syntax. Chicago Linguistic Society Parasession on Agentivity and Causation. CLS 21. Dewell, R. B. 1994. Over again: Image-schema transformations in semantic analysis. Cognitive Linguistics 5–4: 351–380. Evans, V. 2004. The Structure of Time: Language, Meaning and Temporal Cognition. Amsterdam: John Benjamins. Evans, V. 2005. The meaning of time: polysemy, the lexicon and conceptual structure. Journal of Linguistics 41: 33–75. Evans, V. 2006. Lexical concepts, cognitive models and meaning-construction. Cognitive Linguistics 17–4: 491–534. Evans, V. 2009. How words mean: lexical concepts, cognitive models, and meaning construction. London: Oxford University Press. Evans, V. and A. Tyler. 2004. Rethinking English "Prepositions of Movement": The Case of To and Through. In H. Cuyckens, W. de Mulder and T. Mortelmans, eds., Adpositions of Movement, 247-270. Amsterdam: John Benjamins. Fauconnier, G. and M. Turner. 2002. The Way We Think: Conceptual Blending and the. Mind’s Hidden Complexities. New York: Basic Books. Feist, M. I. 2000. On in and on: An investigation into the linguistic encoding of spatial scenes. Ph.D. dissertation, Northwestern University. Fillmore, C. J. 1970. "The Grammar of Hitting and Breaking." In R. A. Jacobs & P. S. Rosenbaum (Eds.), Readings in English Transformational Grammar (pp. 120-133). Waltham, Masschusetts: Ginn and Company. Fillmore, C. J. 1982. Frame semantics. In The Linguistic Society of Korea, eds., Linguistics in the Morning Calm, 111-137. Seoul: Hanshin. Fillmore, C. J. 1985. Frames and the semantics of understanding. Quaderni di Semantica 6-2: 222- 254. Fillmore, C. J. and B. T. Atkins. 1992. Towards a Frame-based organization of the lexicon: the semantics of RISK and its neighbors. In A. Lehrer and E. Kittay, eds., Frames, Fields, and Contrasts: New Essays in Semantics and Lexical Organization, 75-102. Hillsdale: Lawrence Erlbaum. 77 Frazier, L., and K. Rayner. 1990. Taking on semantic commitments: Processing multiple meanings vs. multiple senses. Journal of Memory and Language 29: 181-200. Geeraerts, D. 1993. Vagueness's puzzles, polysemy's vagaries. Cognitive Linguistics 43, 223-272. Geeraerts, D. 1994. "Lexical Semantics." In R. Asher, ed., The encyclopedia of language and linguistics, 2160-2164. Oxford: Pergamon. Gibbs, R. and T. Matlock. 1999. Psycholinguistics and mental representations. Cognitive Linguistics 10-3: 263-269. Gibbs, R. W., and Matlock, T. 2001. Psycholinguistic perspectives on polysemy. In H. Cuykens, and B. Zwada, eds., Polysemy in cognitive linguistics. Amsterdam: John Benjamins. Givón, T. 2001. Syntax. 2 vols. Amsterdam: John Benjamins. Glaser, B., and Strauss, A. 1967. The Discovery of Grounded Theory. Aldine Publishing Company, Hawthorne, NY. Goddard, C. 2000. Polysemy: A problem of definition. In Yael Ravin and Claudia Leacock, eds., Polysemy: Theoretical and Computational Approaches, 129-151. Oxford: Oxford University Press. Goldberg, A. E. 1995. Constructions: A construction grammar approach to argument structure. Chicago: University of Chicago Press. Goldberg, A. 2006. Constructions at work. Princeton University Press. Gries, S.Th. 2006. Corpus-based methods and cognitive semantics: the many meanings of to run. In S. T. Gries and A. Stefanowitsch, eds., Corpora in cognitive linguistics: corpusbased approaches to syntax and lexis, 57-99. Berlin, Heidelberg, New York: Mouton de Gruyter. Gries, S. Th. (2010). Behavioral Profiles: a fine-grained and quantitative approach in corpus-based lexical semantics. The Mental Lexicon, 5(3), 323-346. Gries, S. Th. To appear. Polysemy. In Ewa Dąbrowska & Dagmar S. Divjak (eds.), Handbook of Cognitive Linguistics. Berlin & Boston: De Gruyter Mouton. Gries, S.Th., B. Hampe, and D. Schönefeld. 