Printer Friendly

Too female to be ruthless and too pregnant to argue: semantic conflict and resolution in the [too ADJ to V]-construction/Too female to be ruthless i too pregnant to argue: rjesavanje znacenjskog konflikta u konstrukciji [too ADJ to V].

1. Introduction

Consider these naturally occurring instances of the English [too ADJ to V]-construction, all of which are retrieved from the 2011-section of the Corpus of Contemporary American English, or COCA for short (Davies 2014):

(1) If you're all too scared to go in, I guess that makes me the winner. (COCA 2011 FIC Scholastic)

(2) But I do think it is way too early to think that he is the only viable candidate. (COCA 2011 SPOK Fox_Sunday)

(3) The talent gap was simply too wide to overcome. (COCA 2011 NEWS SanFranChron)

If we take the presence of the degree modifier too to reflect that the adjectival position in the construction expresses SCALARITY, then scared (1), early (2), and wide (3) are all semantically compatible with the construction, because they arguably express gradable attributes. Now, consider this example:

(4) I was too female to be ruthless. (COCA 2011 FIC Bk:DeadMansSwitch)

This example is marked by semantic incompatibility between the absolute semantics of female and the construction itself. Female in (2) appears to be assigned scalar meaning in such a way that female, rather than referring to sexual gender, refers to CULTURAL GENDER and quantifies stereotypical behavioral patterns.

Semantic conflict resolution as such has been addressed before (Goldberg 1997; de Swart 2003; Michaelis 2004; Talmy 2000b: 323-336), and Michaelis (2004: 51) has suggested that an override principle applies in which the constructional semantics overrides the lexical semantics. However, construction-lexeme anomalies in [too ADJ to V], and the semantic resolution strategies that follow, constitute an avenue yet to be explored. The purpose of the present paper is therefore to provide semantic analysis of naturally occurring anomalous instances of [too ADJ to V] with a view to identifying and describing possible strategies of semantic resolution. Drawing on a collostructional analysis from a previous study (Jensen 2013), the present paper isolates eight non-gradable adjectives found among the least attracted lexemes in the ADJ-position. A questionnaire study of native speakers' resolution strategies is presented, in which the respondents were asked to offer their readings of naturally occurring anomalous instances of the eight adjectives. These readings serve as the basis for my analysis of resolution strategies. It should be noted that the empirical nature of the study does not as such hold any authoritative evidentiary value; on the contrary, it is intended to serve as input for empirically-based hypotheses pertaining to the discursive, semantic, and cognitive aspects of the construction which may be rigorously tested in future studies.

2. Construction, construal, and conflict

The broader theoretical framework of the present discussion is cognitive linguistics, in which it is held that semantics is tied in with general human cognitive structures and operations. Moreover, the present discussion presupposes that linguistic competence, like all other human knowledge, is experientially based and intertwined with general human cognition.

2.1 Grammatical constructions

The present discussion draws largely on the construction grammar principles put forth by Croft (2001, 2003, 2005) and Tomasello (2003) as well as the principles associated with usage-based linguistics more broadly (Hopper 1987; Kemmer & Barlow 2000; Croft & Cruse 2004: 291-327), all of which emphasize the experiential nature of language. Since our analysis takes a constructionist approach to language, an overview of the basic principles of construction grammar (Fillmore et al. 1988; Goldberg 1995; Croft 2001; Croft & Cruse 2004: 225-327) is appropriate. (1) The central unit of grammar in construction grammar is the construction, which is a pairing of form and conventionalized meaning. Construction grammarians envision grammar as consisting of networks of constructional templates that pair form and meaning and license instances of constructions in discourse. These networks are essentially organized like the prototype categories known from cognitive linguistics and other cognitive sciences (e.g. Rosch 1973, Lakoff 1987, Geeraerts 1997, Taylor 2003). They display prototype effects and vary in specificity, such that very schematic and very specific constructions and subconstructions may appear in a constructional network.

Recently, construction grammar has taken a turn in the direction of usage-based linguistics, and, in recognition of the intimate relation of mutual influence between language and discourse posed in usage-based linguistics (Hopper 1987; Kemmer & Barlow 2000, Croft & Cruse 2004: 291-327), Croft (2005: 274) suggests a rather general, discourse-based definition of a construction as "an entrenched routine ... that is generally used in the speech community ... and involves a pairing of form and meaning". This definition adds the dimension of the function of language as a means of communication in a speech community and includes the notion of convention as a socio-communicative phenomenon. This is the definition adopted in the present paper.

2.2 Construal and semantic conflict resolution in constructions

In their work on constructions, Stefanowitsch & Gries (2005: 4) propose the principle of semantic compatibility: "words can (or are likely to) occur with a given construction if (or to the degree that) their meanings are compatible". By the logic inherent in this principle, semantic conflict is likely to be found among the least frequent items in a construction. Given the preference for compatibility in construction-lexeme interaction, we can assume that processes of conflict resolution are applied in usage-events that lack semantic compatibility. Semantic compatibility, conflict, and resolution are all aspects of meaning construction in general and involve construal in a number of ways.

Verhagen (2007: 48-49) points to the fundamental nature of construal in meaning construction: "At a very elementary level, construal is a feature of the meaning of all linguistic expressions, if only as a consequence of the fact that languages provide various ways for categorizing situations, their participants, and features, and the relations between them". Most cognitive linguistic research into construal (e.g. Langacker 1987, 1990, 1991; Talmy 2000a, 2000b; Croft & Wood 2000; Croft & Cruse 2004: 40-74; Verhagen 2007) has focused on identification and classification of construal operations. While the classificatory systems and criteria for classification differ from study to study (Verhagen 2007: 54-56), it is commonly held that construal operations are reflections of general psychological processes (Croft & Cruse 2004: 45; Verhagen 2007: 76). Croft & Wood (2000) propose a model, revised in Croft & Cruse (2004: 40-73), of construal operations at play in meaning construction and general human cognition. Their model covers four major categories, each of which is based on an area of experience and its accompanying sets of cognitive abilities: attention/salience, judgment/comparison, perspective/situatedness, and constitution/ gestalt. Attention/salience subsumes construal operations that relate to the ability to distribute one's focus of attention on various details of a scene, such as metonymy, granularity of view, and selection of salient elements. Judgment/ comparison covers construal operations of comparison of experiences on the basis of similarities and differences, such as categorization and metaphor. Perspective/situatedness includes construal operations that enable people to relate to the scene, or context, in which they are situated. Constitution/gestalt subsumes construal operations which allow for the interpretation of the constitution of entities in terms of their physical shapes and spatio-temporal structures. (2)

Interaction between constructions and other linguistic units in discourse involves what Talmy (2000b: 323) calls multiple specification which, more generally, "is applied to the situation where a sentence, or other portion of discourse, provides two or more specifications of the characteristics of the same referent" and is, not surprisingly, commonplace in discourse. There are two logical outcomes of multiple specification: semantic compatibility and semantic conflict. When such conflict occurs, "various processes of conceptual reconciliation can come into play in an addressee under a general cognitive procedure of semantic resolution" (Talmy 2000b: 323). According to Talmy (2000b: 323-336), there are five superordinate cognitive strategies which may be deployed in interpreting instances of semantic conflict: 1) shifting, which involves semantic change in one form to make it more semantically aligned with the other forms; 2) blending, in which semantic features of the conflicting forms are combined into semantic hybrids; 3) juxtaposition, in which the conflicting schemata are activated simultaneously and actually foreground the conflict; 4) schema juggling, which applies to situations where addressees activate strings of various different cognitive schemata to make sense of the linguistic input; and 5) blockage, which is the rejection the linguistic input as meaningless or unacceptable. Semantic construction-lexeme conflict is held to be resolved in accordance with the override principle: "if a lexical item is semantically incompatible with its syntactic context, the meaning of the lexical item conforms to the meaning of the structure in which it is embedded" (Michaelis 2004: 51). Essentially, this means that the construction imposes construal operations upon the lexeme in question, causing the lexeme comply semantically with the construction.

In order for the concepts of conflict and resolution to have any validity, we must assume that linguistic units have what Talmy (2000b: 6-7) calls semantic basicness, and that a basic-divergent model applies in which there are basic uses and divergent uses of linguistic units. If one accepts that the symbolic structure of a linguistic unit conventionally involves semantic content and construal operations that configure the semantic content, then making a case for basicness is quite straightforward. Basicness is thus simply the conventional use and function of a linguistic unit, and divergence from basicness corresponds to unconventional use. Moreover, semantic basicness may be tied in with a prototype-based semantics (e.g. Geeraerts 1997), in which certain form-function combinations in a linguistic unit are considered 'more typical' or 'more normal' than others.

3. Method and data

The present study addresses interpretative strategies of resolution in anomalous instances of [too ADJ to V] in which there is semantic conflict between the lexeme in the ADJ-position and the construction itself. This, of course, requires the analyst to have some idea of what constitutes a 'typical' instance and an 'anomalous' instance of the construction. Jensen (2013), a previous study of the construction based on 2011-component of the COCA (Davies 2014), provides a ranked list of collostructional relations in the construction; see Stefanowitsch & Gries (2003, 2005; Gries & Stefanowitsch 2004a, 2004b) for detailed descriptions of the methodological affinities of collostructional analysis.

In the present study, we will use Jensen's (2013) findings as an indicator of typicality and anomaly, such that the most attracted items in the ADJ-position reflect typicality and the least attracted ones reflect anomaly. From the fifty least attracted items I selected the following eight adjectives:

(5) real, full, correct, impossible, Catholic, female, pregnant, innocent.

The following table accounts for their overall frequencies in the 2011-component of COCA as well as their collostruction strength and ranking, with the lowest rank corresponding to the lowest degree of attraction (thus real is less attracted to the construction than impossible which, in turn, is less attracted than innocent):
Table 1: Frequency data on the eight adjectives (Jensen 2013)

Lexeme       Overall frequency   Collostruction strength   Ranking

real               6130             0.840440686799456        49
full               4080             1.42785638901155         48
correct            1045             3.81690436003222         30
impossible         1161             3.61911360454218         32
Catholic            933              4.0313613348609         24
female             1162             3.61750159129566         33
pregnant            620              4.8146437700432         111
innocent            562             5.00482856112027          9


Each of the lexemes appears only once in the construction.

The questionnaire contains fifteen examples. Eight examples, one example of each of the eight adjectives in the construction, are anomalous. The rest of the fifteen examples are more typical usage-events. The eight anomalous examples are listed as examples (6), (7), (9), (12), (14), (15), (19), and (22) in this paper. The more typical examples were included in the questionnaire for the sake of variation and to prevent biased responses. All examples were imported from the 2011-component of COCA and are thus authentic examples of actual language use. The eight adjectives were selected for the questionnaire by these criteria:

* basicness of non-scalar semantics: the adjectives used in the questionnaire should have non-scalar, or at least non-obvious scalar semantics, as their basic meanings.

* variation: the adjectives used in the questionnaire should preferably be varied enough that the individual respondents may construct different kinds of readings so as to document as many resolution strategies and construal operations as possible; yet they should also overlap to some extent in order to allow for documentation of conflict resolution of similar anomalies.

* relatability of examples: the usage-events in which the construction-lexeme anomalies occur should be relatable enough that the respondents are likely to understand them. They should also be such that the respondents are likely to find them interesting enough to evaluate.

* time (and space): the questionnaire should not be too long, complicated and time-consuming for the respondents. I estimated that fifteen examples, eight anomalous ones and seven typical ones, would not be too time-consuming and that it would take fifteen to twenty minutes to complete the questionnaire. To this end, the anomalous examples included should, in addition to being relatable and interesting, also be fairly simple.

The relatability criterion of course largely rests on my own subjective anticipation of what I think people can relate to and find interesting; another researcher might have chosen different examples.

This being more of a preliminary study, I opted for a small number of respondents. Thirteen anonymous respondents participated. This obviously does not warrant statistical analysis, but it is suitable for qualitative analysis of their readings with a view to inferring applied construal operations and conflict resolution strategies. The respondent group encompasses native speakers of British, American, and Australian English and is fairly heterogeneous in terms of age, education, and gender. Some of the respondents' educational backgrounds have endowed them with fairly advanced metalinguistic understandings and terminologies. Other respondents have had no training in linguistics whatsoever.

The respondents' main task was simply to provide their interpretations of the adjective in that particular usage-event, describing in their own words what they thought the adjective in question meant. The questions were formulated as What does ADJ mean in this particular case (too ADJ to V). For instance, the respondents were shown (15) and asked What does 'female' mean in this particular case (too female to be ruthless)?, thus prompting them to provide an unprimed and free reading of too female to be ruthless. This is, of course, a fairly uncontrolled type of response, but I found that giving the respondents free reign in terms of interpretation was better than priming their answers in a specific direction. Moreover, since the point of the questionnaire is not to generate data for statistical analysis, there is no need for strictly operationalizable information. These native speaker evaluations guide my own application of construal operations in such a way that, in interpreting their interpretations, I could infer possible underlying semantic resolution processes and construal operations. (3)

4. A brief overview of [too ADJ to V] (4)

Formally, the constructional template consists of a lexically open adjectival head premodified by the degree modifier too, which serves as what Paradis (2000: 149) calls a booster. Boosters increase the degree of whatever attribute the adjective expresses (let us call this ADJNESS for short). In the case of too, the boosting function increases the degree of ADJNESS beyond the MAXIMAL THRESHOLD of the adjectival scale in question. Consequently, it also structures the scale as an upper closed scale (Kennedy & McNally 2005), since, in order to exceed the MAXIMAL THRESHOLD of a scale, the MAXIMAL THRESHOLD must logically also serve as the UPPER LIMIT of the scale. The ADJ-position is postmodified by a to-infinitive. in addition to the formal relation of postmodification, there is an underlying semantic relation of force-dynamics between the ADJ-position and the to-infinitive such that the degree of ADJNESS either blocks or enables the situation predicated by the infinitive verb.

Since the construction draws on scalarity of ADJNESS, items in the ADJ-position are also subject to scalar construal operations. The ADJNESS is construed as an upper closed scale and would be included under Talmy's (2000a: 64-66) semantic system of axiality. (5) Scalar construals fall under the heading of gestalt/constitution in Croft & Wood's (2000; Croft & Cruse 2004: 40-74) model of construal operations. Croft & Cruse (2004: 65) describe scalar construals as the application of a SCALE image schema "which provides a gradable dimension to a domain, which may or may not be measurable", and which is contrasted with an absolute structure. According to Johnson (1987: 122), "the SCALE schema is basic to the quantitative and qualitative aspects of our experience". The [too ADJ to V]-construction applies the SCALE image schema to the ADJ-position, imposing a scalar structure upon the lexeme that appears in it. One may argue that, in selecting a position on the SCALE as the MAXIMAL THRESHOLD, the construction also involves the attention/salience-based construal operation of profiling (Langacker 1987).

5. Conflict resolution strategies in construction-lexeme anomalies in [too ADJ to V]

In the following, I will provide a cognitive semantic analysis of construal operations in semantic conflict resolution in the eight adjectives when they appear in [too ADJ to V]. The analysis is, as mentioned, primarily qualitative and based on interpretation of the respondents' readings. For comparison, I retrieved, where relevant, further examples of these specific construction-adjective combinations from COCA in its entirety.

5.1 Real

The basic semantics of real revolves around the state of EXISTING AS A FACT OR TRUTH, which is an absolute relation and not a matter of degree. When real appears in a scalar adjectival construction such as [too ADJ to V], we can expect it to display deviance from its basic semantics due to the construction-lexeme semantic conflict:

(6) And the feelings, too, they were coming back, rather, memories of how it had felt to be so comfortable with someone, so loved and appreciated. She had begun to think of Delphine Crandall with a longing that seemed more than mere nostalgia. It was a longing that finally became too real to ignore. (COCA 2011 FIC Bk:SummerFriends)

Some of the respondents' evaluations of (6) seem to support my claim that it deviates from the perceived basic use of real. Evaluating the use of real in the construction, one respondent notes:

--"'Real' is one of those words that is non-quantifiable--either something is real or it is not. Something cannot be too real. This, to me, conveys a suggestion of bad writing in some trashy romantic novel."

In addition to suggesting that this use of real belongs to a low quality literary genre, the respondent refers to the non-quantifiable nature of real, pointing out the relation of absolute opposition between real and not real. The same respondent moves on to suggesting some alternatives:

--"If I didn't want to convey the sense of the romantic novel, a better word might be 'apparent' or even 'strong'."

Another respondent similarly notes the non-standard use of real here, suggesting an alternative way of expressing the perceived content of the sentence:

--"The sentence doesn't read right to me. I'd have written it as "too much to ignore" instead. Saying "real" implies that feeling nostalgia isn't a real feeling a person can get."

While the latter respondent opts for much, which clearly has a scalar function, the former suggests apparent or strong. Strong arguably is a scalar adjective, as strength is a matter of degree. This suggests some degree of compatibility between the basic semantics of strong and the scalar semantics of the construction. Apparent, on the other hand, cannot be said to be a scalar adjective, but, unlike real, apparent is semantically characterized by a lower degree of certainty in terms of the factual status of what the adjective describes.

The absolute STATE OF EXISTING AS A FACT OR TRUTH is converted into a scale via the construction's imposition of the scale image schema upon the semantics of real. Note also how, in part thanks to the discourse space (6) already established, real expresses EMOTIONAL URGENCY rather than the absolute existing AS A FACT OR TRUTH. Most respondents made this observation. Here are some of their interpretations of the use of real in this sentence:

--"Important or overbearing. The longing began to impact her life in some way."

--"The real in this case means that the female protagonist's character is perceiving her longing as an almost physical manifestation."

--"Too real in this case means that 'She' feels that her thoughts of Delphine are more than just thinking about fond memories and possibly venturing into a sexual, sensual, or love desire."

With scale-oriented descriptors such as more than just thinking, almost a physical manifestation, and important (along with the suggestions of using strong and much), we can infer from the respondents' scalar readings of too real to ignore that EMOTIONAL URGENCY is arguably more likely to be construed in a scalar fashion with degrees of urgency or relevance.

In such readings, it seems that a more or less complete shift is involved in which the binary opposition between TRUE and NOT TRUE is stretched

into a scale, stretching being a resolution process that Talmy (2000b: 326-327) describes as an extension of a semantic feature of an open class unit. Such an operation would fall under Croft & Wood's (2000; Croft & Cruse 2004: 40-74) category of gestalt/constitution. Note that, it could be argued that the TRUTH feature of real is retained in a weakened form, which may be what enables the comparison of longing to a physical manifestation and the description of the longing having gone beyond mere thinking; indeed, one respondent simply described real in too real to ignore as "tangible". A number of other respondents also homed in on this in their readings, arguing that too real to ignore describes the longing as having become "'an actuality'", "true feelings" (as opposed to "fantasy or false desire"), and as "decisive". Such readings suggest that the respondents interpret real in too real to ignore as a gradable expression of RELEVANCE TO ACTUAL REALITY, signaling a very high degree of RELEVANCE on a scale imposed on real via construal operations that include the scale image schema and probably extension. The scalar construal is then guided by the concepts already activated in the discourse space, such as LONGING and NOSTALGIA.

An alternative analysis would involve disambiguation. In this analysis, it is assumed that real is polysemous, and that one of its senses is that of RELEVANCE. The conflict resolution process then consists in the conceptualizer selecting the RELEVANCE sense over the more basic EXISTING AS A FACT OR TRUTH sense. In this process, the RELEVANCE sense is selected because its more scalar nature is compatible with the constructional semantics of [too ADJ to V] (and the current discourse space). As we will see later, disambiguation does seem to figure in semantic conflict resolution, but it requires the sense selected in the construction to be a conventionalized sense of the adjective in question. Real definitely covers senses that are related to RELEVANCE, such as IMPORTANCE, as in the real problem, but it has yet to be documented that RELEVANCE is a conventionalized sense of real (and whether it is limited to real in predicative position or perhaps real in scalar constructions). Hence, this alternative analysis is proposed tentatively. However, if it turns out to be verifiable, it does not rule out stretching as a meaning construction process. it might be the case that frequent application of stretching as a means resolving conflicts between real and scalar constructions would result in conventionalization of stretching as part of the RELEVANCE sense.

5.2 Full

The basic semantics of this adjective expresses the state of a CONTAINER containing the largest amount of CONTENTS possible within its capacity. While fullness [left and right arrow] emptiness is a scale, the concepts are themselves not scalar because they constitute the endpoints of the scale. Thus, full expresses a state that is itself not gradable. Degrees of FULLNESS above FULLNESS are logically not possible. Degrees of FULLNESS below FULLNESS are also impossible, because then the degree of CONTAINMENT would no longer be FULL. We see an example of the adjective in the construction in (7) below:

(7) He gave John the briefest of nods and picked up the fried chicken breast on his plate. Henry said he believed he would walk over to J.W.'s house to stretch his legs. And did Tommy want to join them? Tommy was too full to walk but might meet them there later. (COCA 2011 FIC BlackRenaissance)

Note that full in (7) expresses a metaphorical specification of the FULL-TO-CAPACITY sense, as it describes the STATE OF HAVING EATEN A LOT OF FOOD, which is one of the specialized senses conventionally associated with full.

Not surprisingly, all respondents arrived at this reading. some respondents offered readings in which full was conceptualized in terms of another physical state--namely, BLOATEDNESS as a result of eating. While FULLNESS is not readily an scalar state, BLOATEDNESS is arguably a matter of degree. The fact that these respondents use a gradable adjective in formulating their readings suggests the application of scalarity to full in this construction. A number of readings focus on overeating, suggesting construals of transgression of a MAXIMAL STATE of FULLNESS:

--"'Full' here refers to the feeling of being full of food and unable to eat more. It suggests that Tommy has overeaten."

--"full means 'stuffed' from overeating - both words work in this instance"

--"full-up, eaten too much"

Such interpretations still involve the application of the scale image schema and apply an upper closed scale limit, because it would logically not be possible to exceed a maximal limit of fullness within any other structure than a scalar one. What indicates that these respondents apply scalar readings is the use of the lexeme overeat and the use of the degree modifier too. Finally, some respondents singled out the force-dynamic semantics of the construction by relating Tommy's overeating to its preventive effect on his ability to walk:

--"Tommy gorged on a Thanksgiving feast and couldn't walk because his stomach was full of food."

--"Tommy ate so much that he became immobile"

These readings seem to indicate that, in (7), full is construed as a closed scale whose MAXIMAL LIMIT is transgressed. However, the upper limit is relative to walk, understood such that the scale of FULLNESS itself may technically continue beyond the point of being too full to walk. This way, this understanding of full, as opposed to the one evoked in the three previous readings, does not strictly mean FULL TO CAPACITY, but rather that one has consumed so much food that one is fully satiated. This does not mean, however, that there is no more room in one's stomach for more food, and SATIATION is arguably more likely to be a scalar affair than actual FULLNESS. Note also that, in interacting with the force-dynamic semantic features of the construction, this reading of full may also be argued to involve CAUSE-FOR-EFFECT metonymy within the SATIATION domain. Over-satiation from eating amounts to more than just FULLNESS FROM FOOD. It also involves other physical states and processes which follow from the state of FULLNESS FROM FOOD, many of which can be felt, such as discomfort, tiredness, and abdominal pain. By this logic, it is arguably these manifestations, or EFFECTS, of FULLNESS FROM FOOD that prevent Tommy from walking in (7).

The variability in readings among the respondents suggests, not surprisingly, that interlocutors draw on different strategies in semantic conflict resolution. What seems to have happened in the particular case of (7) is that the respondents have selected one or more of the semantic features of full and applied their semantic conflict resolution strategies to these in slightly different ways. Some have construed Tommy's FULLNESS in terms of the physical state of BLOATEDNESS which may--metaphorically or physically--follow from overeating, while others focused on the overeating itself, and others again on the force-dynamic relation between the ADJ- and the V-positions in the construction.

The application of the scale image schema associated with the construction results in interpretations of the FULLNESS as an absolute point on the scale of CONTAINMENT, such that the notion of FULLNESS is extended into being a scale itself in a process that may be classified as what Talmy (2000a: 62) calls magnification or adoption of a proximal perspective. Magnification is best classified as a subtype of scalar adjustment (Croft & Cruse 2004: 52) which results in a fine-grained perspective on what is viewed. In Croft & Wood's (2000; Croft & Cruse 2004: 40-74) model, scalar adjustment is subsumed under the attention/ salience category, and, consequently, in cases where FULLNESS is stretched from a point on a scale into a scale itself, attention/salience and gestalt/constitution are both at play. It is likely that the FULLNESS scale is expanded from the absolute point of FULLNESS and downwards. Accordingly, in cases like (7), too full expresses a very high degree of FULLNESS, but not necessarily FULL TO CAPACITY. This process of magnification works alongside metonymy in that the stretching is made possible by using the upper end of the CONTAINMENT scale with reference to a portion of the scale just below FULLNESS. This arrangement is then metaphorically projected onto the HUMAN BODY in the construction of the SATIATION or BLOATEDNESS construals.

SATIATION-FROM-FOOD is a conventionalized and specialized sense of full, in which the MAXIMAL STATE OF A CONTAINER scenario is projected onto the human body, as it were. Theoretically, other senses of full may behave differently in the construction. In the following example, a specialized and more scalar sense of full applies:

(8) Sure he'd seen women. Plenty of them. But nothing like this roamed the likes of Bodie. Lips, a shade too full to be fashionable, and high cheekbones accented a pair of piercing whiskey-colored eyes that stole his breath away. (COCA 2012 Bk:Slayer)

In this case, another scalar sense of full applies--namely, LIP VOLUME, which typically applies to full when collocating with lips. I would argue that in a case like (8) there is thus no semantic conflict between full and the construction.

5.3 Correct

Correct and its antonym incorrect (and other lexemes expressing INCORRECTNESS) stand in a relation of absolute opposition, and CORRECTNESS itself is not typically a matter of scale. Correct is thus very likely to be in semantic conflict with [too ADJ to V]. Interestingly, correct does not appear in the construction in COCA without collocating with politically:

(9) America today is too politically correct to acknowledge the reality of Islamic fanaticism, Kharoba said. (COCA 2011 MAG WashMonth)

(10) New York City is too politically correct to be racially profiling in any way. (COCA 2005 SPOK CNN_Politics)

(11) You're caught in the emotional trap of a society too politically correct to validate one perception of reality over another. (COCA 2007 FIC Analog)

Unlike CORRECTNESS as such, POLITICAL CORRECTNESS is not in conflict with the scalar semantics of scalar adjectival construction, as POLITICAL CORRECTNESS is au fond a matter of social behavior. In this case, then, correct is actually not shifted into being in semantic agreement with the construction, because it appears in the constellation politically correct whose semantic content is, to some extent, gradable already.

When asked to interpret correct in (9), most respondents commented on its intimate relation with politically (one respondent pointing out that politically correct is a collocation):

--"correct for me doesn't mean much by itself, I read it as part of the expression "politically correct", meaning without any form of prejudice or stereotyping but to a level where the person blind too reality [sic.], they desire the right image, as someone who is very tolerant of differences, but maybe on a false premise"

--""Correct" on its own in this sentence means little taken out of the context of the "politically" that precedes it. "Politically correct" is a common term."

--"Well, without the qualifier 'politically,' the word 'correct' is sort of meaningless here. Combined with 'politically,' it's the idea of putting extra effort in one's words or behavior so as to not seem Ssic.C to be singling out a particular group. Generally used as a veil for slight racism, to which a preferable alternate would be the foreign ideas of "politeness" and "respect.""

--"'Correct' in this context is part of a collocation that refers to how socially appropriate a particular political view would be. 'Political correctness' refers to a politeness inherent in judgements or assumptions about what might or might not offend a minority group. So 'correct' means 'appropriate' or even 'inoffensive' in this context."

Those who did not comment on its relation with politically described the use of too politically correct as expressing right or INOFFENSIVE BEHAVIOR and EXTREME, BORDERLINE DISHONEST, POLITENESS.

5.4 Impossible

Impossible covers at least two possible absolute senses. In one sense, it stands in absolute contrast with possible, such that something can be either possible or impossible. In the other sense, it expresses an absolute point on a scale of DIFFICULTY, ranging from degrees of EASINESS over degrees of DIFFICULTY to IMPOSSIBILITY. Given that there is a scale of DIFFICULTY, the most obvious reconciliation process when impossible appears in [too ADJ to V] would be magnification, where IMPOSSIBILITY is expanded into a scale, extending from IMPOSSIBILITY and downwards. Indeed this seems to be at play in some instances:

(12) If the first boy you dare love pulls the worst Stephen King Carrie prank in the history of dating, then you run and hide. Because who can love you after that? Maybe your parents. But how can you face them, when you've all spent so much time convincing each other that you're normal? All i'm saying is, if you're me, and you can't reach a gas pump, pay phone, or ATM, and your arms and legs are disproportionately short, and your mouth is too impossible to kiss without it becoming a public carnival, then you don't get to be included in anything but the now obsolete, original meaning of the stupid word normal. (COCA 2011 FIC Bk:BigGirlSmall)

In (12), impossible is converted into expressing, not absolute IMPOSSIBILITY, but a very high degree of DIFFICULTY, bordering on IMPOSSIBILITY, caused by the body structure of the narrator. Indeed, one respondent suggests that impossible describes the shape of the mouth, while another one speculates that the narrator is a woman with dwarfism. one respondent comments on the semantic basicness of impossible:

--"'Impossible' is again something that cannot be qualified. Either something is impossible or it isn't, if one is being pedantic. i have never heard 'too impossible to' do something, although I admit to saying that something is 'really impossible' if I can't do it. 'Impossible' here means 'difficult' or 'awkward' or 'uncomfortable'."

This respondent suggests readings of impossible described by the inherently gradable adjectives difficult, awkward, and uncomfortable, indicating a scalar interpretation of impossible. The respondent's report that he or she has never heard too impossible suggests that the use of impossible in (12) deviates from the perceived conventional use (the fact that the respondent "admits" to having used other degree modifiers with impossible also points in this direction). Another respondent comments: "for me, the more normal expression would be just mouth is impossible to kiss, it is the too that seems odd", also suggesting deviation from semantic basicness. Moreover, one respondent suggests that too should be left out while another respondent evaluates too impossible as "patently incorrect grammar". The scalarization of impossible in (12) into expressing a degree of DIFFICULTY is reflected in other respondents' readings, specifying that kissing the narrator is challenging or very difficult, or that her lips are hard to reach.

One respondent further comments: "This is an exaggeration to express exasperation", suggesting that the use of impossible as a hyperbolic expression of the narrator's frustration with the high degree of DIFFICULTY. This particular respondent has deployed metonymy in his or her construal of impossible in (12) in two ways. Firstly, metonymy is at play in the processing of the scale of DIFFICULTY, such that the highest possible degree of difficulty--namely, IMPOSSIBILITY--becomes representative of a point on the scale that is not impossibility but rather very high DIFFICULTY. Secondly, there is a CAUSE-FOR-EFFECT metonymy at play in this respondent's reading in the form of the high degree of DIFFICULTY representing the EXASPERATION that follows from it.

Interestingly, impossible in [too ADJ to V] seems to have been quite difficult to process to some of the respondents. A number of respondents seem to have opted for canceling the scalar feature of the construction and instead retained the absolute feature of impossible, expressed by readings that include phrasings like unable to kiss and not possible (which is simply a paraphrasing of impossible). This cancellation of the constructional semantics is interesting, because it may be taken as counter-evidence to the override principle (Michaelis 2004: 51), suggesting, at least, that the override principle may not invariably apply to all cases of semantic construction-lexeme conflict. Another piece of counter-evidence could be blockage (Talmy 2000b: 333-334) in which the conceptualizer is not able to resolve the conflict. In evaluating too impossible to kiss, one respondent simply gives up, stating "i don't know what this means." This is an example of complete blockage. There is also an element of blockage in the evaluation by the above-mentioned respondent who rejected the combination of [too ADJ to V] in and impossible (12) as incorrect grammar, and arguably the other respondent who suggested leaving too out also operates with some level of blockage. The usage-event in (12) seems to require a considerable amount of cognitive effort. While some respondents overrode the constructional semantics, and the interpretative efforts of other respondents resulted in blockage, scalar construals of impossible did occur, probably in a process where IMPOSSIBILITY as the upper limit of the scale of DIFFICULTY is expanded via a combination of magnification and metonymy into a scalar structure which signals a high degree of DIFFICULTY rather than IMPOSSIBILITY.

Impossible may appear in [too ADJ to V] and take on a near-modal function:

(13) Doctor Vincent Del: If you do a literature search in the medical literature, there is no Morgellons disease. The skin symptoms fit perfectly with the same group of individual symptoms that we saw in patients with delusions of parasitosis.

Cynthia McFadden (voiceover): Delusions of parasitosis. In other words, it's all in their heads. And despite today's announcement, many doctors remain skeptical that this is a real medical condition at all.

Doctor Randy Wymor: They've pre-decided that, that these symptoms are too impossible to be real, so they have come to this conclusion that it's a delusion. (COCA 2008 SPOK ABC_Nightline)

in this case, the speakers discuss whether or not Morgellons disease actually exists, and when Doctor Randy Wymor describes the symptoms in question as too impossible to be real, he is evaluating their LIKELIHOOD of existing at all. Thus, a scalar structure is applied to the otherwise binary relation of EXISTING versus NOT EXISTING, which results in a scale of LIKELIHOOD--not unlike that associated with epistemic modal verbs. Thus, impossible is interpreted as expressing a very low degree of LIKELIHOOD of EXISTENCE. Technically, this would either be a metaphorical extension from the domain of difficulty into the domain of existence with the scalar nature of difficulty being imposed upon the domain of existence, or a blend, as advocated by Talmy (2000b: 329-332) in his typology of semantic resolution processes.

5.5 Catholic

Due to its basic absolute semantics, this adjective is highly incompatible with [too ADJ to V]. The example in (14) describes an individual of Catholic background who, in reflecting on her perception of the way that catholicism relates to women and homosexuals, is experiencing, perhaps not a crisis of faith, but a crisis of confidence in the church:

(14) As a mother and grandmother and a retired special education teacher, i am a skeptic and critic of many church positions, most notably as they relate to women and gays and others whom God created in God's image and likeness. I am too Catholic to be anything else, but the church hierarchy tries my patience as nothing in my life ever has. (COCA 2011 MAG USCatholic)

One respondent points out how the use of Catholic in (14) deviates from a perceived basic use:

--"Roughly, "I am too engrained in the Catholic church" or "too set in my ways" or "too brainwashed." You either are Catholic or you are not. Sounds like this lady needs a new church."

The respondent's comment of the either-or nature of Catholic and the special communicative status the respondent perceives it to have in (14) indicates a perception of deviation from a more conventional use in (14). The respondent describes Catholic as engrained, set in one's ways, and even brainwashed, all of which are arguably gradable expressions, which suggests that a scalar construal is at play. In fact, the respondent comments on the impossibility of Catholicism being a matter of degree and takes the scalar construal as an indicator of the woman's wavering confidence in the Church. Another respondent comments:

--"This is the first time I hear this, but it makes sense. I think your question is wrong above, I think your question should be what does "too Catholic" mean? But always, Catholic means someone who practices the Catholic faith and can't see themselves practicing another Christian line."

This respondent ultimately seems to select an absolute reading of Catholic, thus canceling out the scalar feature of [too ADJ to V]. Still, in stating that the question in the questionnaire is incorrect, and that it should have asked what too Catholic means, the respondent does operate with the scalar adjectival construction [too ADJ], which of course also serves as a structural element in [too ADJ to V]. What is interesting is that [too ADJ] is per se a scalar construction, and this could be indicative of the respondent engaging in schema juggling. In this case, it seems that the respondent first applied [too ADJ] as a constructional unit, but then canceled it and instead opted for a reading tied in with the semantics of Catholic.

The respondents seemed split in terms of whether their readings involved shifting of Catholic or cancellation of the scalar semantics of the construction. These readings, for instance, seem to favor the constructional semantics, as they all operate with degrees of Catholicism:

--"She is too involved in the religion to identify as anything else."

--"Catholic' with a big 'C' here means a particular Christian religious persuasion. I think the speaker means 'too much of a Catholic' or 'I have been a Catholic for far too long' - meaning that he or she is far too entrenched in the morals and values of the Catholic church to do other than whatever is being discussed."

--"you are brought up heavily on one religion only"

--"a very strict Catholic"

In contrast, another set of readings favors the lexical semantics, simply offering readings that associate Catholic with a religious view or a religious group. One respondent even states that the use of Catholic in (14) is a "perfectly normal use of the name." Another respondent conceptualizes Catholic as referring to behavior, stating: "I still get it but there are "levels" of Catholicism in real world behavior." Rather than expressing direct religious group membership or religious worldview, Catholic, to this respondent, refers to behavior that is perceived as following from the Catholic worldview. Thus, a more scale-friendly construal is involved, as behavior is easier to construe scalarly.

One could argue that metonymy is at play in all readings which favor the constructional semantics such that the denominator of the religion itself is projected onto behavioral patterns associated with Catholicism. In this process, the scalarity of BEHAVIOR is retained. The behavior-oriented reading of Catholic is probably associated with a more general CATEGORY-FOR-BEHAVIOR metonymy, which then is selected when Catholic appears in scalar adjectival constructions.

5.6 Female

Female in [too ADJ to V] is similar to Catholic. Rather than referring to the non-scalar nature of BIOLOGICAL SEX, the focus is moved onto the behavioral aspect of gender stereotypes and involves metonymy and scalarity:

(15) I knew myself to be tenacious, aggressive, and stubborn. The racing world saw me as reserved and feminine, yet competent--and I worked hard for it. But the bottom line, to the good old boys of the racing world, was that I was too female to be ruthless. (COCA 2011 FIC Bk:DeadMansSwitch)

The respondents' readings of (15) tend to focus on culturally based perceptions of the female gender. Some interpret female here as expressing WEAKNESS and POWERLESSNESS, to which scalarity can easily be applied. Others pinpoint more overtly behavioral aspects, thus homing in on perceived degrees of feminine, or female, behavior:

--"femininely gentle"

--"she has too many feminine characteristics, she would be too nurturing and motherly maybe to be cruel"

--"Implicitly highlights particular qualities associated with being female, qualities that are juxtaposed with being ruthless."

Other respondents do not overtly refer to behavior, but do include comments on cultural gender and attitudes towards female gender roles:

--"Semi-normal I guess as unfortunately such sexist attitudes do still exist, and using female in this sense certainly implies a great deal of sexism."

--"female' here means that the speaker is referring to her gender and how that has affected her in her particular sport. It means she harbours feminine traits by which others judge her."

--"it means female, in a negative sense, as in stepping on the toes of men and their domain in typical American gender role-type situation."

It appears that the respondents tend to select associated attributes (Talmy 2000b: 327-379) of the concept of WOMAN as a cultural construct. This is not unlike the resolution strategies applied to Catholic in (14) in which those respondents who did not reject the constructional semantics associated Catholic with a cultural stereotype and then selected from those aspects of the stereotype most compatible with scalarity. Similarly, in their readings of (15), respondents selected various behavioral aspects of the WOMAN stereotype and construe them as the referent attributes of female.

The following examples, which were imported from webpages--as too female to V only occurred once in COCA--seem to further support this resolution type.

(16) Nobody knows the Goldilocks Syndrome better than Hillary Clinton. For as long as she has occupied the national consciousness--as first lady, as senator, and as secretary of state--she has never been quite "right". Too female to be taken seriously (remember that bit of cleavage she revealed on the Senate floor?), she was also dubbed too aggressive for driving health care reform. (http://edition.cnn.com/2013/06/19/ living/women-cheerful-leaders/)

(17) I was perceived as too young and too female to be doing pension trustee work. (http://www.the3rdimagazine.co.uk/2012/09/influential-women-sarah-smart/)

(18) So here is my point: no-one is too female to be a plumber, doctor, astronaut or carpenter, or too male to be a housewife, nurse, teacher or care worker (http://lizterryblog.wordpress.com/2012/01/31/too-female-to-know-about-plumbing/) with references to aggression, ruthlessness, and different job types and social categories, the sentences in (16-18) metonymically encode aspects of a cultural WOMAN stereotype as referents of female. Consequently, the chosen scalar attributes become representative of the female gender.

5.7 Pregnant

Logically speaking, PREGNANCY is a binary affair: either you are pregnant or you are not pregnant. Thus, it is not surprising that, when pregnant appears in a scalar construction, deviation from its semantic basicness occurs and results in semantic conflict:

(19) Unlike Celia's former husband, Jerry kept calling home. After two nights away, he showed up at Sunny's door in time for dinner. Sunny was too pregnant to argue, but Jerry would have to sleep in the family room. Jerry wanted to sleep in bed with her. Sunny said no for almost a week (CoCA 2011 FIC SouthwestRev)

In interpreting the use of pregnant in (19), one respondent asks:

--"With child? But why would that impact her ability to argue?"

This particular respondent identifies the force-dynamic semantics of [too ADJ to V] but seems to find its application illogical in this particular case. This may be the result of schema juggling in which the respondent applies the force-dynamic schema, assigning COUNTER-FORCE to pregnant in the ADJ-position, but then canceling the scalar semantics of the construction, instead opting for the absolute semantics of pregnant, or perhaps blockage, thus violating the override principle (Michaelis 2004: 51). In questioning the validity of the force-dynamic relation the respondent nonetheless signals an awareness, at some level, of semantic deviation.

The remaining respondents, in contrast, offer readings that are more gradably oriented. A number of respondents operate with selection of associated attributes that are more likely to be scalar, such as EMOTIONALITY, HORMONALITY, TIREDNESS, and FATIGUE, with one respondent interpreting (19) more generally as expressing the state of being "Affected by the side-effects of pregnancy." As with Catholic and female, these respondents seek semantic resolution by metonymically selecting attributes associated with PREGNANCY and applying scalar construals to those. One strategy is selection of DEVELOPMENT OF THE FETUS as the focus and specification of an advanced stage of PREGNANCY:

--"It means that the woman is in an advance [sic.] stage of pregnancy. will give birth soon--has a large belly etc."

--"She is constrained in a positive way from arguing. She is really pregnant, but it would have to be somewhere after 4 months, I guess. It would have to show and cause your some discomfort, so she doesn't want more discomfort in her life."

--"almost giving birth"

--"Not sure--is she actually 'pregnant"? Perhaps she is close to having the baby and emotional, not wanting to argue?"

One respondent, who discloses his gender, offers a metonymic reading akin to the ones above, but also questions the logic of the force-dynamic relation and hints at the sentence deviating from his perceived basicness:

--"I don't see how being pregnant would stop Sunny arguing if she didn't want Jerry to stay. she is obviously heavily pregnant though so perhaps she just doesn't want the stress, but this is a strange sentence to me (maybe a woman would identify with it more)."

Compare (19) to the following examples:

(20) It was her father's tenth reunion. Her mother had been too pregnant to make the trip, so her father had taken her along instead. (cocA 1999 FIC Bk:EveningNews)

(21) Why did you wait until I was too pregnant to fly before you went? Were you trying to get away from us? (cocA 1990 FIc ploughshares)

In both cases, and more overtly so than in (19), the focus has moved to a conceptualization of PREGNANCY as a SERIES OF PHASES, ultimately tied down to the growth of the fetus.

5.8 Innocent

Innocent covers a number of conventional senses, ranging from the absolute sense of NOT GUILTY (as opposed to GUILTY) to more scalar ones, such as NAIVE, YOUNG, INEXPERIENCED and the like. In commenting on (22) below, one respondent remarked that the use of innocent in [too ADJ to V] "just sounds strange", suggesting that too young to die would be "more normal":

(22) She had known Maya's shortcomings when she took the ice owl, and never bothered to safeguard against them. she had known all the accidents the world was capable of, and still she had failed to protect a creature that could not protect itself. Now, remorse made her bleed inside. The owl had been too innocent to meet such a terrible end. Its life should have been a joyous ascent into air, and instead it had been a hellish struggle, alone and forgotten, killed by neglect. Thorn had betrayed everyone by letting the ice owl die. (COCA 2011 FIC FantasySciFi)

Other respondents similarly suggest readings of innocent in (22) along more scalar dimensions such as NAIVETY, GOODNESS AS PET, TAMENESS, KINDNESS, STUPIDITY, AGE and PURITY as well as the extent to which the owl deserved its fate. It seems, in resolving this particular semantic conflict, the respondents selected one of the, perhaps less basic, conventional senses that have more scalar structures so they can apply the scalar semantics of the construction.

A few respondents commented on the stylistic function of the use of innocent in (22), suggesting that it serves more literary purposes.

--"This word adds dramatic impact to the text - highlighting how it was unjust that the owl had a terrible end when did not live long enough to even have chance to become aware of life's hardships."

--"It's used in such a way to imply that the person that killed it is guilty, it's [sic.] antonym."

The former, by referring to the life-span of the owl seems to draw on the scalar construal of INNOCENCE as a matter of AGE and EXPERIENCE, but, when stating that the word "adds dramatic impact on the text" and commenting on its highlighting function, the respondent seems to have some notion that innocent is used in a special way in this particular text. The latter, interestingly, opts for a non-scalar construal, setting up an implied contrastive relation between the concepts of innocence and guilt. By suggesting that innocent in (22) serves to set up implicit relations, this respondent also seems to be under the impression that the use of innocent in [too ADJ to V] is, if not a strange one, then at least a special-purpose one.

6. Metonymy and disambiguation in semantic conflict resolution in [too ADJ to V]

There seem to be two recurrent processes--namely, the construal operation of metonymy and the disambiguation process of sense-selection. Given their apparent significance in semantic conflict resolution, this section discusses these two processes.

Metonymy seems particularly pervasive in the respondents' readings of the anomalous instances of [too ADJ to V] in the questionnaire. Metonymy figures prominently in resolution strategies that involve shifts. In the questionnaire data, three types of metonymy were documented:

* UPPER END OF A CLOSED SCALE FOR A LOWER PART OF THE SCALE: this type of metonymy seems to be deployed alongside magnification, as we saw in connection with impossible and full. In this process the lexeme in the ADJ-position conventionally serves as the signifier of the upper end of a closed scale and is recast to represent, not the upper end, but a portion on the scale below the upper end.

* CAUSE FOR EFFECT: in connection with impossible and full, we also saw CAUSE-FOR-EFFECT metonymy at play in some readings in which the lexeme in the ADJ-position was understood as actually representing a state that follows from the otherwise basic semantics of the lexeme. This seemed to be more relevant to shifts which moved the conceptual prominence from the scalar aspects of the constructional semantics to the force-dynamic ones.

* CATEGORY FOR BEHAVIOR: this type of metonymy was prevalent in readings in which the lexeme in the ADJ-position denominated either a perceived social group (Catholic and female), or a physical state associated with certain patterns or aspects of behavior (pregnant). In such cases, selection of associated attributes seems to be applied at a categorial level such that a behavioral attribute associated with the category in question is promoted to the status of referent of the category label. Using a category label with reference to what is really just a perceived behavioral feature of that category is in essence a process of stereotyping. The use of this type of metonymy in conflict resolution is therefore particularly interesting in the perspective of social cognition as it draws on cultural models of stereotypical representation.

The apparent pervasiveness of metonymy probably owes to its underlying relatively economical cognitive mechanism in which the conceptualizer operates within the same domain. When presented with a scalar use of a non-scalar adjective, all the conceptualizer has to look for is a scalar, or at least scale-friendly, feature within the same domain and metonymically apply that as the adjectival referent in that particular usage-event. It is too early to say whether or not there are any patterns of application of metonymy within the overall architecture of semantic conflict resolution, but we may at least hypothesize that the three types of metonymy mentioned above would also surface in a larger and more systematic study of conflict resolution strategies in [too ADJ to V]. It is also very likely that such a study would reveal more types of metonymy and, in addition to falsifying or verifying the ones suggested above, enable the analyst to suggest further relations between conflict resolution types and metonymy types.

The other recurrent strategy was disambiguation via sense-selection. We saw disambiguation at play in connection with full, innocent, and possibly real, as well as, to some extent, correct. With full and innocent (and, more tentatively, real), respondents selected conventionalized meanings within the adjectives' sense networks (7) which were more compliant with the constructional semantics and the current discourse space. Thus, because of the mention of fried chicken in (7), the respondents selected the more specialized satiation sense of full and then applied various construal operations to it to make it comply with the constructional semantics of [too ADJ to V]. With innocent, the respondents were less uniform in their sense selection, but they did tend to opt for readings based on more scalar concepts such as AGE, INTELLECT, NAIVETY, and PURITY. A possible explanation might be that the sense network of innocent is to some extent characterized by vagueness, in the sense advocated by Tuggy (1993). Those units in the sense network of innocent that fit the discursive context in (22) the best are perhaps only vaguely distinctive from one another. For instance, we can assume age, INTELLECT, NAIVETY, and PURITY are intertwined in cultural models associated with MORALITY and that the boundaries between them are perhaps somewhat blurred. This may not completely fit the definition of vagueness as a linguistic form having multiple meanings that are "united as non-distinguished subcases of a single, more general meaning" (Tuggy 1993: 167), but in a model of lexical semantics that operates with an ambiguity [left and right arrow] polysemy [left and right arrow] vagueness continuum (e.g. Deane 1988: 327), the sense network portion in question may be located at the non-extreme end of vagueness, near its transition into polysemy.

Since disambiguation has been documented as a strategy of semantic conflict resolution, we should consider its place in the architecture of such conflict resolution. The respondents who operated with disambiguation, opted for conventionalized, but peripheral meanings. Given that this involves deviation from semantic basicness, it might be tempting to classify this as a type of shift. However, the processes that Talmy (2000b: 324-329) categorize as shifts all involve conceptual manipulation at subsense level, while disambiguation takes place at sense network level. With this in mind, it is perhaps more appropriate to view disambiguation as a semantic conflict resolution strategy on par with shifting, blending, schema juggling, juxtaposition, and blockage. In relating disambiguation to the overall architecture of semantic conflict distribution in [too ADJ to V], we note that there may be an interplay between the accessibility of scalar senses to choose from in the sense network and the choice of strategy in Talmy's (2000b: 323-336) model. For instance, with impossible a respondent gave up interpreting it while other respondents opted for shifts that canceled constructional semantics, thus violating the override principle. This might be a reflection of the sense network of impossible not including readily available scalar senses. Similarly, some of those respondents who did not apply metonymic selection of associated attributes offered readings indicative of the application schema juggling as an attempt to make sense of Catholic and pregnant in (14) and (19) respectively, which in some cases led to cancellation of the constructional semantics rather than semantic adjustment of the adjective. As with impossible, this may owe to unavailability of conventionalized senses that are more compliant with the scalar semantics of the construction.

7. Conclusion

This study addresses conceptualizers' use of conflict resolution strategies in anomalous instances of the relatively unexplored [too ADJ to V]-construction. To this end, an informal questionnaire was made in which native speaker informants were given the simple task of offering their readings of fifteen instances of the construction--all of which were extracted from the 2011-component of COCA. Of these, eight were characterized by construction-lexeme conflict as the scalar semantics of the construction was incompatible with the lexemes in the ADJ-position. A previous empirical study of the construction (Jensen 2013) was used in the identification of typicality and anomaly in the usage of the construction.

The analysis of the evaluations offered by the respondents was of a qualitative nature and focused on inferring processes of semantic conflict resolution from their readings. Relying primarily on Talmy's (2000b: 323-336) model of semantic conflict resolution strategies and Croft & Wood's (2000; Croft & Cruse 2004: 40-74) model of construal operations, I was able to infer a number of possible strategies deployed by the respondents in their interpretation of the anomalous instances. At the level of Talmy's (2000b: 323-336) model of what is essentially a typology of superordinate strategies, we saw that while blockages, blends, and schema juggling did occur, shifts were the most commonly used strategies. Among the shifts applied, we saw stretching, magnification, and selection of associated attributes via metonymy. In addition, disambiguation via sense-selection was applied in cases polysemy in the ADJ-element where a peripheral sense, more compliant with the scalar semantics of the construction than the basic sense of the adjective, was selected. Disambiguation via sense-selection and selection of associated attributes via metonymy seem to be quite pervasive in conflict resolution within the [too ADJ to V]-construction, and should perhaps be included in Talmy's (2000b: 323-336) model of semantic resolution strategies in cases of multiple specification as a sixth resolution type. Given the data generated by this admittedly informal open-ended questionnaire study, it is recommended that a portion of future research into multiple specification and semantic conflict in construction-lexeme relations be focused on these two processes.

Due to the number of respondents, the findings of the present study should not be taken to be indicative of patterns of resolution. Rather, the findings of the questionnaire analysis serve as input to the construction of one or more hypotheses regarding semantic conflict resolution in the [too ADJ to V]-construction, which should be tested rigorously against data sets large enough for reliable statistical analysis. Thus, the contribution of the present study is not an exhaustive analysis of resolution strategies in the construction, but rather intended as input to potentially important research into the construction-recipient interface in cases of construction-lexeme conflict.

UDK 811.111'37

811'367.623+367'625

Izvorni znanstveni clanak

Prihvaceno za tisak 17. 4. 2014.

References

Croft, W.A. (2001). Radical construction grammar: Syntactic theory in typological perspective. Oxford: Oxford University Press.

Croft, W.A. (2003). Lexical rules vs. constructions: A false dichotomy. Cuyckens, H., T. Berg, R. Dirven & K. Panther, eds. Motivation in language: Studies in honour of Gunter Radden. Amsterdam: John Benjamins, 49-68.

Croft, W. A. (2005). Logical and typological arguments for radical construction grammar. Ostman, J., ed. Construction grammars: Cognitive grounding and theoretical extensions. Amsterdam: John Benjamins, 273-314.

Croft, W. A. & D. A. Cruse (2004). Cognitive linguistics. Cambridge: Cambridge University Press.

Croft, W.A. & E.J. Wood (2000). Construal operations in linguistics and artificial intelligence. Albertazzi, L., ed. Meaning and cognition: A multidisciplinary approach. Amsterdam: John Benjamins. 51-78.

Davies, M. (2014). Corpus of Contemporary American English. http://corpus.byu.edu/coca/.

Deane, P. (1988). Polysemy and cognition. Lingua 75: 325-361.

Fillmore, C.J., P. Kay, C. O'Connor (1988). Regularity and idiomaticity in grammatical constructions: The case of let alone. Language 64: 501-538.

Geeraerts, D. (1997). Diachronic prototype semantics. Oxford: The Clarendon Press.

Goldberg, A. (1995). Constructions: A construction grammar approach to argument structure. Chicago: University of Chicago Press.

Goldberg, A. (1997). The relationships between verbs and constructions. Verspoor, M.H., K.D. Lee, & E. Sweetser, eds, Lexical and Syntactical Constructions and the Construction of Meaning. Amsterdam: John Benjamins. 383-398.

Gries, S.T. & A. Stefanowitsch (2004a). Extending collostructional analysis: A corpus-based survey on "alternations". International Journal of Corpus Linguistics 9(1): 97-129.

Gries, S.T. & A. Stefanowitsch (2004b). Covarying colexemes in the into-causative. Achard, M. & S. Kemmer, eds. Language, Culture, and Mind. Chicago: CSLI. 225-236.

Hopper, P. (1987). Emergent grammar. Berkeley Linguistics Society 13: 139-157.

Jensen, K.E. (2013). This construction is too hot to handle: A corpus study of an adjectival construction. Paper presented at 14th Annual Meeting of the Japanese Cognitive Linguistics Association. Kyoto: September 21, 2013.

Johnson, M. (1987). The body in the mind: The bodily basis of meaning, imagination, and reason. Oxford: University of Chicago Press.

Kemmer, S. & M. Barlow (2000). Introduction: A usage-based conception of language. Barlow, M. & S. Kemmer, eds. Usage-based models of language. Stanford, CA: Stanford University Press. vii-xxviii.

Kennedy, C. & L. McNally (2005). Scale structure and the semantic typology of gradable predicates. Language 81: 345-381.

Lakoff, G. (1987). Women, fire, and dangerous things: What categories reveal about the mind. Chicago: University of Chicago Press.

Langacker, R.W. (1987). Foundations of cognitive grammar, vol. 1: Theoretical prerequisites. Stanford: Stanford University Press.

Langacker, R.W. (1990). Concept, image, and symbol: The cognitive basis of grammar. Berlin: Mouton de Gruyter.

Langacker, R.W. (1991). Foundations of cognitive grammar, vol. 2: Descriptive application. Stanford: Stanford University Press.

Langacker, R.W. (2001). Discourse in Cognitive Grammar. Cognitive Linguistics 12.2: 143-188.

Michaelis, L.A. (2004). Type shifting in Construction Grammar: An integrated approach to aspectual coercion. Cognitive Linguistics 15: 1-67.

Paradis, C. (2000). It's well weird: Degree modifiers of adjectives revisited: The nineties. Kirk, J. M., ed. Corpora galore: Analyses and techniques in describing English. Amsterdam: Rodopi. 147-160.

Rosch, E. (1973). Natural categories. Cognitive Psychology 4: 328-350.

Stefan owitsch, A. & S.T. Gries (2003). Constructions: Investigating the interaction between words and constructions. International Journal of Corpus Linguistics 8.2: 209-43.

Stefanowitsch, A. & S.T. Gries (2005). Covarying collexemes. Corpus Linguistics and Linguistic Theory 1.1: 1-43.

de Swart, H. (2003). Coercion in a cross-linguistic theory of aspect. Francis, E.J. & L.A. Michaelis, eds. Mismatch: Form-function incongruity and the architecture of grammar. Stanford: Stanford University Press. 231-258.

Talmy, L. (2000a). Toward a cognitive semantics, vol. 1: Concept structuring systems. Cambridge, MA: MIT Press.

Talmy, L. (2000b). Toward a cognitive semantics, vol. 2: Typology and process in concept structuring. Cambridge, MA: MIT Press.

Taylor, J. R. (2003). Linguistic categorization, 3rd ed. Oxford: Oxford University Press.

Tomasello, M. (2003). Constructing a language: A usage-based theory of language acquisition. Cambridge, MA: Harvard University Press.

Tuggy, D. (1993). Ambiguity, polysemy, and vagueness. Cognitive Linguistics 4(3): 273-290.

Verhagen, A (2007). Construal and perspectivisation. In: Geeraerts, D. & H. Cuyckens, eds. Handbook of cognitive linguistics. Oxford: Oxford University Press. 48-82.

Kim Ebensgaard Jensen

Aalborg University, Denmark

kim@cgs.aau.dk

* I would like to thank Kei Sakaguchi for inspiring me to write this paper and Steven Breunig for providing some useful input. I am also grateful to the two anonymous reviewers who refereed this paper. They provided extremely useful feedback and offered invaluable suggestions that helped me improve this paper immensely. I am, of course, fully responsible for any errors or shortcomings. Finally, I would like to dedicate this paper to the late John M. Dienhart who taught English grammar and linguistics at The University of Southern Denmark.

(1) Although a number of different versions of construction grammar exist, I will refer to them collectively as 'construction grammar' in this article. The overview offered here addresses the basic principles that all, or, at least, most, construction grammars have in common.

(2) With the reservation in mind that "there seems to be no way to organize them all [i.e. construal operations] in terms of an exhaustive classification system" (Verhagen 2007: 76), we will apply Croft & Cruse's (2004: 40-73) updated version of Croft & Wood's (2000) model of construal operations because of its comprehensive nature. For examples of construal operations in language, the reader is referred to Croft & Wood (2000), Croft & Cruse (2004: 40-73), and Verhagen (2007).

(3) We should, of course, keep in mind that, although the examples in the questionnaire prompt the respondents to engage in interpretative processes, questionnaires require respondents to be more conscious and reflective than normal usage-events in natural discourse do.

(4) Based on Jensen (2013).

(5) Although 'scale', 'scalar', 'scalarity' and the like may be interchangeable with 'axis', 'axial', 'axiality' and the like, I find that, for the sake of clarity, it is best to consistently use one set of terms. Given that the 'axis'-set is less widespread in cognitive linguistics in general, I will stick to the 'scale'-derived set.

(6) Defined by Langacker (2001: 144) as "the mental space comprising those elements and relations construed as being shared by the speaker and hearer as a basis for communication at a given moment in the flow of discourse."

(7) See Deane (1988) and Tuggy (1993) for more detailed discussions of polysemy and lexical sense networks.
COPYRIGHT 2014 Croatian Philologic Society
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2014 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:Izvorni znanstveni radovi
Author:Jensen, Kim Ebensgaard
Publication:Suvremena Lingvistika
Date:Jun 1, 2014
Words:11407
Previous Article:Antonio Fabregas i Sergio Scalise: Morphology. From Data to Theories.
Next Article:Investigating meaning construal in the language of the blind: a cognitive linguistic perspective/Istrazivanje konstruiranja znacenja u jeziku...

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters