University of Southern California Dissertations and Theses

Subjectivity, commitments and degrees: on Mandarin hen

(USC Thesis Other)

Subjectivity, commitments and degrees: on Mandarin hen

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Copyright 2021 Huilin Fang SUBJECTIVITY, COMMITMENTS AND DEGREES: ON MANDARIN HEN by Huilin Fang A Dissertation Presented to the FACULTY OF THE USC GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (LINGUISTICS) May 2021 ii Acknowledgements I’d first like to thank my supervisors, Roumyana Pancheva and Barry Schein, who guided me with professionalism and patience throughout the years. Aside from giving professional advice, they cared for me not only as a student, but also as a person. Without their guidance and encouragement, I would not have finished my dissertation. I’d also like to thank my committee members, Audrey Li and Toby Mintz, who helped me put a dissertation defense into shape. Audrey offered invaluable insights on the Mandarin data in my dissertation and other researches I’ve done in the past. I would also like to thank Andrew Simpson, Maria Luisa Zubizarreta, Khalil Iskarous, Hajime Hoji, Louis Goldstein, Elsi Kaiser, Deniz Rudin, Rachel Walker, Elena Guerzoni, Hagit Borer, Sandra Disner, Alexis Wellwood, Robin Jeshion and Elena Herburger. They are the professors I’ve worked with and once taught me and inspired me on pursuing linguistics. Those were among the most valuable experiences during my years in this Ph.D. program. I’d also like to thank the people in the department who helped me make this happen, Lisa Jo Keefer, Guillermo Ruiz and Mary Washburn, and the informants who participated with the collection of my data and helped me sort out the judgements for each sentences. It was not an easy task. Finally, I am grateful for all the support from my family, especially my parents, my brother and my husband, who encouraged me to pursue my dream when I was hesitant to continue the program. Thanks to my friends both in and outside the field of linguistics that encouraged me when I felt that I could never make it. iii Table of Content Acknowledgements ....................................................................................................................................... ii Abstract ......................................................................................................................................................... v Chapter 1 Introduction ............................................................................................................................. 1 Chapter 2 Theoretical notions relevant to hen ......................................................................................... 5 2.1 Degree semantics and the positive degree sentence ...................................................................... 5 2.1.1 Gradable adjectives and scale structures ................................................................................. 5 2.2 Positive sentences and degree modifiers ....................................................................................... 9 2.2.1 Absolute adjectives, relative adjectives and scale structures ................................................ 12 2.2.2 The Mandarin relative adjectives and absolute adjectives .................................................... 17 2.3 Commitments based on certainty ................................................................................................ 20 2.4 Commitments expressed in the truth conditional level ............................................................... 22 2.5 Evidentiality and events .............................................................................................................. 26 2.6 Conclusion ................................................................................................................................... 31 Chapter 3 The puzzles of hen ................................................................................................................. 32 3.1 Overview ..................................................................................................................................... 32 3.2 Previous analyses on hen ............................................................................................................. 32 3.2.1 Hen as a positive degree morpheme ...................................................................................... 32 3.2.2 Hen as a copula ...................................................................................................................... 37 3.2.3 Hen as a degree adverb .......................................................................................................... 43 3.3 The problem of subjectivity ........................................................................................................ 50 3.3.1 Subjectivity in hen ................................................................................................................. 50 3.3.2 Faultless disagreements ......................................................................................................... 52 3.3.3 Embedding under attitude verbs ............................................................................................ 55 3.3.4 The parallelism with other evaluative degree modifiers ....................................................... 58 3.4 Conclusion ................................................................................................................................... 62 Chapter 4 Epistemic commitment .......................................................................................................... 63 4.1 Overview ..................................................................................................................................... 63 4.2 Some notions on epistemic commitments ................................................................................... 64 4.3 Epistemic commitment of hen ..................................................................................................... 67 4.3.1 The hedging effect under negation and euphemism .............................................................. 67 4.3.2 Euphemism like very: a possible analysis ............................................................................. 68 4.3.3 Argument against euphemism: Hen embedded under attitude verb ...................................... 71 4.3.4 An argument against euphemism: focus ................................................................................ 74 4.3.5 Subjectivity and epistemic commitments .............................................................................. 79 4.4 A comparison with the shi…-de form ......................................................................................... 83 4.5 A comparison with the English positive form ............................................................................. 85 4.6 Intermediate summary and further issues .................................................................................... 86 4.7 German eher: a comparison ........................................................................................................ 86 4.8 Conclusion and further issues ..................................................................................................... 89 iv Chapter 5 The licensing condition of mei and the meaning of hen ........................................................ 91 5.1 Overview ..................................................................................................................................... 91 5.2 Traditional views on mei ............................................................................................................. 95 5.2.1 Dynamicity and mei ............................................................................................................... 95 5.2.2 The contrast between mei and bu .......................................................................................... 97 5.3 Stative predicates licensed by mei ............................................................................................. 103 5.3.1 Types of stative predicates .................................................................................................. 103 5.3.2 The possibility of being dynamic ........................................................................................ 107 5.3.3 Further supports of the stative status of these four types of verbs ...................................... 120 5.3.4 A look at hen predicates ...................................................................................................... 123 5.4 Stance taking and temporal boundedness .................................................................................. 124 5.4.1 Overview ............................................................................................................................. 124 5.4.2 Negation and temporal boundedness ................................................................................... 125 5.4.3 Temporal modifiers ............................................................................................................. 128 5.4.4 Attitude verbs ...................................................................................................................... 133 5.4.5 The stance and the evaluation eventuality ........................................................................... 134 5.5 Other supports for the existence of a stance .............................................................................. 138 5.5.1 The complement positions of perception reports ................................................................ 138 5.5.2 Manner adverbs ................................................................................................................... 140 5.5.3 Modification of locative phrases ......................................................................................... 149 5.5.4 Locative modification in Mandarin Chinese ....................................................................... 153 5.6 A generalization ........................................................................................................................ 159 5.7 A look at hen sentences ............................................................................................................. 161 5.8 Mei and the specificity restriction on a discourse situation ...................................................... 162 5.8.1 A comparison with estar in Spanish .................................................................................... 162 5.8.2 The evaluation eventuality in mei sentences ....................................................................... 165 5.8.3 Evidence from the comparison between mei and bu ........................................................... 168 5.8.4 Generic sentences and habitual adverbs .............................................................................. 172 5.9 The scopal property of the evaluation eventuality .................................................................... 176 5.10 A formal account of hen ............................................................................................................ 180 5.10.1 An overview .................................................................................................................... 180 5.10.2 Confidence level and degree morphemes ....................................................................... 180 5.10.3 Contentful events ............................................................................................................ 182 5.10.4 Contentful events and hen ............................................................................................... 184 5.10.5 Formal account of hen ..................................................................................................... 186 5.11 Conclusion ................................................................................................................................. 193 Chapter 6 Concluding remarks ............................................................................................................ 195 6.1 Some issues on other commitment inducing adverbs ............................................................... 195 6.2 A look at quite ........................................................................................................................... 196 6.3 Concluding remarks .................................................................................................................. 199 References ................................................................................................................................................. 200 v Abstract In this dissertation, I explore the nature of a highly controversial particle in Mandarin Chinese, hen. This particle appears in a structure called the positive degree sentence. A positive degree sentence like the English sentence John is tall is used to express the degree of an object that exceeds a certain contextually designated standard. However, in Mandarin Chinese, the same expression is formed differently. The particle hen, which is semantically quite bleached and often translated superficially as very, is obligatory in a positive degree sentence. This particle has several unique properties. First, it seems to have parallels with predicates that are traditionally viewed as dynamic. Second, hen is often linked to a more subjective or personal standard. This standard does not need the assessment of other participants. Putting all these features together, this dissertation concludes that hen is related to an evaluation eventuality linked to an attitude holder. The attitude holder evaluates the property of the object in order to decide if it falls into the positive extension of the gradable adjective. Once an object possesses a significant degree with respect to the dimension of that adjective, it is viewed as a sure case for the attitude holder that allows the attitude holder to commit to the truth of the proposition. Therefore, hen is used to express both a high degree and the attitude holder’s confidence towards the truth of the prejacent. Furthermore, the observations described above show that the traditional views on hen are flawed. In the literature, hen is regarded as the positive degree morpheme in Mandarin Chinese. This is based on the view that English positive degree sentences, such as John is tall, involve a covert morpheme that introduces a contextually given standard. Hen has long been viewed as the overt realization of this particle. However, this analysis would not account for the subjectivity and temporal restrictions of this particle. The study of this particle allows us to look at degree modifiers in a different light. It is well known that degree modifiers can introduce different types of standards. The most basic type of degree modifiers merely modify the standard along the numerical scale. However, there are degree modifiers that introduce standards based on one’s expectations and wishes. What makes hen unique is its relation to an eventuality vi of evaluation. It is possible that there are more degree modifiers with this property that forms a natural class cross-linguistically. 1 Chapter 1 Introduction This dissertation focuses on a specific gradable expression in Mandarin Chinese, the hen form. The particle hen is an adverbial expression that could be literally translated as ‘very’ or ‘quite’ according to native speakers’ intuitions, although both of them are not accurate. An example of a sentence formed with hen is shown in (1). In this dissertation, I analyze it as relevant to evaluativity. It involves an event of evaluation, and it shows the attitude holder’s commitment towards the truth of the proposition. Hence, throughout the dissertation, I will be using the translation certified to illustrate this property of evaluativity for ease of understanding, although certified is not the direct translation, and it does not express native speakers’ intuition. Therefore, throughout this dissertation, I will not translate hen in word-by-word translation, but in the English translation of the entire sentence, the word certified will be used. (1) Afu hen gao. Afu HEN tall. ‘It is certified that Afu is significantly tall.’ The hen form shown above is often regarded as the Mandarin counterpart of the English sentence ‘Afu is tall.’ Yet under native speakers’ intuition, the word hen itself can be superficially translated as ‘Afu is very tall’ because hen is used to denote a significantly high degree possessed by an object. But neither ‘Afu is tall’ nor ‘Afu is very tall’ is an accurate translation. What is unique about this particle is its evaluative nature: what counts as significant is defined by an additude holder based on his or her knowledge and experiences. This feature of hen sets it apart from very, which does not have this evaluative property. In this dissertation, I would argue that this particle hen determines the standard according to an attitude holder’s subjective evaluation. It is done by introducing an eventuality that represents the stance of the speaker or the attitude holder. 2 Before getting into the details, I first clarify some of the basic notions relevant to this dissertation. Adverbial expressions like hen and very are examples of degree modifiers. A degree modifier is used to modify a specific class of expressions, the gradable adjectives. It includes adjectives that denotes properties related to degrees, such as tall, short, angry or healthy. In the literature, gradable adjectives are often regarded as belonging to a different semantic type from typical, non-gradable, adjectives. The former are relativized to an abstract representation of measurement which defines the interpretation of the gradable adjective. This representation, or as scale, is construed of a set of points, each representing a numerical number or a degree. Within this set, the object could be mapped onto a degree on the scale in an order with respect to the dimension of the gradable adjective. This is the degree-based approach towards gradable adjectives, which is first articulated by Cresswell (Cresswell 1972) and Seuren (Seuren 1973), and followed by many others. When the predicate of a sentence is formed by a gradable adjective, as in (1), it needs to denote a certain degree to which the object reaches, and the subject is in the positive extension of this gradable adjective. As a result, a sentence like (1) is called a positive sentence because the height of Afu exceeds the standard for being tall. In a positive sentence, the standard is a degree that serves as the cutoff point that determines the positive extension of the adjective under certain context. For example, this point for a tall person could be 180cm in one context, but 160 in another. Such a standard is often related to a commonly accepted norm under normal circumstances. In the literature, the cutoff point is regarded as determined by a morpheme, the positive morpheme, or the pos morpheme, which picks out a contextually assigned standard. In the English positive form, this morpheme is regarded as covert. Hence, a sentence like ‘John is tall’ contains a positive morpheme, but it is not phonologically realized. Hen is then viewed by some as the overt realization of this morpheme in Mandarin Chinese. If this is the right analysis, then a sentence like (1) would have the truth condition that Afu’s height reaches the cutoff point, or the standard, and nothing more. 3 Degree modifiers are themselves ‘standard-setters’, just like the pos morpheme. The standards determined by these modifiers are based on the standard set by the pos morpheme. For example, a positive sentence like John is tall is true if John’s height exceeds the average of a contextually given set of individuals, say a group of his classmates. However, John is very tall, when under the same context, would require the standard to be set based on the height of only the tall people in the class. Very then selects the average of the tall people in John’s class, and henceforth raises the standard. Now we turn to Mandarin Chinese. As we can see in (1), the formation of a positive sentence in this language requires the existence of the particle hen. Being a marker used to form positive sentences, it has certain puzzling features. Hen appears as an obligatory particle in Mandarin positive sentences, but it is not semantically vacuous. The semantics of it is obscure, and hence there is no direct translation for hen. Intuitively speaking, hen denotes a significantly high degree. But what does a significant degree mean? It seems to make an emphasis that the degree is rather high, to the extent that there is more certainty to the assertion that is made. In this sense, it is not just a positive degree morpheme. It is not a counterpart of English very either. A sentence like ‘John is very tall’ only states that John reaches a higher degree than simply being tall. It does not have this effect of enhancing one’s certainty. Second, hen is licensed under contexts that require more subjective evaluation, while the English positive sentences do not. What I’m proposing in this dissertation is that hen is related to the notion of evaluativity. Evaluativity refers to the judgement based on one’s personal experiences or evidence he or she knows. After the evaluation is made, the speaker then would have the choice of making a commitment to the uttered propositions if he thinks that it is a reasonable conclusion from what he knows. Under my definition, evaluativity is correlated to epistemic modality, but it differs in the respect that there is an act of evaluation when these expressions are used. Yet hen also has a unique feature that the evaluation is bounded to a certain time, space or context. This bounded position could viewed as a stance. Therefore, this dissertation argues that evaluativity is realized as an evaluation eventuality related to the attitude holders’ stance. This eventuality restricts the spatial and temporal location of the evaluation. The attitude holder makes an evaluation for the likelihood 4 of a proposition and then commits to the result of this evaluation based on what he or she observes over that eventuality. The evaluation of a hen sentence is mainly based on the degree the subject possesses under a certain circumstance. Therefore, as hen implies a rather high degree, it is then a degree that is high enough for the attitude holder to make a commitment. The organization of this dissertation goes as follows. In Chapter 2, I discuss some theoretical backgrounds of several aspects relevant to my analysis of hen. I first discuss degree semantics, and focus specifically on the notion of positive degree sentences and semantics of degree modifiers. I further review the notion of subjectivity and epistemic commitment and how they are related to evidentiality. In Chapter 3, I focus on the hen data, and show that there are several puzzles related to hen that cannot be accounted for by previous analyses. Chapter 4 discusses how the notion epistemic commitment could account for the data of hen. In Chapter 5, I discuss the existence of an evaluation eventuality in the hen sentences, and this eventuality occurs in other types of sentences that are also stance-sensitive. Chapter 6 concludes this dissertation. 5 Chapter 2 Theoretical notions relevant to hen 2.1 Degree semantics and the positive degree sentence 2.1.1 Gradable adjectives and scale structures As mentioned earlier, degree semantics focuses on the concept of degrees in natural languages. Gradable adjectives denote properties that come in degrees, which could be ordered on a scale of a certain dimension, such as weight, height or size. Let’s take weight for example. For two objects, if they both possess certain weight, then the weight could be measured and compared and further being ordered according to the weight each possess. An individual object is within the domain of the adjective heavy only if there is a certain standard, a degree, for heavy objects. Such standard is determined by the type of the individual and the objects that the individual is compared to. Sentence (2) is true if and only if the box in question meets the standard of heaviness. (2) This 800g lunch box is heavy. As a result, how the standard is set becomes crucial for the truth condition of this sentence. It would be different when this particular lunch box is compared to a group of ordinary lunchboxes and to a group of concrete bricks, each weighing 2kg. In the former case, an 800g lunchbox is heavy, while in the latter case, it is not. On the contrary, there are properties that classify objects into distinct, non-gradable categories. Take the mathematical notion of even for example. An integer is either even or odd. It is a mathematical concept which is not context sensitive. It does not require a contextually assigned standard for determining the positive and the negative polarities of being even, unlike the concepts like weight or height. These adjectives are non-gradable adjectives. 6 However, there is a subtype of gradable adjectives that is somehow similar to non-gradable adjectives. It refers to those called absolute adjectives (Ungar 1975, Kennedy & McNally 2005). Some examples are shown in (3). One typical feature is that they are still gradable, but they also have absolute standards. (3) Absolute adjectives: flat, straight, bent, full, empty Take flat for example. The difference between flat objects and non-flat objects is more absolute. In one reading, flatness is a mathematical notion that has little space for vagueness. However, among the objects that falls into the set of non-flat objects, there is potential gradation in terms of how much they approximate flatness. Therefore, within this set, these non-flat objects can be ordered with respect to the degree of flatness. (4) a. The top of the desk is flatter than the surface of the playground. b. The surface of a billiard table is flat. The surface of a tea table is flat too. #A billiard table is flatter than that tea table over there. Such adjectives could often be interpreted in two ways, both in the gradable and the categorical sense. In (4)a, the flatness of the top of the desk and the surface of the playground comes in different degrees. However, in the case of (4)b, when we compare the top of a billiard table and that of a tea table, the comparison seems infelicitous. It is because flatness here denotes the general property of the surfaces of billiard tables and tea tables. They are designed as flat objects, and therefore the degree of flatness is not a salient component of the expression. That being said, the differences between absolute and relative adjectives are mainly between how the context plays a role. In the case of flat, flatness may come in degrees, but there seems to be an absolute notion of flatness that does not require a context to determine the cutoff point. For example, if we say that there are two table, which are both flat, and even if one of them is relatively flatter than the other, it is 7 absurd to say that one of them is flat while the other is not, since both of them fall into the category of flat objects. This is shown in the examples given below. In (5)a, there are two objects falling into the category of flat objects. Imagine that the desk is a bit flatter than the table since there is a slight tilt on one side of the table if you look very close. Yet in principle, they are still both flat. Therefore, it becomes awkward when one makes a statement like (5)b, which states that the mug should be placed on the flat one among the two. Since the two are both classified as flat objects, the distention between them is trivial conceptually. (5) a. This desk is flat. That table is flat too. b. I want to put the mug on the flat one. The opposite of absolute adjectives are relative adjectives. They are the typical gradable adjectives like tall, heavy, or fast. These adjectives do not have an absolute standard. Rather, they depend solely on the standard given by the context. For example, when one says that John is tall and Mary is also tall, they could have completely different standards for tallness. Unlike flat, of which people has a sense of consensus on what counts as flat, what counts as tall is totally dependent on the standard given in the context. For example, the standard for a tall basketball player must be different from that of a tall first grade student. Let’s look at the contrast between (5) and (6). Imagine a scenario for (6). The two girls differ in height, but both are considered to be tall girls in this context. (6)b is acceptable while (5)b is not. It is because when the two girls are being compared, due to the height difference, there is still one taller than the other. In (6)b, the standard for being tall is switched to the domain that include the two girls. Therefore, now if we are talking about the tall one among the two, it is no longer about whether the girl counts as tall according to the previous standard, but whether they are considered tall in the context when the two are being compared. The comparison between (5) and (6) shows that the standard for relative adjectives is more determined by the context. 8 (6) a. This girl is tall. That girl is tall too. b. I want to give the flower to the tall one. The degree-based approach (Bartsch & Vennemann 1972, Cresswell 1977, von Stechew 1984, Kennedy 1999, Kenney 2007a, Kennedy & McNally 2005, among many others) introduces the notion of degrees as part of the ontological property of gradable adjectives. In formal semantics terms, a degree is a type, type d, which belongs to the domain of its own, D d. As degrees could be viewed as numbers, the domain D d could be viewed as an ordered pair of ⟨d≻x, ≻ x⟩ (Cresswell 1977). The first element d≻x is the value of an object in a set x under measurement, and ≻ x is the ordering of objects in this set. The alignment of degrees forms a scale. The numbers form a total order. A total order is one that has the features of antisymmetric, transitivity and connectivity. Antisymmetry refers to the relation in which two elements, say a and b, holds in a relation such that if a≤ b, and b≤ a, then a=b. The two elements a and b are in a transitive relation if a≤ b and b≤ c both holds, then a≤ c. Connectivity refers to the relation that for a and b, it is either a≤ b or b≤ a. In other words, degrees can be ordered and compared. The function of gradable adjectives is to map an individual to a degree on the scale. It is represented by the structure shown in (7). The formal description is adopted from Wellwood (2014: 23). (7) A scale is an ordered pair ⟨D ≽ Deg, ≽ Deg ⟩, where D ≽ Deg, is the ordering of sets of degrees, and ≽ Deg is a total order on D ≽ Deg. This scale is the dimension of the scale. A dimension is determined by the lexical semantics of the gradable adjective itself, and in addition, by the context. For example, tall is related to a dimension of height, and heavy is related to that of weight. Some adjectives could have multiple dimensions, such as healthy or smart (Kamp 1975, Klein 1980, Sassoon 2013, among others). The notion of healthiness could be defined in terms of physical or mental healthiness. Smartness could also be defined in multiple ways, 9 such as being street smart or book smart. Objects with these properties could be measured by different dimensions that is often determined by the context. Formally speaking, a gradable adjective is a relation between an individual and a degree, which could be implemented as (8). It is a function that maps an individual x to a degree d on a scale given in the lexical entry of the adjective. In (8), the example of tall shows that the individual is mapped onto the scale of height. (8) ⟦tall⟧ =𝜆d𝜆x.height(x) ³ d ` Other than the degree-based approach, there is also the vagueness based approach. It is proposed by Wheeler (1972), Klein (1980, 1991), Burnett (2014), Larson (1988), among others. This approach views a gradable adjective as having the same semantic type as non-gradable adjectives, not as a measure function that maps an individual to a degree. The comparison between objects is not done by comparing the degrees each have, but by designating a comparison class from the context. The comparison class further determines a norm, represented by some objects in the set. One could make a comparison with respect to the norm of that set to decide whether the object falls in the positive or negative extension of the property denoted by the adjectival expression. In the this dissertation, I will adopt the view that gradable adjectives involve degrees in the lexical entry. 2.2 Positive sentences and degree modifiers A sentence like John is tall is a positive sentence, which states that the height of John falls into the positive extension of tall. The decision is done by deciding a cutoff point determined by the context or indicated by an overt phrase in the sentence. In English, one common way is by adding a prepositional phrase that introduces the set of individuals the subject is compared to. 10 (9) John is tall for a basketball player. In order to introduce a standard and also to saturate the degree argument of the adjective, a positive sentence needs morpheme. the positive morpheme pos to get the job done, as formalized in (10). The adjective measures an object and maps it to a value on the scale. It also introduces a contextually given standard. A positive sentence is true if and only if the degree to which the object measured reaches that standard. The function of pos is to take a gradable adjective and change it into a predicate of individuals. Pos encodes a function standard, which derives a standard degree d if it meets a standard of comparison for an adjective g. Here c represents a variable over properties of individuals whose value is determined contextually. The degree of x that is reached to is at least as high as the standard degree. (10) ⟦pos⟧= 𝜆g𝜆x.∃d. standard(d)(g)(c) & g(d)(x) (Kennedy & McNally 2005: 350) Degree modifiers, such as very, have a similar function as that of pos. In a positive sentence modified by a degree modifier like (11), the addition of very is used to express that the difference between the degree of the subject and the contextually given standard is large. Therefore, it would also infer the existence of a standard introduced by pos. For example, if the standard for tallness is the average of the students’ height in John’s class, then the standard introduced by very in John is very tall would be derived by the set of individuals in the class who are already tall. (11) John is very tall. The way degree modifiers change the standard is illustrated in (12). The standard introduced by pos divides the scale into two categories, one is the positive extension and the other is the negative extension. 11 Very further divides the set of the positive extension into two categories, one is in the positive extension of very-tall, and the other is in the negative extension of that property. (12) In other words, very raises an already given standard and based derives a higher standard. In (13), λy⟦pos(g)(y)⟧ c introduces a contextually given standard with respect to the set of individuals y, which is the set of objects that have degrees reaching the positive standard. In other words, this set is used to derive the new standard for very. In (13), the standard function takes the context c, the adjective g and the set of objects in the positive extension of the standard pos and derives the new standard d for very, and the degree of the subject x is at least as high as this standard d. (13) ⟦very⟧ c = 𝜆g𝜆x.∃d. standard(d)(g)(λy⟦pos(g)(y)⟧ c ) & g(d)(x) Degree modifiers could correspond to different standard degrees in terms of how much the standard is raised to. For example, extremely would be linked to a degree higher than that of very. Some degree modifiers lower the standard. For example, somewhat or half introduce a degree that is lower than a given norm. Standard for pos Standard for very Tall objects Very tall objects Not tall objects Scale of height 12 2.2.1 Absolute adjectives, relative adjectives and scale structures By adopting the view of degree-based semantics, the distinction between absolute adjectives and relative adjectives can be regarded as based on the difference of scale structures. Absolute adjectives are those that have a maximal or a minimal standard degree. They have scales that are either closed at one end or at both ends. On the contrary, relative adjectives have an open scale (see Rotstein & Winter 2004, Kennedy & McNally 2005). The scales of absolute adjectives can be classified into three types of subcategories: lower- closed scales, upper-closed scales and totally closed scales. (14) Totally closed scales: opaque, invisible, full, transparent Upper bound scales: flat, straight, closed, dry Lower bound scales: bent, wet Note that the bounds of the absolute adjectives are not really the ultimate minimal or maximal degrees an object could ever reach. Rather, a bound scale can be regarded as a limited interval on a numerical scale. This interval could potentially be contextually determined. Therefore, when an absolute adjective have upper bounds, it does not mean that it is the highest degree possible. This bound can still be exceeded. This is shown in the case of too full. If the upper bound is the absolute maximal standard for fullness, no object could become too full. But the possibility that one object could become too full implies that the upper bound is not the upper limit for fullness. Another example is that when two objects both fall into the category of full objects. In many cases, one could still distinguish between different degrees of fullness. For example, imagine that there is a glass teapot with 1000ml capacity, and the 1000ml line is marked on the outside of the teapot. In general, the real capacity of such a teapot is larger than 1000ml, since when the water is filled to the 1000ml line, there is still some space above it. If there are two identical teapots, one is filled to the line of 1000ml, and the other is filled to the brim, then the second teapot is fuller than the first one. Yet both are considered full. The second one probably falls into the category of things that are too full, since it not only reaches the standard of 1000ml, but it exceeds it. 13 Yet ontologically speaking, absolute adjectives are more about ‘absolute standards.’ The upper or lower bounds of the scales are where the standards usually lie. In principle, the standards of adjectives with totally closed scales are set on the maximal degrees of the scales when no other adverbs are present. Therefore, when one states ‘The glass is opaque’, it means total opaqueness. However, there are also various degrees of opaqueness. In (15), the degree of opacity of the glass can be minimal, maximal or anywhere in between. Therefore, it can be modified by degree modifiers like totally or completely, which marks a degree on the upper bound of the scale of opacity. Since an object can also have minimal opacity, opaque can be modified by modifiers that marks minimal degrees, such as slightly and a little bit. Finally, since such a scale is bounded, we could use proportional phrases to indicate one proportion of the scale, as in (15)c. (15) a. The glass is totally/completely opaque. b. The glass is slightly/a little bit opaque. c. The glass is 50% opaque. For adjectives with an upper bound scale, the standard is set at the upper bound. Therefore, it can only be modified by modifiers that mark the upper bounds, not by the ones that mark the lower bounds. This is shown in (16)a and (16)b respectively. Slightly is unacceptable because the scale of straight does not have a minimal degree. There is no such as thing as minimal straightness. It is not totally acceptable to appear with proportional expressions. (16) a. The pole is completely/totally straight. b. *The pole is slightly/a little bit straight. c. ?The pole is 50% straight. Finally, if the sentence has a minimal degree, but it does not have an upper bound, the standard is set directly at the minimal degree. It can only be modified by slightly or a little bit, but not by those like 14 completely or totally, which marks the upper bound of a scale. It cannot appear with proportional expressions as well. An example here is dirty. Generally speaking, when an object is slightly stained or contaminated slightly, it becomes dirty. However, it is hard to define if there is maximal dirtiness. This is shown in the example of water in (17). The water could be contaminated by just a little to be viewed as dirty, but it is difficult for the water to be completely dirty, since there is no maximal amount of dirtiness under normal circumstances. Of course, completely and totally could be used when they are referring to the spatial proportions of an object. For example, if now the water here refers to the water in a sink, saying that it is completely dirty would imply the entire sink of water to be dirty. However, it does not really refer to the degree of dirtiness if we take any portion of the water in the sink, since it is hard to define how much dirtiness is required for it to be maximally dirty. (17) a. The water is slightly dirty. b. *The water is completely/totally dirty. c. *The water is 50% dirty. Now we turn to relative adjectives. Different from absolute adjectives, relative adjectives do not have fixed standards that are set normally on one or both ends of the scale. They always rely on contextually determined standard. Relative adjectives cannot be modified by those modifiers that marks the ends of the scale. (18) *John is slightly/totally tall. It seems that there is an asymmetry between the standard setting of absolute adjectives and that of relative adjectives. Kennedy (2007b) argues that the reason for setting the standards on these bounds is pragmatic, and the standard is computed based on an economy principle: since the upper and the lower 15 bounds are what are known as the conventional meaning of an adjectival expression, the standards are set on them. The upper and lower bounds are more canonical interpretations for each absolute adjectives. To sum up, these scale structures are represented in (19) to (21). As shown in (19), an adjective with a completely closed scale would have both minimal and maximal degrees. The standard is usually set at the upper end. (19) Totally closed scales: The cup is full. Implication: the cup reaches the maximal degree of fullness. An adjective with a upper bound scale would set the standard on the upper bound. For example, a pole must reach the maximal degree of straightness on the scale to count as a straight pole. On the contrary, there is no lower bound in terms of the minimal degree. (20) Upper bound scales: The pole is straight. In a lower bounded scale, the standard is set at the lower bound. Therefore, when a stick is bent just minimally, it already counts as bent. 16 (21) Lower bound scales: The stick is slightly bent. Of course, the standard of absolute adjectives can still be context dependent. However, the dependency is often related to a contextually demanded accuracy or certain contextually determined purpose or expectations. Accuracy of the standard is a distinct notion from the contextually determined standard. When we talk about accuracy, it is about the required precision on whether an object actually reaches the standard. Take full for example, when in a context that demands higher accuracy, it only allows an object to be completely full to fit the criteria of fullness. In a more lenient context, it might allow more borderline cases that merely requires the fullness to be close enough to complete fullness. The differences between this degree and complete fullness is too trivial to the extent that it is ignorable. For example, imagine that there is a cup that is 99.5% filled with water. In a context in which 0.5% percent is non-trivial, the cup is not full. But in a more tolerant scenario, it may be ignorable. Furthermore, the standard could be determined by the purpose or expectations demanded by the context. This is often obtained by using a prepositional phrase that marks the purpose for the context (Solt 2012, Schwarz 2010; Bylinina 2013, 2017). In (22), six year old provides a norm for how a 6 year old should find sufficient for a meal, considering the size of the bowl stays constant. (22) The bowl is full for a 6-year old. All in all, the context could introduce a standard that does not fall on the either ends of a bound scale. Aside from an overt prepositional phrase, one way of introducing a standard is through the modification of degree modifiers. A degree modifier can set a standard in between the upper and lower bounds. In the case 17 of quite, it indicates that the bowl is nearly full but not exactly full. In (23)b, the degree of dryness is no longer absolute. Very introduces a contextually given standard for dryness that could be potentially derived from a comparison class. (23) a. The bowl is quite full. b. The glass is very dry. 2.2.2 The Mandarin relative adjectives and absolute adjectives The distinction between relative and absolute adjectives also exist in Mandarin Chinese. The difference is that Mandarin Chinese uses two different positive forms to distinguish the two. The Mandarin relative adjectives uses hen as its default positive form, while the absolute adjectives uses shi…-de form. This contrast seems to suggest that there is an intrinsic difference between the two types of adjectival expressions. Absolute adjectives in Mandarin include those English counterparts like man ‘full’, gan ‘dry’, wan ‘bent’ or touming ‘transparent.’ The three categories of scale structures also applies to Mandarin as well. Totally closed scale: (24) Beizi shi man-de Cup COP full-DE ‘The cup is full.’ (25) Beizi shi touming-de Glass COP transparent-DE ‘The glass is transparent.’ 18 Upper-closed scale: (26) Yifu shi gan-de. Clothes COP dry-DE ‘The clothes are dry.’ Lower-closed scale: (27) Guenzi shi wan-de Stick COP bent-DE ‘The stick is bent.’ In these examples, the standards are set on either bounds of the scale, as shown in the translation. Not surprisingly, Mandarin absolute adjectives can also appear with degree modifiers, which also includes hen. Recall that hen is obligatory with relative adjectives, as shown in (28). However, by default, the absolute adjectives use the shi…-de form. In other words, hen is not obligatory with absolute adjectives. When hen appears with relative adjectives, it yields similar meanings to that of very by allowing the degree to be high as compared to a certain standard. Examples in (29) shows combinations with absolute adjectives with different types of scales. (28) Afu *(hen) gao. Afu HEN tall ‘Afu is tall./ Afu is certified to be tall.’ (29) a. Beizi hen man Cup HEN full ‘The cup is certified to be very full.’ (totally closed scale) 19 b. guenzi hen zhi Stick HEN straight ‘The stick is certified to be very straight.’ (upper bound scale) c. Guenzi hen wan Stick HEN bent ‘The degree of bentness is certified to be very high.’ (lower bound scale) What are the possible reasons for hen to be interpreted as very when they are with absolute adjectives? If we see hen as a particle that introduces a significantly high degree as the standard based on subjective evaluation, this could be accounted for. Note that when hen is present in these sentences, the interpretation varies with respect to the scale structures. In (29)a and (29)b, both with scales that have an upper bound, hen introduces a degree that could be lower than the upper bound, but the degree needs to be close enough. It can also be licensed if the degree is exactly at the upper bound. As for (29)c, the standard is significantly higher than the lower bound, which is the degree the standard is usually set at. Presumably, when hen appear with absolute adjective, the reason that it has as stronger meaning in terms of raising the standard, similar to very, is because it is a standard derived from a comparison class determined by the attitude holder. Take man ‘full’ for example, when an attitude holder compares things that are full, there are objects that are far from full, totally full, or close to full. These things could be ranked according to how close they are to the ideal of being full. If the attitude holder thinks of it as close enough to the prototypical fullness, then it is good enough to have a significant degree. In other words, for adjectives with upper closed scales and totally closed scales, as in (29)a and (29)b, the reference points, or the standards, are the upper bounds. When hen is added, if it requires the standard to be set somewhere no higher than the upper bounds, then a significantly high degree according to typical ontology would be one that is at close to the upper bound. As for an 20 adjective with a lower-closed scale, when an attitude holder compares objects ordered on this scale, there are objects with the degree that falls exactly on the minimal degree, and objects with degrees that are slightly above it or significantly above it. Normally speaking, only those that have a significantly higher degree than the minimal degree would be considered significant and licensed by hen. On the contrary, when hen is used with a relative adjective, there is no given reference points. Therefore, since the expectation is anything goes, it is hard to figure out whether the degree is really significantly higher than any reference points. Hence, the reading that it resembles very in terms of raising the standard is weakened. The above parallelism seems to suggest that the function of hen is just like other degree modifiers in introducing a contextually given standard, which is by default provided by a comparison class. We could assume that hen function similarly to other degree modifiers, such as very and extremely, which modifies the standard introduced by pos. They are all morphemes that appears in the positive form of the adjective and derive a new and higher standard. However, hen has some unique features, which include both the subjective nature and the meaning of commitments, that need to be addressed and will be discussed later. 