2005. Converging evidence: Bringing together experimental and corpus data on the association of verbs and constructions. Cognitive Linguistics 16: 635-676. Gries, S.Th., B. Hampe, and D. Schönefeld. 2010. Converging evidence II: more on the association of verbs and constructions. In J. Newman and S. Rice, eds., Experimental and empirical methods in in the study of conceptual structure, discourse, and language. Stanford, CA: CSLI, 59-90. Gries, S. Th. and A. Stefanowitsch. 2004. Extending collostructional analysis: a corpus- based perspective on 'alternations'. International Journal of Corpus Linguistics 9-1: 97-129. 78 Polysemy and the structure of English THROUGH Hallan, N. 2001. Paths to prepositions? A corpus-based study of the acquisition of a lexico-grammatical category. In J. Bybee and P. Hopper, eds., Frequency and the Emergence of Linguistic Structure, 91-123. Amsterdam: Benjamins. Hampe, B., ed. 2005. From Perception to Meaning: Image Schemas in Cognitive Linguistics. Berlin: Mouton de Gruyter. Herskovits, A. 1986. Language and spatial cognition: An interdisciplinary study of the prepositions in English. Cambridge, England: Cambridge University Press. Jackendoff, R. 1997, The Architecture of the Language Faculty. Cambridge, MA: MIT Press. Jackendoff, R. 2002. English Particle Constructions, the Lexicon, and the Autonomy of Syntax. In N. Dehé, R. Jackendoff, A. McIntyre, and S. Urban, eds., Verb-Particle Explorations, 67-94. Berlin: Mouton de Gruyter. Jakubíček, M., A. Kilgarriff, V. Kovář, P. Rychlý, V. Suchomel. 2013. The TenTen Corpus Family. Paper presented at the 7th International Corpus Linguistics Conference, Lancaster, July 2013. Jespersen, O. 1922. Language: its nature, development, and origin. London: George Allen and Unwin Ltd. Johnson, M. 1987. The body in the mind: The bodily basis of meaning, imagination, and reason. Chicago: University of Chicago Press. Kilgarriff, A. 1992. Polysemy. PhD. Thesis. University of Sussex. Kilgarriff, A. 1993. Dictionary Word Sense Distinctions: An enquiry into their nature.” Computers and the Humanities 26 1-2: 365-387. Kilgarriff, A. 1997. I don’t believe in word senses. In Nerlich, Todd, Herman and Clarke, eds., Polysemy: Flexible patterns of meaning in language and mind, 361-392. Berlin: Walter de Gruyter. Klepousniotou, E. 2002. The processing of lexical ambiguity: Homonymy and. Polysemy in the Mental Lexicon. Brain and Language 81: 205-223. Klepousniotou, E., G.B. Pike, and V. Gracco. 2005. An ERP investigation of meaning activation patterns of homonymous and polysemous words. Poster presentation at BBCS/SCCC 15th annual meeting, Montréal, Québec, July 14-17 2005. Klepousniotou, E., G. B. Pike, K. Steinhauer, & V. Gracco. 2012. Not all ambiguous words are created equal: An EEG investigation of homonymy and polysemy. Brain and language, 123(1), 11-21. Klepousniotou, E., Debra Titone, and Carolina Romero. 2008. Making Sense of Word Senses: The Comprehension of Polysemy Depends on Sense Overlap. Journal of Experimental Psychology: Learning, Memory, and Cognition, Vol. 34, No. 6, 1534–1543. 79 Kövecses, Z. 2002. Metaphor. A Practical Introduction. Oxford and New York: Oxford University Press. Kreitzer, Anatol. 1997. Multiple levels of schematization: a study in the conceptualization of space. Cognitive Linguistics 84: 291–325. Kristoffersen K. E. 2001. Semantic Structure of the Norwegian Preposition mot. Nordic Journal of Linguistics 24-1: 3-28. Lakoff, G. 1987. Women, Fire, and Dangerous Things. Chicago: University of Chicago Press. Lakoff, G. and M. Johnson. 1980. Metaphors We Live By. Chicago: University of Chicago Press. Lakoff, G. and M. Johnson. 1999. Philosophy in the Flesh. New York: Basic Books. Lakoff, G. and M. Turner, 1989. More Than Cool Reason. Chicago: University of Chicago Press. Langacker, R. 1987. Foundations of Cognitive Grammar, vol. 1. Theoretical Prerequisites. Stanford: Stanford University Press. Langacker, R. 1991a. Foundations of Cognitive Grammar, vol. 2. Descriptive Application. Stanford: Stanford University Press. Langacker, R. 1991b. Concept, Image, and Symbol: The Cognitive Basis of Grammar. Berlin and New York: Mouton de Gruyter. Langacker, R. 1999. Grammar and Conceptualization. Berlin and New York: Mouton de Gruyter. Langacker, R. 2005. Dynamicity, Fictivity, and Scanning. The Imaginative Basis of Logic and Linguistics. In Pecher and Zwaan, eds., Grounding Cognition: The Role of Perception and Action in Memory, Language, and Thinking, 164-197. Cambridge, MA: Cambridge University Press. Levin, B. 1993. English verb classes and alternations: A preliminary investigation. Chicago: University of Chicago Press. Lyons, J. (1977) Semantics. 2 vols. Cambridge: Cambridge University Press. Mandler, J. 2004. The Foundations of Mind: Origins of Conceptual Thought. Oxford: Oxford University Press. Mark, D. M., and Egenhofer, M. J., 1994. Modeling Spatial Relations Between Lines and Regions: Combining Formal Mathematical Models and Human Subjects Testing. Cartography and Geographic Information Systems, October 1994, v. 21, No. 4, pp. 195212. Martín, Ángeles. 2000. A cognitive approach to the polysemy of ‘through’ Estudios Ingleses de la Universidad Complutense 12 2000, 8: 11-38. 80 Polysemy and the structure of English THROUGH Menner, R. L. 1936. The Conflict of Homonyms in English. Language 12-4: 229-244 Murphy, G. L. 2010. What are categories and concepts? In D. Mareschal, P. C. Quinn, & S. Lea (Eds.), The making of human concepts. Oxford: Oxford University Press. Nerlich, B. (2003). Polysemy and flexibility: introduction and overview Brigitte Nerlich and David D. Clarke. Polysemy: Flexible patterns of meaning in mind and language, 142, 3. Nerlich, B., Todd, Z., Herman, V., & Clarke, D. D. (Eds.). 2003. Polysemy: Flexible patterns of meaning in mind and language (Vol. 142). Berlin, New York: Mouton de Gruyter. Nightingale, Stephen. 1999. Polysemy and homonymy in Japanese verbal alternations. PhD. Thesis, University of Edinburgh. Ochoa, T. 1998. The many senses of through. Unpublished manuscript, UC Berkeley. Panther, K. U., & Thornburg, L. L. (2002). The roles of metaphor and metonymy in English-er nominals. Metaphor and metonymy in comparison and contrast, 20, 279. Pustejovsky, James. 1998. The Generative Lexicon. Cambridge, MA: MIT Press. Queller, K. (2001). A usage-based approach to modeling and teaching the phrasal lexicon. Applied Cognitive Linguistics II: Language pedagogy, 55-83. Raukko, Jarno 2003. Polysemy as flexible meaning: Experiments with English get and Finnish pitää. In B. Nerlich, Z. Todd, V. Herman, and D. Clarke, eds., Polysemy: Flexible patterns of meaning in mind and language, 161-193. Berlin: Mouton de Gruyter. Ravin, Y. and C. Leacock. 2000. Polysemy: Theoretical and Computational Approaches. Oxford: Oxford University Press. Reddy, M. 1979. The Conduit Metaphor - A Case of Frame Conflict in Our Language about Language. In A. Ortony, ed., Metaphor and Thought, 284-324. Cambridge: Cambridge University Press. Rice, S. 1993. Far afield in lexical fields: the English prepositions. In M. Bernstein, ed., ESCOL '92, 206-217. Ithaca: Cornell University Press. Rice, S. 1996. Prepositional prototypes. In M. Pütz and R. Dirven, eds., The Construal of Space in Language and Thought, 135-165. Berlin and New York: Mouton de Gruyter. Rosch, E. 1978. Principles of Categorization. In E. Rosch and B. Lloyd, eds., Cognition and Categorization, 27–48. Hillsdale, N.J.: Erlbaum. Rudskoger, A. 1952. Fair, Foul, Nice, Proper: A Contribution to the Study of Polysemy. Stockholm: Almqvist and Wiksell. Ruhl, C. 1989. On Monosemy:A study in Linguistic Semantics. Albany: State University of New York Press. 81 Sandra, D. 1998. The study of lexical meaning in cognitive linguistics, cognitive psychology and psycholinguistics. In J. van der Auwera, ed., English as a human language: to honour Louis Goossens, 298-311. München, LINCOM Europa. Sandra, D., and S. Rice. 1995. Network analysis of prepositional meaning: mirroring whose mind - the linguist's or the language user's? Cognitive linguistics 6-1: 89-131. Stefanowitsch, A. 2003. Constructional semantics as a limit to grammatical alternation. In G. Rohdenburg and B. Mondorf, eds., Determinants of Grammatical Variation in English. Berlin: Mouton. Stefanowitsch, A., and S. Th. Gries. 2005. Register and constructional meaning: a collostructional case study. In G. Kristiansen and R. Dirven, eds., Cognitive sociolinguistics: language variation, cultural models, social systems. Berlin, Heidelberg, New York: Mouton de Gruyter. Suttle, Laura and Adele Goldberg. 2011. The partial productivity of constructions as induction. Linguistics 49–6 (2011), 1237–1269. Svartvik, J., & Quirk, R. (1980). A corpus of English conversation (Vol. 56). Studentlitteratur. Talmy, L. 1983. How Language Structures Space. Berkeley Cognitive Science Report No. 4, January. Talmy, L. 2000. Toward a cognitive semantics. 2 vol. Cambridge, MA: MIT Press. Taylor, J. 2002. Cognitive Grammar. Oxford: Oxford University Press. Taylor, J. R. 2003. Polysemy's paradoxes. Language Sciences 25: 637-655. Taylor, John. 2003. Cognitive models of polysemy. In:Polysemy: Flexible Patterns of Meaning in Mind and Language. B. Nerlich, Todd, V.Herman and D Clarke (eds.), Berlin: Mouton de Gruyter (2003), 31-47. Taylor, J. R. 2012. Contextual salience, domains, and active zones. Cognitive Pragmatics, 4, 151. Traugott, E. 1972. The History of English Syntax. New York: Holt, Rinehart and Winston. Tuggy, D. 1993. Conference on the relationship between linguistic and conceptual representation. Notes on Linguistics 60: 45-47 Tyler, A. and V. Evans. 2001. Reconsidering prepositional polysemy networks: The case of over. Language 77-4: 724-65. Tyler, A. and V. Evans. 2003. The Semantics of English Prepositions: Spatial Scenes, Embodied Meaning and Cognition. Cambridge: CUP. Vandeloise, C. 1990. Representation, prototypes, and centrality. In S. L. Tsohatzidis, ed., Meanings and Prototypes, 403-437. London and New York: Routledge. 82 Polysemy and the structure of English THROUGH Zwicky, A. M., and J. M. Sadock. 1975. Ambiguity tests and how to fail them. In J. Kimball, ed., Syntax and Semantics, 1-36. New York: Academic Press. 83

RELATED PAPERS

RELATED TOPICS

Log In

Polysemy and English THROUGH

Polysemy and English THROUGH

Related Papers

RELATED PAPERS

RELATED TOPICS