2.3 Commitments based on certainty The subjectivity of hen is related to commitments. The commitment made by an individual is based on one’s certainty over the possibility of an expression. How is hen linked to certainty? One reason is that the significant degree introduced by hen allows one to see a situation as a certain case based on one’s epistemic knowledge. This section is a summary of previous works on how certainty is linked to epistemic states. Once the attitude holder considers the certainty to be higher than a certain threshold, the speaker is willing to make the commitment. Epistemic commitment is about one’s attitude on evaluating the likelihood of a proposition. We could assume that a speaker is willing to commit to a proposition if and only if the certainty he has based on his knowledge over it is high enough. Commitments could be realized by different expressions. We could often find expressions on certainty expressed by modals. These expressions comes in degrees. Epistemic modals like might and must, and modal adverbs such as certainly, likely, probably, and possibly all express different degrees of likelihood 21 or speaker’s certainty. They often takes scope higher than the matrix sentence. But there are also truth conditional particles that could potentially be used to express epistemic commitments, such as be certain, certified or clear. For example, when one says it is certain that John is home, the speaker is also committed to the truth of this sentence. In the case of hen, we could see the commitments as expressed in the truth condition of the sentence. In the literature on certainty and its relation with epistemic modality, it is often discussed alone the line of possibility measures, which is first brought up by Hamblin (1959). Here is a sketch of his theory. He adopts the notion of mathematical probability to suit the purpose of describing natural languages. This model measures a finite set of worlds W, and assigns each subset of W a probability between [0, 1]. The possibility of a null set of worlds is 0, and the possibility of the entire set W is 1. This view is further adopted by Yalcin (2010). According to him, there is a relation of likelihood, ≿. Assume that Poss(p) standards for the possibility of the proposition p, and Poss(q) is that of the proposition q. Then p≿ q iff Poss(p) ≥ Poss(q). On the same vein, Yalcin provides a probability model based on his view of probability measures. He introduces the notion of probability spaces, which is a pair of ⟨E, Pr⟩, in which E is some subset of W. E is the set of worlds that are epistemically accessible to the evaluation world w. Pr is a function that assigns for each subset of W a probability between 0 and 1. Hamblin’s original view and Yalcin’s modification differ from the extra restriction over the set W. It is a set that is epistemically accessible from the evaluation world. Yalcin’s analysis (Yalcin 2010) of treating graded modals as expressing probability is discussed in many others studies (Lassiter 2010, 2011; Klecha 2012, 2014; Klecha 2012, Portner & Rubinstein 2016, among others). Furthermore, in order to express an attitude holder’s evaluation, Yalcin introduces an independent parameter apart from the world parameter. This new parameter is called the information parameter, which represents the body of information that the attitude holder uses in evaluating a sentence. This body of information is either the knowledge or the evidence one uses as the basis for evaluation. It could be represented by a parameter, which is called an information state. It restricts the worlds that are accessible 22 to the evaluation world, and it also provides quantification over the modal force in restricting a subset of the information worlds. The information state could be seen as the probability state that is used to determine whether the partition of the set of worlds are epistemically accessible from the evaluation world, the set E, is true or false with respect to the asserted proposition p. E could be partitioned into smaller subsets, and for each subset, the asserted proposition p is either true or false. In other words, the evidence and knowledge possessed by an attitude holder can determine whether a proposition is true or false in a given subset of E. Then the probability of p is the sum of the probabilities of each partitions in E. In this dissertation, I will follow Yalcin’s proposal of information parameters loosely for the analysis of hen. This view has the benefit of making evaluation to be over a subset of accessible worlds from the evaluation world. Hen would then be similar to epistemic modals in the sense that it expresses one’s certainty to be high, but in the meantime it expresses one’s commitments. Hence, it is more than a marker about certainty. In short, the relation between epistemic commitments and probability lies mainly in the fact that commitments are based on the evaluation of one’s certainty or probability. Once the probability of a proposition is evaluated to be high enough, the speaker would making the act of committing to the truth of it. Epistemic commitments and probability based on epistemic knowledge are two distinct notions. A commitment, under normal circumstances, are made with the prerequisite that the probability of the proposition is evaluated to be high. When applied to degree modifiers like hen, it would imply that the significantly high degree of the object predicated of the gradable adjective allows one to make a commitment that this object indeed falls into the category that belongs to the positive extension of thea adjective. 2.4 Commitments expressed in the truth conditional level In chapter 1, I mentioned that hen could potentially be translated to certified. The purpose of this translation, aside from the purpose for the ease of interpretation, is to show that one’s commitment is expressed as part 23 of the truth conditional component of the sentence radical. This assumption is against to the intuition that commitments are often a part of the speech act. In the previous section, I discussed how certainly and probability are correlated to commitments. Due to the fact that they are different notions, one would wonder if they are grammatically distinct. There is one enticing proposal for the distinction between the two: commitments are expressed at the illocutionary level, while probabilities are part of the sentence radical. If this dichotomy is on the right track, one could see at first glance that commitments are operators that works on the illocutionary force. Perhaps there is a functional head that represents commitments. On the contrary, we could also assume that probability or likelihood is part of the sentence radical by default. Could it be possible that the two are distinguished along the line that sets apart the dichotomy between illocutionary operators and sentential operators? Yet previous researches on syntactic positions do not show a clear cut distinction between the two. Commitments could be expressed at both levels in terms of the relative scopal property with respect to sentential operators, such as questions, conditionals and the like. For example, according to Krifka (2004), in German, epistemic modal adverbs like wohl cannot be embedded under wenn ‘if’, while another modal adverb with a similar meaning, warscheinlich ‘probably’, could be embedded (Krifka 2004). (30) Wenn es {warscheinlich/??wohl} regnen wird, sollten wir Schmirme. Intended: ‘If it is probable that it will rain, we should take umbrella with us.’ (Krifka 2004: 7) Similarly, English possibly also cannot embed under conditionals either. (31) ?If it is probably going to rain, we should take an umbrella with us. 24 The explanation for the contrast of the German duo is because modals like wohl are interpreted on the level of speech acts. Speech acts could ‘downtone’ or ‘uptone’ the degree of commitments of the speaker, following the view of Vanderveken (Vanderveken 1990). Wohl is a ‘downtoner’ that weakens the strength of the assertion (Krifka 2004: 7). On the contrary, warscheinlich, due to its capability to embed under conditionals, is interpreted as a regular modal that expresses certain degree of likelihood. Krifka argues that the two could be interpreted accordingly in (32). (32) a. Warscheinlich-type: 𝜆w. ∀w’ accessible from w, p(w’). b. Wohl-type: speaker commits to the truth of p. (Krifka 2004: 7) Yet the distinction between the two in (32) is too simplified. Imagine that there are two ways to express one’s commitments. The first one is to express it through a type of speech act with which the illocutionary force of the sentence carries the speaker’s commitments. At the illocutionary level, commitments is not expressed as part of the semantic content in the sentence radical. It could be a modification over the illocutionary force or an illocutionary operator, following the standard speech act theory proposed by Austin (1962) and Searle (1968). Searle (1968) points out that a sentence that expresses promises in the truth condition like (33)a could be viewed as equivalent to the assertion in (33)b. These two sentences show that the same force of making a promise does not always need to be part of the sentence radical. (33) a. I promise to read the novel. b. I will read the novel. 25 If the speaker commits to a proposition John is tall to a certain degree through the placing of an epistemic modal expression, such as possibly, then John is possibly tall would perhaps imply something like (34)b. In this reading, the degree that licenses possibly should be non-trivial. (34) a. John is possibly tall. b. The speaker commits to the fact that John is tall to a certain degree. Yet similarly, one could express certainty and the attitude that one’s certain about the truth of a proposition at the sentential level. This is how words like certified function. For example, if one says a sentence like (35), it implies that there is an attitude holder whose attitude towards the proposition ‘Trump didn’t win the presidential election of 2020’ is certified by him or her. In this case, there is an covert certifier that could be coindexed to either the speaker or someone salient in the context. (35) It is certified that Trump didn’t win the presidential election of 2020. In the case of hen, we also see that it is part of the sentence radical. As shown in (36), hen can be embedded under conditionals. (36) Ruguo Afu hen gao, ta hui da lanqiu. If Afu HEN tall, he will play basket ball ‘If Afu is certified to be significantly tall, he can play the basketball.’ In short, the superficial scopal property is not a good indicator to set commitments apart from certainty or probability. Because the adverb hen is part of the sentence radical, the idea that commitments could be part of the sentence radical becomes relevant for this dissertation. 26 2.5 Evidentiality and events Another aspect that is linked to the topic of this dissertation is evidentiality and its correlation with an evaluative eventuality. Some languages have a system that incorporates the meaning of evidence when making a statement. Originally, evidential markings refers to a set of linguistic markers that indicate the source of the evidence for the core proposition. Indirect evidential expressions, such as inferential evidence markers, are related to an eventuality of observing or evaluation. Hen too implies that there is temporal or spatial anchor at the evaluation. However, it does not mean that expressing evidentiality is the major function of the particle. In other words, the discussion of this section is not meant to suggest that hen itself is an evidential marker in the traditional sense. It is because in the literature on evidential sentences, the focus is mainly placed on the discussion of information sources. The notion of evidence is the knowledge used to arrive at the conclusion that the asserted proposition is correct. In a sense, a sentence that requires some salient evidence is one that is only sufficiently stated if there is some empirical information for concluding the proposition to be true. Therefore, an evidential sentence is not simply an expression of one’s belief, despite the fact that beliefs could be obtained with the existence of evidence. Why is evidentiality worth discussing here? The major reason is that the hen sentences are not simple statements about the facts of the world, nor do they only express the belief of the speaker that is sensitive to a stance. A stance could be viewed as a standpoint for an evaluation based on some known evidence. One unique feature of hen sentences is that such a perspective is often anchored to some spatial or temporal locus that serve as the reference point for evaluation. An example that shows this characteristic is in a scenario in which the perspective of the speaker changes with respect to time or location. Scenario: The speaker is climbing the steps on a hill. Before mounting the hill, there is a 100-meter high monument in the distance that looks pretty tall from this angle. But after he climbs to the top of the hill, that building is already beneath him. 27 (37) Na jinienbei ganggang hen gao, xianzai bu gao. That monument just now HEN tall now NEG tall ‘That monument was viewed as tall, but now it isn’t.’ When spoken out of the blue, the sentence seems a bit off. The reason is that it would be interpreted as if the monument would change its height. But with the added context, it becomes acceptable. It is because we may assume that the height could potentially change when one’s perspective changes. In this context, the standard for a tall building, according to the speaker’s perspective, remains constant, but it is the change in stance that allows the speaker to arrive at different conclusions. In (37), the speaker changes his evaluation over the height of the monument not simply because of the change in an objective standard given by the context. Assume that the monument, being 100 meters high, is tall by any normal standard. However, when the speaker uses the hen form, it is an evaluation based on his own perspective, and in this scenario, it is the relative location and time of the speaker that is determinate in the licensing of hen gao. As the locus of the evaluation is temporally and spatially anchored, a hen sentence implies the existence of an evaluation eventuality, which determines the valid information used for concluding the statement. Adding all these things up, a hen sentence could be conceptualized as a type of evidential expression. This is not really the case in the English counterpart, as shown in (38). The similar sentence is only acceptable under the scenario when the height of the monument indeed changes the height during time. It cannot be interpreted in terms of perspectival change at different times like the Mandarin example in (37). (38) #The monument was tall, but it isn’t tall now. Note that in the literature on evidential expressions, the emphasis is largely placed on two aspects. The first is the types of information from which the evidence is obtained, and the second is how is realized morpho-syntactically. The categorization of evidence earns special attention in earlier studies on evidential sentences (Willet 1988; Aikhenvald 2004, 2013; Aikhenvald & Dixon 2003). Evidence comes in different 28 categories. It could be classified according to how they are linked to the conclusion, be it direct or indirect. They can also be differentiated by the means the speaker or other attitude holders obtains the evidence. Examples of these categorization include direct evidence, such as attested evidence, visual or auditory evidence, and indirect evidence, which include reported or inferential evidence. Inferential evidence are related to inferences from either the results or from reasoning. Some language families, especially Tibetan- Burman, Americans and Semitic languages, are well-known for their elaborate system of evidential markings (see Palmer 1996, Aikhenvald 2004, 2007, Comrie 2000, among many others). All in all, the earlier discussions focus mainly on how to categorize evidential markings in natural languages. While the details of empirical categories and typological aspect of evidential marking is less relevant to this dissertation, there is another aspect that is. It is the correlation between the individual who obtain the information and the eventuality denoted by the proposition. It is the relative stance of the speaker (Mushin 2001, Kockelman 2004, Bergqvist 2018, among others). Stance is the means for an individual to indicate one’s orientation towards the states of affairs as denoted by the sentence (Kockelman 2004). Generally speaking, in some evidential systems in which the stance of the speaker is crucial, different evidential sentences obtain their respective meaning by introducing different relations between the speech event, the narrated event (Jakobson 1990) and the commitment event (Kockelman 2004). The speech event, according to Jakobson, is anchored to the world in which the speech occurs. The narrated event corresponds to the world that is spoken about. The commitment event, according to Kockelman, would then be relevant to the speaker’s commitment over whether a proposition is true in the same world or not. The different correlations between the three types of events is used respectively to distinguish between directness and indirectness of the evidence, whether the speaker participates in the eventuality itself, whether it is factive and so on. An example of the last type of event is often realized in evidentials with counterfactual meanings. Take the Q’eqchi’-Maya example given by Kockelman (2004: 127). This language has several modal clitics that could represent different types of speakers’ commitments. It includes pe’, which is a factive marker that states that in the world of the speech event, the speaker is committed to the proposition p to be true. 29 There are other markers, such as counterfactuals raj or optative taxaq, which are anchored to either a counterfactual world and a wish world respectively. How is the above discussion relevant to our discussion here? As one can see, while unconventional, commitments could be expressed as part of the evidential system in some languages. If we want to make an analogy with the classification of Q’eqchi’-Maya, it seems that what hen could be classified is the factive stance. One may question that may arises is that Mandarin Chinese is not usually considered a language with evidential markings. Yet there are also discussions showing that evidential meaning could be realized in many different categories. They are not necessarily realized as in the grammaticalized evidential system. Instead, some languages allow the implication of evidence to be realized in various grammatical categories. One of the examples is epistemic modality. Epistemic modality itself provides restriction on the modal base. Other categories, such as aspectual markers could also be linked to evidentiality. According to Izvorski (1997), the present perfect aspects in Bulgarian, Turkish and Norwegian are epistemic modals with universal force that encodes the meaning the existence of evidence in the process of justifying the proposition. For example, Izvorski argues that the present perfect aspect could be regarded as indicating the existence that there is indirect evidence for the truth of the proposition. In English, it is argued that epistemic modals like must provide some restriction over the information that is used to justify the prejacent. This set of proposition are either regarded as indirect evidence (von Fintel & Gilles 2008) or inferential evidence (Rett 2013). We could see that modal expressions in languages without an established evidential system do not directly provide the information about how the evidence is obtained, following Willet’s categorization. However, they may express whether the evidence is direct or indirect through the restriction on the modal base by providing a presupposition, as in the case of epistemic must in English. Other examples of non-grammaticalized lexical words that expresses evidentiality include verbs and modal adverbs. For instance, perceptual verbs assert that the speaker has perceptual evidence. English adverbs like apparently and clearly also indicates the existence of evidence (Izvorski 1997, Barker 2009, Crone 2018), although the types of felicitous evidence are also underspecified in these expressions. 30 Perceptual verbs and evidence types: (39) a. I hear John singing. (direct evidence: auditory) b. I hear that John was singing. (indirect evidence; report) (Izvorski 1997: 4) Adverbs presupposing the existence of evidence (underspecified evidence): (40) a. John was apparently singing. b. John is clearly home. c. John is obviously a teacher. d. John is evidently home. (Izvorski 1997: 4) Hen can be viewed as a particle that draws resemblance to evidential markers like evidently, which is a particle that represents the assessment based on the attitude holder’s possessing of evidence. It is worth noting that there is a debate over whether the meaning of evidentiality is expressed through propositional or an illocutionary level operator. Across languages with evidential systems, this distinction may be typological (Faller 2002, 2006): some languages utilize propositional level operators while others may express evidentiality through illocutionary operators, such as Cusco Quechua, and some utilize both. According to Faller, the major support for the distinction comes from scopal properties of particles in different languages. It is argued that languages like Cusco Quechua, which has a grammatical evidential system, evidential enclitics are realized as illocutionary modifiers (Faller 2002, 2006). Different from Cusco Quechua, German reportative modal sollen ‘said’ is argued to be a propositional level operator. Take English adverbs such as clearly and evidently for example. These expressions contributes to the propositional content, and could have narrow scope with respect to some sentential operators. Modal adverbs such as probably or possibly are less acceptable when embedded. 31 (41) a. The airplanes were not clearly/evidently bombers. b. ??The airplanes were not possibly/probably bombers. As mentioned before, the evidential scope of hen would then be closer to the latter case: it is an evidential marker that operates on the sentential level, not on the illocutionary level. 2.6 Conclusion This chapter provides a sketch of what may be linked to the unique features of the particle hen, which draws resemblance to evaluative sentences, subjective judgment and certain aspects to evidentiality. All these features could be subsumed under the notion of speakers’ commitments, an attitude based on certainty. 32 Chapter 3 The puzzles of hen 3.1 Overview In the literature, there are several different accounts on Mandarin hen. One of the mainstream views is that it is the overt realization of a positive morpheme, which is phonologically covert in many other languages, such as English. Some argues that this morpheme is simply a degree modifier that modifies a standard given by pos. However, both views have their shortcomings. This chapter points out some of the puzzles that cannot be accounted for by previous analyses. 3.2 Previous analyses on hen 3.2.1 Hen as a positive degree morpheme In the literature, hen is a marker that has long been considered as the realization of the positive degree morpheme (Sybesma 1999, Kennedy 1999, Kennedy & McNally 2005, Huang 2006, Liu 2010). A positive degree morpheme pos serves two functions 1 . First, it introduces a contextually fixed standard that sets the degree for comparison. In other words, it determines the set of individuals that reaches this standard to count as having the property described by the adjective. Second, it resolves the type mismatch between a gradable adjectives, which is of type <d, <e, t>>, and an individual-denoting subject, which is of type e. This analysis sees pos morpheme as obligatory in a positive degree sentence. Based on the assumption that pos should exist in positive degree sentences across the board, it has been regarded as a puzzle over why pos is not phonologically realized, at least in some of the most well- studied languages like English. In a sense, the necessity to posit a morpheme hen here in the Mandarin case 1 Here I follow the proposals of Cresswell (1977), Bierwisch (1989), Kennedy (1999, 2005), Kennedy and McNally (2005) and Graff (2000) on the function of pos. 33 becomes a welcoming result supporting the view that pos indeed exists (Kennedy 1999, Liu 2010, Zhang 2015). One of the major advocates of the pos analysis is Liu (Liu 2010). Liu clearly points out that hen is the positive degree morpheme. It occupies the head of the degree phrase, DegP. His main argument is based on the fact that hen is obligatory, and the meaning is quite bleached when it is in the positive degree form. He presents some parallelism between Mandarin hen and English pos sentences. A positive degree sentence with hen shows the same property in data like implicit comparison, which has some unique features mentioned by Kennedy (2007b). Implicit comparison refers to a type of comparative form that does not involve an overt comparative morpheme within a sentence, but is instead done by a biclausal structure, and one of which is the positive form, as shown in (1). (1) Compared to Mary, John is tall. An implicit comparative form is semantically different from ordinary comparatives like John is taller than Mary. Aside from their syntactic properties, the major distinction is that in a sentence like (1), the degree that John exceeds Mary is significantly high. Therefore, if John’s height is only 1mm taller than Mary, and if 1mm is trivial in the context of utterance, (1) is infelicitous. Similarly, in a hen sentence, we also see the same licensing condition. In (2), Afu’s height exceeds Lisi’s significantly. (2) Gen Lisi bi, Afu hen gao. To Lisi compare Afu HEN tall ‘Compared to Lisi, it is certified that Afu is certified to be tall.’ Another analysis that sees hen as a pos-like argument is from Huang (2006). He provides a predicate- forming theory on hen, which resembles the standard account of positive degree sentences. Huang argues that simple adjectives in Chinese are nominal expressions of type e. Therefore, they need to combine with 34 a type-shifting element to become a predicative expression. In Mandarin Chinese, this particle is hen. Huang’s analysis is based on the assumption that Mandarin does not distinguishing between the semantic types of nominal and adjectival expressions. Both of them share the same morphological forms. For example, hong ‘red’ could refer to the nominal red color or the adjectival expression of red. This view presupposes that simple adjectives in Mandarin are not gradable, and hence do not contain a degree component. Therefore, the function of hen is to type shift the type e individual into a type <e, t>. However, it is different from the type shifting mechanism of pos, which shifts a gradable adjective into a predicate of individual of type <e, t>. The pos analyses gives one major prediction: hen itself does not have much added meaning except for the type shifting function that shifts an adjectival expression into a predicate. However, this analysis is unable to account for some features of hen. The first problem comes from the intuition that hen describes a significantly high degree, making it patterns with degree modifier that functions as an operator that raises the standard. This is especially obvious when there is clearly a stipulated standard already given in the context. The hen form is used to describes a situation in which the standard is significantly exceeded Context: the degree for tall people is 180cm. (3) Afu hen gao. Felicitous context: Afu’s height ≫ 180cm Infelicitous context: Afu’s height = 180cm In contrast, English positive degree sentences do not have the requirement that the degree is well above some given standard. This is shown in the contrast with (4) under the same context. 35 (4) John is tall. ✓Afu’s height ≫180cm ✓Afu’s heigh = 180cm As shown above, despite the parallelism between English pos and hen, the data of implicit comparison is not a good argument to see them as identical. If hen is not as bleached as one might think, and if it refers to some significant degree, we could easily explain the condition imposed on (2). The English sentence could be true of a standard that is explicitly given in the context while the Mandarin hen cannot. Furthermore, if we simply assume that hen is pos, there is the question of entailment. In Mandarin Chinese, there is another positive expression, the shi…-de form. It implies a categorical conceptualization of adjectival expressions. It is the default form of non-gradable adjectives, as shown in (5), which is a nominalized expression that describes a categorical concept of the group of people that are tall. It requires a fixed standard already given in the context. When a speaker chooses to use shi…-de, the standard is presupposed and backgrounded. An example like Afu shi gao-de ‘Afu belongs to the category of tall people’ implies that the participants of the conversation agree on a collectively known standard in mind, and Afu’s height exceeds that standard. It is the default positive form for absolute adjectives, but it is a marked form for relative adjectives. When it occurs with relative adjectives, it implies that there is a contextually determined standard that divides a given set of individuals into two categories, the ones that reach that standard, and the ones that do not. There is no ordering between the objects in these two categories. For example, in (6), Afu belongs to the category of tall individuals, of which the standard of being tall is contextually determined. There is no salient ordering of degrees in these two sets. Therefore, a shi…-de form cannot be modified by a degree modifier. 36 (5) Absolute adjectives (default with shi…-de form): Shi-de ‘wet’, zhi-de ‘straight’, yuan-de ‘round-shaped’, fang-de ‘square-shaped’, wan-de ‘bent’ Shi…-de with relative adjectives: (6) Afu shi gao-de. Afu COP tall-DE ‘Afu belongs to the category of tall individuals.’ (7) *Afu shi feichang gao-de. Afu COP very tall-DE Intended: ‘Afu belongs to the category of very tall individuals.’ 2 There is another distinction between the hen form and the shi…-de form. The shi…-de form entails the the proposition that the subject reaches a certain standard degree, but it does not entail the degree to be any higher than the standard, let along being significantly higher. Therefore, the hen sentence entails the shi…- de form, but not vice versa. (8) a. Afu shi gao-de ↛ Afu hen gao b. Afu hen gao → Afu shi gao-de 2 In (7), the sentence is not ungrammatical per se, but it is interpreted as a cleft sentence in which feichang gao de is a relative clause. The particle -de here is not an adjectival marker. It is attached to the entire predicate phrase feichang gao. The sentence has a focus reading which could be translated as ‘Afu is indeed very tall.’ 37 Furthermore, the pos analysis of hen faces another problem when it comes to negation. The first is that when the hen form is negated, hen does not need to be present. I call this uninflected adjective form the bare form. The negated sentence with the bare form simply asserts that Afu does not reach the standard. It has the same meaning as the English positive sentence in (10). Note that the negation form used here is bu, the canonical negation for stative predicates. There is another negation marker mei. It appear with dynamic predicates and certain subtypes of stative verbs that requires hen to be present. The use of mei is one of the major problem with hen. It will be discussed in chapter 5 as well. (9) Afu bu gao. Afu NEG tall. ‘Afu is not tall.’ (10) Afu is not tall. Meaning: Afu does not reach the contextual standard for tall individuals. The fact that bu does not require the presence of hen in order to negate the positive form is quite interesting. The contrast between mei and bu suggests that hen introduces an evaluative eventuality. 3.2.2 Hen as a copula The second analysis for the status of hen is to see it as a special type of copula. Huang’s view discussed in the previous studies could also be regarded as seeing hen as a type of copula that shifts a nominal expression into a predicate. However, he does not argue that hen itself is syntactically a copula. Yet this view is argued by He & Jiang (2011), who discuss a special use of hen that coerces a noun into a gradable expression. 38 (11) Xiaohua hen shunü. Xiaohua HEN lady-like ‘It is certified that Xiaohua is ladylike.’ (He & Jiang 2011: 892) Usually shunü is used as a nominal form ‘lady.’ But when hen is present, it is interpreted as a gradable adjective that denotes a typical property of being a lady. For example, the property could refer to having the demeanor of a woman from high society, or to behave in a refined manner. Furthermore, the property coerced from the nominal expression is not an intrinsic property of the object. It is an extrinsic one, a distinction first brought up by Lewis (1983). An intrinsic property is one that an object has of itself, which is held independent of other things. An extrinsic property is one of which the property is fully or partially dependent on other existences. Therefore, intrinsic properties are often related to kinds, while extrinsic properties could be viewed as related to a stage of that individual. In (11), the lady-like-ness is only a temporary property based on the behavior of Xiaohua at a given time or situations. It could be translated literally as ‘Xiaohua is in a state of being lady-like.’ He & Jiang argues that nouns in Mandarin Chinese are always kind denoting. Therefore, hen takes a kind noun and derives a salient property related to the noun, and the property is only stage-level. He & Jiang’s work presents an interesting perspective: a hen sentence expresses a property that is potentially temporally bounded. However, they do not propose that hen necessarily forms a bounded predicate. Under specific contexts, the stage of being lady-like could be permanent, although it is more difficult to obtain. Their analysis also suggests that hen, at least in this particular use, could be viewed as some sort of copular expression that forms a predicative phrase. Hen is a type shifter that makes a nominal expression to become a predicatable category. Now two questions arise. Is He & Jiang’s copula analysis on the right track? If seeing this particular use of hen as a copula is correct, we would expect it to be also applicable to the typical use of hen we are discussing here. The answer to both is ‘no.’ He & Jiang’s analysis faces a very straightforward problem. 39 One argument is that hen is not obligatory if we want to coerce a reading from a nominal expression. This is shown in the bare form in (12). Although the bare form is quite marked, it could be used in contrastive focus structures. In (12), the property for Xiaohua being lady-like is compared with the property of Lisi being gentlemanly. Shenshi is also a nominal expression, and it is coerced into an adjectival expression ‘gentlemanly’ in this use. (12) Xiaohua shunü, Lisi shenshi. Xiaohua lady-like. Lisi gentlemanly ‘Xiaohua is lady-like.’ Furthermore, nominal expressions could be coerced into ordinary gradable adjectives even in comparatives. (13) Afu bi Xiaohua shunü. Afu COMP Xiaohua lady-like ‘Afu is more lady-like than Xiaohua.’ If we look at (12) and (13) together, we can conclude that there should be another particle in charge of the coercion. In other words, we could either assume that there is an independent particle which is phonologically covert that triggers the coercion, or there are three different particle in each of these structures, the hen form, the bare form, and the comparatives, that could get the job done. In either cases, it would not directly support the view that hen is a copula with this unique type shifting function, since the same effect could be obtained elsewhere, as shown in (13). It is not a sound argument if we consider hen in this particular use is a copula simply because it has this type shifting function. It would be based on the assumption that all the particles given above, including comparative forms and the bare forms, all are capable of coercing a noun into a predicate. It sounds unlikely. Either there are other grammatical particles 40 that are involved in the coercion or these nominal expressions are not really nominal. They could be homophonic adjectives in disguise. The data with nominal expressions are not enough to argue that hen is a copula. Perhaps even more problematic with the copula analysis is that hen in this particular use does not show the same pattern as the main copula shi in Mandarin Chinese. This is shown in sentences in which hen co- occur with both nominal expressions and absolute adjectives. Shi could be repeated and form a yes-no question by using a reduplicated form, which is the A-not-A structure. According to Huang (1982), the A- not-A form is licensed when there is a question feature on Infl or C. Only copulas, modals, adjectival or verbal elements could be reduplicated. The repeated part in a A-not-A form are usually located at the top of the predicate. (14) a. Afu xihuan bu xihuan xigua? (Verbs) Afu like NEG like watermelon ‘Does Afu like watermelon?’ b. Afu shi bu shi laoshi? (Copula)c Afu COP NEG COP teacher ‘Is Afu a teacher?’ c. Afu hui bu hui shangban? (Modals) Afu will NEG will work ‘Will Afu go to work?’ However, the hen form cannot appear in A-not-A form, showing that it is not a copular particle (Zhang 2015). 41 (15) *Afu hen bu hen gao/ shunü? Afu HEN NEG HEN tall/lady-like Intended: ‘Is Afu tall?/ Is Afu lady-like?’ (16) *Xiaohua hen bu hen shunü? Xiaohua HEN NEG HEN lady-like ‘Xiaohua is ladylike.’ Another counter-evidence comes from ellipsis in answering A-not-A questions. The part of the predicate following shi can be elided in answering a question, while in a hen sentence, the part following hen cannot be deleted, as shown in (18), hen cannot be stranded alone. Other heads of the predicate, such as verbs and modals, could both be stranded, as in (19). Copula shi: (17) A: Afu shi bu shi laoshi? Afu COP NEG COP teacher ‘Is Afu a teacher?’ B: Afu shi. Au COP ‘Afu is.’ Hen: (18) A: Afu hen congming ma? Afu HEN smart Q ‘Is Afu smart?’ 42 B: *Afu hen. Intended: ‘Afu is.’ Modals and verbs: (19) a. A: Afu hui bu hui shangban? Afu will NEG will work ‘Will Afu go to work?’ B: Afu hui. Afu will ‘Afu will.’ (20) b. A: Afu xihuan bu xihuan xigua? Afu like NEG like watermelon ‘Does Afu like watermelon?’ B: Afu xihuan. ‘Afu likes (watermelons).’ Therefore, it is counterintuitive to view hen as a copula, and it would be more natural to see it as modifier of the adjective. This is shown in the fact that hen could be easily replaced by other degree modifiers, as in (21), and they do not juxtapose, as in (22). (21) Afu hen/ feichang/ xiangdang gao. Afu HEN/very/quite tall ‘Afu is very/quite tall.’ 43 (22) *Afu (feichang/ xiangdang) hen gao. Afu very/quite HEN tall In principle, if hen is indeed a copula, we would need to view these degree modifiers as copulas as well. This would be unlikely. 3.2.3 Hen as a degree adverb In this section, I show analyses that see hen as a degree modifier. hen patterns more with degree modifiers in Mandarin Chinese, but it cannot be translated as the counterpart of very. In the literature, hen is sometimes viewed as a kind of degree modifier. There are two major perspectives supporting this view. The first is to see it as a semantically bleached degree modifier (Li & Thompson 1981, Grano 2012). Li and Thompson points out that when hen has a neutral meaning, it is a dummy marker. Yet when it is focused, it has the meaning of very. The presence of hen serves the function of making the gradable adjective predictable to the individual subject. According to Grano (2012), hen is a degree modifier that is projected in order to meet certain syntactic restrictions. Grano argues that Mandarin Chinese has a language specific requirement: the direct object of Tense need to have a [+V] feature, which could be projected either by a verbal expression or an extended projection of a verb. A gradable adjective in Mandarin Chinese does not have a [+V] feature on its own. It need to merge with a comparative morpheme or a degree adverb like hen to fulfill the requirement. I agree with Li & Thompson and Grano’s views in terms of the syntactic category that hen belongs to. In many ways, hen do behave like an ordinary degree modifier, at least syntactically. However, I do not agree with their views that they see it as a particle with little semantics of its own. It is easy to see why this is problematic. First of all, it is obvious that hen raises the standard to a significantly high degree. Furthermore, it does not function merely to raise the standard. We could compare hen with English very, a degree modifier that is used to raise the standard only without other added meanings. However, the data 44 from the entailment relations shows that hen and very are quite distinct. As very raises the standard, a sentence with very entails the positive form. (23) John is very tall ⟶ John is tall. Subsequently, when the sentences are under negation, the scale is reversed. If someone is not as tall as 180 feet (the degree of being tall), then he cannot be 185 (the degree of being very tall). (24) John is not tall ⟶ John is not very tall. Of course, when the two propositions are reversed, the entailment relation does not go through, since John can still be tall while not being very tall. (25) John is not very tall. ↛ John is not tall. Furthermore, it is felicitous to utter a sentence like (26) without causing contradiction. (26) John is not very tall. Well, I mean he is still tall. This is not the pattern we see in hen. When hen is present, it has a different meaning from the negation of a simple positive form. According to the judgment from native speakers, the use of hen tones downs the assuredness, which seems to provides a sense that the speaker does not want to make a strong commitment over whether Afu is indeed not tall. This is shown in (27). 45 (27) Afu bu (shi) hen gao. Afu NEG COP HEN tall ‘Afu is not very tall.’ (Canonical negation: the bu form) As shown in (28), if hen is any kind of degree modifier that raises the standard based on the one provided by pos, one would expect the negated bare form (the positive form being negated) to entail the negated hen form. Surprisingly, this entailment relation does not hold (28) a. Afu bu gao. ↛ Afu bu hen gao. Afu NEG tall. Afu NEG HEN tall ‘Afu is not tall.’ ‘Afu is not certified (by someone/the speaker) to be very tall.’ b. Afu bu congming ↛ Afu bu hen congming. Afu NEG smart Afu NEG HEN smart ‘Afu is not smart. ‘Afu is not certified to be very smart.’ The hen sentence in (28) seems to have a different meaning from English ‘Afu isn’t very tall.’ There is less certainty over whether the object actually passes the significant standard set by hen, and it does not entail that Afu is tall. The best way to interpret the sentence is to see it as negating a certification from some individual, which is most likely the speaker. In that case, we can explain why Afu hen gao does not intuitively entail Afu bu hen gao. The reason that the entailment of (28) does not go through could be explained if we see hen as introducing a certification of some individual but not by the general public. If it is a certification of the general public, the entailment relation actually holds. Similar to English, the reverse does not follow, as in (29). 46 (29) Afu bu hen gao ↛ Afu bu gao. Now let’s turn to the compatibility between the shi…-de form and the hen form. If hen is indeed very, we would expect the hen form to entail the shi…-de form as well. This prediction is born out. (30) Afu hen gao ⟶ Afu shi gao-de Afu HEN tall Afu COP tall-DE ‘It is certified that Afu is tall.’ ‘Afu belongs to the category of tall people.’ However, when under negation, the entailment relation between the hen and the shi…-de forms is not reversed. (31) Afu shi gao-de ↛ Afu bu hen gao Finally, the negation of shi…-de form is not quite compatible with the negated hen form, as shown in (32). The intended meaning for hen is that it is interpreted as very, which is apparently not the case. This is distinct from the English near counterpart in (33). (32) a. #Afu bu shi hen gao, eh, ta hai shi gao-de. Afu NEG COP HEN tall well he still COP tall-DE Intended: ‘Afu is not very tall. Well, he is still tall.’ 47 b. #Afu bu shi hen congming, eh, ta hai shi Afu NEG COP HEN tall well he still COP congming-de smart-DE Intended: ‘Afu is not very smart. Well, he is still smart.’ (33) a. John isn’t very tall. I mean he doesn’t belong to the category of tall people. b. John isn’t very smart. I mean he doesn’t count as a smart person. In fact, the degree modifier feichang, which is superficially translated as very, patterns more with very. When it is negated, it is entailed by the bare form (positive form). (34) Afu bu gao. ⟶ Afu bu shi feichang gao. 3 Afu NEG tall Afu NEG COP very tall ‘Afu is not tall.’ ‘Afu is not very tall.’ Likewise, the opposite direction of the entailment is also the same as the English counterparts. 3 Here the copula shi emerges when the sentence is negated by the negation marker bu perhaps for phonological reasons. When the degree modifier is bi-syllabic or has more than two syllables, shi emerges to satisfy some language- specific prosodic requirements. This is shown in the contrast between a sentence with monosyllabic modifiers and bi- syllabic modifiers. There might be reasons other than phonological constraints that influences the distribution. I leave this for future studies. (1) Monosyllabic modifiers: bu (shi) ting gao; bu (shi) hen gao NEG COP rather tall; NEG COP HEN tall Bi-syllabic modifiers: bu *(shi) feichang gao; bu *(shi) xiangdang gao NEG COP very tall NEG COP quite tall 48 (35) a. Afu bu shi feichang gao. ↛ Afu bu gao. Afu NEG COP very tall Afu NEG tall b. Afu bu shi feichang congming. ↛ Afu bu congming. Afu NEG COP very smart Afu NEG smart The negated form of a feichang sentence is compatible with the sentence in the shi…-de positive form. (36) Afu bu shi feichang gao, eh, ta hai shi gao-de. Afu NEG COP very tall well he still COP tall-DE ‘Afu isn’t very tall. Oh well, he still belongs to the category of tall people.’ Table 1, 2 and 3 below are summaries of the entailment relations of very, Mandarin hen and Mandarin feichang ‘very’ respectively. The three darker columns in the three tables shows the major differences between the three. Very and feichang have the same entailment relations. Sentences contains these two are both entailed by the negated positive sentences. Hen, however, is different from the two. Table 1 Entailment relations for very Entailment (from {A, B} ⟶ {I, II}) I. ¬𝑝os-Adj. II. ¬very-Adj. A: ¬𝑝os-Adj. Yes Yes Ex. John is not tall. ⟶ John is not very tall. B: ¬very-Adj No Ex. John is not very tall. ↛John is not tall. Yes 49 Table 2 Entailment relations for hen Entailment (from {A, B} ⟶ {I, II}) I. ¬∅-Adj. II. ¬ℎen-Adj. A: ¬∅-Adj. (bare form; Equivalent to English ¬𝑝os-Adj.) Yes No Ex. Afu is not tall ↛ (Afu bu gao) Afu is not hen-tall. (Afu bu shi hen gao) B: ¬hen-Adj. No Afu is not hen-tall. ↛ (Afu bu shi hen gao) Afu is not tall. (Afu bu gao) Yes Table 3 Entailment relations for feichang Entailment (from {A, B} ⟶ {I, II}) I. ¬∅-Adj. II. ¬feichang-Adj. A: ¬∅-Adj. (bare form: Equivalent to English ¬𝑝os-Adj.) Yes Yes Ex. Afu is not tall . ⟶ (Afu bu gao) Afu is not feichang-tall. (Afu bu shi feichang gao) B: ¬feichang-Adj. No Afu is not feichang-tall. ↛ (Afu bu shi feichang gao) Afu is not tall. (Afu bu gao) Yes The above data show that hen does not pattern with the English pos and the English very, nor does it pattern with Mandarin feichang. Rather, very patterns more with feichang. If we see pos and very as the type of degree modifiers that function mainly on switching the standard upward along the scale without other significant added meanings, then hen certainly does not belong to this category. 50 3.3 The problem of subjectivity 3.3.1 Subjectivity in hen One important feature characteristic of hen is its association with the speaker or an attitude holder’s perspective. The notion of perspectives is related to subjectivity. Subjectivity is a cover term for expressions that are associated with personal opinions, knowledge and attitudes. Hen can be used to express evaluation inferred from direct experiences. It can also be related to a comparison class derived from personal preferences that does not need to be based on direct experiences. There are several different types of subjectivity. It could be expressed through epistemic modality and evaluatives. There is a certain subclass of adjectival expressions used to express subjective opinions, such as Predicates of Personal Tastes (PPTs) (Lasersohn 2005, 2009, 2017; Stephenson 2007, 2010; Sæbø 2009; Stojanovic 2007; McNally & Stojanovic 2017; Pearson 2013, 2015; among many others), and aesthetic predicates (McNally & Stojanovic 2014). The source of subjectivity also varies. The first type is related to opinions, which is about feelings or an inference derived from direct personal experiences. An expression with this type of subjectivity does not require any assessments from other individuals. PPTs are one of those that express such pure subjective opinions based on direct experiences. Furthermore, epistemic expressions that are based on ‘subjective’ knowledge are also considered as subjective expressions. Finally, subjectivity could also be realized in the standard setting of gradable adjectives, especially in the selection of comparison classes. For example, in a sentence like John is tall, the standard for tallness could be determined by an attitude holder. This standard could vary from person to person according to their understanding of what counts as tall. Therefore, a sentence like the basketball player is tall could be regarded as subjective if we take into consideration of how the standard is selected. Perhaps for one person, a tall basketball player should be at least two meters tall. Perhaps for another person, 1.9 meters is sufficient. Therefore, even with adjectives that are related to an objective dimension, such as height, it could also be subjective. But since the subjectivity lies in the individual preferences for comparison classes, it is still possible that people have consensus or disagreements on these comparison classes once its denotation is made explicit. This latter example of 51 subjectivity is a feature of vague predicates, which often involve covert indexicals selected by different individuals in the truth conditions. What are the subjectivity involved in hen sentences? Aside from the fact that the standard could be subjectively selected, hen can also be used to express inferences derived from personal or even direct experiences. To summarize, the subjectivity of hen lies in several aspects. Its subjectivity is not only related to a contextually selected comparison class chosen by some individuals. It could also be linked to direct personal experiences. The latter is shown in the licensing of hen sentences in a scenario that requires a unique experience that is not able to be experienced by other individuals. In the latter case, the sentence only needs the attitude holder’s personal assessment. Imagine a scenario like this. When a speaker says someone is hen-tall, he can think of a personal experience purely about one’s feelings. The speaker could at some point stand close by a person and suddenly feels that the person’s height to be overwhelmingly tall because he overshadows the attitude holder. This experience could draw the inference that the degree of that person is indeed significant. But such an inference is only assessed by the speaker. Furthermore, in the case of hen, its function is not merely to introduce a standard, or to provide a new standard based on the positive standard. Rather, it also provides the meaning that the degree is high enough for one to present it as a sure case. It resembles the using of the words like plainly or clearly, as in John is plainly/clearly tall. The degree is high enough for the speaker to think of the object as falling into the positive extension of the adjective, although it is not a requirement for it to be universally acknowledged for John to be tall. Hen also involves an event of evaluation. It could license the scenario in which the evaluation is a unique experience to the speaker, which is about the matter of opinion. In this respect, it resembles the Predicates of Personal Tastes. Aside from faultless disagreements, other pieces of evidence including hen sentence could be embedded under the expression gangxiang shi, ‘the opinion is’, an expression that strictly requires the 52 embedded proposition to express one’s opinion. This is discussed in 3.3.3. Finally, in 3.3.4, I will discuss how hen parallels with, quite and sort of. 3.3.2 Faultless disagreements A diagnostics for opinions is related to faultless disagreements. This could be used as a test to see if the sentences expresses personal opinions. Specifically, it is related to the type of personal experience that is derived from direct evidence. Faultless disagreement usually refers to the situation in which least two parties disagree with each other, with at least one participant states that a proposition p is true, while the other participant(s) states that non-p is true. While there is an contradiction, both parties are not making false statements, at least on the surface. Faultless disagreements are often linked to the diagnosis of Predicates of Personal Tastes (PPTs), the type of adjectives like fun or tasty. These adjectives are judgments based on direct experiences. For example, for a person to think of a cake to be tasty, that person need to have tasted the food. If this attitude holder does not taste the cake, he or she cannot make a claim like ‘the cake is tasty.’ (42) Speaker A: The cake is tasty. Speaker B: No! It’s not. It isn’t tasty. In (42), an example with a PPT, both parties are not at fault since they are only expressing their respective opinions. The faultlessness disappears when the adjective is not a PPT. In (43), either A or B is correct once the standard for tallness is contextually given. (43) Speaker A: John is tall. Speaker B: No! he is not. 53 Perhaps the reason that causes this disagreement is due to a less discussed aspect on PPTs. In my opinion, aside from the requirement of direct experiences from the attitude holders, PPTs are licensed when they are experiences unique to each attitude holder, which cannot be replicated or re-experienced by other individuals. Judgments like fun or tasty are conclusions derived directly, or being generalized, from these experiences. Therefore, when there are two different people disagreeing over whether a cake is tasty, it is impossible for them to actually replicate each other’s experiences. The term faultless disagreements is used only to describe the superficial characteristics defined by the seemingly endless arguments between two parties. There are various arguments over whether faultless disagreements really exist or whether it is simply a type of misunderstanding, which could include situations like the lacking of contextual information (see Kölbel 2004): there could be covert indexicals that is not otherwise expressed in the sentence, such as the standard for fun or tasty, resulting in the misunderstanding. In other words, the faultlessness might only be superficial. PTTs may sound faultless because the criteria for judging what is considered tasty is not explicitly given. However, in my opinion, the test of faultless disagreement is a legitimate test because PTTs are used to express direct experiences exclusive to the attitude holder. Now we turn to the hen sentences. Hen sentences is subjective in two ways. First, it is the same as other degree modifiers or pos. They could be involved with a comparison class determined by personal opinions. However, it could also be used to represent opinions derived from experiences exclusive to the speaker. This is shown in the example in (44). Scenario: Afu is a man whose height is slightly below average. But speaker A had an experience to be standing by Afu, and since Afu was taller to than A and overshadowed her, A found Afu to be tall because of that moment. In short, the inference that Afu is tall is based on A’s personal experience. (44) Speaker A: Afu hen gao ‘Afu is certified (by me) to be tall.’ 54 Speaker B: Bu-dui, wo cong bu juede. Meiren juede. NEG-right I never NEG think nobody think ‘Wrong. I don’t think so. Nobody ever think so.’ Speaker B cannot direct deny A’s opinion that Afu is tall, since A is referring to a unique experience that she finds Afu to be tall based on that experience. Unless B can have the same experience as A, B cannot have a say in how A feels. Therefore, even if B wants to justify his argument by addressing how the general public might think of Afu’s height, it is useless because that is not what A bases her view on. In comparison, in English positive sentences, it is harder to license a sentence like ‘Afu is tall’ or ‘Afu is very tall’ in the same scenario as that in (44). In English, one cannot ‘feel’ that someone is tall to call him tall. That person needs to fit into some sort of criteria that can be examined and assessed by other individuals. Furthermore, when a non-subjective adjective like gao is in the comparative form, it does not trigger faultless disagreements. The reason is obvious: if a sentence is used to express one object to be taller than the other, it is an objective fact. It is not an inference based on personal experiences. This is no different from the English comparative counterpart. A comparative sentence is not vague in terms of standard setting. It only compares two degrees, one corresponding to the object often denoted by the sentential subject, and the other corresponding to the object that is being compared to. Given the fact that the dimension is height, which is not subjective, the assertion that Lisi is taller than Afu is not questionable. (45) Speaker A: Gen Lisi bi, Afu bijiao gao. With Lisi compare Afu in comparison tall ‘Compared to Lisi, Afu is taller.’ Speaker B: Bu-dui. Lisi bijiao gao. NEG-right Lisi in comparison tall’ ‘Wrong. Lisi is taller.’ 55 3.3.3 Embedding under attitude verbs A diagnosis often mentioned in the literature of subjective predicates like PPTs is the embedding under attitude verbs. Some attitude verbs only takes subjective predicates. According to Sæbø (2009), Swedish verb synes ‘seem’, Geman finden and English find are one of those examples. For example, find is infelicitous when the embedded sentence denotes a factual or a non-debatable proposition. In this section, I show that hen patterns with other subjective predicates in embedding under attitude verbs. Hen licenses a non-subjective adjective to appear under an attitude expression ganxiang shi, ‘the opinion is.’ It is used as an adjunctive phrase for expressing personal opinions only. As shown in the contrast between (37)a and (37)b, the expression haochi, ‘tasty’, is an opinion, while ‘Afu is at home’ can only be an objective fact. Only (a) is acceptable. (37) a. Wode ganxiang shi zhe lamian haochi. My opinion COP this ramen tasty. Lit. ‘My opinion is that this ramen is tasty.’ b. *Tade ganxiang shi Afu zai jia. His opinion COP Afu at home Lit. ‘He has the feeling is that Afu is home.’ The hen form can license ganxiang shi, when showing that (38)a patterns with the subjective predicates. On the contrary, the shi…-de form does not. Imagine a scenario in which the speaker sees Afu for the first time, and he describes how he feels about Afu. It is more acceptable to use the hen form. (38) a. Wode ganxiang shi Afu hen gao. My opinion COP Afu HEN tall Lit. ‘My opinion is that Afu is tall.’ 56 b. *Wode ganxiang shi Afu shi gao-de. My opinion COP water COP gao-DE Lit. ‘His opinion is that Afu is tall.’ Furthermore, the comparative form sounds funny if it is an unquestionable fact that Afu is heavier than Lisi. Compare (38)a with (39). (39) ??Wode ganxiang shi Afu bi Lisi zhong. My opinion is Afu COMP Lisi heavy Lit. ‘My opinion is that Afu is heavier than Lisi.’ (38) is more felicitous in situations in which the speaker actually involved in weighing Lisi, or get to realize that Lisi’s weight by an evaluation or inference. In these cases, it is a result based on experiences. Ganxiang shi is also licensed when an epistemic modal expression is present, even when the prejacent is not subjective. (40) a. Wode ganxiang shi, Afu you keneng zai jia. My opinion COP Afu has possibily at home ‘My opinion is that Afu might be at home.’ b. Wode ganxiang shi, Afu keneng bi Lisi zhong. My opinion COP Afu possibly COMP Lisi heavy ‘My opinion is that Afu might be taller than Lisi.’ Note that in (40), the sentence is most natural when the epistemic modal is interpreted subjectively. Subjective epistemic modals refers to those that are relativized to personal beliefs (Lyon 1977, Papafragou 57 2006, Rett 2012). It is distinguished from objective epistemic modals. The latter is regarded as taking more generally known knowledge from a relevant community as the basis for the assertion. According to Stephenson (2007), one way to disambiguate the two types of epistemic readings is to see how they are interpreted under attitude verbs. English epistemic modals are interpreted subjectively when embedding under attitude verbs. (41) Mary is certain that it might lrain. Stephenson argues that in a sentence like the above, the subjective interpretation of might is linked to the sentential subject Mary. In other words, the likelihood for the weather to be cold is based on what is known to Mary and no one else. However, Stephenson’s view may be too simplified. We could still come up with a reading in which might is based on community knowledge: Mary thinks that it might be the case based on some knowledge the community knows but she herself does not know. Therefore, a sentence like (42) is acceptable. (42) Mary thinks that John might be the tallest. Now if we look at the Mandarin ganxiang shi sentences in (40), what is interesting is that the epistemic modals are expected to be evaluated based on subjective knowledge only. The word ganxiang ‘opinion’ restricts the worlds compatible with the attitude holder’s, or the sentential subject’s in this case, knowledge, not what is known publicly. This is shown in the infelicitous sentence of (43). In this sentence, the knowledge is based on those from the general public, as indicated by dui dajia lai-shuo ‘for everyone.’ 58 (43) ??Wode ganxiang shi, Afu dui dajia lai-shuo, My opinion COP Afu for everyone to-say keneng bi Lisi gao. possibly COMP Lisi tall ‘My opinion is that Afu might be taller than Lisi according to everyone’s opinion.’ In short, the data from ganxiang shi supports the fact that hen is more subjective than the shi…-de form and the comparatives. 3.3.4 The parallelism with other evaluative degree modifiers In the previous section, I showed how hen has some typical features of a subjective predicate. This feature again sets it apart from typical degree modifiers like very and feichang. Therefore, hen has a different entailment pattern, as shown in 3.2.3. This could be explained if we see hen as an evaluative degree modifier. Now if we compare hen with degree modifiers that have a more evaluative nature, they seem to be more similar. One such category that expresses one’s evaluation is the class of hedges, in the sense of Lakoff (1973), which include expressions like rather, quite and sort of. Hedges are in general used to ‘de- intensify’ a proposition. They express a less committed form towards the degree that the subject is true of. For example, in (44)a, the use of rather implies that John has a height which perhaps exceeds a borderline height for tallness, but the height does not reach the typical degree of being tall. In (44)b, sort of has an even stronger hedging effect. The intelligence of John is only borderline smart. Quite is similar to rather, but the hedging effect is not as strong. John is still considerably tall, but perhaps not tall enough for him to be deemed as a canonical case of tall people. (44) a. John is rather tall. b. John is sort of smart. c. John is quite tall. 59 On the surface, hedges do function as if they lower the standards. It is shown in the examples in (45), which includes three pairs of sentences. In each pair, the first sentence is a proposition that includes a hedge. The second sentence provides a more precise description related to the subject: the degree to which the subject holds the property is not high enough for the subject to be considered as having that property in a typical sense. For example, in (45)a, John’s height is somewhere near the standard of tall people, but if it is examined under scrutiny, John does not fall into the tall category. This fact is made precise by intensifiers like really, or by adding focal stress, as in the case of (45)c. (45) a. John is rather tall. But I wouldn’t say he is really that tall. b. John is sort of smart. But I wouldn’t say he IS smart. c. John is quite tall. I wouldn’t say he is really a tall person though. All the sentences with hedges in (45) seems to suggest that the standards are lowered. If this is indeed the case, we would expect them to hold certain entailment relations with the positive form. They are expected to be entailed by the positive counterparts. This prediction is not born out. As we can see in (46), the positive form does not entail the hedged forms, and the reverse does hold either. (46) a. John is tall ↛ John is {rather/quite/sort of} tall. b. John is {rather/quite/sort of} tall ↛ John is tall. As these degree modifiers are linked to subjective evaluation from the speaker, one would expect them to show the properties of typical subjective predicates. Not surprisingly, rather, quite and sort of could trigger faultless disagreements and they could embed under find. In (47), what is being disagreed is due to the vagueness of the standard introduced by rather. 60 (47) A: John is rather smart. B: No, he isn’t. He’s hardly smart at all. A: I don’t agree! Furthermore, all these degree modifiers are more acceptable when they are embedded under find compared to the positive form. (48) a. I find John rather tall. b. I find John sort of smart. c. ?I find John quite attractive. d. ?I find John tall. The reason for hedges to become subjective is because when it is added to a sentence, it implies that the individual is not a typical case in the category denoted by the gradable adjective. For example, John is sort of smart is evaluated by assuming a hypothetical category of smart individuals, which in turn includes a canonical example of a smart individual that the speaker has in mind that serves as the standard for comparison. The smartness of the individual John is not close enough for it to be a typical case of a smart individual, and hence it results in the lessened certainty implied by sort of. In Lakoff’s original work (Lakoff 1973), he sees sentences with hedges as fuzzy expressions. Fuzzy expressions introduces a fuzzy set. At the core of it, there is a class of object that represents the prototypical case of that category. Other objects are ordered with respect to how close they are to this typical case. For example, the category of birds includes different subtypes of birds. They could be ranked according to how close they are to the prototypes of birds. When one thinks of a bird, what usually comes to mind is a bird like a robin, not a penguin or an ostrich. When a hedge is used, it indicate that the individual is not as close to the prototype as one would expect. For example, in the case of John is rather tall, the height of John 61 does not makes him a quintessential example in the category of tall people. Therefore, the determining of a quintessential example in that category could also be subjectively determined. (49) Afu mei *(hen) gao. Afu NEG hen tall ‘Afu is not certified to be significantly tall.’ The distinction between the bu and the mei forms, as shown in the contrast between (48) and (49) in negating the gradable adjective predicates lies in the existence of a salient time during which the state as being significantly high holds. Yet it does not imply that the property is true only within that time frame. For example, in (49), Afu is significantly tall during a specific time frame, but it does not mean that such evaluated property only holds within the time frame of this salient time. The details of this feature will be discussed in chapter 5. Bu can also appear with dynamic predicates. However, it has a strong habitual or generic reading, which expresses that it is generally not the case for such an event to happen. When bu appears with a stative predicate, it is not interpreted as a generic sentence. (50) Ta bu chi yao. He NEG take pills ‘He doesn’t take pills (generally).’ In the literature, a sentence with mei asserts the nonexistence or the unrealization of an event (Lin 2003). Therefore, under this analysis, (51) would mean that the event of winning the ball game did not occur or is not realized. 62 (51) Ta mei ying na-chang qiu sai He NEG win that-CL ball game ‘He didn’t win the ball game.’ The general observation of mei is that it negates a dynamic eventuality, in the sense of Comrie (1976) and Vendler (1967) (see also Smith 1991, Bach 1986, a.m.o.). Dynamicity is defined as requiring ‘input of energy for obtaining the state. Yet the data from hen is an obvious counterexample for this generalization of mei. In the end, we either need to stick to the view that mei selects eventive expressions across the board, or we will need to revise the generalization. In chapter 5, I will discuss the problem of mei and dynamicity in more details. 3.4 Conclusion In this chapter, I showed some of the major problems related to previous analyses on hen and other unsolved problems, including the subjective nature of hen and the problem of negation mei. A way to incorporate this two features is to see hen as introducing an evaluative eventuality that could represent an attitude holder’s perspective over a potentially temporally limited situation. 63 Chapter 4 Epistemic commitment 4.1 Overview In this chapter, I discuss how the notion of epistemic commitment could account for the puzzles presented by hen. It is about a speaker’s evaluation based on what he believes to be true given the evidence or knowledge he has. It is related to the speaker’s attitude towards the truth of the content of a proposition that is regarded as certain or factual (Lyons 1977). When one has enough evidence, the speaker may want to commit to the truth of the proposition. Commitments could be expressed in various forms. In many cases, when a speaker utters a neutral declarative sentence, such as ‘John is home’, it is by default a statement believed by the speaker himself or herself. Hence the speaker commits to the truth of it. Modals shows various degrees of commitments. For example, when one uses might, the speaker is showing less certainty towards the truth of the proposition when compared to using must. Another example comes from factive predicates. Sentences with verbs like regret, or recognize all presuppose that the embedded proposition is what the subject believes to be true (Kipasky & Kiparsky 1971; Karttunen 1974; Lasersohn 2009, a.o.). Verbs as such do not really imply certainty, but they still imply commitments from the part of the sentential subject. In the previous chapter, I presented some puzzles of hen. The first is the hedging effect under negation. The second is the subjective nature of a hen sentence. The third is the obligatory occurrence of hen under negation marker mei. In this chapter, I show that these phenomena can be accounted for if we see hen as a marker that also indicates the speaker’s commitment. Recall that hen provides the meaning that the degree to which the subject has the relevant property is significantly high. In plain words, we could say that due to the high degree that an object has, the speaker is committed to the statement that this object falls into the category expressed by the adjective. In (1), Afu’s height reaches a significant degree. This fact ensures that Afu falls in the positive extension of tallness objects. As a result, it allows the speaker to commit to the statement that Afu is tall. 64 (1) Afu hen gao. Afu HEN tall ‘Afu is tall.’ Therefore, when hen is negated, we would expect it to reach a degree that is less than the one the attitude holder or the speaker is willing to commit to an asserted proposition. As a result, it indirectly denies the commitment made by the speaker. For example, if p is Afu is tall, and the speaker says not hen p – the speaker is not only denying that the degree reaches a certain height, but also the commitment made about the truth of p. The result is a hedging reading from negating the speaker’s full commitment over the statement. (2) Afu bu/mei hen gao. Afu NEG/NEG HEN tall Lit. ‘It isn’t the case that Afu is tall.’ Implication: The speaker is not fully committed to the statement that Afu falls into the category of tall objects. The second puzzle, the subjectivity of hen, is accounted for by this analysis as well. Since speaker’s commitment is based on one’s knowledge base, it is accounted for by the theory of epistemic commitment. 4.2 Some notions on epistemic commitments The notion of epistemic commitments appears in several different linguistic inquiries, such as illocutionary acts, studies on dialogue, modality, and evidentiality. The one I’m adopting here is more linked to the latter two concepts. I translated hen as a particle that could be loosely translated as certified. This translation 65 would imply that there is an attitude holder who makes certifications based on his or her epistemic knowledge. 4 Of course, a commitment is not identical to beliefs. Hamblin (1970) views the commitment of a speaker as a persona or a façade of a sentence for the sake of keeping one’s statements consistent within certain conversations. Lyons (1995) also holds the same opinion towards epistemic commitment. To make a statement is to express a proposition and simultaneously to express a particular attitude towards it. I will call this attitude, for reasons which will be clearer when we look at the notion of modality, epistemic commitment…. Anyone who states a certain proposition is committed to it, not in the sense that they must in fact know it or believe it to be true, but in the sense that their subsequent statements – and anything that can be legitimately inferred from their accompanying and subsequent behavior – must be consistent with the belief that it is true. (1995: 253-254) Following Hamblin and Lyons, I view hen as a particle that specifically expresses epistemic commitment, which is based on one’s belief but is also about one’s attitude of affirming it. Therefore it could not be seen as a particle identical to epistemic modals, in which commitment is just a secondary meaning. Previously I mentioned how hen could be linked to epistemic commitments due to its nature as a degree modifier. When an object reaches a degree that is high enough, one could say it’s a certain case for 4 Yet it is not identical with the commitments expressed by epistemic modals, of which commitment is often viewed as an inseparable component expressed by epistemic modals (Halliday 1970, Lyon 1977, Palmer 2001, Bybee and Fleischman 1995) and some others, epistemic modality often also indicates the attitude the speaker has towards the truth of the propositional content of his prejacent. However, in the case of hen, when translated as certified, the commitment is not a secondary meaning. 66 one to make a commitment. Therefore, we could say that there are two components in the extension of hen. One of them is the degree component, which function just like a degree modifier. The other component is the commitment part. Presumably, when hen appears under negation, there would be two things that could be negated. The first is the lowering of the degree, and the second is the hedging effect resulted from the negation of the proposition that an object exceed a certain degree. This is shown in the compatibility of the two scenarios (i) and (ii) below. (3) Afu bu hen gao. The speaker’s attitudes: (i) The speaker thinks that Afu is not tall at all. There is no vagueness in terms of whether Afu falls into the category of tall people. But Afu wants to hedge his assertion. (ii) The speaker thinks that Afu is not tall enough for him to commit to the fact that he is tall. In scenario (i) of (3), the speaker thinks that it is clear that Afu’s height does not fall into the category of a tall person. Yet the use of hen make it sounds less assertive. In scenario (ii), on the contrary, the speaker’s attitude is more in line with the degree of commitment in the statement (3)a, who is hesitant in committing to the proposition ‘Afu is tall.’ When being negated, Afu’s height falls into the range between the standard for tall objects and the standard newly introduced by hen, which is a degree that the speaker is unable to commit to either a positive or a negative statement of tallness. Therefore, negating hen also indirectly negates the full commitment when compared to (i). Lastly, I will also address the issue of German comparative particle eher. Eher is regarded as a comparative morpheme used for comparing the degree of commitment of the attitude holder has towards two propositions (Herburger & Rubinstein 2018). Eher could be viewed as the comparative counterpart of hen, which will be discussed in 4.7. 67 The organization of this chapter goes as follows. 4.3 illustrates how epistemic commitment could account for the two puzzles, hedging under negation and subjectivity, of hen. 4.4 illustrate the comparison between hen and the shi…-de form. 4.5 is a comparison with English positive form. 4.6 is an intermediate summary. 4.7 is a comparison between eher and hen. 4.3 Epistemic commitment of hen 4.3.1 The hedging effect under negation and euphemism In the previous chapter, I discuss up the puzzle of hen under negation, and why it obtains a hedging effect when it is negated. In principle, hen under negation poses a problem for its analysis as a covert positive degree morpheme. If hen were a positive degree morpheme, one would expect it to pattern with the English positive sentence under negation. Therefore, the negated sentence with hen, as in (4)a, should be identical with the English counterpart in (4)c. However, (4)b, the counterpart without hen, is in fact closer to the English one. (4) a. Afu bu hen gao. b. Afu bu gao. c. Afu is not tall. Yet when under negation, hen (4) provides some sort of hedging effect by making the assertion sounds less assertive for a native speaker of Mandarin Chinese. What is the possible explanation for it? An immediate explanation for the hedging effect is that hen is used for euphemism. It is not uncommon for degree modifiers to be used as a tool to express euphemism. For example, when one utters ‘He isn’t super tall’, it is perhaps an euphemism for a more direct expression, ‘He isn’t tall.’ As a degree modifier, hen should also trigger such inferences. However, seeing hen as expressing euphemism alone cannot account for some of the patterns of hen when compared with English. 68 4.3.2 Euphemism like very: a possible analysis One might wonder whether analyzing hen as related to epistemic commitment is actually a stretch. A possible example of the hedging effect could be the result of euphemism, in which the speaker does not want to sound that he is making a strong statement. Therefore, he or she uses the negating of a positive form to express the negative meaning. Sentences with degree modifiers, when under negation, are often used as a means to express euphemism as well. For example, if a person wants to say that John is dumb, it would often be expressed like (5). In a sense, the sentence would have a hedging effect by attenuating the strength and the impact for the addressee. (5) John is not very smart. Could it be possible that when hen is being negated, what we see as a hedging effect is also a case of euphemism for ordinary degree modifiers like very? Euphemism, in its most general sense, refers to the speech use in which a speaker refers to an object with a camouflage without technically referring to it (Horn, 1989). The purpose of euphemism is often out of politeness and appropriateness. A common method to express euphemism is to mask a contrary with a contradictory (Horn 1989, Bosanquet 1911), as in (6). (6)b is the contradictory of (6)a. By asserting the contradictory (6)b, what is actually implicated is (6)c. (6) a. I’m thrilled by your advice. b. I’m not thrilled by your advice. c. I dislike your advice. (Horn 1989: 17) The contradictory of a proposition p, p, is the superset of the contrary of p. Therefore, the contrary of a given proposition p is a stronger expression than the contradictory of p. Using set theory notions, this 69 effect is illustrated in the Venn diagram in (7). The contradictory of the set of worlds of thrilled is the part is the densely striped part. The contrary of thrilled, hostile, is a subset of thrilled. (7) By negating thrilled, the speaker’s real intention is to express the stronger contrary dislike. Such euphemism is quite ubiquitous across negation with gradable expressions, especially with those that raises the degree to a significantly high level. Take (8) for example. The speaker intends to utters (8)b or (8)c by adding super, as in (8)a, as an euphemism. (8) a. John isn’t super tall. b. John is not tall. c. John is short. Euphemism in degree modifiers like (8)a work in such a manner. Degree modifiers raises a given standard by different contextually determined differences. They can be ordered with respect to how much each raises the standard. Compare very and super for example. Assume that the function of the two modifiers is purely about standard raising for the time being and that other inferences, such as personal evaluation or emotions, are to be ignored. By comparing the two, the amount of degree raised by very is smaller than that of super. Assume that the standard introduced by the positive degree morpheme is d 0. The 70 new standard obtained by very is d 1 and that obtained by super is d 2. Since d 2> d 1> d 0, ‘John is super tall’ entails ‘John is very tall’, it in turn entails ‘John is tall.’ However, since euphemism is expressed through implicature, it could be cancelled. As a result, (9)a could be followed by a statement confirming that John is tall, as shown in the acceptability of (9)b. Of course, the implicature could be made explicit by another interlocutor, as in (10). (9) a. John isn’t super tall. b. John isn’t super tall. He’s still tall though. (10) Speaker A: John isn’t super tall. Speaker B: You mean he isn’t tall at all. He is short right? Now we come back to hen. Despite its other inferences, since hen is a degree modifier that raises the standard to a significant degree, the negation of it could also result in euphemism in a similar pattern. Let’s assume that the speaker wishes to express the sentence Afu hen ai ‘Afu is short’, as in (11)a. He could either choose to make an euphemism by negating the contrary of ai, which is gao ‘tall’, as in (11)b. If he wishes to make the expression more indirect, he could add a degree modifier hen, which expands the domain of the contradictory proposition when it is under negation. Assume that the standard introduced by the positive degree morpheme in (11) is d 0. Hen raises the standard to d 1, and d 1> d 0. The contradictory reading is obtained by negating gao, as in (11)b, which has the denotation that Afu’s height is somewhere below d 1. By raising the standard with the addition of hen, the domain of contradictory is expanded as compared to the counterpart without hen. Therefore, (11)a becomes a smaller subset upon uttering (11)c when it is compared to (11)b. 71 (11) a. Afu hen ai. Afu HEN short ‘Afu is short.’ b. Afu bu gao. (Afu is shorter than d 0) c. Afu bu hen gao. (Afu is shorter than d 1) Could the attenuation of assertiveness in the hen sentence be a result of euphemism, just like very or super given above? Of course, as hen is a degree modifier, it is expected that when under negation, the existence of hen indeed provides such an inference. However, it is not the entire picture. There is more to hen than simply just a modifier that raises the standard degree. One support comes from the embedding under attitude verbs, and another comes from focus sentences. 4.3.3 Argument against euphemism: Hen embedded under attitude verb An interesting aspect of hen that set it apart from ordinary degree modifier is shown when hen is in the complement position of an attitude verb. Not surprisingly, when in the embedded position, the commitment provided by hen is switched to the sentential subject. (12) Lisi juede Afu hen congming. Lisi think Afu HEN smart. ‘Lisi thinks that Afu is certified (by Lisi) to be smart.’ However, what is interesting is that when a sentence with hen embedded under an attitude verb is being negated, it shows a different pattern from sentences without hen. There is an interesting feature for a sentence with an attitude verb like think to get negated. It seems that whether the negation marker is placed in the matrix or the embedded position, there is no significant truth conditional differences between the two. This is often referred to as negative raising. Therefore, the 72 two sentences in (13) are viewed to be equivalent. Note that in (13)a, there are two readings. The first reading is that Mary doesn’t have any beliefs. The second reading is that she believes the movie to be not good. Only when (13)a is interpreted as the second reading would the two become equivalent. (13) a. Mary doesn’t think that this movie is good. b. Mary thinks that this movie isn’t good. Mandarin Chinese also has the same negative raising pair when attitude verbs are used. This is shown when the negation marker bu is used. By using bu, we only get the reading that the sentential subject indeed has the belief denoted by the embedded sentence. The two sentences in (14) would be truth conditionally equivalent. (14) a. Afu bu juede Lisi zai jia. Afu NEG think Lisi at home ‘Afu doesn’t think that Lisi is at home.’ b. Afu juede Lisi bu zai jia. ‘Afu thinks that Lisi isn’t at home.’ We would expect one to find the same equivalent pair in sentences with degree modifiers. Specifically, we would expect the same hedging effect to show up when hen is being negated whether the negation is in the embedded or the matrix position. At least this is the pattern we find in English. (15) a. Mary doesn’t think that John is super tall. b. Mary thinks that John isn’t super tall. 73 For one thing, if hen is simply a degree modifier like very or super, we would expect a sentence with hen to show the same pattern as (15). However, it is not the case. When the negation is in the embedded position negating hen directly, it obviously has a different meaning from the proposition with matrix negation. In (16)a, when the negation marker is in the matrix sentence, it is negating the certification by either the speaker, or some third party. This is the de re reading. But when it is in the negated sentence, we get the reading that what is being negated is the certification from the embedded subject Lisi. This is the de dicto reading. (16) a. Afu bu juede Lisi hen gao. Afu NEG think Lisi HEN tall ‘Afu doesn’t think that Lisi is certifiedly tall.’ (hen anchored to the speaker/others) b. Afu juede Lisi bu hen gao. Afu think Lisi NEG HEN tall ‘Afu is not committed to the case that Lisi is tall.’ (hen anchored to Afu) The effect of hen in the two sentences in (16) is that the hedging of (16)a is not identical to that of (16)b. In (16)a, Afu does not think that Lisi’s height meets the standard that is certified by others. Therefore, there is no hedging in terms of the commitments given by Afu. In (16)b, we get the expected meaning that Afu thinks that Lisi’s height does not reach the degree certified by Afu. The distinction between the two is obvious. We also get the same hedged meaning that hen is negated in the matrix sentence: Afu does not want to commit to the proposition that Lisi’s height reaches a degree that makes him a tall person. 74 4.3.4 An argument against euphemism: focus There is a second piece of argument against seeing the hedging effect as ordinary euphemism. This is shown in the contrast between hen and very when they are under focus. Since hen is not just a degree modifier that simply raises the degree, unlike very, we would expect it to introduce different alternatives when it is assigned focal stress. Hence, when under negation, we would not be expecting it to have the same exact euphemism triggered by very. Following the well-known Roothian focus semantics, when we place focal stress on the degree modifier, it would trigger a set of focus alternatives. (Rooth 1992, 1985; Roberts 2012; Bürning 2012; Beaver & Clark 2008, a.m.o.). (17) a. ⟨cold, cool⟩ b. ⟨certainly p, probably p, possibly p⟩ In an ideal situation, the propositions including these different expressions are involved in an entailment relation, such as (18). (18) a. x is cold. ⇒ x is cool. b. John is certainly rude. ⇒ John is probably rude. In cases like very and hen, the focus alternatives are comprised of propositions that replaces the focused words for others degree modifiers. The set of degree modifiers is determined by the lexical semantics of the focused word as well as by the context, following the rules of Horn Scale (Horn 1973, Levinson 1983). (19) John is [very] FOC smart. (20) Afu [hen] FOC congming. 75 If hen is merely a degree modifier that raises the standard and function as nothing else, we would expect that when it is focused, it would trigger a scalar focus alternative set that only varies with respect to the degrees they are raised, just like very. This set would only differ in terms of the degrees each gradable adverbs provide. Depending on the context, the Horn scale could contain various sets of elements. For example, very could be formed with other degree modifiers that express different degrees the standard is raised or lowered to, with very being in the middle of the scale, as in (21)a. (21) a. ⟨extremely, super, very, pos, somewhat⟩ b. John is extremely smart⇒ John is super/very/pos somewhat smart. c. John is very smart ⇒ John is smart. Very could also introduce a scale in which very takes the highest order on the scale, as in (22)a. In this case, if very is focused, it emphasizes the fact that the degree is high, as opposed to the lower degrees denoted by pos and somewhat. (22) ⟨very, pos, somewhat⟩ In each of the alternatives, the position being focused is replaced by other degree modifiers on the scale. The propositions in the set has the same semantic content as that of the assertion except for the syntactic position occupied by very. For example, they may include alternatives like very, pos and extremely. These propositions therefore forms a scale by means of pragmatic entailment, as shown in (23)b. The higher the standard is raised to, the stronger the proposition becomes. In the positive form, the stronger propositions pragmatically entails the ones ordered lower. The order is flipped under negation, as in (24). 76 (24)b represents a downward entailing scale, in which the propositions with weaker modifiers entailing those with stronger ones. Positive form: John is [very] FOC smart. (23) a. Alternative set: {John is friendly, John is very friendly, John is super friendly, John is extremely friendly} b. Positive form: extremely≻super≻very≻pos Negative form (24) a. Alternative set: {John is not friendly, John is not very friendly, John is not super friendly, John is not extremely friendly} b. Negated form: ¬pos ≻ ¬very ≻ ¬super ≻ ¬extremely Now we turn to hen. If hen were simply a degree modifier like very, we would expect hen to have the above interpretations under focus. That being said, what could be potentially focused would only be the part on standard raising or lowering. This is not the case. Instead, when hen is focused, it allows an extra focus meaning, which could be roughly described as the focus on assertiveness, despite that it could also have the same focus meaning as very, which emphasizes that the degree for Afu to exceed the standard is a lot. Here the translation of hen is intended as very, while of course, hen is not very. Scalar focus: Assuming hen is very (25) Afu [hen] FOC congming. Afu HEN smart Intended: ‘Afu is VERY smart.’ 77 This is shown in the example below. The preceding sentence, Afu bu shi zhiyou congming, ‘Afu isn’t just smart’, indicates that there is perhaps a scale of some sort of ordering between possible degrees of smartness. Afu is not just smart, he is very smart. Hen under focus with a degree meaning: (26) Afu bu shi zhiyou congming, Afu [hen] FOC congming. Afu NEG COP only smart Afu HEN congming. ‘Afu isn’t just smart, he is VERY smart.’ However, there is this extra focus interpretation that occurs when the speaker stresses that Afu is indeed tall. This reading is more obscure and it requires more contextual support to emerge. It often appears when the statement is contrasted with another statement of which the speaker denies the opposite. Take (27) for example. Speaker A assumes that B know that Afu is not smart. The question particle ba here is a marker that introduces a special presupposition. The speaker assumes that it is more likely for the addressee to believe the asserted proposition Afu is not smart. In the response by B, he denies A’s assumption, and stresses that he believes Afu to be smart by putting a focus intonation on hen. Focus on commitment: (27) A: Ni juede Afu bu congming ba? You think Afu NEG smart Q. You think Afu HEN dumb ‘Didn’t you think that Afu was dumb?’ B: Caiguai! Afu [hen] FOC congming. Wrong Afu HEN smart ‘Nope! Afu IS indeed smart.’ 78 Apparently, the answer of B in (27) cannot be an answer to the scalar focus meaning when very is being focused. The focus intonation placed on hen is a strengthening over speaker B’s own personal opinion. One might argue, however, that what is being introduced is a polar focus alternative set. Polar focus introduces two opposite alternatives, one is the positive alternative, and the other the negative alternative. In English, when a sentence with a copula is focused, two different opposite alternatives are provided. One is the positive extension, John is tall, and the other is the negative extension, John is not tall. (28) John [is] FOC tall. There is a possible explanation for B’s reply in (27). Maybe what is expressed is not the speaker’s own commitment, but simply an emphasis on the positive statement ‘Afu is smart’, as opposed to the alternative proposition ‘Afu is not smart.’ Yet this analysis is problematic. It seems that hen cannot trigger the desired polar focus reading. This is shown in the test regarding question-answer pair in (29). (29) a. Afu shi bu shi [hen] FOC congming? Afu COP NEG COP HEN smart Intended reading: ‘IS Afu smart?’ b. #Bu shi. Ta bu congming. NEG COP he NEG smart ‘Nope. He isn’t smart.’ In short, the focus interpretations of a hen sentence show that there are at least two types of focus semantics related to hen, one is the typical scalar reading, and the other is the commitment reading. 79 Presumably, the emergence of these two reading is determined by the context. This again shows that we cannot see the hedging effect of hen as a result from the same effect of negating ordinary degree modifiers like very, which merely provides the meaning of modifying the standard degree. 4.3.5 Subjectivity and epistemic commitments Throughout this dissertation, I’ve mentioned how hen sentences are viewed as subjective in an intuitive sense. Subjectivity is another aspect of hen that is obviously linked to epistemic commitments. Epistemic commitment entails evaluations based on one’s epistemic knowledge. In chapter 2 and 3, I discussed how hen shows typical features of a subjective predicate. In this sense, we could view hen as requiring the same type of knowledge base as epistemic modality when they are interpreted subjectively. If this view on epistemic commitment is on the right track, we would expect a sentence with the speaker’s commitment to also pass the typical subjectivity tests. An example is shown in the adverb totally. Totally has one use for the speaker to express a high degree of commitment (Beltrama & Bochnok 2015; Beltrama 2016, 2018; Irwin 2004) or ordinary epistemic modals. (30) #He totally clicked that page, but I don’t think it’s likely that he did. The above sentence is degraded due to the fact that the first clause expresses high degree of commitment to the truth of the proposition, and such commitment should be made according to what is regarded as likely based on the knowledge one has at the speech time. Therefore, it conflicts with the second statement that the speaker does not actually consider it to be true. Similarly, the Mandarin hen sentence shows this trait as well. Recall that there are two forms with gradable adjectives, the hen form and the shi…-de form. The shi…-de form is used to express the categorical conceptualization of an adjectival property, and the criteria for determining the categorization is well- accepted in the Common Ground, meaning that the standard is either presupposed or backgrounded. It is not used for expressing one’s assessment, which is the opposite of the hen form. 80 Since only the hen form is linked to the an attitude holder’s assessment, we would then expect one to find a sentence in which the speaker’s opinion is different from the public opinion. The shi…-de form is often linked to a publicly known standard. Yet this standard may not fit the one the speaker would commit to. There is a dichotomy between the standard given in the Common Ground and what is provided by the speaker’s personal belief. Presumably, the standard given in the Common Ground could be viewed as fitting a more objective norm, but not the one that is certified to be true according to the speaker’s own standard that allows him to make the assertion. In (31), the second sentence refers to the disagreement of such a standard. (31) Dajia shuo Afu shi gao-de. Suiran wo bu Everyone say Afu COP tall-DE although I NEG tongyi nage biaozhun agree that standard ‘Everyone says that Afu belongs to the category of tall people (in a given context). Although I don’t really agree with that standard.’ In contrast, it is strange for the speaker to deny a property in a hen sentence when it is in the matrix sentence. (32) #Afu hen gao. Suiran wo bu tongyi. Afu HEN tall Although I NEG agree ‘Afu is certified to be tall. Although I don’t think so.’ Another example comes from certain fixed nominal expressions that corresponds to a stereotypical standard in a given context. Gao-gezi denotes the set of individuals who are generally considered tall. The standard is not limited to the immediate contexts the speakers are referring to, nor is it determined by 81 personal opinions. It is related to some general understanding of what tall people should be like. Sometimes the standard itself is even understated. For example, one would say Afu is a gao-gezi because his figure would be generally regarded as tall. However, as generalization is not without exceptions, such general understanding will not always be agreed by everyone, especially if the general standard in gao-gezi is regarded as unfit for a certain scenario, it may not be agreed by the speaker. Obviously, since gao-gezi is related to the standard based on generalization, it is obviously distinct from the standard introduced by hen. Gao-gezi does not require the speaker’s assessment. It could be viewed as assessed by the general public. This feature is shown in (33). (33) could be easily licensed under a situation in which there is generally accepted standard for tall people, and Afu reaches that standard. However, the speaker does not need to agree with it, since he may have his own standard. (33) shows hen itself as marking the assessment from the speaker. Therefore, it does not conflict with the ‘objective’ beliefs accepted in the Common Ground. (33) Afu shi gao-gezi, dan wo bu juede hen gao. Afu COP tall-person but I NEG think HEN tall ‘Afu is a tall person, but I don’t think he is certified to be significantly tall (by me).’ Furthermore, epistemic commitments are about certainty based on personal evaluation, which is linked to a personal epistemic base. Therefore, this feature of hen is reminiscent to epistemic modals when interpreted as ‘subjective.’ The distinction is often resulted from the selection of modal bases. As I mentioned earlier, a modal relativized to a modal base with personal evidence is often regarded as a subjective modal (Lyons 1977, Papafragou 2006). Yet compared to a vague expression like hen, epistemic modals seems less likely to trigger faultless disagreements. While faultless disagreements are still possible, the debate no longer is faultless if there is more evidence to clarify the matter. Scenario: Speaker A and B are talking on the street. 82 (34) A: Afu yiding/yinggai zai jia. Afu must/should at home ‘Afu must/might be home.’ B: Ni cuo-le. Ni kan, ta bu jiu zai You wrong-PERF you look he NEG then at duimian jie-shang ma? opposite street-on Q ‘You’re wrong. Look! Isn’t he just standing there across the street?’ On the contrary, disagreements involving hen can maintain faultless even though one of the participants knows more about the situation. This is simply due to the fact that according to each individual’s modal bases, different standards could be selected. Scenario: Afu is a 5 th grader. Speaker A knows better about the general height of people of Afu’s age, while B has no knowledge about it. (35) A: Afu hen gao. ‘Afu is tall.’ B: Ni cuo-le. Ni kan, Afu bi You wrong-PERF You see Afu compare same-age-group short tong-nian ai ‘You’re wrong! Afu is shorter than anyone of his age.’ 83 In previous chapters, I mentioned that hen does not trigger ‘real’ faultless disagreements. I argued that one of the reading of hen that trigger a seemingly effect of faultless disagreement is its vagueness in the criteria for determining the standard. In a disagreement, the attitude holder of a hen sentence can easily make clarifications the comparison class to make the disagreement disappear. Take tall for example. As there is no absolute standard for tallness, for an object that possesses certain height, one could always selects a comparison class that works for the statement. Unless we assume that different people highlight the same set of knowledge in making the evaluations, it is less likely that two people can make the exact same evaluation, given that they have different knowledge base in the first place. In this sense, the subjectivity of hen is similar to that of subjective epistemic modals, which introduces knowledge and evidence that could be known to the attitude holder only. However, since hen only subcategorizes gradable adjectives, hen is always context sensitive. Epistemic modality, on the contrary, does not necessarily vary from context to context, at least not in the way that is relativized to context-wise standard setting as hen does. Yet there is a correlation in terms of the means of making the commitment, which is related to the speaker’s certainty. 4.4 A comparison with the shi…-de form It is obvious that the shi…-de form is not related to commitments. It stresses that the subject indeed reaches the known standard. As opposed to hen, the shi…-de form is objective: it does not trigger faultless disagreements, as discussed in chapter 2, and it cannot co-occur with ganxiang shi ‘one’s opinion is.’ Not only does the shi…-de form fail in passing these subjective tests, it does not express one’s commitment either. This is shown in (31), in which the speaker does not need to agree with an individual having the property denoted with a shi…-de form. There is no attenuation in the assertiveness or one’s commitments when the shi…-de form is not negated. It only means that the subject does not belongs to the category denoted by the adjective. 84 (36) Afu bu shi gao-de. Afu NEG COP tall-DE ‘Afu does not belong to the category of tall people.’ Another difference is that the shi…-de form cannot be followed by a sentence that uses the fixed nominal form to refute the previous sentence, as shown in the underlined part in (37). As the nominalized form refers to a well-defined category, it implies a well-accepted standard. For a given context, there is an entailment relation between the nominal form gaogezi and the shi…-de form: If an object belongs to the category of gaogezi, it belongs to the category denoted by the shi…-de form. Since the standard introduced by the shi…-de form is one that is accepted by the Common Ground, and a well-defined standard for the fixed nominal form should be accessed by the Common Ground as well. Therefore, the first sentence with the shi…-de form is entailed by the second sentence, which results in the contradiction we see in (37). (37) #Afu shi gao-de. Suiran wo geren bu juede Afu COP tall-DE although I personally NEG think ta shi gaogezi He COP tall-person ‘Although Afu belongs to the category of tall people (under this context), he is not a tall person.’ Another support that there is no commitment involved in the shi…-de form comes from focus structures. The focus stress posed on shi triggers a polar focus set and asserts that Afu is indeed tall. It cannot be interpreted as if the speaker is emphasizing the strength of his or her belief, contrary to the situations in which hen is being focused. 85 (38) Afu [shi] FOC gao-de. Afu COP tall-DE ‘Afu IS tall.’ Alternatives: {Afu is tall, Afu is not tall} In conclusion, the shi…-de form is not only relatively objective, it does not involve personal commitments as well. 4.5 A comparison with the English positive form Assume that the English positive degree sentences are statements without the meaning of the speaker’s commitment. If this the case, it is predicted that when the speaker uses a positive form, it is compatible with the denial of an expression that is relativized to a standard assigned by the context not based on the speaker’s own judgment. As a result, it seems felicitous to utter (39). (39) John is tall. But it’s not based on my personal opinion. (39) does not patterns with the Mandarin hen sentences. This is shown in the unacceptability of (40). English adjectives in the attributive position, as in (40), has its own domain for standard assignment. This standard could either be assigned in a specific context or as reflecting a norm accepted by the general public. Nevertheless, when there is little context to pinpoint the comparison class to which the attributive expression is relativized, it is more likely to be defaulted to the matrix context. Hence (40) sounds contradictory. (40) #John is tall. But he is not a tall person. 86 The Mandarin hen form can co-occur with the nominalized form gaogezi, as in (37). This again shows that hen indeed has a subjective component that is otherwise lacking in English positive degree sentences. 4.6 Intermediate summary and further issues The argument that hen expresses a speaker’s commitment is based on two major observations: the hedging effect under negation and the subjective feature of hen. I showed that the hedging effect of hen can be accounted for if hen expresses the speaker’s commitment. If we view hen as merely a degree modifier without the evaluative quality, this cannot be explained. This argument is further supported by the fact that hen has an obvious subjective connotation, especially when we compare it with the shi…-de form and the English positive degree sentences. In the case of negation, the hedging effect of hen is not identical to that of euphemism of ordinary degree modifiers despite the fact that hen indeed triggers euphemism as other degree modifiers do. When under focus, hen has two types of focus interpretations, the ordinary scalar one and one that place focus on one’s commitment. The two interpretations do not emerge simultaneously. While the scalar focus interpretation is no different from placing focus intonation in sentences with other ‘ordinary’ degree modifiers, the focus on one’s commitment is obtained through the meaning provided by hen, which provides a standard that is significantly high to such an extent that for any objects that falls in the positive extension with respect to this cutoff point is to be counted as a certain case for the attitude holder to commit. 4.7 German eher: a comparison The German particle eher, ‘more’, is regarded as a comparative morpheme about epistemic commitment (Herburger & Rubinstein 2014, 2018), which could be seen as the comparative counterpart of hen. On the surface, eher could be misinterpreted as an ordinary comparatives. It is because when they appear with gradable adjectives, the difference in meaning between (41)a and (41)b seems subtle. 87 (41) a. Eva ist eher groß als Maria Eva is more tall than Maria ‘I’m more inclined to think that Eva is tall than that Maria is.’ b. Eva ist größ-er als Maria Eva is tall-COMP than Maria ‘Eva is taller than Mary.’ (Herburger & Rubinstein 2018) However, eher can appear with non-gradable adjectives, such as schwager ‘pregnant.’ This cannot be accounted for if we see eher as similar to ordinary comparatives that compares two degrees. (42) Maria ist eher schwager/*schwager-er als Eva Maria is more pregnant/pregnant-COMP than Eva (With eher): ‘I’m more inclined to believe that Maria is pregnant than that Eva is.’ (Herburger & Rubinstein 2018: 6) Furthermore, eher could appear in a comparative sentence with two non-gradable adjectives, such as regionaler ‘regional’ and staatlicher ‘national.’ What is being compared to is obviously not the degree of being regional or the degree of being national. (43) Dies ist auf regionaler Ebene eher mölich als auf staatlicher. This is on regional. Level EHER possible than on national ‘It’s more likely that this is possible on a regional than on a national level.’ (Herburger & Rubinstein 2018: 4) 88 For (42), the sentence compares the two situations of two women getting pregnant. Since pregnancy does not come in degrees, it must be something else that is being compared. Likewise, (43) does not mean that the thing is more of a regional nature than it is of a national nature. Rather, it is a comparison on how confident the speaker is over the two situations. What is the generalization for these data? Herburger & Rubinstein argues that eher compares two different degrees of epistemic commitment towards the propositions. According to them, epistemic commitment refers to one’s willingness to believe a statement and the certainty one has over it. It is distinct from pure certainty or possibility. The support that eher is used to express the speaker’s confidence comes from sentences used to express first person wishes. Under this circumstance, it is infelicitous to use eher. It is because it would sound bizarre to have any preferences over one’s wish. This is shown in (44). While it is felicitous to use lieber ‘rather’, it is not felicitous to use eher. Examples like (44) supports the view that eher is not just about simply expressing different degrees of probability or certainty. (44) Ich will lieber/#eher nach Wien fahren als I will RATHER/EHER to Vienna travel than in Bregenz bleiben. In Bregrenz stay (With lieber)‘I prefer to go to Vienna than stay in Bregrenz.’ Eher is comprised of two parts, the morpheme eh- and the comparative morpheme -er. The -er part in the same morpheme for typical comparatives, as in größ-er ‘taller.’ The comparison on epistemic commitment is derived compositionally from eh and -er, with eh relating propositions to degrees of beliefs. Herburger and Rubinstein adopt Yalcin’s formalization on information states (Yalcin 2007). An information state is a parameter that is only referenced by modals and attitude expressions, but it is idle in other non-intensional expressions. An information state is a set of possible worlds compatible with the 89 knowledge of an attitude holder. It is a triple of 〈Π,𝜋,𝑃𝑟〉. Π is the partition of all possible worlds. Each partition is a cell with a set of worlds. Π recognizes a proposition p, which is the subject matter, as an alternative. For each cells in Π, it is either a set of p worlds or a set of ¬𝑝 worlds. Furthermore, 𝜋 is a subset of Π. It represents worlds that are actually possible under the information state. As for 𝜋, it is dissected into various cells. Pr is a function that assigns every cells in 𝜋 a numerical value between 0 and 1, which is the probability value for each of the cells. The probability of a proposition p is the sum of all the probability of the cells. Herburger and Rubinstein also view epistemic commitment as the probability evaluated from an individual. The morpheme eh- is to be combined with a degree d, which will be introduced by er. This degree d represents the probability of the proposition being compared. In (45), z is the attitude holder in the evaluation world w. The Bel function measures the strength of one’s confidence towards the proposition based on the information state S z. (45) ⟦eh-⟧ c, S z , w =λp.λd. d≤Bel S z (p) (Herburger & Rubinstein 2018: 8) The epistemic commitment proposed by Herburger and Rubinstein on eher is gradable. Consider an expression that ‘the speaker is more committed to p than to q.’ When the cells in the information state has higher probability for p to be true, the higher the degree of the commitment the speaker has. In a sense, we could see hen as a particle that is similar to eher, except that it functions like a positive degree morpheme in the sense that it introduces a standard over which the degree of the subject exceeds. In a sense, we could see hen as having the meaning of adding eh with a positive morpheme. 4.8 Conclusion and further issues In this chapter, I argue that hen is linked to epistemic commitments. It can account for the hedging effect of hen under negation. It is a result due to the lowering of one’s commitments towards the proposition that 90 is indirectly linked to the lowering of the degrees. If it exceeds some threshold, then it is considered to be a sure case for the speaker to commit. In other words, the decision of making a commitment covaries with the degree of the object. Furthermore, I compare hen with German eher. While eher and hen both express commitments, hen is used to modify positive degree sentences while eher is used as a comparative from. 91 Chapter 5 The licensing condition of mei and the meaning of hen 5.1 Overview In this chapter, I discuss the semantic contribution of hen by looking at the licensing condition of mei. There is an interesting puzzle of hen is that that a gradable predicate can be licensed under the more ‘dynamic’ negation marker mei when hen is present, which seems to suggest that the presence of hen makes a gradable predicate more dynamic than what is expected from a typical stative adjective. What I am proposing here is that hen introduces an eventuality of evaluation that is placed over the matrix eventuality modified by the adjective. This eventuality corresponds to the stance of the attitude holder, and it is often temporally and spatially bounded. A puzzle related to Mandarin hen is the pattern of negation licensed by hen. There are two types of negation markers in Mandarin, the bu form and the mei form. The distinction between the two is often regarded as based on the differences of the aktionsarten of the sentence (Huang & Magione 1985, Huang 1988, Ernst 1995, Lin 2003, among others). In general, the bu form occurs mainly with stative sentences. Therefore, it is used to negate sentences with the copula shi, which include nominal predicates, as in (1), and adjectival predicates with shi, such as absolute adjectives in (2), and relative adjectives with the marked form, as in (3). (1) Afu bu shi laoshi. Afu NEG COP teacher ‘Afu is not a teacher.’ (2) Shui bu shi man-de Water NEG COP full-DE ‘The water is not full.’ 92 (3) Afu bu shi gao-de Afu NEG COP tall-DE ‘Afu does not belong to the category of tall individuals.’ There is another negation marker, mei, which is regarded as negating more ‘dynamic’ predicates. It negates that a dynamic eventuality has ever occured. As a result, it is often natural to translate a mei sentence into one with past tense. Mei is infelicitous in negating a sentence in which copular shi is used. (4) Afu mei chi fan. Afu NEG eat meal ‘Afu didn’t eat meal.’ (5) a. *Afu mei shi laoshi Afu NEG COP teacher b. *Shui mei shi man-de Water NEG COP full-DE c. *Afu mei shi gao-de Afu NEG COP tall-DE As gradable adjectives are stative expressions, we would expect them to license the use of bu. This prediction is correct. However, the default negated form of a gradable adjective is without hen. (6) Afu bu gao. Afu NEG tall ‘Afu is not tall.’ 93 When hen is present, it shows the hedging effect, which shows that the speaker is less willing to commit to the proposition that Afu is not tall, as shown in (7). The translation of very in (7) is only literal. The meaning of hen here is not vacuous. (7) Afu bu hen gao. Afu NEG HEN tall ‘Afu is not certified to be very tall.’ While the above pattern regarding the lacking of hen in the default negated form is confusing on its own, there is an even more puzzling issue related to mei. As shown in (6) and (7), When hen is present, it in fact could license the use of mei. When hen is absent, it is infelicitous. This is quite counter-intuitive, since mei in general only licenses expressions that are regarded as eventive. If hen is an ordinary degree modifier like very, it should not render the predicate eventive. Recall that hen can be negated by two different negation markers, bu and mei. Bu is usually viewed as a the negation from that takes only stative predicates, while mei takes dynamic ones. Hen predicates, on the surface, are adjectival, and is thus expected to be licensed under bu. However, the presence of hen, as opposed to the lacking of it, actually licenses the use of mei. What is the contrast between the use of bu and the use of mei? (8) a. Afu bu hen gao. Afu NEG HEN tall b. Afu mei hen gao. Afu NEG HEN tall ‘Afu is/was not certified to be significantly tall.’ 94 Intuitively speaking. there are differences between the two interpretations. When using mei, one tends to interpret the sentence as negating a situation that holds at a bounded time or location. Yet this contrast is quite subtle, and temporal boundedness is not a feature unique to mei. Both the bu and the mei forms are compatible with the reading such that the situation is to be held at a specific time. This is shown in the fact that a temporal or a locational modifier could appear in both forms with a temporal modifier that marks temporal boundedness of the eventuality. (9) Afu na nian bu/mei hen gao. Afu that year NEG HEN tall ‘It wasn’t the case that during that year Afu was certified to be significantly tall.’ (8)b can be translated as ‘In a certain salient time frame, Afu is/was tall.’ Not only does hen license mei, so do other degree modifiers, such as feichang ‘very’, xiandang ‘rather’ and chaoji ‘extremely.’ (10) a. Afu mei feichang/xiangdang/chaoji congming. Afu NEG very/quite/extremely smart ‘Afu wasn’t very smart. / It wasn’t the case that Afu was quite smart. /Afu wasn’t extremely smart.’ What is the common feature of these degree modifiers that allow the adjectival expression to be licensed under mei? Apparently, the presence of these degree modifiers are linked to personal stances, especially temporal or spatial locations. We could see an evaluation as an act that corresponds to a salient speaker or the attitude holder of the sentence based on some stances. Hence, each act of evaluation are potentially anchored to some time or spaces. In other words, this analysis requires an independent eventuality of evaluation that is anchored to some salient spatial-temporal spaces. 95 While this assumption seems likely, we need to first examine the general licensing conditions of mei when it subcategorizes stative predicates. Presumably, if temporal and locational anchor is what really licenses the use of mei, we would expect other stative predicates licensed by mei to also show some sort of temporal or locational anchor, at least in the cases when mei is used. It is indeed the case. There are at least four types of stative expressions that could be licensed under mei, which include location verbs, verbs of property description, judgment verbs, and verbs of measurements. This is by no means an exhaustive list, but they do share some similar properties. In principle, when co-occurring with mei, all these four types of verbs are correlated with a situation that is being evaluated with respect to a certain stance. I call this event the evaluation eventuality, which could be seen as an attitude eventuality that serves that purpose of picking out a certain stage of the states, and there is an evaluation involved over that stage by the evaluation event. Being evaluative, a hen expression is correlated with a perspective, which is the subjective evaluation about the salient degree over which the attitude holder is willing to commit to the proposition that the property holds of the subject. The existence of this event makes the sentence more likely to be temporally bounded. The organization of this chapter is as follows. 5.2 explains the traditional view between the contrast of mei and bu. 5.3 discusses the four types of stative predicates that are licensed under mei. 5.4 and 5.5 discuss the temporal boundedness of these four types of predicates, and how such a property is correlated to a stance. In 5.6, is an intermediate summary that provides a generalization for the cases of the four types of predicates. 5.7 returns to hen sentences. 5.8 discusses how mei is related to a specificity restriction on a discourse situation. 5.9 discuss the scopal property of hen. 5.10 is the formalization. 5.11 conclude this chapter. 5.2 Traditional views on mei 5.2.1 Dynamicity and mei The licensing conditions of mei is often discussed along the line of dynamicity. But before we jump to the conclusion, we need to first look at how dynamicity is defined. If we follow Vendler’s classification 96 (Vendler 1957, Smith 1991), dynamic verbs refers to those like activity verbs, achievement verbs, accomplishment verbs and semulfactives. Activity verbs include those like run and walk, which are continuous and homogeneous eventualities that does not have defined starting points and end points. Accomplishment verbs are eventualities that has a certain duration and a goal. Examples of accomplishment verbs include those like destroy or bake. Achievement verbs are similar to accomplishment verbs, but they do not have a significant duration before reaching the goal. This class includes verbs like find or arrive. Finally, the semelfactives are verbs that are punctual and instantaneous. Yet unlike achievement verbs, they are atelic. Examples of such verbs include blink, sneeze and knock. According to Comrie (1976), dynamic eventualities are those that are ‘continually subject to new input of energy’ (Comrie 1976: 49). Therefore, if the energy does not continue, such event ceases to hold. The opposite of dynamic predicates are the stative predicates. They do not require continuous energy inflow for the state to hold, nor does it necessarily require a state to be obtained as the result of certain force (see Comrie 1976, Smith 1991, Vendler 1967, Bach 1986, a.m.o.).Taylor (1977) argues that although ‘they obtain in time, they do not take time.’ (Taylor 1977: 206). There is an in-between category, the dynamic states (Carlson 1981, Bach 1986), such as stand and sit, that has both the properties of states and dynamic predicates. They need continuous input of energy for the situation to hold and to start. Yet similar to ordinary states, they possess what is called the subinterval property. For such an eventuality, any subinterval of that eventuality has the same property of the entire eventuality. Therefore, they do not have a non-dissectible minimal unit. They are also imperfective, meaning that they do not have natural end points. Generally speaking, Mandarin mei is licensed under all of the above dynamic predicates and dynamic state verbs. This is shown in Table 4. 97 Table 4 Mei and dynamic verbs/ dynamic states Dynamic eventuality type Examples Activity: pao ‘run’, shuo ‘talk’ Afu mei pao. Afu NEG run ‘Afu didn’t run.’ Accomplishment: kao ‘bake’, hui ‘destroy’ Afu mei kao dangao Afu NEG bake cake ‘Afu didn’t bake a cake.’ Achievement: zhaodao ‘find’ Afu mei jaodao beizi Afu NEG find cup ‘Afu didn’t find the cup.’ Semelfactives: chiao ‘knock’, kesou ‘cough’ Afu mei kesou. Afu NEG cough ‘Afu didn’t cough.’ Dynamic states: zuo ‘sit’, tang ‘lie’ Afu mei zuo zai yizi shang. Afu NEG sit at chair on ‘Afu didn’t sit on the chair.’ Yet dynamic predicates are not the only ones that license mei. In the next section, I show that there are some stative predicates that can still be licensed by mei, and this phenomenon need to be accounted for. I will first discuss how the problem is shown in the general pattern of the negation system. 5.2.2 The contrast between mei and bu The common misconception for the distinction between mei and bu is that they are licensed by different aktionsarten. Yet in fact there is no obvious distinctions in this respect for the predicates subcategorized by mei and bu. The difference lies in the fact that when mei is used, the sentence would have a salient temporal or spatial restricted phase. Here is a recap of the general patterns of the two negation markers that I discussed partially in Chapter 2. More often than not, bu negates stative predicates, which include adjectival expressions and nominal expressions. (11) Ta bu shi laoshi. He NEG COP teacher ‘He isn’t a teacher.’ 98 (12) Ta bu gao. He NEG tall ‘He isn’t tall.’ Yet bu can also appear with dynamic predicates. However, it has a strong habitual or generic reading. When under negation, the generic operator takes scope over negation. What it expresses is that it is generally the case that he does not take pills. (13) a. Ta bu chi yao. He NEG take pills ‘He doesn’t take pills (generally).’ b. Generic interpretation: GENe [It is the case e such that he doesn’t take pills.] On the contrary, in the literature, a sentence with mei asserts the nonexistence or unrealization of an event (Lin 2003). Under this analysis (14) would mean that the event of winning the ball game did not occur or is not realized. (14) Ta mei ying na-chang qiu. He NEG win that-CL ball game ‘He didn’t win the ball game.’ Furthermore, mei only appears with the copula you, as in (15). There are two major copulas in Mandarin. The first is shi, which is the default copula that appears with nominal predicates in the positive form, and while it appears in the negated forms, only bu is acceptable. On the contrary, you often serves as the copula in existential sentences, but it does not emerge in the positive form. Rather, it emerges only under negation with mei. It also has certain aspectual implications (Wang 1965, Teng 1973, Lee & Pan 99 2001, among others). It has been observed that you and a perfective marker, le, appears in complementary distributions. You only emerges under negation, as in (15), and it is ungrammatical for it to appear with le in the same sentence, as in (16). (15) Zhangsan mei you zai jia. Zhangsan NEG COP at home ‘Zhangsan wasn’t home.’ (16) Ta (*you) ying na-chang qiusai le. He COP win that-CL ball game PERF ‘He has won that ball game./He won that ball game.’ As mei only appears with you, it does not license sentences with the main copula shi in any case. Shi appears with indicative sentences, as in (17), non-gradable adjectives as in (11), and gradable adjectives with shi…-de form, as in (19). It cannot appear with the shi…-de form even with gradable adjectives either. (17) *Afu mei shi laosho. Afu NEG COP teacher Intended: ‘Afu is not a teacher.’ (18) *Zhe zhi niao mei shi huo-de. This CL bird NEG COP alive-DE Intended: ‘This bird was not alive’ (19) *Afu mei shi gao-de. Afu NEG COP tall-DE Intended: ‘Afu was tall a tall person.’ 100 It cannot appear with the ...-de form by itself either, even when the copula shi is not present. (20) *Afu mei gao-de. Afu NEG tall-DE Intended: ‘Afu was not a tall person.’ Furthermore, it cannot occur with the bare gradable adjectival form, as in (21). (21) *Afu mei gao. Afu NEG tall Intended: ‘Afu was not tall.’ Due to the selectional restrictions between mei and you and the replacement of you with le in non- negated sentences, we need to take a look at the meaning of le in order to understand the function of mei. There of two types of le’s, and the one in question appears in the post-verbal position, which is usually called le 1. This le is different from the other le, le 2, that appears in the sentence final position. The interpretation of le 1 varies with respect to the verb types. If it appears with a telic verb, le 1 is used for marking the completion of the entire eventuality. If the verb is non-telic, then it is used to express that the eventuality has started. The examples are given below. Le 1: Telic verbs: chi-wan ‘finish eating’, da-si ‘kill by beating’ (22) Afu chi-wan le 1 yu. Afu eat-finish PERF fish ‘Afu has finished the fish.’ Atelic verbs: pao ‘run’, shui ‘sleep’ 101 (23) Afu shui-le 1 Afu sleep-PERF ‘Afu has started sleeping.’ For both types of verbs, a sentence with the presence of le 1 entails the initial state of the eventuality to have already happened. The negation of le 1, therefore, would suggest that that an eventuality has never started. If you is the counterpart of le 1 under negation, and since you can only be negated by mei, we may arrive at the conclusion that mei negates the initiation of an eventuality. This conclusion does conform to the general observation that mei negates dynamic events. It is because dynamic eventualities usually has an initial stage, which is the beginning of the entire eventuality. In contrast, non-dynamic expressions, such as states, has understated initial stages. Yet this is not the entire picture. In fact, mei seems to be licensed by some of the stative eventualities. There are two types of non- dynamic predicates in question. The first is the type of adjectival expression compatible with an inchoative reading in certain contexts. The second one is the progressive sentence. The examples of these two are given in (24) and (25) respectively. In (24), ‘old’ can be interpreted as a change of state predicate because the property ‘old’ is obtained through time. Therefore it is not a stable property. On the contrary, a more stable property, such as one’s intelligence, is less likely to undergo change. Therefore, a stable property congming ‘intelligent’ would not be allowed to appear with mei, as shown in the unacceptability of (25). (24) Ta mei lao. He NEG old(er) ‘He didn’t grow old. Adjectives that does not undergo change: (25) *Ta mei congming. He NEG smart 102 Progressive predicates are not typical dynamic predicates. They describe a homogeneous property that does not develop or change in time. Yet they still requires some input of energy for maintaining the status. Lin (2003) follows the view of Bertinetto (1994) in seeing progressives as statives that requires a continuous input of energy, and hence could be licensed by mei (Bertinetto 1994). Progressive: (26) Ta mei zai chifan. He NEG at eat ‘He is not eating.’ Therefore, it seems that the case of statives licensed by mei are not really stative after all. Either they are borderline dynamic predicates like progressives, or they are transformed into inchoatives by perfective aspects. Yet the data from hen is an obvious counterexample for this generalization, since mei in Mandarin Chinese should be stative, and it does not allow a change of state reading, as in (27). Presumably, if hen only adds to the meaning that Afu’s height is in the positive extension of gao, it should be quite compatible with the inchoative reading. (27) Afu hen gao le. Afu HEN tall PERF ‘Afu is already tall, (which is unexpected.)’ NOT ’Afu has become tall.’ Furthermore, as we already know, when hen is negated, it could function as a hedge. This feature holds when the negation is done by both bu and mei. This cannot be explained if we see mei as negating dynamic predicates only, unless we want to go all the way saying hen renders the situation dynamic. 103 (28) Afu mei hen gao. Afu NEG HEN tall Intended: ‘Afu was not very tall.’ As mentioned in the previous chapter, this hedging effect is a result that hen expresses one’s commitments over the truth of the proposition. Therefore, the negation of hen lessens one’s commitment of the sentence. Now if we try to jump to the conclusion here, we could assume that hen as expressing one’s commitment is a dynamic predicate. But why does such an interpretation make a dynamic predicate? Commitments can be expressed through non-dynamic expressions, such as certain modals. There seems to be no immediate answer to that. In the next section, I show that aside from the examples given in the literature, such as statives formed by progressives and inchoatives, there are more stative predicates licensed under mei. The non-dynamic predicates licensed under mei all have one common feature. They are linked to an evaluation based on some individual’s stance, and this stance could be viewed as anchored to an eventuality of evaluation. 5.3 Stative predicates licensed by mei 5.3.1 Types of stative predicates Contrary to the generalization made in the literature, quite a few of the verbs that mei can license are viewed traditionally as non-dynamic. Stative verbs licensed by mei include those given below. (29) Four types of stative predicates licensed under mei a. Location verbs: zai ‘at/ located in’, weiyu ‘located at’, chuyu ‘in the position of’, jieyu ‘situated in between’, zhiyu ‘being put at’ b. Verbs of property description: xiang ‘resemble’, xiangsi ‘resemble’, chengxianchu ‘show/presented as’, xiande ’appear’, 104 c. Judgment verbs: quexin ‘firmly believe’, yiwei ‘think’, renwei ‘think/regard’ d. Verbs of measurement: zhong ‘weigh’, gao ‘measure in height’, shen ‘measure in depth’ Aside from the fact that they are all classified as static states, in the sense of Bach (Bach1986; also see Dowty 1975), they seem to show features of the category that overlaps with what is often called Kimian States or K-states by Maienborn (see, Maienborn 2003, 2004, 2008; Katz 2003; Mittwoch 2005; Geuder 2006; Rothmayr 2009; Moltmann 2013; Ernst 2016, among many others). According to Maienborn (2003), K-states is regarded as an opposite to the Davidsonian states, or D-states, which are used to denote non- dynamic state verbs, such as own, weigh and resemble (Maienborn 2003). In contrast, D-states refer to those like dynamic states, such as sit or lie. K-states not require continuous input of energy. They are unlike many ordinary dynamic eventualities that are anchored in a restricted spatial and temporal location. Here is a non- exhaustive list of the states that mei licenses. They can be categorized into location verbs, verbs of property description, judgment verbs, and verbs of measurement. Normally these sentences are negated by bu by default. However, they can also appear with mei. Location verbs refers to those that indicate the position of an object, and the relative position with respect to other objects. Location verbs (30) Diqiu mei (you) weiyu taiyangxi. Earth NEG (COP) located at solar system ‘The Earth is/was not located in the solar system.’ (31) Afu mei (you) chuyu buli-de qingshi Afu NEG (COP) in the position of bad-MOD situation ‘Afu is/was not in a bad position’ (32) Na beizi mei (you) zhi-yu zhuozi-shang. That cup NEG (COP) put-on table-top ‘That cup was not put on the table.’ 105 The second class, verbs of property description, are those that are description of an object as perceived by an outsider. These verbs involve those that indicates the similarity of two objects, as in (33) and (34), and those that describe what they are perceived at a stage of the state the object is having, as in (35). Verbs of property description: (33) Afu mei (you) xiang Lisi Afu NEG (COP) resemble Lisi ‘Afu didn’t resemble Lisi.’ (34) Afu gen Lisi mei (you) xiangsi. Afu and Lisi NEG (COP) resemble ‘Afu didn’t resemble Lisi.’ (35) Zhe zhaopian mei (you) chengxianchu yi-zhong xinfu. This photo NEG (COP) show one-CL happiness ‘This photo didn’t show/is not showing any sense of happiness.’ Now we turn to judgment verbs. Here they refer to the subclass of the attitude verbs that shows judgment and decision towards a situation made specifically by an attitude holder. This set includes a specific subset of propositional attitude verbs: those that show the subject’s degree of certainty towards the situation, as in (36) to (37). For example quexin ‘firmly believe’ is a modal used to express the attitude holder’s strong belief in the prejacent. Yiwei ‘think (counter to the fact)’ is a counterfactual modal, which is used when the knowledge holder faulsely believe that the prejacent is true. In contrast, typical epistemic modals, which refers to those that do not show a strong personal judgments of the subject attitude like zhidao ‘know’ and ‘believe’, are less acceptable with mei, as shown in (38). Note that these verbs are used to express judgments that seems to anchor to certain situations, not the general perspective of this attitude holder. 106 Judgment verbs: (36) Afu mei (you) quexin Lisi shi haoren. Afu NEG (COP) firmly believe Lisi COP good person ‘Afu firmly believed that Lisi is a good person.’ (37) Afu mei (you) renwei Taiwan hen da. Afu NEG (COP) think Taiwan HEN big ‘Afu didn’t think that Taiwan was big.’ (38) ??Afu mei (you) zhidao/xiangxin Lisi zai jia. Afu NEG (COP) know/believe Lisi at home Intended: ‘Afu didn’t know/believe that Lisi was at home.’ The last case, verbs of measurement, are verbal expressions that have adjectival counterparts that share the same phonological forms. They express how an object is measured with respect to the dimension corresponding to the adjective. Verbs of measurement: (39) Afu mei (you) zhong wushi gongjin Afu NEG (COP) weigh fifty kilos ‘Afu didn’t weighs fifty kilos.’ (40) Afu mei (you) gao lian gongchi. Afu NEG (COP) measure-in-height two meters ‘Afu didn’t two meters tall.’ (41) Shui mei (you) shen yi-bai mi. Water NEG (COP) measure-in-depth one-hundred meters ‘The water wasn’t 100 meters deep.’ 107 Used as adjectives, the word zhong ‘weigh’ literally means ‘heavy’, gao is ‘tall’ and shen means ‘deep.’ Intuitively speaking, the uses in (39) to (34) behave like verbs. But they cannot appear with both perfective and non-perfective aspectual markers, such as experiential perfect guo and progressive zhe. (42) *Afu zhong-le/guo/zhe wushi gonjin. Afu weigh-PERF/EXP.PERF/PROG fifty kilos*Shui shen-le/guo/zhe yi-bai mi. Water measure-in-depth-PERF/EXP.PERF/PROG one-hundred meter Obviously, the view that mei only licenses dynamic predicates would become problematic with these unquestionably stative predicates unless we want to embrace the view that these statives are in fact dynamic. In the next section, I discuss how they could potentially be interpreted dynamically and then conclude that it is still not likely to be the right analysis. 5.3.2 The possibility of being dynamic It is tempting to think that there is still a possibility that the Mandarin verbs in (22) are still dynamic after all, as the group are non-homogeneous and one need to examine them one by one. Aside from the fact that such verbs are traditionally classified as static states or K-states, the answer is no. One of the widely accepted tests used in English for identifying dynamic predicates is the progressive form test. Only dynamic predicates and dynamic states can have progressive forms. As for Mandarin Chinese, I propose that there are two tests: (a) the progressive test, and (b) the perfective marker le 1 test. Mandarin Chinese has two forms that could be regarded as similar to English progressives (Smith 1991, Klein et. al 2000). The first one, zai, occurs with canonical dynamic predicates, and the second one, zhe, appears with dynamic states, such as zuo ‘sit’ and tang ‘lie’, or chuan ‘wear.’ Zhe is also a durative marker. 108 It indicates that a situation is enduring or continuing. Zai is not allowed to appear with dynamic states, while zhe cannot appear with canonical dynamic situations. (44) Mandarin progressive markers: Zai (canonical dynamic situations): Dynamic states: zai-chifan ‘eating’, zai-shuijiao, ‘sleeping’ Non-dynamic states: *zai-zhan ‘standing’, *zai-tang ‘lying’ Zhe (dynamic states): Non-dynamic states: zhan-zhe ‘standing’, tang-zhe ‘lying’ Dynamic states: *chifan-zhe ‘eating’, *shuijiao-zhe ‘sleeping’ Furthermore, if verbs are distinguished with respect to the dichotomy of events and processes, only zai can appear with events, while zhe only appears with processes. A contrast between the two can be shown below. Zai-progressive: (45) Lisi zai chuan yi-jian qunzi. Lisi ZAI wear one-CL skirt ‘Lisi is putting on a skirt.’ Zhe-progressive: (46) Lisi chuan-zhe yi-jian qunzi. Lisi wear-ZHE one-CL skirt. ‘Lisi is wearing a skirt.’ (Klein et al. 2000: 727) 109 None of the four types of verbs are compatible with zai or zhe, as shown in (47) to (50), and (51) to (53). Therefore, they do not pattern with dynamic verbs or dynamic states. Zai-progressive: (47) *Diqiu zai weiyu taiyangxi. Earth ZAI located solar system (location verbs) (48) *Afu zai xiang Lisi Afu ZAI resemble Lisi (verbs of property description) (49) *Afu zai zhong wushi gongjin Afu ZAI weigh fifty kilos (verbs of measurement ) (50) ??Afu zai quexin Lisi shi haoren. Afu ZAI strongly-believe Lisi COP good person (judgment verbs) Zhe-progressive: (51) *Diqiu weiyu-zhe taiyangxi. Earth be-located-in-ZHE solar system (location verbs) (52) *Afu xiang-zhe Lisi Afu resemble-ZHE Lisi (verbs of property description) (53) *Afu zhong-zhe wushi gongjin 110 Afu weigh-ZHE fifty kilos (verbs of measurement ) (54) ??Afu quexin-zhe Lisi shi haoren. Afu strongly-believe-ZHE Lisi COP good person (judgment verbs) The second test, the two le’s test, could distinguish between two classes, dynamic predicates and dynamic states as the first category, and non-dynamic statives as the second. Earlier I mentioned that there are two types of le’s , le 1 and le 2. Le 1 is the perfective marker that occurs directly after verbs. Le 2 contributes to either a change of state meaning or the implication of contrary-to-expectations. Below is the summary of the two le’s with respect to different predicate types. (55) The distribution of two le’s and verbal predicates Position With (canonical) dynamic With dynamic states With stative verbs Le 1 after the verb perfective perfective inchoative Le 2 Sentence final change of state change of state Gradable adjectives: Contrary to expectations Non-gradable adjectives: Change of state Recall that le 1 could be used to form inchoative expressions. To be more precise, le 1 has two readings (Smith 1991). When le 1 appears with non-stative predicates, such as dynamic verbs and dynamic states, it 111 yields a perfective reading. If it appears with stative predicates, there is only an inchoative reading, which states ‘a coming about of the situation’ (Smith 1991: 346) 5 as in (56). (56) Afu pang-le Afu fat-LE 1 ‘Afu has become fat.’ Le 1 is not acceptable with most of the four predicate types. It is ungrammatical with location verbs, as in (57), verbs of property description, as in (58), and the judgment verbs, as in (60). Le 1 and verbs in (29): (57) *Diqiu weiyu-le taiyangxi. Earth be-located-in- LE 1 solar system (location verb) (58) ??Afu xian-le Lisi Afu resemble Lisi (verbs of property description) (59) a. Afu zhong-le wushi gongjin Afu weigh- LE 1 fifty kilos 5 Other authors have proposed a different licensing condition for the inchoative reading of le 1. For example, Shih (1990) argues that the distinction is about the telicity of the event. When it appears with telic events, the sentence gets the perfective reading. Otherwise it gets an inchoative reading. This observation is consistant with Comrie’s analysis (Comrie 1976) that perfective markers across languages have this function of turning a stative verb into an inchoative verb. 112 ‘Afu has gained fifty kilos.’ b. Shui shen-le wu gongchi. Water deep- LE 1 five meter ‘The water has become five meters deeper.’ (verbs of measurement) (60) a. ??Afu quexin-le Lisi shi haoren. Afu strongly-believe-LE 1 Lisi COP good person b.??Afu yiwei-le Lisi shi haoren. Afu think-falsely- LE 1 Lisi COP good person (judgment verbs) The only acceptable cases are the verbs of measurement, as in (59), which yields the inchoative reading: the sentence is interpreted as Afu would start to be in the new state of having fifteen more kilos. None of the examples with le 1 show that they pattern with ordinary dynamic predicates. Now let’s turn to the problem of le 2 that appears at the sentence final position. It has two perhaps correlated meanings: (i) the change of state meaning and (ii) contrary to expectation meanings (Li & Thompson 1981; Sybesma 2013; Soh 2009). The licensing condition for both readings is that the eventuality is not obtained at some time prior to the reference time. In other words, sentence (61)b would be infelicitous if there is no state such that the Earth did not circling the sun. (61) a. Diqiu rao taiyang xuanzhuan. Earth circle sun turn ‘The Earth circles around the Sun.’ b. Diqiu rao taiyang xuanzhuan le Earth circle sun turn LE 2 113 ‘The Earth circles around the Sun (, which, it did not before/contrary to what one may expect.)’ (Soh 2009: 633) According to Soh (2009), the two different readings come from whether the newly updated Common Ground of assertion p accepts or negates the ¬p presupposition in a previous time before the currently updated Common Ground. Assuming that there is a stage ¬p prior to the speech time t s. The Common Ground at t 0 is that everyone accept ¬p to be true. In a later time t 2> t 1, the Common Ground is updated. If the participants of the Common Ground at t 2 accept ¬p to be true at t 1, it yields a change of state reading. If the participants reject ¬p at t 1, and asserts p at t, we get the contrary to expectation reading. In the latter reading, I propose that it could be interpreted that the time of the change from ¬p to p is earlier than one might expect. While Soh’s analysis captures the two meanings in a uniform manner by introducing Stalnaker’s Common Ground and dynamic updates, it is worth noting that the emergence of each reading is distinguished by different aktionsarten of the verbs. Although a detailed examination of the aktionsarten in Mandarin verbs is beyond the scope of this work, it is generally the case that canonical dynamic predicates and dynamic states tend to be interpreted with a change of state reading, which often accompanies a weak, if not completely non-existing meaning that implies unexpectedness. This is shown in the sentences below. Dynamic predicates: (62) Afu chi-fan le. Afu eat-meal LE 2 ‘Afu has eaten meal (, which, he has not before.)’ (63) Afu paobu le. Afu run LE 2 ‘Afu has run (, which, he has not before.)’ 114 Dynamic states: (64) Afu zhan yi-xiaoshi le Afu stand one-hour LE 2 ‘Afu has been standing for an hour (, which, he has not reached one hour earlier.)’ (65) Afu zai chuang-shang tang yi-tian le. Afu on bed-above lie one-day LE 2 ‘Afu has been lying on the bed for a day (, which, he has not reached the length of a day earlier.)’ The contrary to expectation meaning arises most naturally in just one subset of stative predicates, the gradable predicates, as in (56). However, in other non-gradable expressions, the existence of this inference is unclear. Instead, it is the change of state reading that emerges. Non-gradable stative expressions: (66) Zhangsan xiangxin Lisi le. Zhangsan believe Lisi LE 2 ‘Zhangsan believed Lisi (, which he did not before).’ Apparently, if one only resorts the two meanings to the distinction between accepting or negating the presupposition of a prior ¬p stage, there is no mechanism to prevent the emergence of the contrary to expectation reading in all the predicate types in (62) to (65). Now I return to the four types of stative verbs. They all show two features. First, not all of them are compatible with le 2. When le 2 appears with verbs of location like weiyu ‘located at’, as well as judgment verbs and verbs of property description, the grammaticality margins from borderline acceptable to ungrammatical, as in (67) and (68). The unacceptability probably comes from unnaturalness of change in 115 these verbs, such as in the case of location verbs, only if the location is potentially changeable, and in the case of verbs of property description, only when the look can change easily with time would it be natural for them to be interpreted as changeable predicates. (67) ??Na-tai che weiyu Diwudadou le That-CL car locate Fifth Avenue LE 2 ‘The car is already located on the Fifth Avenue (, which, was not the case before/unexpected.)’ (location verb) (68) ?Afu xiang Lisi le Afu resemble Lisi LE 2 ‘Afu resembles Lisi (, which, was not the case before/unexpected.) ’ (Verbs of property description) (69) a. ??Afu quexin Lisi shi haoren le Afu strongly-believe Lisi COP good person LE 2 ‘Afu strongly believed that Lisi was a good person (,which, was not the case before/unexpected.)’ b. ??Afu yiwei Lisi shi haoren le. Afu think-falsely Lisi COP good person LE 2 ‘Afu thought falsely that Lisi was a good person (, which, was not the case before/unexpected.’) (judgment verbs) As for measurement verbs, le 2 is more acceptable. However, the sentences are still marked, as shown in the ‘?’ marking below. Verbs of measurement entail that there is a prior state that the object being measured is less in quantity or degrees. For example, (70)a implies that Afu was not as heavy as fifty kilos in the past. Another reading for le 2 here is to act as a scalar particle that operate on a numerical degree scale, and is related to an assertion that counts as high on the scale. This analysis makes le 2 similar to English already (Soh 2009), specifically, as the scalar subclass of already. 116 (70) a. ?Afu zhong wushi gongjin le Afu weigh fifty kilos LE 2 ‘Afu is fifty kilos, (which, was not the case before).’ b. ?Shui shen wu gongchi le. Water deep five meter LE 2 ‘The water is fifty meters (, which, was not the case before).’ (verbs of measurement) Below is a summary of the results of the progressive tests and the le tests. (71) The distribution of le 1 and le 2 with the four types of predicates Dynamic verbs Location verbs Property description Judgment verbs Verbs of measurement Progressives OK ✗ ✗ ✗ ✗ Le 1 OK ✗ ?? ?? ? Le 2 OK ? ? ? ? To sum up, the progressive test clearly show that the four types of verbs do not pattern with canonical dynamic predicates. As for the two le’s tests, only verbs of measurement are more compatible with both, but such sentences are still marked. These all show that the four types of verbs are not dynamic predicates at all. Aside from the lacking of parallelism with the dynamic sentences under the progressive test and the two le’s tests, one might want to look at whether the semantics of these verbs are natural for a dynamic interpretation in the semantic sense. I first look at the location verbs. Now I turn to the question on whether location verbs can be interpreted as dynamic if we see the situation as requiring continuous energy to be placed in a certain locations. 117 To view things this way is to say that for an object to be located in one place, it is either (a) an action of the object itself, or (b) the result of some outside force that places the object in that location. Yet both seems to have some problems. For the first option, while the idea makes perfect sense for verbs with an agentive or sentient object, it does not seem to hold for non-sentient objects. In fact, the verbs used to describe an agentive or sentient object locating at a certain place are those like zhan ‘stand’, tang ‘lie’, or li ‘stand erect.’ These verbs are typical dynamic states, and they can all appear with progressive zhe, while verbs of location cannot, as in (44). As for the second possibility, it would be refuted if such verbs are known facts or does not have a prior ¬p state. A good example that demonstrates verbs of location as distinct from ordinary dynamic state verbs come from expressions that describe rules, physical or mathematical facts, which include those that hold true even across worlds. Take (72) for example. The verb jieyu, ‘situated between’, can be used to denote a relation between abstract numbers. There is no input of energy that puts 7 between 6 and 8. It would be totally counterintuitive if we view it as an action of placing 6 between 7 and 8. The position of a number on a number line is not an action of 7, either. Facts and location verbs: (72) Zai ziran shu zhong, shuzi qi jieyu In natural number number seven situated between shuzi liu he ba zhijien. Number six and eight middle ‘Among natural numbers, the number 7 is between number 6 and 8.’ The next verb categories that have the taste of being dynamic are the verbs of measurement and the judgment verbs. Verbs of measurement can be regarded as those that involve some action of measuring. If one put it this way, the verb zhong ‘weigh’, for example, can be reinterpreted as the object undergoing measurement. With this analysis, however, one would expect that the measurement can be temporally 118 bounded, and it might have a starting point and perhaps an end point as well. However, it is not the case. One cannot use verbs like kaishi, ‘to start’ for those verbs. (73) ??Afu kaishi zhong wushi gongjin. Afu start weigh fifty kilos Intended: ‘Afu now starts to weigh fifty kilos.’ (74) ??Shui kaishi shen wu gongchi. Water start measure-in-depth five meter Intended: ‘The water starts to be the case that the water is measured as fifty meters deep.’ Now I move to judgment verbs. Judgment verbs seem to be the most likely category to be dynamic, especially if one sees mental activities as actions that requires constant energy flow. However, even among judgment verbs, or say epistemic verbs in general, there might be different degrees of the existence of the notion of mental energy. For example, verbs like rending and duanding, both translated as ‘firmly believe’, and typical epistemic modals like ‘believe’ xiangxin are more ‘dynamic’ than verbs like juede ‘think’ , renwei ‘think’ and yiwei ‘think falsely.’ The contrast is similar to the contrast between English think and believe. In the case of think, it could refer to a specific mental activity, which could obtain continuous energy from the agent. It could refer to multiple activities or one continuation of thinking activity. However, in the case of believe, it is more continuous, and one’s belief is less likely to be modified throughout time. Therefore, I conclude that there are at least two types of judgment verbs, despite all being statives, with one class leaning stronger towards a more dynamic reading, and the other towards a more stative reading. The examples below show some of words of the two categories. All these verbs can be negated by mei. (75) a. ‘Dynamic’ judgment verbs: xiangxin ‘believe’, duanding ‘firmly judge’, panduan ‘to judge’, rending ‘firmly believe’ 119 b. ‘Non-dynamic’ judgment verbs: juede ‘think’, renwei ‘think’, yiwei ‘think falsely’, ganjue ‘feel/find’ On the surface, the distinction between the two categories in (75) seems to be based on whether the evidence is more subjective or more objective. While this may be part of the distinction, it is still related to the aktionsart of the verbs. The verbs in group (a) in (75) can appear with le 1 to get a perfective reading. (76) Afu duanding/panduan/rending-le Lisi shi xiongshou. Afu firmly judge/judge/firmly judge-LE 1 Lisi COP murderer. ‘Afu firmly made a judgment that Lisi was the murderer.’ To conclude, as the four types of verbs do not perform like typical dynamic predicates, they do not appear in progressive forms, nor do they have a perfective reading when they appear with le 1. The generalization is given below. (77) Features of stative predicates licensed under mei (verbs in (22)): a. They do not appear with progressive zai or zhe. b. They do not appear with perfective le 1 to yield a perfective reading. c. Le 2 is compatible with all the verbs if the properties can change with time. It has an extra meaning of scalar already when it occurs with verbs of measurement. d. Intuitively speaking, they do not fit into the criteria of dynamic verbs, which typically require continuous input of energy. Among the three conditions, (77)b is the one that sets ordinary stative verbs apart from those licensed by mei. Only stative verbs can appear with le 1 and yields a perfective reading. To conclude, the lengthy discussion given above shows that dynamicity is not the defining property of the licensing condition of mei. 120 5.3.3 Further supports of the stative status of these four types of verbs Here are two further test for the stative nature of these verbs. The first is the still and no longer tests proposed by Katz (2003). English still and no longer can appear only with stative verbs. Consider the contrast below. (78) a. John kissed Mary no longer. b. John no longer owns a car. (Katz 2003: 465) Similarly, in Mandarin Chinese, the counterpart of still and no longer, which are hai and buzai respectively, can occur with all the four types of Mandarin predicates. (79) Na-jian fanzi hai weiyu Diwudadao That-CL house still located at Fifth Avenue. ‘That house is still on Fifth Avenue.’ (location verb) (80) Afu hai xiang Lisi. Afu still resemble Lisi ‘Afu still resembles Lisi.’ (Verbs of property description) (81) Afu hai zhong wushi gongjin Afu still weigh fifty kilos ‘Afu still weigh 50 kilos.’ 121 (verbs of measurements) (82) Afu hai shi quexin Lisi shi haoren Afu still COP strongly-believe Lisi COP good person ‘Afu still strongly believed that Lisi was a good person.’ (judgment verbs) A second piece of evidence is shown in the test using ‘in the state/situation of.’ This phrase is not acceptable to occur with dynamic predicates. This test is used by Fábregas et. al. (2012) on Spanish psych verbs and nouns. The Mandarin counterpart is zai…-de qingxingxia, ‘under the circumstance/state of.’ (83) ??Zai Afu chi-le hanbao de qingxingxia, ta chi-bu-xia At Afu eat-PERF hamburger DE circumstance he eat-NEG-able qita dongxi. other thing Intended: ‘Under the circumstance that Afu had eaten a hamburger, he could not have eaten anything else.’ (84) a. Lisi hai zai jia de qingxingxia, Lisi xianran shuohuang Lisi still at home DE circumstance Lisi obviously lie le. PERF ‘With the state that Lisi was still at home, he obviously lied.’ 122 (85) b. Zhe-dong dalou weizai diwudadao de qingxingxia, buneng This-CL building located-at Fifth Avenue DE circumstance cannot suibian chai. sloppily demolish. ‘Under the circumstance that the building is on Fifth Avenue, it cannot be demolished sloppily.’ (location verbs) (86) Zai fenzi A he B xiansi de qingxing-xia, At molecule A and B similar DE circumstance-under A bi B gen you huoxin. A COMP B more have activeness ‘Under the circumstance that molecule A resembles molecule B, A is more active.’ (87) a. Zai Lisi duanding zhe gu you-du de At Lisi judge this mushroom poisonous DE - qingxing-xia, bie chi. circumstance-under better-not eat ‘Under the circumstance that Lisi firmly believes that the mushroom is poisonous, it’s better not to eat it.’ b. Zai Lisi renwei mien haochi de qingxingxia, chushi hui At Lisi think noodle tasty DE circumstance cook will yiran yong tongyang de shipu. Still use same DE recipe ‘Under the circumstance that Lisi thought that the noodles were tasty, the cook would continue using the same recipe.’ (property description verbs) 123 (88) a. Zai Afu zhong wushi gongjin de qingxingxia, ta pao-de-wan At Afu weigh fifty kilos DE situation he run-could-finish malasong. Marathon ‘Under the circumstance that Afu weighs fifty kilograms, he can finish a marathon.’ (verbs of measurement) Once again, we could see that it is unnatural to see the four types of verbs as dynamic. Viewing a hen sentence as simply dynamic would not be an easy fix for why it licenses mei. 5.3.4 A look at hen predicates The hen form shows the same patterns of those in (77) as other four types of predicates. It would not be surprising to see that hen cannot appear with progressives, as in (89), and not having a typical dynamic meaning. (89) a. *Afu zai hen gao. Afu PROG HEN tall b. *Afu hen gao zhe Afu HEN tall PROG The hen form cannot occur with le 1 if it is placed after hen. (90) *Afu hen-le gao Afu HEN-LE 1 tall 124 Yet hen can appear with le 2. As shown in (91). However, it only yields the contrary to expectation reading. In comparison, when hen is absent, the change of state emerges. The contrast between the two sentences below show that when hen is there, it patterns with location verbs, verbs of property description, and judgment verbs. (91) Afu hen gao le. Afu HEN tall LE 2 ‘Afu is already tall (, which is unexpected)’ (92) Afu gao le. Afu tall PERF ‘Afu has become tall.’ 5.4 Stance taking and temporal boundedness 5.4.1 Overview In the previous sections, I looked into why it is an oversimplification to see dynamicity as the licensing condition of mei. In this section, I show temporal boundedness, instead of dynamicity would be one of the necessary features for the licensing condition of mei. The temporal boundedness comes from the existence of an evaluation eventuality that is temporally anchored. The temporal boundedness is not part of the nature of the eventuality predicated by the matrix predicate. It temporal boundedness were indeed the feature of the eventuality itself, there could be two types of sources for this boundedness: (a)we could see the predicate as a stage level predicate (SLPs) themselves, and (b) we could see them as either stage level predicates or individual level predicates (ILPs), but what is being negated is just one temporally bounded stage of that property. As the four types of stative predicates licensed by mei could be ILPs, the second approach should be more likely. Here is a summary of the two possibilities. 125 (93) Possible sources of temporal boundedness when mei is used (to be rejected): For a given property p true of an eventuality e… (a) if p is a stage level predicate (SLP), then ¬p with mei negates the non-permanent property p as a whole. (b) if p is an individual level predicate (ILP), then ¬p with mei negates a stage of p. However, in the sections below, I will show that neither cases would work. While it is obvious for (a), since not all sentence negated by mei are SLPs, it is equally problematic for (b) as well. This point will be explained the following sections. 5.4.2 Negation and temporal boundedness When a sentence with property p is negated by mei, the negated proposition ¬p is true for a given time frame. Under this analysis, a sentence like (94)a is interpreted as (94)b. (94) a. Afu mei zhu jia-li. Afu NEG live home-in ‘Afu was/is not living at home.’ b. There is a salient time t such that during t Afu was/is not considered to be living at home. One might wonder why we cannot see a negated sentence ¬p negated by mei as related to temporal boundedness reminiscent to the contrast between the notions of SLPs and ILPS. Yet it is not difficult to see why (93)a is problematic. The distinction between SLPs and ILPs is about the duration of a property that holds true of the object itself. SLPs refers to those that can be true to an ontologically non-permanent state (Carlson 1977, Milsark 2014, Kratzer 1994, Diesing 1992, Jäger 2001, among others), while ILPs usually denote permanent properties. Some typical examples of SLPs include those like available or hungry. They are only properties true of a certain temporally bounded stage of the subject. On the contrary, ILPs include 126 those like altruistic, smart or intelligent. ILPs often denote generic or characteristic properties. Of course, all sorts of states can have the potential to be true only with respect to a certain stage only if we only look at one temporal slice of that state. In other words, ILPs have the potential to be interpreted as SLPs under certain contexts. For example, modifiers that marks temporal frames could potentially change an ILP into a SLP. For example, in (95), smart is interpreted as a SLP. (95) He was smart when he talked to you. Yet some ILPs are less capable of being interpreted as stage level even with the presence of this type of temporal modification. For example, attitude verbs like believe and know cannot occur with temporal adverbs. Could it be the case that mei only negates SLPs, as suggested in the first possible generalization in (93)a? This view could be easily refuted. Mei could be used with permanent states like the examples with location verbs, as in (96). This property of Polaris, for not being in the solar system, is obviously not an SLP. (96) Beijixing mei weiyu taiyanxi. Polaris NEG located at solar system ‘Polaris is/was not located in the Solar System.’ Apparently, the interpretation of (93)a is rejected. How about in (93)b? Under (93)b, Polaris is not in the Solar System at a salient time t. In theory, such a reading is truth conditionally sound, since if Polaris is not in the Solar System at t, it is compatible with the truth condition that it is not in times other than t. However, the problem with this reading is that when there is a temporal restriction, it often triggers the implicature that this property could be temporary. This is shown in the example (97). Here I use the more 127 neutral negation marker bu. The property of kindness is more naturally interpreted as true of one stage of Afu simpliciter, not a characterizing property of him. (97) Afu na shihou bu shanliang. Afu that time NEG kind ‘Afu was not kind.’ Interestingly, when (96) is modified by a temporal modifier, as shown in (98), the same implicature is also present. The sentence sounds unnatural unless one expects Polaris to be at some point capable of being in the Solar System. (98) Beijixing na shihou mei weiyu taiyangxi. Polaris that time NEG located at solar system ‘Polaris is/was not located in the Solar System at that time.’ Another counterexample for (93)b comes from sentences with kind subjects. When a subject is kind denoting, the predicate is more naturally interpreted as ILPs. In (99), since the predicate hibernated is a property of the category of bears, it refers to a characteristics of the bear type. Therefore, to hibernate is an ILP under this context, as shown in (99). However, when Mandarin mei appears with kind subjects, it does not output an ILP reading, as shown in (100). (100) is not interpreted as if there was a time during which it was generally the case that bears didn’t hibernate. It only refers to a single instance in which bears during that time didn’t hibernate. (99) Bears hibernated. 128 (100) Xiong-lei mei dongmian. Bear-kind NEG hibernate ‘Bears didn’t hibernate.’ In short, if what mei is negating is the property that holds at a certain time duration, the sentence would likely to trigger the implicature that there is a possibility that such a property does not hold in some other times. This would run into problems with examples like (97) and (98). Therefore, the second possibility given in (93)b is rejected. If the two possibilities in (93) are both rejected, what does it negate? That’s where the event of evaluation gets into the picture. To conclude, the above examples from kind subjects show that what is being negated is not just one temporal slice of the ILP. This would support the view that there is indeed an eventuality of evaluation, and what is being negated is the property as perceived by the attitude holder. 5.4.3 Temporal modifiers The above assumption is supported by the fact that the four types of predicates licensed by mei are all natural with temporal modifiers, but less so when they are negated by bu. This suggests that they are more likely to refer to a temporally bounded situation. As mentioned earlier in this chapter, there is an obvious contrast between a sentence negated by bu and mei. When negated by bu, it is a negation of the characteristic property of the subject, and it also often yields a generic reading. When negated by mei, on the contrary, it negates a situation or property that has existed in a specific time frame. This time frame would then coincide with the time during which that stance held of the attitude holder. This observation is supported by data in which these two negation markers co-occur with different temporal adverbs. The function of a temporal modifier is to bind a time during which a certain state holds. As shown in (101) to (105), they could be modified by temporal modifiers that are used to express either a time point or a time frame. Xianzai ‘now’ refers to a time point, while zheme jiu yilai ‘for such a long time’ and yizhi yilai ‘all the time’ both refer to time durations. 129 Temporal modifiers: (101) Diqiu {xianzai/ zheme jiu yilai dou} weiyu taiyangxi. Earth now / such long since all located at solar system ‘The Earth is {now/has long been} located in the Solar System.’ (Location verbs) (102) Afu {xianzai/yiqian/ yizhi yilai} dou xian Lisi Afu now /before now/all the time always resemble Lisi ‘Afu now resembles Lisi/Afu resembled Lisi in the past/Afu has always resembled Lisi.’ (103) Zhe gongyuan {xianzai/zhiqian/yizhi dou} chengxianchu This park now/before now/all the time all show yi-zhong pingjin. one-kind serenity ‘This park now shows an atmosphere of serenity/ This park used to show an atmosphere of serenity/ This park has been shown an atmosphere of serenity.’ (Verbs of property description) (104) Afu {xianzai/zhiqian/yizhi} renwei Taiwan hen da. Afu now/before now/all the time think Taiwan HEN big ‘Afu now thinks that Taiwan is big /Afu thought that Taiwan was big before/ Afu has always thought that Taiwan is big.’ (Judgment verbs) 130 (105) Afu {xianzai/zhiqian/yizhi/zuotian} gao lian gongchi. Afu now/before now/all the time/yesterday measure-in-height two meters ‘Afu is now/was/ two meters tall./Afu has always been two meters tall./Afu was two meters tall yesterday.’ (Verbs of measurement) Both bu and mei are acceptable. This is shown in (106) to (110). (106) Diqiu {xianzai/zheme jiu yilai dou} mei/bu weiyu taiyanxi. Earth now/such long since all NEG located at solar system ‘The Earth is {now/has long been} located in the Solar System.’ (Location verbs) (107) Afu {xianzai/yiqian/yizhi yilai dou} bu/mei xian Lisi Afu now /before now/all the time all NEG resemble Lisi ‘Afu now resembles Lisi/Afu resembled Lisi in the past/Afu has always resembled Lisi.’ (108) Zhe gongyuen {xianzai/zhiqian/yizhi} bu shi/mei chengxianchu This park now/before now/all the time show NEG COP/NEG yi-zhong pingjing. one-kind serenity ‘This park now shows an atmosphere of serenity/ This park used to show an atmosphere of serenity/ This park has been shown an atmosphere of serenity.’ (Verbs of property description) 131 (109) Afu {xianzai/zhiqian/yizhi} bu/mei renwei Taiwan hen da. Afu now/before now/all the time NEG think Taiwan HEN big ‘Afu is not thinking that Taiwan is big now/Afu didn’t think that Taiwan was big before/ Afu has always think that Taiwan is not big.’ (Judgment verbs) (110) Afu {xianzai/zhiqian/yizhi/zuotian} bu/mei gao Afu now/before now/all the time/yesterday NEG/NEG measure-in-height lian gongchi. two meters ‘Afu isn’t two meters tall now./ It is not the case that Afu has always been two meters tall./Afu was not two meters tall yesterday.’ (Verbs of measurement) However, in (106) to (110), the two negation markers give rise to slightly different interpretations. With temporal modifiers, when bu is used, the sentences implicate that the negated property only holds of the time restricted by the temporal adverb. In other words, it is more natural to yield an SLP reading when bu is used in tandem with temporal modifiers. Take (111) for example. Since it emphasizes that it is only during that time that Afu was not two meters tall, it provides an implicature that the height of Afu could be different from time to time. (111) Afu na-shi bu shi gao lian gongchi. Afu that-time NEG COP measure-in-height two meters ‘During that time, it was not the case that Afu was two meter tall.’ Implicature: It was not necessarily the case that Afu remains the same height throughout time. 132 However, when bu is replaced by mei, such an implicature disappear. (112) Afu na-shi mei you gao lian gongchi. Afu that-time NEG COP measure-in-height two meters ‘During that time, it was not the case that Afu was two meter tall.’ Since height could change as time progresses, if we want to see the contrast more clearly, we should look at predicates that are more strictly interpreted as ILPs. Location verbs and verbs of measurement are the two categories among the four that tends to be interpreted as ILPs. Therefore I only discuss the examples with these two. Compare the two sentences in (113) below. Both sound unnatural out of the blue. However, when they are placed in a context in which the situation is restricted to the time denoted by the temporal adverb, the mei form of (113)a sounds better than the bu form of (113)b. Here is the scenario. Imagine that the speaker is talking about a periodic table. A few minutes ago the speaker checked the periodic table on a website, and he saw that that the chemical zinc was not in the first row. Now it is felicitous for him to utter (113)a, a sentence negated by mei, but not (113)b with bu. For the sake of showing the difference, here I translate the sentence roughly as ‘was not shown’ to indicate feature that this property of zinc is not permanent in (113)a, although it may not be the most accurate translation. (113) a. Ganggang xin mei zai zhouqi biao diyi hang shang. Just then zinc NEG at periodic table first row on ‘Just then, the chemical zinc was not shown on the first row of the periodic table.’ b. ?Ganggang xin bu zai zhouqi biao diyi hang shang. ‘Just then, the chemical zinc was not on the first row of the periodic table.’ Why is (113)a more acceptable than (113)b? The explanation is that it is not the property itself being restricted temporally by ganggang. Rather, it restricts the time when the speaker’s observation is made. 133 Such an option is provided by mei, but less likely by bu. It seems that bu tends to negate the property denoted by the predicate that holds within the time restricted by the temporal adverbs. Similar contrast is shown in measurement verbs. They denote permanent properties when the unit of measurement is fixed. Imagine a scenario like this. The speaker was looking at a well a few minutes ago, which was a time before the utterance time. He concluded that the well was not 10 meters deep. Assume that the speaker thinks that the depth of the well does not change. Now it is felicitous for him to utter a sentence with mei, such as (114)a, but not one with bu. The negation of bu implies that that the feature ‘not 10 meter deep’ is only restricted to the time ‘just then.’ (114) a. Ganggang na jin mei shen shi gongchi. Just then that well NEG measure-in-depth ten meter ‘Just then, the well was not shown as if it were ten meters deep.’ b. ??Ganggang na jin bu shi shen shi gongchi. Just then that well NEG COP measure-in-depth ten meter ‘Just then, the well was not ten meters deep.’ The contrast between bu and mei could be explained if mei is licensed by a time anchored to the time of the evaluation is also shown in the examples of judgment verbs. 5.4.4 Attitude verbs Attitude verbs that represent one’s subjective judgements are involved with a temporally bounded stance. Therefore, they could also serve as a test for the temporal boundedness in sentences that license mei. I mentioned earlier that attitude verbs can be classified into two subtypes according to whether they could be anchored to a stance. When it is used to refer to constant or permanent beliefs, such as xiangxin ‘believe’, it is less likely to be bounded by time. On the contrary, verbs like rending ‘firmly believe’ provides extra meaning related to the speaker’s subjective evaluation and the strength of belief, which could change with 134 time more easily. The same goes with judgment verbs that also encodes meanings of the attitude holder’s feelings, such as ganjue ‘feel.’ In comparison, if the attitude is held more ‘objectively’, which means they refer to one’s belief or knowledge from the outside world or from inferences, rather than about one’s personal opinions at some given point, it is often more stable. This is especially true about knowledge. Once a person knows something, it is usually for good unless certain unnatural conditions occur. If this dichotomy between the two categories makes certain sense, one would expect that the Mandarin counterpart for English know, zhidao, would be bad under mei. On the contrary, judgment verbs that changes with the attitude holder’s stance or perspectives are more acceptable to be negated by mei. This prediction is born out, as shown in the contrast between (115) and (116). (115) *Wo mei zhidao Lisi zai jia. I NEG know Lisi at home Intended: ‘I didn’t/don’t know that Lisi is home.’ (116) Wo mei rending Lisi zai jia. I NEG firmly believe Lisi at home ‘It wansn’t the case that I firmly believed that Lisi was home.’ The other three types of verbs, verbs of property descriptions, location verbs and verbs of measurement often denote individual level properties. Yet they are still licensed under mei despite the fact that an ILP interpretation is more natural. 5.4.5 The stance and the evaluation eventuality The arguments in the previous sections refute the two possible sources of temporal boundedness given in (93). The solution here is that there is an eventuality that represents the stance or the evaluation of an attitude holder in sentences licensed by mei. I assume that the eventuality could be introduced independently in 135 sentences with different aktionsarten. In my opinion, regarding the four types of sentences licensed by mei, such an eventuality is also introduced independently. Yet the nature of these predicates allow them to be more natural with the presence of such an eventuality. Let’s first take a look at location verbs. An object can change its location, which renders the predicate temporary. However, even if the context does not allow the object to move, it can still ‘change’ its location according to the change of point of view. Take (117) for example. Imagine that there are three people, A, B and C. They are sitting in a straight row. All of them are facing south. If the observer is seeing the three people from the south, B is between A and C. If the observer is looking from the east side of the three people, then B is right behind C. The relative position of the observer determines the relative locations of A, B and C. This the allocentric frame of reference based on the perspective of the observer. (117) A B C East South Another possible source for temporal boundedness of location verbs is the change in the duration of certain perception. For example, in the example (113) above, the speaker saw that zinc was not on the first row of the periodic table. If the speaker is only referring to what she has been observing during a specific time duration, it is felicitous to use mei. It does not mean, however, that the fact is viewed as true only during that time. The third category, verbs of property description, could also vary with respect to the speaker’s stance. Take (33) and (34) for example, repeated here as (118) and (119). (118) Afu mei(you) xiang Lisi Afu NEG resemble Lisi ‘Afu didn’t resemble Lisi.’ 136 (119) Afu gen Lisi mei(you) xianssi. Afu and Lisi NEG resemble ‘Afu didn’t resemble Lisi.’ Whether Afu resembles Lisi can vary with respect to time and context, as well as the sentence experiencer’s subjective perspectives. Different people might have different opinions towards whether Afu and Lisi are alike. Similarly, context might play a role in determining both the dimension and the standard of resemblance. A speaker might think that Afu resembles Lisi because they both have big eyes and thick brows, while another speaker might not agree, since Afu might be chubby while Lisi slender. The change in the criteria for evaluating the properties may vary with respect to time. Another example is the verb chengxian ‘show/presented as’, which is about how the object is viewed from an outside perspective. It could also vary with respect to time and context. Finally, verbs of measurements are the most problematic category to be interpreted as stance sensitive. An measurement of an object’s weight, depth, and length are not easily changeable. If the measure units also stay constant, it seems that as long as an object doesn’t change its intrinsic properties, and the criteria of measurement is not changed, the property measured is not changeable. Yet when mei is used, there is a tendency that the predicate holds only with respect to the time when the measurement is made. There are two possible scenarios for using mei instead of bu. First, the speaker assumes that measurements can be inaccurate, and the standard for accuracy is determined by the context. Some contexts require high accuracy, as in a science lab, but in most cases, the requirement on accuracy is lower. In other words, different measure eventualities can have different results. Even weight, length and depth can vary under different pressure, humidity or outside causes, and under different gravity of each planet. In fact, if the properties are considered constant across worlds and different types of outside conditions, such as mass, it would be infelicitous with mei. As shown in (120)b, if the sentence describes mass, which does not have a unit when expressed in Mandarin Chinese, it is less acceptable. 137 (120) a. Na yeti zhiliang zhong yi. That liquid mass weigh one ‘The mass of that liquid is 1g/cm 3 .’ b. ??Na yeti zhiliang mei zhong yi. That liquid mass NEG weigh one Intended: ‘The mass of that liquid wasn’t 1g/cm 3 .’ Below is a summary of the different interpretations and licensing conditions of mei and bu with the four predicate types in (21). (121) Mei and bu co-occurring with temporal modifiers: *t refers to the time bounded by the temporal modifier or some covert time. Location verbs mei Stance: It is anchored at certain spatial-temporal location. ILPs or SLPs: Both possible bu No salient stance anchored at t: The subject does not belong to the category denoted by the predicate throughout the time t. ILPs or SLPs: SLPs are more natural Property description mei Stance: it is relativized to a standard of resemblance, which is determined by the context and subjective experiences. ILPs or SLPs: SLPs are more natural bu No salient stance anchored at t: The subject does not belong to the category denoted by the predicate throughout the time t. ILPs or SLPs: SLPs are more natural Measurement verbs mei Stance: it is relativized to a standard of resemblance, which is determined by the context at a certain time and location determined by the speaker. bu No salient stance anchored at t: The subject does not belong to the category denoted by the predicate throughout time t. ILPs or SLPs: SLPs are more natural Judgment verbs mei Stance: it is relativized to the speaker’s certainty or doxastic state at a certain time. bu No salient stance anchored at t: The subject does not belong to the category denoted by the predicate throughout t. ILPs or SLPs: -SLPs are more natural than ILP. -Attitudes less changeable are not good: *xiangxin ‘believe’ *zhidao ‘know’ -Attitudes encode extra meanings, such as certainty or the means of obtaining the evidence are more acceptable: OK quexin ‘firmly believe’, rending ‘firmly believe’, ganjue ‘feel’ 138 In short, from the data given above, we could see that the licensing condition of mei is related to a salient time and a stance, which could well be represented by an event of evaluation. When a temporal modifier is present, the salient time coincides with the time bounded by that modifier. This salient time is further codependent with the time a certain stance is taken by the speaker. The negation marker bu does not introduce a salient time for the perspective to hold. Rather, it only states that the property or the eventuality holds as a general case throughout the entire time denoted by the temporal modifier. Therefore, we do see the effect that these predicates under bu as interpreted as SLPs. 5.5 Other supports for the existence of a stance I turn to some other tests for the temporal boundedness and the existence of a stance related to the four types of predicates, which include being the complement position of perception report verbs, and the modification of manner adverbs. 5.5.1 The complement positions of perception reports A standard test for temporal bounded predicates, the embedded position of perception verbs. The four types of verbs can appear in the complement of perception verbs, which implies that they are temporally bounded eventualities. Generally speaking, the compliment of a perceptual report verb depicts an eventuality that occurs simultaneously with the perceptual eventuality in the matrix clause. This is shown in the English sentence below. In (122)a, the speaker sees the complete event of drawing a circle, and in (122)b, there is a stage of the entire drawing event that is being perceived by the speaker. There are certain aspectual distinctions between the two. (122) a. I saw John draw a circle. b. I saw John drawing a circle. 139 Mandarin perceptual reports include those like kan-dao/kan-jian ‘see’, ting-dao/ting-jian ‘hear’, wen- dao ‘smell.’ These verbs do not select ILPs or permanent situations. An example is given in (123). ILPs under Mandarin perceptual reports (123) a. #Afu kan-jian/kan-dao Lisi shi Meiguoren. Afu see Lisi COP American Intended: ‘Afu saw that Lisi was American.’ b. #Afu tin-jian Lisi shi dasangmen. Afu hear Lisi COP loud person Intended: ‘Afu hear that Lisi was a loud-speaking person.’ Jiao and dao are both aspectual markers that implies the predicates to be telic. As most Mandarin verb stems, these perceptual verbs need to co-occur with aspectual markers. The two markers, jian and dao add subtle meanings to the sentences. Jian, literally translated as ‘to meet’, implies that the agent is taking an active role in perceiving, rather than taking a passive stance of receiving sensory stimuli from the outside world. It also renders the event punctual. As for dao, it is regarded as a ‘high transitivity marker’ (Dao & Chen 2014), which make an emphasis that there is a heightened sense from the perceiver. It therefore is used as an emphasis that the speaker indeed notices the occurrence of the situation. These aspectual particles intensify the meaning that what is being perceived is only a stage of the object’s state or the event he is involved in. The four types of stative verbs are all compatible with perception verbs. (124) Afu kan-jian Diguodaxia weiyu yan-qian. Afu see Empire State Building located eye-front ‘Afu saw The Empire State Building was in front of him.’ (Location verbs) 140 (125) Afu ting-jian Lisi rending ta shi jiandie. Afu hear Lisi firmly believe he COP spy ‘Afu heard that Lisi firmly believed that he was the spy.’ (Judgment verbs) (126) Afu kan-jian Lisi zhong liang-bai bang. Afu see Lisi weigh two-hundred pounds ‘Afu saw that Lisi was two hundred pounds.’ (verbs of measurement) (127) Afu ka-jian Lisi xiang Ali. Afu see Lisi resemble Ali ‘Afu saw that Lisi resembled Ali.’ (verbs of property description) Note that the perspective of the embedded propositions often conform with the one taken by the subject of the perceptual verbs. For example, location verb in (125) describes a relative location from the perspective of Afu. In the case of property descriptions in (127), the evaluator on whether Lisi and Ali are alike is also the matrix subject Afu as well. This uniformity of the perspectives is reminiscent to the embedding of PPTs under perception verbs. 5.5.2 Manner adverbs Manner adverbs could be used as a test for the temporal and spatial boundedness of the eventuality denoted by the predicate. Specifically, there are some manner adverbs that could appear with statives or copular sentences. They depicts the manner of an entire eventuality or a certain stage of an eventuality. Therefore, they become a good indicator of whether an eventuality is itself temporally bounded, which could indirectly 141 imply the existence of a certain salient stage that could be bounded by the modifier. The four types of verbs can be modified by a subset of manner adverbs, especially those that imply spatial or temporal locations, and they can also be modified by those that depicts the manners being perceived by the speaker. Specifically, they are more compatible with speaker oriented manner adverbs, which also suggest that they are prone to have an evaluative reading. Yet some may argue that manner adverbs are indicators for eventive, or non-stative predicates. This is what is proposed by Maienborn (2008), despite some superficial counterexamples in the literature, as shown below. (128) a. John was a Catholic with great passion in his youth. (Jäger 2001) b. The board is coarsely grooved. (Parsons 2000) Maienborn argues that these examples are not real counterexamples. Her explanation is that these manner adverbs coerce an eventive reading (as opposed to abstract K-state reading in statives and copular sentences) from the states as denoted by the predicates (see also Maienborn 2003). For example, (128) can be interpreted as the case that there is an event of John being a Catholic at his youth, and it is an event with great passion. She points out that there is a contrast between (128)a and (129). In the latter example, the feature of being a relative does not have an eventive reading. As a result, with great passion does not coerce the K-state into a Davidsonian state. (129) *John was a relative with great passion. However, I do not think that Maienborn’s analysis is really a counterexample for the legitimacy of the manner adverb test. First of all, her data is limited. There are in fact many counterexamples of stative verbs 142 or even copular sentences that could be modified by manner adverbs. According to Ernst (2016), there are at least three major types of modifiers that are capable with modifying copular sentences. (130) Tyep I: Evaluative adverbs They are related to feelings or evaluation induced in the speaker, the agent, or the subject. It has three subtypes. a. Speaker’s feelings: pleasantly archaic, obnoxiously idiosyncratic, tediously slow, mysteriously beautiful b. Agent-oriented: politely attentive, quirkily personal c. Subject-oriented mental adverbs: cheerfully selfish, proudly assertive, happily sore (131) Type II: Adverbs that refer to semantic properties of the event themselves: evenly blue, seamlessly ubiquitous, blotchily white Type III: Independents: They are less tied to the semantics of the adjectival properties, but requires the predicate to be overtly manifested in a prototypical way. Temporal adverbs: consistently attentive, enduringly rich, lastingly penetrating What these modifiers in common is that they would provide certain temporal or even spatial anchor for the eventuality of the statives. They could be anchored temporally to the perspective of the speaker. This feature is especially prominent in speaker-oriented manner modifiers. Take type I adverb for example, as in (132). The modifier tediously, according to Ernst’s classification, represents the speaker’s feeling towards an eventuality. Hence, the situation holds true as long as the speaker holds the opinion. 143 (132) John’s writing speed is tediously slow. Furthermore, type II modifiers are often about how an eventuality is realized in space and time. For example, in (133), evenly describes the spatial distribution of the color blue on the shirt. (133) The shirt is evenly blue. Now let’s turn to the case of Mandarin manner modifiers and stative predicates. It is not surprising that many manner adverbs are not compatible with statives and copula sentences. For example, the copula shi sentence is quite bad to co-occur with the three types of adverbs in (130). The type I adverb with speaker- oriented reading, such as lingrentaoyende ‘obnoxiously’ and the adverb like kaikaixinxinde, which describes the manner of the subject, are both incompatible with copular sentences. (134) *Na-dong fanzi lingrentaoyende shi hongse-de. That-CL house obnoxiously locate red-DE Intended: ‘It is obnoxiously so that the house is red.’ (135) *Afu kaikaixinxinde shi laoshi Afu happily COP teacher Intended: ‘Afu is happy being a teacher.’ In theory, among the adverbs given above, the type of adverb that are quite compatible with statives are perhaps the speaker-oriented ones. The reason is that the speaker’s attitude is usually independent of the actual manner the event or state has taken place. In other words, such adverbs are less linked to the properties of the states or events themselves. Rather, they are emotions or judgments on the probability or degree of expectedness triggered by the events when the speaker perceives it. Agent and subject oriented 144 adverbs, on the contrary, depicts the manner of the eventuality that an agent or a subject possesses during the occurrence of that eventuality. Therefore, the properties described by these adverbs are more likely to affect or actually are a part of the contingent proponent of the event happened. They are less likely to be correlated to perspectives. Aside from the distinction between speaker and agent/subject orientedness, there is a second aspect which is also important for our discussion. It is on whether the event of the adjectival expression can have a pattern related to the subparts in time or space. Such properties are more linked to the physical distribution adverbs in Type II and the temporal adverbs in Type III. They could provide good tests for identifying whether the predicate is temporally or spatially anchored. For example, evenly and spotchily imply that the state or event have locational subparts. Some of the adverbs could also be interpreted as spreading throughout time, such as regularly curvilinear. It could depict several curvilinear eventualities that occurs spatially in a regular or predictable manner, or temporally, if the curvilinear property occurs in uniform intervals. Now I come back to the four types of verbs to see how they are combined with manner adverbs with the two aspects given above. I start with the first aspect, speaker-orientedness versus subject/agent- orientedness of Type III verbs. The first category is the location verbs. They are generally only compatible with speaker-oriented adverbs, as in (136), and are less acceptable with agent/subject oriented adverbs, as in (139). Speaker-oriented adverbs (136) Na-dong fanzi yiwaide weiyu lesechang-pang. That-CL house surprisingly locate landfill-side ‘That house is surprisingly located right beside the landfill.’ 145 (137) Na-ke shu rerenyende weiyu yuanzi zhongyang That-CL tree annoyingly locate garden middle ‘Annoyingly, that tree is located right in the middle of the garden.’ (138) Zhexie lesetong yizhi weizai renxindao nei-ce These trash cans consistently locate sidewalk inner-side ‘These trash cans are consistently located in the inner sides of the sidewalks.’ (Location verbs) Agent/subject-oriented adverbs: 6 (139) ?Maomi baozaode zai wuzi-li. Cat grumpily at house-in Intended: ‘The cat is grumpy while staying at home.’ (140) ?Afu kaikaixinxinde zai jia. Afu happily at home Intended: ‘Afu is staying at home happily.’ (Location verbs) 6 While these two sentences are not completely bad, they get significantly better if the verbs are changed to more dynamic verbs, such as zhang‘ stand’ or tang ‘lie.’ 146 Verbs of property description and verbs of measurement show a similar pattern. Only speaker-oriented adverbs are more acceptable, while agent/subject-oriented adverbs are not. Speaker-oriented adverbs: (141) Afu yiwaide/shengchide xian ta baba. Afu surprisingly/miraculously resemble his father ‘Surprisingly/miraculously, Afu resembles his dad.’ (Verbs of property description) (142) Zhe-bei yiti yiwaide/bukesiyide zhong san gongjin. This-CL liquid surprisingly/unbelievably weight three kilo ‘Surprisingly/unbelievably, this glass of liquid weighs three kilos.’ (Verbs of measurement) Agent/subject-oriented adverbs: (143) #Afu buqingbuyuande/ kaikaixinxinde xian ta baba. Afu unwillingly /happily resemble his father (verbs of property description) (144) #Afu buqingbuyuande/kaikaixinxinde zhong jiushi gongjin. Afu unwillingly/happily weigh ninety kilo (verbs of measurements) Judgment verbs, however, are compatible with both speaker-oriented adverbs and agent/subject-oriented adverbs. 147 Speaker-oriented adverbs: (145) Afu yiwaide/bukesiyide duanding/yiwei Lisi shi Afu surprisingly/unbelievably firmly judge/think falsely Lisi COP xiongshou murderer ‘Surprisingly/unbelievably, Afu {made a firm judgment/think falsely} that Lisi was the murderer.’ (Judgment verbs) Agent/subject-oriented adverbs: (146) Afu kaikiaxinxinde yiwei Lisi shi xiongshou Afu happily judge-falsely Lisi COP murderer Intended: ‘Afu happily made a false judgment that Lisi was the murderer.’ (Judgment verbs) As for the second aspect, the feature of having subparts, there are cases in which temporal or spatial adverbials can co-occur with all these four types of verbs. In (147), the adverb sipingbawende ‘balanced’ is usually used to describe a furniture-like element that has four feet, or could stand on the ground by its four sides. This adverb requires the four corners or four feet to be placed in a balanced and symmetric manner. (147) Zhe youtong sipingbawende zai jiedao pangbian. This postbox balanced at street side Intended: ‘This postbox is situated on the side of the street in a balanced manner.’ (148) Huicheng banbode zai/weiyu kuzi-shang. Dust splotchily at/situated at pants-on Intended: ‘The dust is spread on the pants in a splotchy manner.’ 148 (Location verbs) (149) Afu de biaoqing shibushide xian baba. Afu GEN expression randomly/from time to time resemble dad ‘Afu’s expressions look like his dad’s from time to time.’ (150) Zhe-kuai yu shibushide chengxianchu guanze. This-CL jade random/from time to time show luster Intended: ‘This piece of jade shows luster in a random manner/from time to time.’ (Verbs of property description) (151) Afu wenwende zong yi-bai gongchi This-CL stably weigh one-hundred kilo ‘Afu has been steadlily measured as 100kg.’ (Verbs of measurements) (152) Afu shibushide duanding xiongshou de shengfen. Afu regularly firmly judge murderer GEN identity ‘Afu makes judgment about who the murderers are regularly.’ Scenario: Afu’s job is to look through murder cases and identify the murderers for each of them. (153) Afu shibushide yiwei ziji hen congming Afu from-time-to-time think-falsely himself HEN smart Intended: ‘Afu falsely thinks that he is smart from time to time.’ (Judgment verbs) 149 To sum up, the four types of verbs can all occur with speaker-oriented adverbs, but not necessarily with subject-oriented ones. Only judgment verbs are capable to occur with both subject and agent oriented adverbs. This is not surprising, as one would expect that the subjects with attitude verbs are also agents, and their attitudes would determine the way they make judgments. As for the second aspect, the adverbs of temporal or spatial distribution and subparts, it is not surprising either. These verbs can occupy temporal spaces as well as spatial spaces, and hence they could perform in a certain manner as depicted by the manner modifiers. In the next section, I show that this second aspect is further supported by modification of locative phrases. 5.5.3 Modification of locative phrases An important characteristics that for these four type of verbs is that they can occur with locative modifiers. It is commonly believed that locative modifiers are known to supply a temporal anchor for the event or the state. These predicates are capable of being modified by typical locative phrases. An anchored location may also imply the existence of a certain salient time in which the event occurs. Therefore, they are incompatible with individual level predicates (Kratzer 1995, Chierchia 1995, Fernald 2000, Maienborn 2001). Locative phrases oftentimes seem to be a good diagnostic for temporal or spatial boundedness. Here I show that this test could also support my view that the four types of verbs are related to an eventuality that has an independent temporal/ spatial anchor when being negated by mei. However, before adopting this test to Mandarin, there is a need to distinguish between ordinary locative modifiers and frame-setting locative adverbs. Frame-setting locative adverbs provides a spatial-temporal frame for the situation to hold without restricting the time of the event itself. They indicate the broader background in which the event occurs. The cases with frame-setting modifiers are given in (154) (Maienborn 2005 (15)), adopted from Fernald (2000: 24). (154) Carol was tired/hungry/nervous in the car. 150 Her argument is that the locative in the car is not an ordinary locative expression that describes where the event occurs. Rather, it is a frame-setting adverbs that are generally correlated with a discourse topic that sets the whole setting of the eventuality. There are other two types of locative expressions, the external modifiers and internal modifiers, as shown in (155)a and (155)b respectively. Both of them are related to the occurrence of the event itself. The external locative modifier often denotes the location in which the eventuality happens. The internal locative modifier denotes the theme, instruments or goal related to that event. They are both related to the forms in which the eventuality happens. Therefore, they are event-related. The locative expression in Argentina in (155)a serves as a topic that address the location where the entire situation described by the sentence comes about. The external reading of in Argentina is related to where the event occurs, as opposed to the internal reading of on the last page, which is the theme or goal for the event of signing. (155) a. In Argentina, Eva still is very popular. (Frame-setting modifier) b. Eva signed the contract in Argentina. (External modifier) c. Eva signed the contract on the last page. (Internal modifier) Maienborn proposes that frame-setting modifiers are basically situated in C, restricting the topic of the sentence, which could either be a discourse topic or a sentential topic. When it is a discourse topic, it provides a locative constraint on a semantically underspecified referent embodied in the discourse topic (Maienborn 2001: 232). When it is a sentential topic, it serves the function of singling out a topic time, which is similar the topic time in the literature on aspects, following Klein’s view (Klein 1994). What we are concerned here is of course the event related modifiers. Tests related to frame-setting modifiers and external modifiers can be used to distinguish between the two. However, as frame-setting adverbs and external adverbs are not necessarily distinguished with respect to their overt sentential positions, we need to find tests for distinguishing the two. There are two major differences between frame- setting modifiers and event-related external modifiers. First, frame-setting modifiers have a different 151 entailment relation. If we compare the case of external modifiers, a sentence with a frame-setting modifier does not necessarily entail the counterpart in which the modifier is absent. Frame-setting modifiers (156) In Argentina, Eva still is popular ↛ Eva is popular. (Maienborn 2001: 194) However, an external and internal modifier do entail the propositions without the modifiers. (157) Eva signed the contract in Argentina → Eva signed the contract. (External modifiers) (158) Eva signed the contract on the last page. → Eva signed the contract. (Internal modifiers) The second feature of frame-setting modifiers is that it can occur with ILPs like English cost and resemble. Maienborn’s explanation, following Kratzer (1995), is that ILPs do not have an event argument. Therefore, the modifier could not possibly be about the location where the event occurs. A frame-setting modifier is therefore not event related. It does not denote the property of that eventuality itself. Instead, it provides an underspecified domain restriction for the overall proposition. The domain is quite flexible, which allow a frame-setting locative modifier to be interpreted in different ways, such as a certain context, a location or a time. For example, in (156), in Argentina could imply the time the subject was in Argentina or the location Argentina itself. It could also refers to the standard of popularity set in Argentina. Now we turn to external modifiers. External modifiers are about the location where the eventuality occurs. It denotes a property about the eventuality itself, rather than a frame in which the event occurs. In Maienborn’s theory, statives and copular sentences cannot appear with external modifiers. 152 (159) *Carol war im Auto blonde/intelligent. Carol was in car blonde/intelligent (Maienborn 2005: 288 ) Under Maienborn’s system, the locative data serves as a piece of evidence that suggest that there is no event argument in these stative sentences. I will not adopt her view here. Instead, I will follow the general view that there is an eventuality, a state argument to be specific, even in non-dynamic sentences. One of the analyses that retains the presence of eventuality in copular sentences is proposed by Ernst (2015), who follows Chierchia’s view (Chierchia 1995). He proposes that copular sentences, which denote states, are eventualities that have a stable property across worlds. In other words, when an external modifier is used, it points out specifically which location it occurs at. Therefore, the location of the state is not a salient feature. Following general pragmatic principles, a sentence that is compatible with an external locative expression would trigger the implicature that this eventuality does hold in other possible locations. However, since states are assumed to be constant across locations, the presence of an external modifier would restrict the state to be true only of that location, which is the opposite of what states are characteristic of. In sum, locative modification can be distinguished between frame-setting modifiers, external modifiers and internal modifiers. For any given locative modifier, it could potentially be interpreted as either one of them. In the case of English, when it occurs in copular sentences, it is more likely to be interpreted as a frame-setting modifier or an internal modifiers, not an external modifier. On the contrary, for a given type of predicate, if a modifier could be interpreted as an external modifier freely, we could conclude that the predicate is more likely to have a temporally bounded component. I will use the test to see if the four types of verbs can all occur with external locative modifiers. But before applying the test, we need to see how Mandarin locative modifiers are interpreted in different sentential positions. 153 5.5.4 Locative modification in Mandarin Chinese Here I show that the four types of verbs are more compatible with event-related locative modifiers in languages like English. Mandarin Chinese also contrasts between external modifiers and frame-setting adverbs. They can be distinguished by the positions in a sentence. Locative expressions in Mandarin Chinese can be placed in at least two places, sentence initial and post-subject (Tai 1975, Chao 1968). (160) Zai chufang-li, ta zai ku. At kitchen-in he PROG cry ‘In the kitchen, he is crying.’ (Sentence initial locative modifier) (161) Ta zai chufang-li ku. He at kitchen-in cry ‘He cried/is crying in the kitchen.’ (Post-subject locative modifier) (Tai 1975: 155) Intuitively speaking, locative adverbs in the sentence initial position is often interpreted as frame- setting adverbs, while those in the post-subject position are interpreted as both frame-setting and external modifiers. In copular sentences, only sentence initial adverbs, which are interpreted as frame-setting adverbs, are more compatible. In eventive predicates, locative adverbs can occur in both positions, meaning that both interpretations are acceptable. On the contrary, since sentence initial locative modifiers are interpreted as frame-setting, they could appear with both states and dynamic predicates. We first look at sentences with sentence initial locative modifiers in terms of the entailment relations. When the truth condition of the predicate that are context-sensitive, such as gradable adjectives, the 154 sentence does not entail the meaning without the locative modifier. It is because the standard is also restricted by the frame-setting modifier, which also determines the truth condition of the sentence. Gradable adjectives: (162) Zai Meiguo, Afu hen ai. ↛ Afu hen ai In The United States Afu HEN short Afu HEN shore ‘In the United States, Afu is short.’ ↛ ‘Afu is short.’ (Sentence-initial modifier ) However, this effect disappears when the predicate is not context sensitive. This is shown in the example of ordinary eventive predicates. The sentence could only denote the location where the event happens. In this case, the sentence entails that Afu ate the bison meat hamburger. Yet zai meiguo in (163) is still interpreted as frame setting, since it restricts the time in which the eating happens: it was during Afu’s visit in the US that he ate bison meat hamburger. Eventive predicate: (163) Zai Meiguo, Afu chi-le yi-ge yeniu-rou hanbao. In the United States Afu eat-PERF one-CL bison-meat hamburger ‘In the United States, Afu had a bison meat hamburger.’ ⟶ Afu chi-le yi-ge yeniu-rou hanbao. ‘Afu had a bison meat hamburger.’ (Sentence initial modifier) Furthermore, sentence initial modifiers can appear with characterizing copula sentences. There is an implication that it encodes a topic a time during which the subject is in that location. In other words, it can 155 be translated as ‘during Afu’s time in the United States, he was a teacher.’ It does not imply that he is a teacher only in the United States because he fits certain criteria for a teacher there. Characterizing copula sentences: (164) Zai Meiguo, Afu shi laoshi. In the United States Afu COP teacher ‘Afu is a teacher in the United States.’ =Afu is a teacher during his time in the United States. ≠Only when Afu was in the United States did he fit the criteria of being a teacher. (Sentence-initial modifier) Now let’s turn to post-subject modifiers. When such a modifier appears with gradable adjectives, it does not provides a topic situation that sets the standard. Therefore, it entail the sentence without the modifier. In (165), Afu is only short according to the standard set in the United States. Rather, zai Meiguo cannot be interpreted as an external modifier in which the eventuality happens. (165) Afu zai Meiguo hen ai. ⟶ Afu hen ai. Afu at The United States HEN short =’Afu was short by the standard of The United States.’ ≠′Only when Afu was in The United States was he short.’ The sentences shown above lead to two conclusions. First, the Mandarin sentence-initial locative modifier has the effect of restricting the state as unique to a certain situation introduced by the modifier, similar to the frame-setting adverbs mentioned by Maienborn. This situation could often be a temporal restriction, but it could also be a topicalized location where the situation happens, which subsequently 156 restricts the criteria specific to the location in which the situation holds. Second, the post-subject modifier, however, merely tells that the situation holds in that location when the predicate is a copular predicate. On the contrary, when the main predicate is an eventive one, the post-subject modifier could be interpreted as an external modifier. (166) Afu zai Meiguo chi-le yi-ge yeniu-rou hanbao. Afu at The US eat-PERF one-CL bison-meat hamburger ‘Afu ate an bison hamburger in The US.’ The generalizations are given in (167). (167) Restrictions on locative modifiers Sentence initial modifier: a. It is similar to a frame-setting modifier in English and German. b. It restricts the location, time or context for the states to hold. Post-subject modifier: a. It could be interpreted as both frame-setting and external modifiers. b.When it appears with dynamic predicates, it could be interpreted as external modifiers. With the contrast in mind, I now look at the four types of verbs. Both sentence-initial and post- subject modifiers are acceptable. When they occur with post-subject modifiers, the modifier could be interpreted as external. This feature shows that they can all be anchored in time or space. (168) a. Zai Meiguo, zhe-dong lou weizai diwudadao In The United States this-CL building located Fifth Avenue. ‘In The United States, this building is located on the Fifth Avenue.’ 157 b. Zhe-dong lou zai Meigue weizai diwudadao. This-CL building at The United States situated Fifth Avenue ‘This building is situated on the Fifth Avenue in the United States.’ Or ‘In The United States, this building is located on the Fifth Avenue.’ (169) a. Zai Meiguo, Afu duanding/rending At the United States Afu judge firmly/firmly believe Lisi shi fanren. Lisi COP suspect ‘When Afu was in the United States, he made a judgment/firmly believed that Lisi was the suspect.’ b. Afu zai Meiguo duanding/rending Lisi shi fangren. Afu at the United States judge firmly/firmly believe Lisi COP suspect ‘Afu firmly made a judgment/firmly believed that Lisi was the suspect in the United States.’ (judgment verbs) (170) a. Zai Meiguo, Afu xiang Lisi. 7 At the United States Afu resemble Lisi ‘Afu resembled Lisi by the standard in The United States.’ b. ?Afu zai Meiguo xiang Lisi. Afu at the United States resemble Lisi. ‘Afu resembled Lisi when he was in The United States.’ 7 The interpretations in the two sentences in (170) emerge under the context that the looks of Afu and Lisi do not change with respect to time or location. It is the standard that changes. 158 (Verbs of property descriptions) (171) a. Zai Meiguo, Afu zhong wushi gongjin. In the United States Afu weigh fifty kilos ‘During the time when Afu was in The United States, Afu weighed fifty kilos. (172) b.Afu zai Meiguo zhong wushi gongjin. Afu at the United States weigh fifty kilo ‘He weighed fifty kilos when he was in The United States.’ (measurement verbs) The data shows that all the four types of verbs are acceptable with post-subject modifier. Property description verbs are the only exception. It seems that these verbs are more acceptable when they are with sentence-initial modifiers. The reason is probably because the verbs that describe resemblances are gradable. A topic situation provided by the sentence initial adverb can provide the standard and criteria for determining the meaning of the sentence. On the contrary, (170)b is less acceptable because looks does not change easily. Furthermore, the post-subject modifier only indicates where the state or the event holds. It does not imply any standard for similarities. But it does not mean that post-subject adverbs are not totally unacceptable. Under the scenario in which the look of Afu and Lisi can indeed change, the sentence of (170)b is more acceptable. Therefore, I put one ‘?’ before the example. However, in this case, resemblance is no longer a stative feature that does not change with time easily. Therefore, under this interpretation, it lead to the contrast of the entailment relation between the two below. 159 (173) Afu zai Meiguo xiang Lisi. → Afu xiang Lisi. Afu at the United States resemble Lisi. Afu resemble Lisi ‘Afu resembles Lisi when he was in the United States.’ → Afu resembled Lisi. (174) Zai Meiguo, Afu xiang Lisi. ↛ Afu xiang Lisi At the United States Afu resemble Lisi Afu resemble Lisi ‘In the United States, Afu resembled Lisi.’ ↛ Afu resembled Lisi.’ Scenario for (173): Afu or Lisi’s looks can change when they are in the United States that renders them looking similar only when they are only in that location. Therefore, the first sentence in (173) is interpreted as Afu resembles Lisi when they are in the United States. In (174), the locative modifier is not truth preserving. It is because the locative modifier zai Meiguo ‘in the United states’ provides a situation that could determine the standard for what counts as similar in appearances. Therefore, when the modifier is absent, the standard and the criteria is obtained elsewhere in the context. The second sentence Afu xiang Lisi is true of a standard different from that assigned by locative modifier. In conclusion, there is a consistent pattern that the four types of verbs are more compatible with the post-subject modifiers. Locative modifiers in this position restricts the time during which the sentence is true. Therefore, although some of these Mandarin adverbs have counterparts as individual level predicates, as in the case of resemble, they are more tolerant to temporal and spatial restrictions. This observation is consistent with the previous result with the temporal modifier test, both showing that the expressions licensed under mei tend to be temporally or spatially bounded than ordinary copular sentences. 5.6 A generalization The discussion above points out several properties of these four types of verbs. (a) they do not pattern with ordinary dynamic predicates, including dynamic states. For example, they can be modified by manner and 160 locative modifiers, showing that they are quite natural to be anchored to certain time or spatial locations. (b) They can be perceived and hence can be embedded under perceptual verbs. (c) They can all be modified by temporal modifiers. This property indicates that they naturally possess temporally bounded stages, which set them apart from individual-level predicates. In other words, it is not difficult for these verbs to come up with a context that licenses the situation that holds with respect to a bounded temporal stage only. Yet it does not mean that the sentences with temporary properties are SLPs. My claim is weaker than that. Being an SLP implies that the properties hold only temporarily and the termination of the state is naturally conceived. In this respect, the four types of verbs can be ILPs. Personal judgments can change with respect to time, but can also last during the entire time of the existence of the attitude holder if he does not change his perception about the world. Similar cases apply to verbs denoting resemblance and measurements. If the properties that the subjects possess are not temporally bounded, the source of temporality should come from elsewhere. There are several possible candidates that show the possible sources of temporal boundedness, as generalized in (175). (175) Possible types of temporal bounded stages Location verbs -Different perceptual stages: Ex. Seeing/knowing something being in the location -Different perspectives: The same object can be viewed from different relative locations Property description -The standard of being similar/dissimilar can change with respect to context. Judgment verbs -Change of perspectives Measurement verbs -Measurements as evaluation that changes with perspectives Regarding locative verbs and property description verbs, it is the contextually determined perspectives that can change with respect to time. In the case of judgment verbs, it is the simple change of perspectives. As for verbs of measurement, it could be viewed as restricted by the time where a perspective holds as well. Verbs of measurement is the most difficult one to pinpoint. I do not have a good explanation for it right now. I leave this for future research. 161 5.7 A look at hen sentences We now turn to hen sentences. It expresses a stative predicate. Yet it shows certain features of temporal or spatial boundedness presented in the other four types of verbs as well. However, different from these predicates, hen does not allow manner modification, as shown in (176). Manner modification and hen: (176) *Ta de sudu wenwende hen kuai. He GEN speed stably HEN fast ‘His speed is certified to be stably fast.’ The reason for the ungrammaticality of the above example could be due to the fact that what is being described by the manner adverb is about an internally changeable state or a stage that intrinsically has internal stages. The gradable property of hen are still homogeneous states. Yet there are other types of evidence, especially from temporal modifiers. For example, yixiazi…yixiazi ‘suddenly… suddenly’ is used for describing the situation that has two repeating events that are juxtaposed together. Yixiazi... yixiazi licenses the use of hen form even when the predicate denotes an individual-level one. On the contrary, the shi….-de form is infelicitous. In (177), assume that the standard degree for the size of the house is prevalent only to one stage of the house. (177) Afu fanzi yixiazi hen da yixiazi hen xiao. Afu house suddenly HEN big suddenly HEN small ‘Afu’s house is certified to be big at one moment and small at another.’ In (177), a natural scenario to account for the change of standards could be the change of personal perspectives. At one moment the speaker consider the house big, while at another he considers it small. 162 In short, hen sets the standard that could be true only of a temporally bounded stage. This does not mean that hen sentences are always SLPs. The temporal boundedness is related to the time duration of the evaluation over the situation. The evaluation most naturally comes from the speaker. 5.8 Mei and the specificity restriction on a discourse situation 5.8.1 A comparison with estar in Spanish In a sense, we could view mei as introducing a specific discourse situation, which could again fit into the generalization that there is an event of evaluation that holds during a specific time, and is relativized to the relative location of the speaker. The restriction that the property applies to a discourse situation is reminiscent to the treatment of estar proposed by Maienbourn (2004) and further followed by Deo et. al (Deo et. al 2016). In traditional literature on estar, this copula is viewed as one that takes a temporally bounded predicate. Contrary to this analysis, according to Deo et. al (2016), estar sentences have several different uses, such as the change from previous knowledge to new knowledge, unexpectedness, to express the exceeding of contextual expectations, and the change of standard when the asserted proposition is a gradable adjective. In general, the asserted proposition in an estar sentence is related to a specific situation or circumstance that is being compared with other realistic or hypothetical situations. In other words, estar introduces some sort of context restriction that is otherwise missing in ser sentences. According to Deo et. al., this can be modeled by making reference to the truth values of estar sentences across possible circumstances of evaluation. Here are some of the uses of estar that shows this feature. In terms of knowledge change, estar is used when the speaker has a different understanding prior to the situation. In contrast, es here refers to the property that Juan has always possess. 163 Knowledge change: (178) Context 1: The speaker has prior knowledge that Juan was not fat. a. Juan está gorgo. Juan ESTAR.PRES.3SG fat ‘Juan looks fat.’ b. Juan es gorgo. Juan is fat ‘Juan is fat.’ (Deo et. al 2018: 11; Rodán 1974: 72) The relational use of gradable adjectives refer to adjectival expressions that is interpreted with respect to a degree dependent on the denotation of a nominal expression. This nominal expression is either saliently given in the context, or overtly given in the sentence. For example, ‘the curtain is big for the bathroom’ is about a property that is true of a degree when we consider the condition of the bathroom in question. In such uses, when estar is used, it is related to a nominal expression that is linked to a salient situation. As in the case of (179), the curtain is too big because it is related to size of the shower room. Relational use of gradable adjectives: (179) Context: The speaker is looking for a shower curtain. He gets one and think it will fit. Unfortunately, it is too big for the purpose. La cortina está larga The curtain ESTAR.PRES.3SG large 164 The use in changed contextual standards refers to the inclusion of the subject in the positive polarity of the property denoted by the predicate by lowering a preexisting contextual standard. Changed contextual standards: Context: The speaker used to think that the New York skyscrapers are the standard for tall buildings. But when he is in New Haven, where the buildings are uniformly lower, the speaker lowers his standard by calling a building that is taller than the other a tall building. (180) Ok, ese edificio está alto Ok that building ESTAR.PRES.3SG tall ‘Ok, that building is tall.’ In order to account for the various uses of estar, Deo et. al. argues that an estar sentence generates a set of alternative circumstances, which are taken into consideration by the interlocutors of a conversation. Formally speaking, circumstances are defined as alternative propositions of which the respective truth conditions vary with one specific parameter. For example, when estar is used temporally, it denotes an evaluation over a proposition content that holds within a certain time t as opposed to circumstances holds of different times. Likewise, the alternative circumstances can be about propositions true of different locations, worlds or of different delineations. These parameters are represented as 〈w, t, l, a, c〉, in which l stands for location, a for agents, and c for contextually determined standards. Different circumstances can be compared by ‘strength.’ The strength between two circumstances is defined through entailment relations. It could be a realistic ordering source, as in the case of w, or mereological inclusion, as in the case of t and l. Let’s discuss only the entailment relation and the mereological inclusion cases, which are relevant to the discussion of this chapter. 165 In the case of c, strength of delineation parameter c is determined by entailment relations via the standards assigned, where i is stronger than i’ if and only if the proposition true of i entails that of i’. Between two propositions, the one assigned with a lower standard is stronger than the one with a higher standard. Therefore, c i ≫c i’ if and only if the standard assigned by c i is lower while that by c i’ is higher. Take John is tall for example. Imagine that there are two circumstances. In the first one, the standard for tallness is assigned by comparing the group of basketball players, and in the second one, comparing the group of elementary students. Let c i denote the context that assigns the standard of tallness in ‘John is tall for a basketball player’, and let c i’ denote the standard for ‘John is tall for an elementary student’. Since ‘John is tall’ relative to c i entails ‘John is tall’ relative to c i, c i ≫c i’, we could assume that the c i is at least as strong as c i’. In the case of mereological inclusion, time t i is as strong as t i if and only if t i includes t i’. Similar case applies to the location parameter. A larger location l i is at least as strong as l i’ if and only if the l i contains l i’. According to Deo et. al., a sentence with estar require the asserted circumstance to be bounded. It refers to two restrictions over the set of alternative circumstances: (a) if the asserted proposition is true of i, then there exists some alternative circumstances i’ such that i’ is at least as strong as I, and (b) for all circumstances i”, if p(i”) then i must be at least as strong as i”. In other words, an asserted proposition p in an estar sentence is true with respect to a restricted time, location, world, or delineation. This is the boundedness of estar. Es, on the contrary, is neutral in this respect. By providing a restriction over the alternative circumstances, estar sentences picks out a unique circumstance that the speaker considers salient. 5.8.2 The evaluation eventuality in mei sentences By looking at the Spanish estar and es data, the idea that there is a salient circumstance seems intuitively applicable to the four types of stative predicates under mei. However, there is a major difference between 166 the licensing conditions of mei and estar. In the estar sentences, the major function of using estar instead of es is that the property denoted by the predicate in estar sentences is only a subpart of the circumstances being considered. For example, in the estar sentences involving change of knowledge, there is a distinction between prior knowledge and the knowledge the speaker has at the reference time. In the case of the changing of standard, there is a new standard introduced as opposed to the older one. Such boundedness is not expressed in a mei sentence. The property is not regarded as being contrasted or highlighted as opposed to a set of alternative circumstances. Let’s take a look at the sentence in (181). It is perfectly acceptable if the speaker believes that it is not the case for Afu to be tall for an extended time period. Even if the speaker thinks that the standard of tallness and Afu’s height remain the same, and hence there is no potential change for the property ‘tall’ to apply to Afu, the sentence is still felicitous. (181) Afu mei hen gao. Afu NEG HEN tall ‘Afu is/ was not certified to be tall.’ It becomes more obvious if the predicate is an individual level one. As shown in example (30), repeated here as (182). (182) Diqiu mei weiyu taiyanxi Earth NEG situated at The Solar System ‘The Earth is not situated in the Solar System.’ For the speaker, there is no presupposition that there is at least one circumstance at a different time, world or location such that the Earth is part of the Solar System. It does not presuppose that not being in the Solar System is only under the current circumstance either. 167 Furthermore, a mei sentence does not trigger the implication that the property only holds within a restricted circumstance. Take the time parameter for example. The strength of the circumstances that vary with respect to time are ranked by mereological inclusion. For two time durations t and t’, if t is stronger than t’, then t’⊆t. Assume that the asserted proposition p is true of t, then for all t” such that t”⊂t, p is not true of t”-t. It is because scalar implicature would negate p to be true of any superset of t that does not contain t. Again, a mei sentence does not trigger this implicature. In (183), the speaker states that at the time of yesterday Afu was not at home, but it does not negate Afu to be home throughout a larger time duration that include yesterday. The assertion following the first is all natural. It is not an attempt of cancelling the implicature. (183) Afu zuotien mei zai jia. Ta zhe zhenzi dou bu zaijia Afu yesterday NEG at home he this while all NEG home ‘Afu was not home yesterday. He hasn’t been home recently.’ The above analysis conforms to the view that the proposition licensed by mei introduces a topic situation an eventuality which anchored always to a perspective from the standpoint of the speaker or some salient attitude holder. Here a perspective involves the anchoring of multiple parameters, which requires the fixing of an agent, a time and a location where the evaluation takes place. Such feature can be easily explained if we introduce an eventuality that stands for the event of evaluation, which immediately provides a temporal reference. Furthermore, the anchoring of the location is situated at a reference point that often coincide with the location of the agent. These parameters are fixed by the location of the event. (184) ∃e∃e’[Agent(e’, SPEAKER) ∧ Perceive(e’, e) ∧ Theme(e, Afu) ∧ in-front(e)] 168 The event of perception is an abstract mental act. However, the evaluation is given by an agent. Therefore, the eventuality is indirectly related to the location of the agent. All in all, the evaluation takes place by taking into consideration of the time of the evaluation and the location of the agent, and where the world in which such an evaluation is made. 5.8.3 Evidence from the comparison between mei and bu When we compare sentences negated by mei and bu, we could clearly see the contrast that only mei requires an eventuality of evaluation or a stance, while bu does not. A stance would be relevant to two parameters. There are two pieces of evidence showing this contrast. The first is the perspective of the speaker or other salient individuals, and the second is the temporal duration that the perspective holds. In sentences negated by bu, in contrast to those negated by mei, both parameters are underspecified. Therefore, when mei is used as opposed to bu, it always implies the existence of a specific situation that is salient in the context. This is shown in some of the contrasts between mei and bu. In the case of location verbs, when bu is used, the sentence expresses a point of view that is not anchored to any perspectives. This is shown in the contrast between the two sentences below. (185) Na dong lanse fanzi bu zai diwudadao shang. That CL blue house NEG at Fifth Ave. on ‘The blue house is not on Fifth Avenue.’ (186) Na dong lanse fanzi mei zai diwudadao shang That CL blue house NEG at Fifth Ave. on ‘There was a salient situation of which the blue house was not on the Fifth Avenue’ 169 In (185), when bu is used, the location of the blue house does not change with perspectives. However, in (186), when mei is used, it implies that the house not being on Fifth Avenue is only restricted to certain circumstances. Therefore, when a sentence is uttered out of the blue, as in (187), the use of mei is bad. Out of the blue context: (187) Diguodaxia *mei/bu zai bailaohui shang. Empire-state-builiding NEG/NEG at Broadway on. ‘The Empire State Building was not on Broadway.’ The oddness comes from the world knowledge that the location of Empire State Building has been at the same spot, and it still exists. On the contrary, when the context provides a temporal interval that restricts the state to hold only within that interval, as Afu zai niuyue de shihou ‘the time when Afu is in New York’, interestingly, bu becomes less acceptable, and mei becomes more acceptable. (188) Afu zai Niuyue de shihou, Diguodaxia ?bu/mei Afu at New York DE time Empire State Building NEG zai Bailaohui shang. at Broadway on ‘When Afu was in New York, the Empire State Building was not on Broadway.’ The contrast between mei and bu, again, shows that the existence of perspectives is crucial in licensing a mei sentence. Another aspect that shows the contrast between mei and bu is the co-occurrence with the experiential perfect marker, guo, it licenses stative predicates, and when it appears with mei, it does not sound totally acceptable. Guo is licensed when the eventuality has occurred in the past (see discussions from Chao 1968, Smith 1991, Iljic 1990, Lin 1979, Ma 1977). Guo can appear not only with non-homogeneous eventuality, 170 but also with states or activities. It is licensed when the situation has occurred in the past and no longer holds at the reference time. (189) Afu pao guo paodao. Afu run Exp. PERF track ‘Afu have had the experience of running on the tracks.’ Guo is licensed only under mei, but not bu. Bu zai jia ‘not at home’ here refers to a state that does not entail an endpoint, and the sentence becoms infelicitous. (190) Afu mei zai jia guo. Afu NEG at home Exp. PERF ‘Afu has not been at home.’ (191) *Afu bu zai jia guo. Afu NEG at home Exp. PERF There are many different accounts on the more specific licensing condition of this particle. Yeh (1996) argues that it has three properties: First, it need to have at least one occurrence in the past. Second, it is only allowed with events that can reoccur. Third, it allows a ‘class’ meaning, which means it implies the existence of a kind of events to have happened or potentially to have happened in the past. Guo involves discontinuity. According to Wu (2008), This particle has only one restriction, the terminability constraint. It requires that either the activity or the state has terminated, or if there is a result state, it ends before the reference time. An example that shows this requirement comes from the verb si ‘die.’ It is unacceptable to use si with guo, since the result state of si would be a unless it is uttered in a world that allow people to resurrect from death repeatedly. 171 (192) *Afu si guo. Afu dead Exp. PERF Intended: ‘Afu died (before).’ Here I adopt the view that guo requires the situation to be terminable. However, I do not make any conclusion whether it is the sole requirement, as proposed by Wu (2008). I also think that terminability is the only constraint that is certain. I assume that the idea of terminability is related to a temporal interval before the reference time, in which the situation can only hold in that interval, but not later. The contrast could be interpreted as such. In (192), the sentence being negated is ‘Afu has been at home’, and when it is marked by guo, it denotes a situation that is terminable before the reference time, and mei negates that such event occurred during that time. On the contrary, I assume that bu does not negate telic eventualities unless they are habitual or generic. Therefore, it clashes with guo. Second, guo licenses out of blue context like (187), as shown in (193). When guo is added, it yields the reading that there is a salient time before the reference that The Empire State Building was never on Broadway during that time. (193) Diguodaxia mei zai Bailaohui shang guo. Empire-state-builiding NEG at Broadway on Exp. PERF. ‘The Empire State Building has never been on Broadway.’ Guo is licensed in a hen sentence, as in (194). (194) Afu hen renjen guo. Afu HEN hard-working Exp. PERF ‘There has been a time when Afu was hardworking.’ 172 In contrast, the shi…-de form, which is used to describe a defining adjectival property, cannot occur with guo. (195) *Afu shi gao-de guo. Afu COP tall-DE Exp. PERF. Intended: ‘There has been a time in which Afu belonged to the category of tall people.’ We could see guo as introducing a bounded temporal interval in which the eventuality of evaluation could occur. Below shows that the time t as introduced by guo is before the reference time t @, and the evaluation eventuality e is contained within time t. There could be multiple occurrences of evaluation eventuality during t, but each of them is bounded, given that t is also bounded. (196) ∃t∃e[percieve(e) ∧ 𝜏(e)⊆t ∧ t≤t @] The data of guo again shows that mei introduces a bounded situation. 5.8.4 Generic sentences and habitual adverbs The last piece of evidence that mei introduces a specific situation is the eventuality of evaluation comes from generic and habitual sentences. Early on in this chapter, I mentioned that bu is interpreted as a negation marker that expresses a generic statement, especially when the predicate is dynamic. But mei can appear with generic and habitual sentences when there are certain habitual modifiers in the sentence, as shown in (197). Since habitual and generic sentences are not episodic, this seems to contradict with what I’m proposing here about mei. 173 When mei occurs with habitual or generic modifiers, it yields a specific restriction over eventualities quantified by the habitual adverb. It restricts the times in which the eventualities adverbs occurs. The habitual adverb quantifies a set of situations that are distributed in time in a somewhat even manner, as in the case of changchang, and in a somewhat dense manner, as in the case of zongshi, across this specific time interval. (197) Afu mei changchang/zongshi zai jia. Afu NEG often /always at home ‘Afu is not often at home/Afu is not always at home.’ In the literature, habitual sentences are a subclass of generic sentences. They usually appear with episodic verbs, and marks the general situation over sets of episodes (see Rooth 1995; de Swart 1991; Rimell 2004; Banvise 1978; Barwise and Perry 1983; Portner 1992). Here I will simply treat situations as eventualities. One standard view on habitual sentence is illustrated in (198). Here OFTEN quantifies over the set of salient events. C is a contextual restriction on appropriate eventualities. Among them, the majority are those that Mary eats roast beef. (198) Mary often eats roast beef sandwiches. OFTEN e[C(e)][Mary eats roast beef in e] (Revised from Rimell 2004: 3) When mei is used, as in (197), it implies that the generic situation only lasts for a certain duration. Therefore, it is felicitous if the sentence is followed by a sentence that the general situation does not hold. In (199), it implies that there is a specific time t 1, such that for all contextually salient situations contained 174 in t 1, there is no event of Afu being at home, and a time t 2 after t 1, there is (at least) one event of Afu being at home. Houlai anchors the time of the second sentence with respect to the specific time t 1. (199) Afu mei (you) changchang/zongshi zai jia. Dan houlai you. Afu NEG COP often/always at home but later COP ‘Afu was not always at home / Afu was not at home in general (at that time), but later he was (often at home).’ In contrast, when the negation is bu, it is less felicitous with that same following sentence. This is because bu does not introduce a bounded salient time. It is licensed when the embedded clause is either an atelic eventuality, or a habitual predicate. Therefore, the second sentence sounds off, since the expression houlai has no specific time to take as reference. (200) #Afu bu changchang/zongshi zai jia. Dan houlai hui. Afu NEG often/always at home but later would Intended: ‘Afu was not always home/Afu was not at home in general (at that time), but later he would (be often at home).’ The second piece of evidence comes from the relative position of habitual adverbs. When the habitual adverb takes scope over negation, as in (201), the sentence is unacceptable. (201) ??Afu changchang/zongshi mei zai jia. Afu often/always NEG at home Intended: It is always the case that Afu was not hungry (at a given time/ at several given times). 175 If one assume that mei introduces one contextually restricted salient time, there would be a conflict between introducing a set of eventualities quantified by habitual adverbs, and hence the unacceptability of (201). There is a possible interpretation to avoid this clash, however, is for the sentence to have a distributed reading over eventualities by adding a morpheme dou ‘all.’ It is used to provide a quantification over multiple objects in a given set. The function of dou is that it indicates the distribution of the predicate over a plural expression preceding it (Xiang 2008). An example is given below. (202) Afu zhe ji tian dou zai jia. Afu these couple day all at home ‘Among these couple of days, Afu is home in all of them.’ In the case of a habitual sentence, dou provides a maximal quantification reading, similar to all, over the set of eventualities quantified by the habitual adverbs. Take (203) for example. (203) Afu zongshi dou zai jia. Afu always all at home ‘Afu is always at home.’ When dou is used with habitual adverb and mei, it is to assume that there are multiple events quantified by changchang/zongshi. For each of these events, there is one specific time t that Afu is not home. This reading is in fact acceptable. (204) Afu changchang/zongshi dou mei zai jia. Afu often/always all NEG at home ‘Afu is always/often not at home.’ Implications: There are multiple events, and in each event, there is a time that Afu is not home. 176 In short, the above data shows that mei indeed would introduces a salient time interval in which the events quantified by the habitual adverbs fall within that time. 5.9 The scopal property of the evaluation eventuality In this section, I discuss the scopal property of the eventuality introduced by hen. It is situated at a rather high position in a sentence. It takes scope over mei. Take (205) for example. A sentence like (205) is true if there exist an evaluation such that Afu is not in front of the agent of the evaluation eventuality, not that there is no such an evaluation. This shows that mei does not negate the existence of the evaluation eventuality. (205) Afu mei zai qianmian. Afu NEG in front ‘Afu is not in the front.’ Further evidence comes from sentences with temporal and locative modifiers at the sentence initial position. The location or the time of this eventuality could be anchored to the time denoted by adverbs located at this position. This temporal adverb marks the time where Afu was observed as not being in the front. Yet during that observation, Afu is indeed not in the front. Therefore, the two times overlap. Imagine that the time t 1 refers to na shihou ‘that time’, and t 2 refers to the time when Afu is not in the front. The relation between t 1 and t 2 is that t 2 may include t 1, or they may coincide. (206) Na shihou, Afu mei zai qianmian. That time Afu NEG in front ‘At that time, Afu is not in the front.’ 177 There are some temporal expressions that only modifies the evaluation event. Take na nian lai-shuo in (207) for example. The particle lai-shuo ‘in terms of’ marks that the sentence is about the speaker’s attitude. The evaluation time t 1 evaluates Afu’s situation according to the standard of t 2 of na nian ‘that year.’ (207) a. Na nian lai-shuo, Afu hen gao. That year in terms of Afu HEN tall ‘In terms of that year, Afu was tall.’ b. Evaluated at t 1 at that year t 2, Afu was tall at t 3. The time t 1 is the time that shows the evaluation of the attitude holder or the speaker. It is an independent time that differs from t 2, although t 1 is not explicitly expressed in the sentence. If there is only the time of the standard t 2 as denoted by na nian, we would expect that if we add a different evaluation time t 3, such as jinnia ‘this year’, a sentence like (208) is acceptable. However, it is inconsistent because there are two evaluation times. (208) #Na nian lai-shuo, Afu mei hen gao. Wo shi shuo That year in terms of Afu NEG HEN tall I COP mean zhao wo jinnian kan-lai. according-to I this year see-accordingly ‘In terms of that year, Afu wasn’t viewed as certifiedly tall. I mean, it’s because it’s due to my perspective of this year.’ 178 Since a stance is often anchored to a location, now I turn to locative modifiers. When a locative modifier is used, as in (209), the evaluation event could refer to both the location where the locus of the evaluation is or the location of the eventuality. (209) Zai wo na-ge weizi, Afu mei zai qianmian. At my that-CL location Afu NEG in front ‘At my location, Afu is not in the front.’ Similar to temporal modifiers, locative expression zai wo na-ge weizhi ‘at my location’ could refer both to the speaker’s evaluation stance only, but it could refer to the stance where both the speaker and Afu is situated. Again, we could find parallel examples like (208) for (209). Imagine a situation in which the evaluation eventuality is anchored to the location denoted by zai wo na-ge weizi. Let’s assume that this location is l 1. If the evaluation eventuality has lower scope than that of the negation, the sentence is true in at least two conditions. The first is that there is no such an event of evaluation, and the second is that there is an evaluation but it is not at location l 1. Both would face the same problems in sentences with temporal modifiers. Thus, we could conclude that the evaluation eventuality has a higher scope than the negation. Furthermore, while the evaluation eventuality takes higher scope than negation, it is lower than illocutionary operators and tense operators, such as question and conditionals. The examples with temporal modifier, as in (206), already shows that tense is higher than the scope of the eventuality. In the exclamative sentence in (210), the exclamation is based on an evaluation about a certain time. Similarly, in the interrogative sentence in (211), the evaluation is also being questioned. It is an interrogative that questions whether Afu is home under a specific situation. In the conditional sentence, the evaluation eventuality is also embedded under the antecedent, which denotes a condition that is evaluated with respect to a certain situation. 179 Exclamatives: (210) Laotian! Mao mei zai jia. Gosh! Cat NEG at home ‘My gosh! The cat isn’t home!’ Interrogatives: (211) Afu mei zai jia ma? Afu NEG at home Q ‘Isn’t Afu regarded as home?’ Conditionals: (212) Ruguo Afu mei zai jia, ta jiu hui zai xuexiao. If Afu NEG at home he then will on campus ‘if Afu isn’t at home (at a specific situation), he will then be on campus.’ Recall that in the previous chapter, I showed that what hen introduces is an evaluation eventuality related to one’s commitments. The scopal feature of the commitment eventuality in hen sentences conforms with the scope of the evaluation eventuality under the mei sentences. Hen also takes narrow scope with respect to negation, exclamative expressions, interrogatives and conditionals. There is an obvious restriction on the time of the evaluation eventuality: the time of the evaluation should overlap or to be contained in the time that the state holds. In short, the temporal relation between the evaluation eventuality and the state is given below. The temporal duration of the evaluation eventuality e is a subset of that of the state e’, where e’ is the state Afu is in. R denotes a relation of perception from the evaluation eventuality used to evaluate the state denoted by the predicate e’. (213) [∃e∃e’: 𝜏(e) ⊆ 𝜏(e’)][Eval(e)∧ HOLDER(e’, Afu) ∧ hen-tall(e’) ∧R(e, e’)] 180 This formalization is just an first attempt. We need to take into consideration the part of commitment expressed by hen, which will be discussed in the next section. 5.10 A formal account of hen 5.10.1 An overview The formalization of hen need to take into consideration one property: it introduces an evaluation event that represents the speaker or any salient attitude holder’s knowledge anchored at a certain time point or time frame. In a sense, we could formally see this eventuality as linked to the information state of the attitude holder, which refers to the set of knowledge compatible with what the speaker knows. Yet the traditional notion of an information state itself is not identical to the evaluation eventuality, since the information state represents the knowledge that the speaker possesses as a whole, and it is not anchored to a particular time or space. Therefore, I introduce the notion of a contentful event, a la Hacquard (2006, 2010, 2013). For Hacquard, such an event represents the perspective of the attitude holder that is responsible for deriving the modal base. The contentful event provides a possible temporal and spatial anchor that is not otherwise provided by merely adopting the notion of an information state. 5.10.2 Confidence level and degree morphemes The commitment made by hen could be seen as related to the confidence level of the attitude holder. Following Herburger & Rubinstein (2018), the confidence level of a proposition p could be measured by the function Bel Sz(p), which takes a proposition p and measures its degree of credibility based on the belief state of an individual. (214) Bel Sz(p) is the credibility of a proposition p with respect to the belief state of the individual z. 181 Herburger & Rubinstein use this notion to explain German particle eh-. In their view, eh- is used to introduce the degree d that represents the degree of confidence level of a proposition p with respect to S z. (215) ⟦eh-⟧ c, S z , w =λP <d, <e, t>> λd <d> . d≤Bel s z (p) In a sense, we could assume that degree morphemes used to express commitments are about whether the confidence level exceeds a certain compared degree. The information state becomes ‘activated’ when the degree modifier is evaluative. It becomes ‘dormant’ when the degree modifier or the positive degree morpheme is not related to commitments or confidence levels. Let’s make an assumption that English rather is also related to confidence level. Therefore, we could assume that the same formalization would be adopted as we introduce a standard of confidence. Below is the illustration of how rather could be interpreted if we adopt the same notion of Bel Sz(p) given above. A Hypothesis for rather: (216) ⟦rather⟧ c, S z , w =λp <d, <e, t>> λx <e> .∃d.Bel s z (p)≥d c The above formalization reads that p, which could be seen as the prejacent of rather, has a degree of credibility that exceeds a contextually given standard for accepted credibility relevant to p. The semantics of hen would provide a similar meaning But in order to accommodate the fact that hen is sensitive to the temporal or spatial anchor linked to the evaluation eventuality, we need to make certain modifications. One is more technical. We need to incorporate events into our semantics. Events are understated in typical degree semantics, as shown in (215) and (216). Furthermore, we also need to put temporal and spatial restrictions on the information state. This is how the notion of a contentful event become useful. 182 5.10.3 Contentful events The evaluation eventuality could be viewed as the realization of a contentful event in Hacquard’s sense (2006, 2010, 2013). An attitude expression has contents, which is about the set of knowledge, beliefs, desires or wishes that one has at a certain time. The content that comes from the evaluation eventuality, similarly, could be seen as evidence and beliefs used to justify the propositions. If we draw parallelism between epistemic modality and hen, we could see this eventualilty as a means to derive the modal base. This idea for an eventuality to be responsible in deriving a modal base is brought up by Hacquard (2006, 2013). According to her (Hacquard 2006), the speech-act level and the VP level of a sentence each introduce an eventuality. The eventuality at the speech level is related to the perspective during the speech time, while that introduced in the VP is related to that of the event time. The speech event is responsible for the interpretation of epistemic modality, while the VP level event corresponds to root modals. Similarly, an attitude verb also introduces an attitude event at the predicate level. Both the speech and the attitude events are about the ‘content’ of that eventuality, while the event introduced by the root modal is based on circumstantial evidence. What does the content of an eventuality mean here? The content of an contentful event refers to intentional contents. It could be viewed as the ways of which an attitude holder’s thoughts are about the objects being considered or perceived. The intentional content of an attitude verb comes from the attitude event of perceiving or believing, and other thinking and perception related activities. More specifically, according to Hacquard (2013), the content of the speech event could be viewed as the origin of the speaker’s commitments. Other than content-based modal bases, the speaker’s view on events corresponding to the root modal is related to the circumstantial evidence for justifying the prejacent. Syntactically speaking, Hacquard’s system would require the eventuality to be introduced by the modals or the attitude verbs themselves, which means that they are merged at the syntactic position of the modal expression. But the interpretation of these events are determined through a binding relation between this event and another event, which could determine the modal bases. In general, the contentful events for 183 epistemic modals are bound by the speech event when it is in the matrix predicate, and when in the embedded context under an attitude verb, it is bounded by the matrix attitude event. Events introduced by root modals, on the contrary, are bound by the aspect that quantifies over the VP event. In Hacquard’s system, there is a dichotomy between the contentful modal bases and the factual modal bases (Hacquard 2013). (217) a. High modal: f content(e)={w: w is compatible with the content of e} Examples: speech event (epistemic modals), attitude verbs The binding event: speech event e 0 (matrix context), attitude event/viewpoint aspect (embedded context) b. Low modal: f fact(e)={w: w is compatible with the circumstances of e} Example: root modals The binding event: Aspect What is more related to our discussion is of course the contentful event corresponding to the high modal, which is influenced by the presence of the speech event. It determines the epistemic basis of the speaker. In Hacquard’s view, if the sentence is an assertion, this speech event is the assertion event. This event is positioned at the CP level quantified by the ASSERT function, which quantifies over the speaker’s doxastic alternatives as determined by e 0 below. The CON(e 0) refers to the doxastic alternatives of an individual based on the speech event e 0. Formally, this is shown below. (218) ⟦ASSERT e o ⟧=λp <w, t> λw <s> .Assert ' (e o , w) & ∀w ' ∈∩CON(e o ):pIw ' J=1 184 Where ∩CON(e o ) is the doxastic alternative of the unique x such that x is the speaker and the holder of the event e 0. (Hacquard 2010: 38) Hacquard’s studies are not the only ones that endorse an event-based view on modality. Similar to Hacquard’s view, Kratzer (2013) also discusses the possibility that a modal base could be derived from eventualities merged at different syntactic locations. As shown earlier, the events introduced by modals could appear below viewpoint aspects, such as locating at the root modal positions. It could also be realized in the aspectual realm simpliciter. For example, it is said that the English progressives invoke a modal reading that requires inertia worlds (Portner 1998). In some languages, modals could even be realized in verbal affixes. It would not be surprising if a degree modifier would also carry the same function as these particles. 5.10.4 Contentful events and hen I argue that hen, which functions as a degree modifier, also introduces an event like the contentful event with universal modal force, and this contentful event could be seen as the event of evaluation, or observation itself. The modal base derived from this event is anchored to the speech event. Assume that the contentful event is e, and the modal base derived from that event is CON(e). Then for all worlds in the set of propositions in the content of e, there is some event e’ such that the state P holds of e’. Furthermore, following (213) above, there is a relation that holds between e and e’, which could be regarded as s perceptual or evaluation relation. The event e is used for evaluating the event e’. (219) a. Afu hen gao. ‘Afu is tall.’ b. There is an event e, the event of evaluation Eval(e), which is the contentful event from the asserting event’ 185 c. ∀w∈CON(e): ∃e’[e’ in w ∧ tall(e’) ∧ HOLDER(e’, Afu) ∧ R(e, e’)] As hen sentences are based on subjective evaluation, we could see the modal base derived from this event as equivalent to the information state possessed by the speaker or the attitude holder. The propositions in the content of e may include information like knowing Afu’s height, knowing the standard, or determining a relevant world or time for the suitable standard. It could also be about the conceptual notion of being ‘standing out’ from other objects in the comparison class. Finally, the binding between the speech event and the evaluation eventuality could explain for the high scope of the event corresponding to the making of the commitment in a hen sentence. Following the traditional Cinquean treatment of left-periphery (Cinque 1999), the ordering of epistemic modality is higher than tense and negation. This is what we have observed in the part of commitments in hen sentences. This is shown in a sentence in which there are two different modifiers, one modifies the speech event and the other tense. When they coexist, we could see the temporal anchor of the evaluation event takes higher scope over tense. This is shown in the example in which the time-related perspectival modifier dangshi lai-shuo ‘in terms of that time’ modifies the speech event, and xiaoshihou ‘childhood’ modifying tense of the matrix sentence. In (220), the evaluation for whether Afu’s tallness is significant is determined by the time denoted by dangshi, not by xiaoshihou. (220) Dangshi lai-shuo, Afu xiaoshihou mei hen gao. Then in terms of Afu childhood time NEG HEN tall ‘In terms of that time, Afu wasn’t certified to be tall during his childhood.’ This feature again conforms to the general observation that the syntactic positions of Mandarin quantifiers and modifiers reflect the scopal relations in the logic forms of the sentences (Huang 1982, Tang 1988). In other words, Mandarin Chinese tends to show isomorphism between the surface form and the 186 logic form when it comes to quantifiers. This feature comes in handy for figuring out the scope of the temporal anchor of the evaluation eventuality. 5.10.5 Formal account of hen With all the issues settled, now we finally can turn to the lexical semantics of hen. Our formalization here need to take into consideration the notion of events in order to account for the relations between different eventualities salient in the meaning of a hen sentence. In chapter 2, I discussed the standard analyses on the formalization of gradable adjectives. While they are descriptively adequate in expressing the gradability of such adjectives, they do not take into consideration of the eventuality of such adjectives. We need to incorporate events into the semantics. The traditional view of Davidsonian and Neodavidsonian event semantics sees an adjective as a predicate of an eventuality. In cases of stative verbs and states, the eventuality could be regarded as including a different ontological subcategory, the state argument s, as opposed to the event argument e. Here I will not get into the debate of whether statives and states require a different sort of argument. I use events to denote all eventualities across the board for now. (221) is an illustration of how adjectival expressions are often enunciated in event semantics. There is no distinction between gradable and non- gradable adjectives. They are both predicates of events. The adjective tall is a function of type <v, t>, in which ‘<v>’ represents the eventuality type. Apparently, this treatment is different from the standard view on gradable adjectives, as proposed by Kennedy, who sees gradable adjectives as having an extra degree component. (221) λe.TALL(e) If we want to see tall as gradable and to account for the fact that it requires a degree argument, a major issue one need to deal with is how the degree argument is introduced. There are two possible accounts. For 187 one, we could get it done by adding an extra degree argument in the lexical entry, in which the measure function µ that measures the event with respect to the salient feature of tall. (222) λdλe.Height(e) & μ(e)≥d Yet as a gradable adjective is traditionally viewed as having the same semantic type as non-gradable predicates, we could also assume that the degree argument is introduced from another particle. Therefore, gradable adjectives has the same logical type as non-gradable adjectives, both sharing the same form given in (221). This view is endorsed by Wellwood, Hacquard & Pancheva (2012) and Wellwood (2014). In their system, the gradable adjectives themselves do not provide a measure function for mapping the object onto the scale. Instead, other particles, such as much, is added for making the measurement. The function of much is to measure an event with respect to certain dimension related to one of the properties it has. The result, however, is not much different from what we get with a semantics without much in (222). The major difference is that now it is ⟦much tall⟧ that represents the function λdλe.Height(e) & μ(e)≥d, in which the measurement is done by much over the function of tall, λe.Height(e). Yet we could also view the gradable adjective as an independent category that requires an extra degree argument, as proposed by Kennedy & McNally (2005). The differences between the two is not crucial for the discussion here. For simplicity, I will adopt the view that gradable adjectives still have an extra degree argument in the lexical entry in (222), which parallels Kennedy’s original view on gradable adjectives better. Likewise, we could better follow Kennedy’s and Kennedy & McNally’s view on the treatment of positive degree morphemes and degree modifiers alike. However, when it comes to those that are used to express commitments, we need to incorporate the Bel function into the semantics. Recall that degree modifiers are those that modifies a standard introduced by pos. Now we simply change the modification of an individual variable into one that also modifies an event variable. The standard 188 function takes the abstracted function λe.⟦pos⟧ c, w (p)(e) to derive a standard d based on the set of eventualities with the property denoted by p. (223) ⟦very⟧ c, w =λp <d, <v, t≫ λe <v> ∃d. (p(e)(d) & standard(d)(λe. ⟦pos⟧ c, w (p)(e)) We could also see hen as one that has the same function of raising the standard. This would put hen in line with degree modifiers like very. The difference would lie in the added meaning contributed by Bel. In general, we could see Bel as a function that takes a proposition p and returns the degree to which the is confident towards that proposition based on the contentful event e. The contentful event takes the role of the information state, and the attitude holder ‘retrieves’ the knowledge and propositions responsible for making the evaluation from that event. Therefore, Bel takes an extra eventuality that represents the contentful event. This is given below. This shows that there the confidence level of p with respect to the contentful event is exactly d. (224) ⟦Bel⟧ c, w =λe <v> λp <d, <v, t≫ .ιd. d= confident(p)(CON(e)) Furthermore, we need to introduce a function similar to a positive degree morpheme, ⟦pos⟧ c, w , which introduces a contextually given standard d c. This standard can either denote the typical standard for gradable adjectives or the standard for credibility. (225) ⟦pos⟧ c, w =λe <v> λp <d, <v, t≫ λe <v> .∃d(p(d)(e))& d > d c Now we can combine the above function with hen. The lexical entry is shown in (226). The lexical entry of hen involves two ⟦pos⟧ c, w functions written in the lexical entry. The first one combines with the gradable adjective, the eventuality of the gradable adjective and derives a standard as a cutoff point for the 189 positive extension of that gradable adjective. This is the standard use of pos in gradable adjectives. This is shown in the part of (⟦pos⟧ c, w )(p)(e)) in (226). Furthermore, it combines with the Bel function and derives a confidence degree based on the degree of the gradable adjective. Here e i is the contentful event. The Bel function takes the proposition denoting the positive extension of p, ⟦pos⟧ c, w (p)(e)), and returns a degree of credibility, and such a degree exceeds the contextually given standard of credibility provided by this second pos. This lexical entry shows that when the degree of the object is high enough, the confidence level would also be high enough. The lexical entry of hen: (226) ⟦hen⟧ c, w =λe i <v> λp <d, <v, t≫ λe <v> .⟦pos⟧ c, w (⟦Bel⟧ c, w (e i )(⟦pos⟧ c, w )(p)(e))) Now I turn to the compositionality of the hen sentence. I follow the standard view that degree modifiers occupy the head of DegP. It then merge with the light verb v, which further introduces the external argument. At the CP level, an event of assertion could be viewed as binding the event introduced at the DegP level. This allows the evaluation eventuality to take wide scope over sentential operators. (227) [ CP ASSERT(e i) [ TP Afu [ vP v [ DegP [[ Deg’ hen e i][ AP gao]]] The function of the light verb v is to introduce an external argument for the predicate. Therefore, the lexical entry of v is a two place function that takes an individual and an event to truth values. As shown in (228). The external argument is introduced as the HOLDER of the state denoted by the predicate. (228) λxλe.HOLDER(x)(e) 190 In principle, if we follow the Neo-Davidsonian view, each arguments of a verb or an adjective are combined separately through conjunction, the external argument need to be combined with the predicate by conjunction as well. However, traditional Heim & Kratzer’s treatment of compositional semantics (Heim & Kratzer 1998) does not provide a rule for the external subjects to be introduced through conjunction. Therefore, Kratzer (1996) introduces a separate rule, Event Identification (EI). It takes two functions, one of type <v, t>, in which v stands for events, and another with type <e, <v, t>> to yield a predicate of events. (229) is an example given by Kratzer (1996: 122). EI enables the external subject to be joined to the matrix predicate VP (λe <v> [feed(the dog)(e)]) simply by conjunction. This rule would allow the external subject as a separate component from the predicate for not being one of the arguments of the verb itself. (229) Event Identification (EI): f + g →h <e, <v, t>> <v, t> <e, <v, t>> λx <e> λe <v> [Agent(x)(e)] λe <v> [ feed(the dog)(e)] →λ𝑥 <e> λe <v> [Agent(x)(e)∧ feed(the dog)(e) ∧ Agent(x)(e)] (from Kratzer 1996: 122) In the above example, the function f of type <e, <v, t>> represents the light verb head. The function g represents the matrix predicate, which is a predicate of event. When the two are combined, the result is a function with the same type as g, which is a function from an individual to truth values, and the truth condition is the conjunction of the two functions. I assume that the same rule, Event Identification, could be applicable for gradable predicates as well, since the degree argument is saturated at the level of DegP. 191 Finally, we need to revise the formalization for the asserting function, ASSERT function, which takes an event e o that is bound by the speech event. (230) 8 Lexical entries: ⟦gao⟧ c, w = 𝜆dλe.Height(e) & μ(e)≥d ⟦hen⟧ c, w = λe i <v> λp <d, <v, t≫ λe <v> .⟦pos⟧ c, w (⟦Bel⟧ c, w (e i )(⟦pos⟧ c, w )(p)(e))) ⟦pos⟧ c, w = λp <d, <v, t≫ λe <v> ∃d. (p(d)(e))& d > d c ⟦Bel⟧ c, w =λe <v> λp <d, <v, t≫ .ιd. d= confident(p)(CON(e)) 8 Thanks to Barry Schein (p.c.) for suggestions on the formalization. DegP AP TP vP e o CP v Afu ASSSERT(e o ) Deg’ hen A gao λe o 192 ⟦v⟧ c, w = λxλe. HOLDER(x)(e) ⟦Afu⟧ c, w =Afu ⟦ASSERT⟧ c, w =λe <v> λp <v, t> ∃e'. Assert(e) & ∀w ' ∈∩CON(e):p(e')=1 in w' Derivations: SDeg ' T c, w =⟦hen⟧ c, w (e o ) =λp <d, <v, t≫ λe <v> .⟦pos⟧ c, w (⟦Bel⟧ c, w (e o )(⟦pos⟧ c, w )(p)(e))) (By Function Application) ⟦DegP⟧ c, w =SDeg ' T c, w (⟦gao⟧ c, w ) =λe <v> .⟦pos⟧ c, w (⟦Bel⟧ c, w (e o )(⟦pos⟧ c, w )⟦gao⟧ c, w (e))) (By Function Application) ⟦vP⟧ c, w =⟦DegP⟧ c, w (⟦v⟧ c, w ) =[λe <v> ⟦pos⟧ c, w (⟦Bel⟧ c, w (e o )(⟦pos⟧ c, w )⟦gao⟧ c, w (e)))](λxλe.HOLDER(x)(e)) =λx <e> λe <v> .⟦pos⟧ c, w (⟦Bel⟧ c, w (e o )(⟦pos⟧ c, w )⟦gao⟧ c, w (e))) & HOLDER(x)(e) (By Event Identification) ⟦TP⟧ c, w =[λx <e> λe <v> .⟦pos⟧ c, w (⟦Bel⟧ c, w (e o )(⟦pos⟧ c, w )⟦gao⟧ c, w (e))) & HOLDER(x)(e)](⟦Afu⟧ c, w ) = λe <v> .⟦pos⟧ c, w (⟦Bel⟧ c, w (e o )(⟦pos⟧ c, w )⟦gao⟧ c, w (e))) & HOLDER(Afu)(e) ⟦ASSERT(e o )⟧ c, w =[λe <v> λp <v, t> ∃e'.Assert ' (e) & ∀w ' ∈∩CON(e):p(e')=1](e o ) 193 = λp <v, t> ∃e'.Assert ' (𝑒 % ) & ∀w ' ∈∩CON(e):p(e')=1 in w' ⟦CP⟧ c, w =[λp <v, t> ∃e'.Assert ' (e o ) & ∀w ' ∈∩CON(e o ):p(e')=1] (λe <v> .⟦pos⟧ c, w (⟦Bel⟧ c, w (e o )(⟦pos⟧ c, w )⟦gao⟧ c, w (e))) & HOLDER(Afu)(e)) = 1 iff ∃e o .∃e'. Assert ’ (e o ) & ∀w ’ ∈∩CON(e o ):⟦pos⟧ c, w (⟦Bel⟧ c, w (e o )(⟦pos⟧ c, w )⟦gao⟧ c, w (e'))) & HOLDER(Afu)(e')) (By Function Application) In short, the truth condition of the sentence is that there exist an assertion event e 0, which is the contentful eventuality, and based on this eventuality, the attitude holder has confidence over the proposition that Afu’s height exceeds the positive standard. In the above derivation, we can see that hen retains the syntactic status of a typical degree modifier while introducing certain restrictions over the proposition predicated of the upcoming gradable adjectives. This restriction is done by imposing a credibility function that operates on the proposition of which the observed event is predicated. On the contrary, ordinary degree modifiers only provide the comparison between two degrees, the one possessed by the subject, and that of the standard. To sum up, this section presents an analysis based on the view that an eventuality could be used as a means to derive an information state, following Hacquard’s insight on the derivation of modal bases derived by contentful events. The information state is an index that is active when the sentence about one’s judgment based on his or her personal knowledge, and become ‘dormant’ when it is in objective and non-intensional expressions. Perhaps we can view this evaluation event as merged either at some functional level above the matrix predicate, such as the CP-periphery. This treatment is inspired by Percus’s view, who argues that there are situation variables located at the phrasal periphery (Percus 2000). 5.11 Conclusion 194 So far I’ve presented some empirical data of sentences licensed under mei. The generalization is that mei only subcategorizes a predicate with an event that represents an individual’s stance or attitude. This evaluation event could either be the one that updates the information state, as in the cases involving some individual’s attitudes, or it could be non-intentional, which is simply linked to an eventuality that represents one’s stance anchored at a time or space. This chapter does not deal with the detailed syntactic location of such an eventuality. I assume that it is introduced at the predicate level, but is perhaps coindexed to the speech event on the CP layer. More researches are required to clarify this point. 195 Chapter 6 Concluding remarks 6.1 Some issues on other commitment inducing adverbs In this dissertation, I show that different from typical degree modifiers, hen has two properties. One is that it introduces an eventuality of evaluation, and the other is related to the relative degree the object possesses that is determinate for making a commitment. It is possible that there are similar degree modifiers that could potentially form a natural class both within Mandarin Chinese and cross-linguistically. In Mandarin Chinese, hen is just one of the degree modifiers that could potentially express either evaluativity or an stance that could be explained by the existence of an evaluation event. Recall that in chapter 3, Other degree modifiers, such as feichang ‘very’ or ji ‘extremely’, presumably provide the same meaning. However, I showed that feichang ‘very’ shows a different entailment pattern from hen: it seems to be a ‘simpler’ degree modifier of which the function is to raise the standard, just like English very. Yet we cannot simply see it as the exact counterpart of English very either. It is because it still shows some characteristics similar to Mandarin hen. They are all licensed under subjective contexts. For example, they can both be licensed under ganxiang shi ‘the opinion is….’ And they could be licensed by mei. This is shown in (1) and (2). In the negated sentences (2), they also show hedging effect similar to hen. Subjective contexts: (1) Wo ganxiang shi, Afu xiangdan/ji gao. My opinion COP Afu quite/extremely tall ‘My opinion is that Afu is quite/extremely tall. (2) Afu mei you feichang/ji gao Afu NEG COP very/extremely tall ‘Afu wasn’t/isn’t extremely tall.’ (1) 196 Presumably, the Mandarin particle could all belong to the same category of degree modifiers that introduces an extra event argument, which serves the function of representing an attitude holder’s stance. Each of such degree modifiers provide some modifications on the standard set by the positive degree morpheme. For example, the degree modifier feichang could refer to a degree that is significantly higher than the standard used for denoting the positive extension, and the difference between the standard set by pos and that of very is higher than that between hen and pos. As for extreme adjective like ji ‘extremely’, it would introduce a standard within the set of degrees at the extreme end of the scale. Presumably, these degree modifiers, just like hen, introduce an eventuality that are bound by a higher speech event. The commitment is based on the fact that the degree of the target object reaches a threshold. Aside from the Mandarin examples, we may find a cross-linguistic category that shows the same use as related to one’s commitments. English degree modifiers like quite is also a possible candidate to have the same property of hen. I assume that it is similar to hen in terms of providing a commitment based off the degree it raises to. This view is distinct from previous analyses in many respects. 6.2 A look at quite The particle quite is regarded as evaluative when it is used to modify an adjectival or adverbial expression. Similar to very, it seems to introduce a degree standard that is significantly high. Yet the standard it raises to is not as high as those raised to by very or really. (3) John is quite tall. Quite and very are different in a second aspect. It can appear with non-gradable adjectives. When it does, it coerce them into gradable expressions. Klecha (2014) provides a comparison between the two sentences, showing that quite could appear with non-gradable correct, while other degree modifiers like too are ungrammatical. If we replace quite with very, it sounds funny. 197 (4) a. Your response is quite correct. b. *Your response is too correct. (Examples from Klecha 2014: 178) Regarding the feature of standard raising, it seems that quite introduces a degree that could be situated higher than the positive standard, but it is not as high as the one assigned by very. This is shown in the contrast between quite and very below. (5) John is quite tall, but I wouldn’t say he is VERY tall. Furthermore, when quite co-occur with non-gradable adjectives, it seems that it provide some kind of emphasis that the property denoted by the adjective is indeed the case. For example, in (4)a, what is expressed by quite correct is that the person is indeed correct. Similarly, quite right is nearly equivalent to saying that it is being absolutely right. In other words, the use of quite with non-gradable adjectives seems to suggest that it is not about providing a standard, but about emphasizing the proposition to be true. Therefore, it is possible that such emphasis is also linked to commitments. Furthermore, a sentence with quite seems to be vaguer than a sentence with a gradable expression modified by very. Here I do not refer to the run-of-the-mill vagueness which is present as a general property of gradable expressions. When we speak of vagueness in gradable expressions, it is often either about context variability or the uncertainty of the required degree of accuracy. It could also refer to the fuzziness with the various degree of tolerance for borderline cases. Vagueness is a general property for gradable adjectives and sentences with ordinary degree modifiers across-the-board. It is possible that one of the sources for vagueness induced by quite is due to the arbitrariness of the dimension, which is often coerced by certain restrictions in the context. When one says that ‘your response is quite correct’, it does not just express that for a given context, the correctness exceeds a standard by a lot, as in the case of ordinary sentences modified by very, since correctness itself is not usually a gradable notion. Rather, the 198 corresponding gradable notions is an emphasis that the expression reaches the anticipated correctness. In this sense, it seems that it could be associated with the notion of stereotypical correctness. Perhaps what quite corresponds to is a scale that ranks objects based on how correct it is stereotypically. If we assume that quite is still gradable even when it can occur with non-gradable adjectives, we could now move on to the problem of how its interpretation is related to standard raising. Presumably, quite introduces a standard that is above the positive standard, but it is not as high as that given by very. Yet it is raised to a degree that still exceeds the threshold enough for one to make a commitment. One thing that is worth noting is that there seems to be an intuition that the commitment made by quite is not as strong as those given by those like really or indeed, which provide an extreme degree of intensification, intuitively speaking. Unlike a sentence modified by very, really and indeed actually provides an even higher degree of commitment. For example, if one says ‘John is really tall’, what the speaker intends is to put extra emphasis on the high certainty he has towards the proposition, while ‘John is quite tall’ only states that John’s being tall is a certain case. If what differentiates really and indeed from quite lies in the degree of commitment, we may assume that the degree of commitment is codependent with the degree raised by the degree adverb. Presumably, the scale is an ordering of the closeness to the ideal of expected correctness, and it is linked to the scale relevant to the dimension of the adjective. This is illustrated below. The longer bar represents the set of degrees that are high for one to marked commitment over John is tall, and the short bar represents the set of degrees that are not high enough for one to make a high degree of commitment with really or indeed. (6) Scale of height quite/rather really, indeed etc. full commitment pos high degree of commitmeent 199 Of course, the above analysis is just an assumption. We still need to look into the possible correlation between the degree the standard being raised to and the potential different degrees of commitments. 6.3 Concluding remarks In this dissertation, I discuss how a commitment account could explain some of the puzzles related to the Mandarin particle hen. It allows one to explain why it seems more ‘subjective’ than its English counterpart, the positive degree form, and how it is related to temporal boundedness. Within the language itself, the contrast becomes striking when we compare hen with the shi…-de form. We would expect similar particles to appear in other languages as well. In English, quite could potentially having the same function that might form a natural class. Yet we still need further studies to fully understand the nature of this cateogory. 200 References Adger, D. (1994). Molly Diesing, Indefinites. (Linguistic Inquiry Monograph 20.) Cambridge, MA: MIT Press, 1992. Pp. xiv 175. Journal of Linguistics, 30(2), 570-571. Aikhenvald, A. I. (2004). Evidentiality. Oxford University Press. Aikhenvald, A. Y. (2013). Possession and ownership: A cross-linguistic perspective. Possession and ownership: A cross-linguistic typology, 1-64. Aikhenvald, A. Y., & Dixon, R. M. (Eds.). (2003). Studies in evidentiality (Vol. 54). John Benjamins Publishing. Anand, P., & Hacquard, V. (2008, October). Epistemics with attitude. In Semantics and linguistic theory (Vol. 18, pp. 37-54). Austin, J. L. (1962). 1962: How to do things with words. Oxford: Clarendon Press. Bach, E. (1986). The algebra of events. Linguistics and philosophy, 9(1), 5-16. Banvise, J. (1978). Monotone quantifiers and admissible sets. Studies in Logic and the Foundations of Mathematics, 1-38. doi:10.1016/s0049-237x(08)70925-x Bartsch R & Vennemann T (1973). Semantic structures: A study in the relation between syntax and semantics. Frankfurt: Atha¨enum Verlag. Beaver, D. I., & Clark, B. Z. (2009). Sense and sensitivity: How focus determines meaning (Vol. 12). John Wiley & Sons. Beltrama, A. (2016). Bridging the gap: Intensifiers between semantic and social meaning (Doctoral dissertation, The University of Chicago). Beltrama, A. (2018). Totally Between Subjectivity and Discourse. Exploring the Pragmatic Side of Intensification. Journal of Semantics, 35(2), 219-261. Beltrama, A., & Bochnak, M. R. (2015). Intensification without degrees cross-linguistically. Natural Language & Linguistic Theory, 33(3), 843-879. Bergqvist, H. (2018). Evidentiality as stance. Evidence for evidentiality, 61, 19. Bertinetto, P. M. (1994). Temporal reference, aspect and actionality: their neutralization and interactions, mostly exemplified in Italian. Tense, aspect and action: Empirical and theoretical contributions to language typology, 113-137. Bierwisch, M., & Lang, E. (1989). Somewhat longer–much deeper–further and further: Epilogue to the dimensional adjective project. Dimensional Adjectives: Grammatical Structure and Conceptual Interpretation. Berlin et al.: Springer, 471-514. Bosanquet, B. (1911). Logic: Or, The Morphology of Knowledge (Vol. 1). Clarendon Press. Büring, D. (2012). Focus and intonation. Routledge companion to the philosophy of language, 103-115. Burnett, H. (2014). A delineation solution to the puzzles of absolute adjectives. Linguistics and philosophy, 37(1), 1- 39. 201 Bybee, J. L., & Fleischman, S. (Eds.). (1995). Modality in grammar and discourse (Vol. 32). John Benjamins Publishing. Bylinina, L. (2013, August). Degree infinitival clauses. In Semantics and Linguistic Theory (Vol. 23, pp. 394-411). Bylinina, L. (2017). Judge-dependence in degree constructions. Journal of Semantics, 34(2), 291-331. Carlson, G. N. (1977). Reference to kinds in English (Doctoral dissertation, Thesis - University of Massachusetts, 1977). Amherst, MA. Carlson, L. (1981). Aspect and quantification. In Tense and aspect (pp. 31-64). Brill. Chao, Yuen-ren (1968). A Grammar of Spoken Chinese. Berkeley: University of California Press. Chierchia, G. (1995). Individual-level predicates as inherent generics. The generic book, 125. Cinque, G. (1999). Adverbs and functional heads: A cross-linguistic perspective. Oxford University Press on Demand. Comrie, B. (1976). Aspect: An introduction to the study of verbal aspect and related problems (Vol. 2). Cambridge university press. Comrie, B. (2000). Evidentials: Semantics and history. Evidentials: Turkic, Iranian and neighbouring languages, 1, 12. Cresswell M J (1977). ‘The semantics of degree.’ In Partee B (ed.) Montague grammar. New York: Academic Press. 261–292. Crone, P. (2019). Assertions of clarity & raising awareness. Journal of Semantics, 36(1), 53-97. De Swart, H. (1991). Adverbs of quantification: A generalized quantifier approach (Vol. 1). Rodi. Deo, A., Sanchez-Alonso, S., & Pinango, M. (2016). Alternative circumstances of evaluation and the ser/estar distinction in Spanish. Ms., Yale University. URL https://ling. auf. net/lingbuzz/003543. Dowty, D. R. (1975). The stative in the progressive and other essence/accident contrasts. Linguistic Inquiry, 6(4), 579-588. Ernst, T. (1995). Negation in mandarin Chinese. Natural Language & Linguistic Theory, 13(4), 665-707. Ernst, T. (2016). Modification of stative predicates. Language, 92(2), 237-274. Fábregas, A., Marín, R., & McNally, L. (2012). From psych verbs to nouns. Telicity, change, and state: A cross- categorial view of event structure, 162-184. Faller, M. (2002). Semantics and pragmatics of evidentials in Cuzco Quechua (Doctoral dissertation, stanford university). Faller, M. (2006). Evidentiality below and above speech acts. Special issue of Functions of Language on evidentiality. Fernald, T. B. (2000). Predicates and temporal arguments. Oxford University Press. Geuder, W. (2006). Manner modification of states. In Proceedings of Sinn und Bedeutung (Vol. 10, No. 1, pp. 111- 124). 202 Grano, T. (2012). Mandarin hen and universal markedness in gradable adjectives. Natural Language & Linguistic Theory, 30(2), 513-565. Hacquard, V. (2006). Aspects of modality. PhD diss. MIT. Hacquard, Valentine (2009). On the Interaction of Aspect and Modal Auxiliaries. Linguistics and Philosophy, 32, 279-312. Hacquard, V. (2010). On the event relativity of modal auxiliaries. Natural language semantics, 18(1), 79-114. Hacquard, V. (2013, December). On the grammatical category of modality. In Proceedings of the 19th Amsterdam colloquium. Halliday, M. A. (1970). Functional diversity in language as seen from a consideration of modality and mood in English. Foundations of language, 322-361. Hamblin, C. L. (1959). The modal" probably". Mind, 68(270), 234-240. He, C., & Jiang, Y. (2011). Type shifting, Chinese hen+ N structure, and implications for semantic parameters. Lingua, 121(5), 890-905. Herburger, E., & Rubinstein, A. (2014, August). Is' more possible'more possible in German?. In Semantics and Linguistic Theory (Vol. 24, pp. 555-576). Herburger, E., & Rubinstein, A. (2019). Gradable possibility and epistemic comparison. Journal of Semantics, 36(1), 165-191. Horn, L. R. (1989). A Natural History of Negation. Chicago: University of Chicago Press. Huang, C. R., & Mangione, L. (1985). A reanalysis of de: Adjuncts and subordinate clauses. In Proceedings of the fourth west coast conference on formal linguistics (pp. 80-91). Huang, C. T. J. (1982). Move wh in a language without wh-movement. The linguistic review, 1(4), 369-416. Huang, S. Z. (2006). Property theory, adjectives, and modification in Chinese. Journal of East Asian Linguistics, 15(4), 343-369. Iljic, R. (1990). The verbal suffix-guo in Mandarin Chinese and the notion of recurrence. Lingua, 81(4), 301-326. Irwin, P. (2014). SO [totally] speaker-oriented: An analysis of “Drama SO”. Micro-syntactic variation in North American English, 29-70. Izvorski, R. (1997, November). The present perfect as an epistemic modal. In Semantics and Linguistic Theory (Vol. 7, pp. 222-239). Jäger, G. (2001). Topic-Comment Structure and the Contrast between Stage Level and Individual Level Predicates. Journal of Semantics, 18, 83-126. Kamp, H. (2013). Two theories about adjectives. In Meaning and the Dynamics of Interpretation (pp. 225-261). Brill. Kaplan, D. (1989). Demonstratives: An essay on the semantics, logic, metaphysics and epistemology of demonstratives and other indexicals In J. Almog, J. Perry & H. Wettstein (Eds.), Themes from Kaplan (pp. 481- 563). Oxford University Press. Karttunen, L. (1974). Presupposition and linguistic context. Theoretical linguistics, 1(1-3), 181-194. 203 Katz, G. (2003, August). A modal account of the English present perfect puzzle. In Semantics and Linguistic Theory (Vol. 13, pp. 145-161). Kennedy C (1999). Projecting the adjective: The syntax and semantics of gradability and comparison. New York: Garland Press. Kennedy, C. (2005). Parameters of comparison. Unpublished manuscript, University of Chicago. Kennedy, C. (2007a, January). Modes of comparison. In Proceedings from the annual meeting of the Chicago Linguistic Society (Vol. 43, No. 1, pp. 141-165). Chicago Linguistic Society. Kennedy, C. (2007b). Vagueness and grammar: The semantics of relative and absolute gradable adjectives. Linguistics and philosophy, 30(1), 1-45. Kennedy, C., & McNally, L. (2005). Scale structure, degree modification, and the semantics of gradable predicates. Language, 345-381. Klecha, P. (2012). Positive and conditional semantics for gradable modals. In Proceedings of sinn und bedeutung (Vol. 16, No. 2, pp. 363-376). Klecha, P. (2014). Bridging the divide: Scalarity and modality: University of Chicago dissertation. Klecha, P. (2014). Bridging the divide: Scalarity and modality. University of Chicago, Division of the Humanities, Department of Linguistics. Klein, E. (1980). A semantics for positive and comparative adjectives. Linguistics and philosophy, 4(1), 1-45. Klein, E. (1991). Comparatives. Semantik: Ein internationales handbuch der zeitgenössischen forschung, ed. by Arnim von Stechow and Dieter Wunderlich, 673–91. Klein, W., Li, P., & Hendriks, H. (2000). Aspect and assertion in Mandarin Chinese. Natural Language & Linguistic Theory, 18(4), 723-770. Kockelman, P. (2004). Stance and subjectivity. Journal of Linguistic Anthropology, 14(2), 127-150. Kölbel, M. (2002). Truth without objectivity. Routledge. Kölbel, M. (2004, June). III—Faultless disagreement. In Proceedings of the Aristotelian society (Vol. 104, No. 1, pp. 53-73). Oxford, UK: Oxford University Press. Kpasky, P., & Kiparsky, C. (1971). Fact. In D. D. Steinber & L. A. Jakobovits (Eds.), Semantics: An interdisciplinary reader (pp. 345-369). Cambridge: Cambridge University Press. Kratzer, A. (1994). On external arguments. University of Massachusetts occasional papers in linguistics, 20(1), 7. Kratzer, A., & Heim, I. (1998). Semantics in generative grammar (Vol. 1185). Oxford: Blackwell. Krifka, M. (2004). Semantics below and above speech acts. Handout from a talk at Stanford University, April, 9. Lakoff, R. (1973, April). The logic of politeness; or, minding your p’s and q’s. In ninth regional meeting of the Chicago Linguistic Society (Vol. 8, pp. 292-305). 292J305. Chicago: Chicago Linguistic Society. Larson, R. K. (1988). Scope and comparatives. Linguistics and philosophy, 11(1), 1-26. 204 Lasersohn, P. (2005). Context dependence, disagreement, and predicates of personal taste. Linguistics and philosophy, 28(6), 643-686. Lasersohn, P. (2009). Relative truth, speaker commitment, and control of implicit arguments. Synthese, 166(2), 359- 374. Lassiter, D. (2010, August). Gradable epistemic modals, probability, and scale structure. In Semantics and Linguistic Theory (Vol. 20, pp. 197-215). Lassiter, D. (2011). Measurement and modality: The scalar basis of modal semantics (p. 9). Ph. D. thesis, New York University. Lewis, D. (1983). Extrinsic properties. Philosophical Studies, 44(2), 197-200. Li, C. N., & Thompson, S. A. (1989). Mandarin Chinese: A functional reference grammar (Vol. 3). Univ of California Press. Lin, J. W. (2003). Aspectual selection and negation in Mandarin Chinese. Linguistics, 41(3), 425-459. Lin, W. C. (1979). A descriptive semantic analysis of the Mandarin abstract-tense system (Doctoral dissertation, Cornell University). Liu, C. S. L. (2010). The positive morpheme in Chinese and the adjectival structure. Lingua, 120(4), 1010-1056. Lyons, J. (1977) Semantics. 2 vols. Cambridge: Cambridge University Press. Ma, J. H. S. (1977). Some Aspects of the Teaching of. Journal of the Chinese Language Teachers Association, 12(1), 14-26. MacFarlane, J. (2005). The assessment sensitivity of knowledge attributions. Oxford studies in epistemology, 1, 197- 233. Maienborn, C. (2001). On the Position and Interpretation of Locative Modifiers. Natural Language Semantics, 9(2), 191-240. Maienborn, C. (2003b). A discourse-based account of Spanish ser/estar. To appear in Linguistics 43(1). Maienborn, C. (2004). On the Limits of the Davidsonian Approach: The Case of Copula Sentences. To appear as target article in Theoretical Linguistics. Maienborn, C. (2008). On davidsonian and kimian states. In Existence: Semantics and syntax (pp. 107-130). Springer, Dordrecht. McNally, L., & Stojanovic, I. (2017). Aesthetic adjectives. Semantics of Aesthetic Judgements, 17-37. Milsark, G. L. (2014). Existential Sentences in English (RLE Linguistics D: English Linguistics). Routledge. Mittwoch, A. (2005). Do states have Davidsonian arguments? Some empirical considerations. Event arguments: Foundations and applications, 69-88. Moltmann, F. (2013). On the distinction between abstract states, concrete states, and tropes. Genericity, 43, 293. Mushin, I. (2001). Evidentiality and epistemological stance: Narrative retelling (Vol. 87). John Benjamins Publishing. 205 P. (2017). Subjectivity and perspective in truth-theoretic semantics (Vol. 8). Oxford University Press. Palmer, F. (2001) Mood and Modality. 2nd ed. Cambridge: Cambridge University Press. Palmer, F. R. (2001). Mood and modality. Cambridge University Press. Palmer, G. B. (1996). Toward a theory of cultural linguistics. University of Texas Press. Papafragou, A. (2006). Epistemic modality and truth conditions. Lingua, 116(10), 1688-1702. Pearson, H. (2013). The sense of self. Cambridge, Mass.: Harvard University dissertation. Pearson, H. (2015). The interpretation of the logophoric pronoun in Ewe. Natural Language Semantics, 23(2), 77- 118. Percus, O. (2000). Constraints on some other variables in syntax. Natural language semantics, 8(3), 173-229. Perry, J., & Barwise, J. (1983). Situations and attitudes. Cambridge, MA: MIT Press. Portner, P. (1998). The progressive in modal semantics. Language, 760-787. Portner, P. H. (1993). Situation theory and the semantics of propositional expressions. Portner, P., & Rubinstein, A. (2016). Extreme and non-extreme deontic modals. Deontic modals, 256-282. Rett, J. (2012). On modal subjectivity. UCLA Working Papers in Linguistics, 16, 131-150. Roberts, C. (2012). Information structure: Towards an integrated formal theory of pragmatics. Semantics and Pragmatics, 5, 6-1. Roldán, M. (1974). Toward a Semantic Characterization of" Ser" and" Estar". Hispania, 57(1), 68-75. Rooth, M. (1992). A theory of focus interpretation. Natural language semantics, 1(1), 75-116. Rooth, M. (1995). Indefinites, adverbs of quantification and focus semantics. The generic book, 265-299. Rooth, M.: 1985, Association with Focus, PhD thesis, GLSA, University of Massachusetts, Amherst. Rothmayr, A. (2009). The structure of stative verbs (Vol. 143). John Benjamins Publishing. Rotstein, C., & Winter, Y. (2004). Total adjectives vs. partial adjectives: Scale structure and higher-order modifiers. Natural language semantics, 12(3), 259-288. Sæbø, K. J. (2009). Judgment ascriptions. Linguistics and philosophy, 32(4), 327-352. Sassoon, G. W. (2013). A typology of multidimensional adjectives. Journal of semantics, 30(3), 335-380. Seuren, P. A. (1973). The comparative. In Generative grammar in Europe (pp. 528-564). Springer, Dordrecht. Schwarz, B. (2010). A note on for-phrases and derived scales. Handout for talk at Sinn und Searle, J. R., & Searle, J. R. (1969). Speech acts: An essay in the philosophy of language (Vol. 626). Cambridge university press. 206 Shi, Z. (1990). Decomposition of perfectivity and inchoativity and the meaning of the particle le in Manarin Chinese /汉语中完成体和始成体的分解及 “了” 的意义. Journal of Chinese Linguistics, 95-124. Soh, H. L. (2009). Speaker presupposition and Mandarin Chinese sentence-final-le: a unified analysis of the “change of state” and the “contrary to expectation” reading. Natural Language & Linguistic Theory, 27(3), 623-657. Solt, S. (2012). Comparison to arbitrary standards. In Proceedings of Sinn und Bedeutung (Vol. 16, No. 2, pp. 557- 570). Speas, P., & Tenny, C. (2003). Configurational properties of point of view roles. Asymmetry in Grammar1, 315-345. Stephenson, T. (2007). Judge dependence, epistemic modals, and predicates of personal taste. Linguistics and philosophy, 30(4), 487-525. Stephenson, T. (2010). Control in centred worlds. Journal of semantics, 27(4), 409-436. Stojanovic, I. (2007). Talking about taste: Disagreement, implicit arguments, and relative truth. Linguistics and philosophy, 30(6), 691-706. Sybesma, R. (2013). The mandarin VP (Vol. 44). Springer Science & Business Media. Tai, J. H. (1975). On two functions of place adverbials in Mandarin Chinese. Journal of Chinese Linguistics, 154- 179. Tang, C. C. J. (1988). Wh-topicalization in Chinese. Ms., Cornell University, Ithaca. Taylor, B. (1977). Tense and continuity. Linguistics and philosophy, 1(2), 199-220. Unger, P. (1978). Ignorance: A case for scepticism. OUP Oxford. Vanderveken, D. (1990). Meaning and speech acts: Volume 2, formal semantics of success and satisfaction (Vol. 2). Cambridge University Press. Vendler, Z. (1957). Verbs and times. The philosophical review, 66(2), 143-160. Vendler, Z. (1967). Causal relations. The Journal of philosophy, 64(21), 704-713. von Stechow A (1984a). ‘Comparing semantic theories of comparison.’ Journal of Semantics 3, 1–77. Wang, William S.-Y. (1965). Two aspect markers in Mandarin. Language 41, 457–470. Wellwood, A. (2014). Measuring predicates: University of Maryland. College Park PhD dissertation. Wellwood, A., Hacquard, V., & Pancheva, R. (2012). Measuring and comparing individuals and events. Journal of semantics, 29(2), 207-228. Wheeler, S. C. (1972). Attributives and their modifiers. Noûs, 310-334. Willet, T. L. (1988). A Cross-Linguistic Survey of the Grammaticalization of Evidentiality. Studies in Language, 12-51. Wu, J. S. (2008). Terminability, wholeness and semantics of experiential guo. Journal of East Asian Linguistics, 17(1), 1-32. 207 Xiang, M. (2008). Plurality, maximality and scalar inferences: A case study of Mandarin dou. Journal of East Asian Linguistics, 17(3), 227. Yalcin, S. (2007). Epistemic modals. Mind, 116(464), 983-1026. Yalcin, S. (2010). Probability operators. Philosophy Compass, 5(11), 916-937. Yeh, M. (1996). An analysis of the experiential guo exp in Mandarin: A temporal quantifier. Journal of East Asian Linguistics, 5(2), 151-182. Zhang, N. N. (2015). Functional head properties of the degree word hen in Mandarin Chinese. Lingua, 153, 14-41.

Abstract (if available)

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Superlative ambiguities: a comparative perspective

PDF

Comparative iIlusions at the syntax-semantics interface

PDF

Building adjectival meaning without adjectives

PDF

Narrowing the focus: experimental studies on exhaustivity and contrast

PDF

Where number lies: plural marking, numerals, and the collective-distributive distinction

PDF

Reasoning with uncertainty and epistemic modals

PDF

The morphosyntax of states: deriving aspect and event roles from argument structure

PDF

Number marking and definiteness in Bangla

PDF

A unified syntactic account of Mandarin subject nominals

PDF

When things are left unsaid: existential and anaphoric implicit objects in discourse

PDF

Syntactic and non-syntactic factors in reflexive pronoun resolution in Mandarin Chinese

PDF

Processing the dynamicity of events in language

PDF

Reference time in the dynamics of temporal dependency in Korean

PDF

Exploring the effects of Korean subject marking and action verbs’ repetition frequency: how they influence the discourse and the memory representations of entities and events

PDF

Towards the unity of movement: implications from verb movement in Cantonese

PDF

Constraining assertion: an account of context-sensitivity

PDF

Towards a correlational law of language: three factors constraining judgement variation

PDF

Meaningfulness, rules, and use-conditional semantics

PDF

Considering the effects of disfluent speech on children’s sentence processing capabilities and language development

PDF

Silence in answers: a study of ellipsis in Hindi

Asset Metadata

Creator Fang, Huilin (author)

Core Title Subjectivity, commitments and degrees: on Mandarin hen

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Linguistics

Publication Date 04/25/2021

Defense Date 03/05/2021

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag adjectives,commitments,comparatives,degree semantics,epistemic commitments,event semantics,events,formal semantics,gradable adjectives,Linguistics,Mandarin Chinese,OAI-PMH Harvest,positive degree morpheme,positive degree sentences,Scales,semantics

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Pancheva, Roumyana (committee chair), Schein, Barry (committee chair), Li, Audrey (committee member), Mintz, Toben (committee member)

Creator Email demitri1011@gmail.com,fhuilin@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-452495

Unique identifier UC11669034

Identifier etd-FangHuilin-9538.pdf (filename),usctheses-c89-452495 (legacy record id)

Legacy Identifier etd-FangHuilin-9538.pdf

Dmrecord 452495

Document Type Dissertation

Rights Fang, Huilin

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA