The Syncategorematic Nature of Neo-Aramaic and English Antonyms

 

Ala Al-Kajela

McMaster University

 

Antonyms have always been considered the starting point for language learners; therefore, they are familiar cross-linguistically. In this research, we try to provide a semantic description of antonymy in Neo-Aramaic (a member of the Semitic family) as it has not been put under scrutiny in the literature. Second, we analyze the semantic features of Neo-Aramaic antonyms according to two criteria, viz., markedness and committedness. We try to answer questions such as which member neutralizes the opposition in questions and whether nominalizations of these adjectives follow the same pattern as to markedness and committedness. The study sheds some light on the universality of these criteria and how they correlate in some cases but dissociate in others. Our analysis is, in most part, context-bound and shows that adjectives tend to change their semantic features due to the influence of the quantified noun. The analysis has revealed some striking differences between Neo-Aramaic and English, for example hot and cold are not prototypical equipollents in Neo-Aramaic. Nominalizations of the adjectives are morphologically derived; suppletive nominalizations do not exist in the grammar of this language. A preference for using yes/no questions has been noticed as a maneuvering technique with some uncalibrated attributes.

 

1. Introduction[1]

 

Opposites can be considered one of the most essential lexical relations in the semantic system of every language. Language users would more easily comprehend and manipulate opposites than other sense relations such as synonymy and hyponymy. In the Book of Genesis, we encounter good and evil as an ideal instantiation of oppositeness. Greek philosophers such as Socrates, Plato, Aristotle and others have also tackled this binary relation. In Phaedo, one of Plato’s dialogues on soul’s immortality, Plato (360 B.C.) argues that everything has a quality and comes to be from the opposite of that quality; therefore, something that comes to be larger must have been smaller and if someone is alive, they must have been dead before, and will return to being dead later. In addition, there are processes involved in these opposite states for example the processes increase / decrease and coming -to-life / dying accompany larger / smaller and alive / dead respectively.

 

In chapters 10 and 11 of Categories, Aristotle (350 B.C.) gives a more delineated description than that of Plato and classifies opposites into four major categories: (i) correlatives to one another, (ii) contraries to one another, (iii) privatives to opposites, (iv) as affirmatives to negatives. Here, we give a brief account of Aristotle’s insights. First, Aristotle introduces double as the opposite of half and he considers them as interdependent terms because they refer to one another. In other words, if X is the double of Y, then Y is half of X. Second, contraries are non- relational because the existence of something cold does not necessitate or guarantee the existence of something hot. A pair of contraries can be: a) binary if there are two terms and no other options between them b) necessary where specific things can have one of two appropriate properties but not both and never neither. There is compelling evidence that Aristotle treats such opposites as having positive and negative poles; cold is symbolically associated with death whereas hot is associated with life. Similarly, old and dry are conceived of as negative whereas young and wet as positive. Third, this category can also be referred to as possession and deprivation as it implies natural features that things possess. Aristotle introduces sight and blindness as an example for this category. According to Aristotle, this category refers to one thing that either is deprived of or possesses a natural feature, for example, a man can either see or be blind and thus these two features are opposite for him. It is essential to state that being deprived of a certain feature is not the same as not having it, only those things that are supposed to have sight can be deprived of it. Therefore, doors cannot be blind because they do not have sight in the first place. Fourth, unlike the other categories, the last category implies pairs of sentences or statements where one of them is affirmative and the other is negative. In other words, the relation that holds between these full sentences is contradictory. What makes this category distinct from the others is that one of the statements is always true and the other is always false, for example, Socrates is standing or he is not. The other opposites can be used in contradictory sentences, but the output is not always one true and the other false.

 

2. Oppositeness of meaning: some preliminary observations

 

In ordinary parlance, fuzziness has characterized the use of the term antonymy as nonprofessionals, dictionaries and thesauri have used it as a cover term for oppositeness of meaning (in the broad sense of the term). This fuzziness stems from the inclusion of other opposite relations such as converses exemplified by follow: precede; reversives as in rise: fall; and antipodals as in top: bottom. In this paper, antonymy is going to be discussed in its narrow sense following Cruse (1976, 1986), Lehrer and Lehrer (1982), Lehrer (1985), and Lyons (1963, 1968, 1977). Antonymy is a binary or dichotomous relation that holds between gradable adjectives. It is a binary relation manifesting opposition between two terms on a single scale or dimension. Gradability implies that the two terms denote a variable property or an attribute envisaged on a virtual scale with two endpoints. These endpoints are sometimes interposed by a neutral area (i.e. less extreme terms) that cannot be modified by qualifiers such as very, fairly, extremely but they are normally used with comparative and superlative forms.

 

(1) John is short.

(2) John is very tall.

(3) My truck is extremely heavy.

(4) This pot is hotter than this one.

(5) She does not like sweet foods.

(6) Lily is more beautiful than Susan.

(7) The water is very/extremely warm.

 

2.1 Contrariety and contradiction

 

The opposed, for example long: short, hot: cold and heavy: light, terms constitute a contrariety relation. Two terms, contrary and contradictory, originated in logic are used to draw a line between gradable and ungradable antonyms. Two terms are contrary when both cannot be true yet could possibly both be false: John is tall and John is short cannot both be true, but can both be false: John is neither tall nor short is a well-formed sentence. Contrariety relations are characterized by the following properties (entailments):

 

(8) (John is X) → (John is ¬Y) e.g. John is tall → John is not short.

(9) (John is ¬X) ↛ (John is Y) e.g. John is not tall↛ John is short.

(10) (John is Y) → (John is ¬X) e.g. John is short → John is not tall.

(11) (John is ¬Y) ↛ (John is X) e.g. John is not short ↛ John is tall.

 

Contradictory terms, on the other hand, cannot both be true and neither can they be both false, but one of them is definitely true:

 

(12) (John is X) → (John ¬Y) e.g. John is alive → John is not dead.

(13) (John is ¬Y) → (John is X) e.g. John is not dead → John is alive.

(14) (John is ¬X) → (John is Y) e.g. John is not alive → John is dead.

(15) (John is Y) → (John is ¬X) e.g. John is dead → John is not alive.

 

Accordingly, one of the propositions in (16) must be true, (17) is semantically impossible whereas (18) is semantically acceptable:

 

(16) John is dead and John is alive.

(17) #John is neither dead nor alive.

(18) John is neither short nor tall.

 

Therefore, gradable antonyms are contrary to each other whereas ungradable antonyms or complementaries (see Lyons 1968: 461) are contradictory. Succinctly, complementaries are not par excellence opposites as they lack mid terms such as warm: cool and they can neither be normally modified by intensifiers nor can they be used in the comparative and superlative degrees; therefore, the following sentences are interpretable but would still sound odd to the ears of an ordinary English speaker:

 

(19) #John is very dead.

(20) #She is more married than her sister.

(21) #Lily is extremely female.

 

It is worth noting that Cruse (1980) has noted that there is a set of adjectives that can be subsumed under the category of complementaries and he dubbed it gradable complementaries. This subcategory includes adjectives such as clean: dirty, safe: dangerous, pure: impure, rough: smooth, etc. However, it is undeniable that these adjectives are underlyingly gradable and readily reveal their antonymic nature as in (22)-(25) but still have the ability to reflect their complementary effect as in (26) and (27):

 

(22) This room is cleaner than that one.

(23) How clean is it?

(24) It is very dirty.

(25) It is neither clean nor dirty.

(26) He took off the dirty shirt and put on a clean one.

(27) This shirt is clean.

 

In a prototypical antonymic pair, one term has “a positive and the other a negative polarity” (Lyons 1968: 467). In other words, there is a presupposition that the property exists to a greater or lesser degree. This means that one term has more quantity or dimension, rather than less, of the scaled property, viz. short has less length than long, heavy has more weight than light. Accordingly, long, heavy, clever and beautiful are termed ‘Q-positive’ (where Q denotes quantity) whereas short, light, stupid and ugly are ‘Q-negative’; hence Q-polarity. However, Q- polarity can also be referred to as ‘natural’ polarity where the absence of the property denotes natural negative and its presence denotes positive (Cruse 1986). The Q-positive terms are further subdivided into: a) neutral measures such as length, weight, size, etc. b) desirable measures properties such as intelligence, beauty, etc. this can be referred to as evaluative polarity (E- polarity) as it is contingent on the speakers’ intuitions and evaluations. Klooster (1972) has termed those in (a) objective gradables since they are measured by using conventional units (they are in fact universal) of measure such as centimetres, kilograms, cubic meters, etc., whereas the terms in (b) have been described as subjective gradables where no such standard units are used. How does this taxonomy apply to the aforementioned gradable complementaries? Consequently, clean, safe, pure, etc. denote the absence of dirt, danger, and blemishes respectively and are thus deemed Q-negative. On the other hand, our evaluations of the inherent semantic properties of these adjectives reveal their E-positive nature. Their counterparts dirty, dangerous, and impure are Q-positive and E-negative.

 

2.2 Committedness and markedness

 

I think a note on the categorization of these adjectives is in order before we lay the subtleties on the table. Here, we retain Cruse’s (1986) classification of antonyms, which was based on the committedness of the comparative forms. Antonyms as in (1)-(3) are termed polar antonyms and can be measured by using conventional units, equipollent (see Trubetzkoy 1939 for terminology) antonymic pairs express emotions and sensations such as those in (4) and (5), and overlapping antonyms, as in (6) and pairs such as kind: cruel, polite: rude, good: bad, are characterized as being evaluatively positive or negative. I repeat these examples below for the sake of convenience:

 

(1) John is short.

(2) John is very tall.

(3) My truck is extremely heavy.

(4) This pot is hotter than this one.

(5) She does not like sweet foods.

(6) Lily is more beautiful than Susan.

 

It is expedient to explain what (un)committed adjectives mean in this context. The adjective is considered committed when its comparative form entails the scaled property and uncommitted when it does not. However, the English layout of (un)committedness system is not necessarily universal but it may overlap with other languages in many respects. In this regard, there is no one-to-one mapping cross-linguistically because antonymous adjectives bear disparate conceptualizations which are highly contingent on culture immersion. Cruse (1976) has already pointed out that terms of polar antonyms are both uncommitted (impartial) as their comparatives yield symmetrical entailments as in (28)-(29). The comparatives do not entail the scaled property as in (30)-(31).

 

(28) John is taller than Jack ╞ Jack is shorter than John.

(29) The red table is heavier than the white one╞ The white table is lighter than the red one.

(30) John is taller than Jack but both are short.

(31) The red table is heavier than the white one but both are light.

 

Equipollent antonyms, as the name of this category suggests, have both terms committed or biased (see Bolinger 1977 for the use of biased); therefore, the entailment in (32) fails and, unlike polar antonyms, the comparative entails the scaled property, which results in the anomaly of (33)-(34).

 

(32) Lily is happier than Sally ⊭ Sally is sadder than Lilly.

(33) #Lily is happier than Sally but both are sad.

(34) #Lily is sadder than Sally but both are happy.

 

The third category, overlapping antonyms, is distinct from the previous ones in that one term is committed and one is impartial; the entailment is unidirectional. This asymmetry stems from the failure of the entailment in (35) and the eligibility of (36). The comparative does not entail the scaled property, hence the uncommittedness of (37), but where the committed term is involved, oddity emerges as in (38) and the comparative entails the base form of the adjective.

 

(35) John is kinder than Sam ⊭ Sam is crueler than John.

(36) John is crueler than Sam ╞ Sam is kinder than John.

(37) John’s words are kinder than Sam’s but both are cruel.

(38) #John’s words are crueler than Sam’s but both are kind.

 

Markedness is an inherent feature of antonymic pairs. We will follow the general trend in using this notion. Accordingly, one member of the pair is described as unmarked (e.g. long, heavy) and the other as marked (e.g. short, light), at least in the case of polar antonyms; the former is more frequent and neutral than the latter. It is also clear that the unmarked term has more quantity of the property in question; therefore, long and heavy have quantitatively more length and weight respectively. Lyons (1968), Bolinger (1977) and other linguists have noticed that the unmarked member neutralizes the contrast in questions and nominalizations (i.e. the question is not loaded with any assumptions or suppositions which reveal the identity of the measured property therefore the contrast).

 

(39) How long is it?

(40) Is it long?

(41) How short is it?

(42) Hów long is it?

(43) The length of the movie surprised the audience.

(44) When the doctor checked his weight, he asked him to follow a special diet.

(45) The shortness of the movie surprised the audience.

 

The answer to the questions in (39) and (40) is not restricted to long objects but it includes short objects as well; thus the object of the inquiry “is completely open, or ‘unmarked’, as to the expectations of the inquirer” (Lyons 1968: 466). Lyons (1968) and Ljung (1978) have pointed out that shifting the nucleus stress from the adjective to hów in (42) overrides the unmarked reading in (39) and yields a marked question similar to the one in (41). The normality of (43) and (44) show that nominalization of the unmarked member yields a neutral question; (43) induces two legitimate interpretations a) the movies was long and boring and the other induces b) the movie was extraordinarily short as compared to average movies. The same applies to (44) as weight has not revealed whether the person suffers from anorexia or bulimia. The unmarked member resulting from nominalization of polar antonyms is either a suppletive form (i.e. alien to the paradigm): weight, age and size (temperature is associated with equipollents), or morphologically derived from the scaled property for example length, width, height, depth. We do not get the same effect when the marked member is nominalized as in (45) because there is a presupposition that the movie is short which yields a biased reading. The marked member of polar antonyms is Q-negative whereas the unmarked is Q-positive, but both terms are E-neutral.

 

We noted earlier that members of equipollent antonyms are both committed and thus both yield a biased question when each member operates on its own scale (i.e. the scale is exhaustively used) as in (46)-(48). Some fuzziness deranges the consistency we have already noticed in polar antonyms due to some counterexamples (see Lehrer 1985 for more examples) and contextual pressure. We have noticed above (32)-(34) that happy and sad are both committed and they should have a biased how-question, which is not the case in (49) and (50). The committed term that constitutes the neutral question is usually E-positive; therefore, such a term has a wider scope of interpretation (or a mental parameter) unlike the narrower scope of its peer E-negative term as shown in (51)-(52).

 

(46) How hot is it?

(47) How cold is it?

(48) How beautiful is she?

(49) How happy is Lily?

(50) How just was the ruler?

(51) How sad is Lily?

(52) How cheap was the motel you stayed at?

 

Turning to the other point, pragmatic factors override the biased readings by using nouns with inherent features; for example, lava and North Pole are described as being inherently hot and cold respectively (Ljung 197; Cruse 1976). In this case the antonyms are not operative on the whole scale but rather on a scale peculiar to the specified feature without extending into the part where the other feature is operative on the temperature scale (see Cruse 1992). Consequently, the terms used in (53)-(55) are all unmarked because their equivalent opposing peers are excluded from these readings due to inherency, which restricts the functionality of each term to less than a half of the scale. In other words, these nouns are inherently marked for the property in question and when we associate them with an appropriate adjective, they yield unmarked questions. Inherently negative nouns (i.e. negative connotations) yield anomalous questions when used with E-positive members, such as happy, beautiful, etc. in (56)-(58). These and the following cases in this section delineate the deviant semantic behaviour of antonyms in various contexts. Some aspects of the semantic properties of antonymic adjectives, viz., markedness are contingent on the nouns they modify. The noun functions as a parameter that determines whether the adjective yields a marked, unmarked or even anomalous question. This nominal manipulation motivates and explains the syncategorematic leaning of antonyms and undergirds the claim that antonyms context-sensitivity can be a universal feature, as we shall see in next section.

 

There are few cases where one of the terms is uncommitted, but they both yield a biased question, as noted by Lehrer (1985). Consider the example in (59) and (60) where wine can be either sweet or dry but the inquirer presupposes the domination of sweetness in the former and dryness in the latter. However, the questions in (61) and (62) override this prediction by using an inherently sweet and dry items respectively, which neutralize the question (i.e. the opposition), as the inquirer does not have to form any presuppositions. Succinctly, Lehrer’s prediction is not borne out completely.

 

(53) How hot is the lava?

(54) How cold is the North Pole?

(55) How beautiful is Miss Universe?

(56) #How nice were the accusations?

(57) #How happy were the mourners?

(58) #How beautiful is the crone?

(59) “How sweet is the wine?” (Lehrer 1985:404).

(60) How dry is the wine?

(61) How sweet is that honey?

(62) How dry is that Chardonnay?

 

As mentioned earlier, overlapping antonyms are evaluative in nature; one term is evaluatively positive and the other is negative, for example, good, kind, polite and pretty are positive whereas their opposing counterparts are negative. In line with Cruse’s (1976) prediction, the positive term is uncommitted and yields a neutral question as in (63) but the negative committed term yields a biased question (64). In accordance with the equipollents, nouns inherently marked for a specific property, which varies from personal matters to natural phenomena (assault, earthquake, flood, accident, famine, etc.), yield unmarked questions when either term is used (65) and (66). In addition, evaluatively positive terms collocate oddly with inherently bad nouns (67) and (68). Some speakers, due to cultural, social, or occupational factors can relativize inherently bad nouns. It would sound quite normal for two detectives to utter a sentence like the one in (69), but unlikely for the parents of an abused child.

 

(63) How good is the news?

(64) How bad is the news?

(65) How rude was that boor?

(66) How polite was that gentleman?

(67) #How good was the earthquake?

(68) # How polite was that boor?

(69) How bad was the assault?

 

English does not have a complicated morphological system, which makes it an effortless job to attach antonymic prefixes such as un- or dis- to adjectives in order to create their opposite. Zimmer (1964) has noted that there is a tendency that bans the use of negative affixes with adjectival stems that are evaluative negative. This restriction governs the application of such affixes to E-negative opaque adjectives (i.e. morphologically underived) such as sad, false, rude, etc. but applies safely to their unmarked E-positive counterparts: unhappy, untrue, and impolite. However, this restriction cannot be generalized to apply across the board to unmarked terms such as good and pretty.

 

3. Neo-Aramaic antonymous adjectives

 

Unlike other Semitic languages, viz., Arabic and Hebrew, Neo-Aramaic[2]dialects of Iraqi Christians have not received the required scholarly attention to survive the looming extinction. However, I must admit that there is an extensive literature on Neo-Aramaic. Unfortunately, most of this literature, compared to Arabic, Hebrew and Jewish Neo-Aramaic dialects, focuses on historical and descriptive aspects of language (to underpin this claim see Cohen 2012; Hoberman 1988, 1989; Khan 1999, 2002, 2004, 2008, 2009; Kroktoff 1982; Mutzafi 2004, 2008; Napiorkowska 1989 to name a few). Recently, researches by prominent linguists have heralded a new era in tackling Neo-Aramaic. In other words, these scholars have taken Neo-Aramaic to a higher level of linguistic analysis and theoretical research (for more on related topics see Coghill and Duetscher 2002; Doran and Khan 2012). Other fertile fields of linguistic research such as language processing, bilingualism, language acquisition are yet to be investigated.

 

The first part of the study focuses on gathering adjectives from native informants. These informants come from two small Iraqi northern towns: Bartella and Mangesh. Dialectal variation does not have a significant impact on the results of the current study and falls beyond its scope. Furthermore, this variation has no effect on the linguistic intuition of the informants. Twenty native speakers volunteered to participate in the present study (ten males). Participants have been interviewed individually to provide adjectives and rate sentences on acceptability scale. Each interview lasted between sixty and ninety minutes.[3] The range of their ages has been between 40 and 60 years. Age is an important factor in reducing Arabic interference, because Arabic has significantly affected the younger generation’s mother language (i.e. Christian Neo-Aramaic in Iraq). This can be attributed to the fact that this variety is only spoken and has not developed from the written standard variety, which is used and taught in churches. On the other hand, most, if not all, Neo-Aramaic speakers learn and use Arabic in their daily interactions as it is the official language of schools. Neo-Aramaic speakers constitute ethnic minority within national entities with hegemonic majorities (i.e. Arabs and more recently Kurds).

 

The gathered list of adjectives reveals crucial facts about the primitive nature of this variety. The bulk of the collected adjectives is related to the basic needs of their daily life; if push comes to shove they have recourse to their second language (i.e. Arabic).

 

3.1 Polar antonyms

 

The second part of the study deals with the applicability of committedness and markedness criteria. In case of polar antonyms, we refer to the degree of salience of the scaled properties rather than their presence or absence. Both members of the pair are uncommitted, for example (70) and (71), due to the normal reversibility of the comparative form (cf. 28 and 29). We need to point out that bɪš ‘more’ is used to form the comparative degree with both masculine and feminine forms of adjectives though we tried to stick to masculine adjectives for the sake of consistency. However, there was one counterexample to this pattern; the majority of the speakers considered (72) normal but found (73) odd. For Neo-Aramaic speakers, the inherent attributes of raqiqa[4] ‘shallow’ blocks any kind of association with things characterized by being deep. raqiqa in this sense is analogous to equipollent antonyms (see Table 1 below).

 

(70)	ʔāða guda bɪš jārixale māða bas tɪrwaθlehen kɪrjena.
	‘This wall is longer than that one but both are short’.

 

(71)	ʔāða guda bɪš kɪrjale māða bas tɪrwaθlehen jarixena.
	‘This wall is shorter than that one but both are long’.

 

(72)	ʔāða bɪš ʕamuqale māða bas tɪrwaθlehen raqiqena.
	‘This is deeper than that one but both are shallow’.

 

(73)	#ʔāða nera bɪš raqiqale māða bas tɪrwaθlehen ʕamūqena.
	‘This river is shallower than that one but both are deep’.

 

The marked member denotes less of the scaled property whereas the unmarked denotes more. In other words, the latter is Q-positive and the former is Q-negative. The unmarked member neutralizes the opposition in questions and nominalizations. In (74), the inquirer does not carry any presupposition as to the length of the object in question, which has the flexibility to range from centimetres to kilometers relative to what is being inquired about. Therefore, the object referred to can be either long or short- the answer to such questions encompasses a full range of long and short things; therefore, tall is operative on the entire scale (see Diagram 1). On the other hand, the inquirer presupposes that the measured object in (75) is short rather than long relative to other objects belonging to the same field (i.e. the negative member is not operative on the whole scale). We will show below, from a theoretical point of view, how the marked member becomes operative on half of the scale when the quantified noun has some inherent properties.

 

(74)	māqadale jārixa?
	‘How tall/long is he/it’?

 

(75)	māqadale kɪrja?
	‘How short is he/it’?

 

A hybrid construction, namely (76) and (77), places the nominal in a question. The same effect of (74) and (75) as to the assumptions of the inquirer is replicated here. Unlike English, Neo-Aramaic does not have suppletive nominals such as weight, size, speed, etc. Where a nominalization of the antonym is required, it is morphologically derived from the adjective; otherwise, a structural gap emerges. Half of the marked adjectives do have derived nominalizations. In measure phrases, the nominalization of the unmarked member in (78) neutralizes the contrast whereas the marked one in (79) does not. It is worth pointing out that there is an interesting morphological phenomenon here; the nominalizations behra ‘light’ and rɪћqa ‘farness’ share the characteristic of lacking the nominalization affix ūθa which the rest of the adjectives have (see Table 2). This morphological asymmetry explains the anomalous use of these nominalizations in questions (80 and 81) and measure phrases (82).

 

(76)	māqadala jrxūθeh?
	‘What is its length’?

 

(77)	māqadala kɪrjūθeh?
	‘What is its shortness’?

 

(78)	jrxūθɪd gūda ṭlāθa metārela.
	‘The length of the wall is three meters’.

 

(79)	kɪrjūθd gūda ṭlāθa metārela.
	‘The shortness of the wall is three meters’.

 

(80)	#māqadale behreh?
	‘What is its lightness’?

 

(81)	#māqadale rɪћqeh?
	‘What is its farness’?

 

(82)	#rɪћqeh ʔarba metārale.
	‘Its farness is four meters’.

 

Table 1 Polar antonyms and committed members.

[Number 1 indicates that the member on the left is committed, number 2 refers to the other member, and number 3 indicates that both are uncommitted.]

 

We have already stated that the unmarked (i.e. the Q-positive) member in how-questions does not provide any clue to form presuppositions as to which part of the scale is involved. The answer to such a neutralized question accepts both ends of the scale. On the other hand, the Q- negative member carries a presupposition as to which part of the scale is involved thus the markedness. Cruse (1995: 120) has pointed out that the relative scale of polar antonyms “bears some resemblance to an equipollent system, in that it has two counter-directional subsidiary scales with a kind of neutral zone in between”. The relative scale of length, for example, has two gradable members (i.e. long and short), each of which operates on its respective subsidiary scale, and the member denoting more of the scaled property names the absolute scale. Relative scale is potentially similar to Diagram 2. Absolute scale is calibrated in conventional units of measurement (e.g. foot, yard, meter, gram, etc.), the salient property names the whole scale, and the scale has an end-point as in Diagram 1. That said, it is theoretically appealing to assume that the marked term of polar antonyms, for example kɪrja ‘short’, qalūla ‘light’, raqiqa ‘thin’, etc. can yield a neutralized question in the same fashion of marked terms of equipollent antonyms (bad, rude, cruel). It is well motivated that two negative qualities collocate normally to produce a neutral question. By virtue of this assumption, midget, feather and paper are, by definition, inherently marked for the quality in question (i.e. shortness, lightness and thinness). By this logic, kɪrja ‘short’ in this case is operative on the neutral-short scale rather than tall-short scale, viz., the whole scale. In other words, it is not operative on the whole scale. The only difference between the two is that the marked equipollent term is committed and negative whereas the polar term is marked but uncommitted. It is an essential characteristic feature of polar antonyms that the marked member, even when nominalized, is refractory to neutralization due to the presuppositions accompanying it (see 75, 77, 79)- contrary to Ljung’s (1974: 86) claim, “a marked noun like shortness can become unmarked”. Possible nominalizations of the marked members maintain their marked identity in how-questions, but most informants rated (87) unacceptable, which resonates with the natural corollary of the morphological gaps in Table 2. Neo-Aramaic speakers have not found (83-86) odd and they kept the essence of their interrogative setup intact. They have been inquiring rather than exclaiming, as they already knew about the association between the scaled property and the quantified noun. However, these examples are apparently marked because the scaled property is Q-negative.

 

(83)	māqadale kɪrja ʔāða qazam?
	‘How short is this midget’?

 

(84)	māqadala raqɪqta ʔāði waraqa?
	‘How thin is this paper’?

 

(85)	māqadale qalūla ʔāða parra?
	‘How light is that feather’?

 

(86)	māqadale nqiða ʔāða gðaða?
	‘How thin is this string’?

 

(87)	#māqadala kɪrjūθɪd ʔāða qazqm?
	‘What is the shortness of this midget’?

 

Table 2 Morphologically derived nominals for polar antonyms.

 

3.2 Equipollent antonyms

 

Equipollent antonyms in Table 3 have a common feature that makes them distinct from other antonyms- both members are committed. For the sake of convenience, we repeat the test frame used with polar antonyms to clarify the committed nature of equipollent antonyms. Comparatives of this group are not transposable and thus the entailments in (88) and (89) fail; there is a bidirectional failure of entailments. The anomaly of (90) and (91) is the by-product of malfunctioning entailments and that equipollent comparatives entail the base form of the adjective.

 

(88)	A bɪš ћaluja mɪn B ⊭ B bɪš marirale mɪn A.
	‘A is sweeter than B ⊭ ‘B is bitterer than A’.

 

(89)	A bɪš marirale mɪn B ⊭ B bɪš ћalujale mɪn A.
	‘A is bitterer than B’ ⊭’B is sweeter than A’.

 

(90)	#A bɪš ћalujale mɪn B bas tɪrwaθlehen marirena.
	‘A is sweeter than B but both are bitter’.

 

(91)	#A bɪš marirale mɪn B bas tɪrwaθlehen ћalujena.
	‘A is bitterer than B but both are sweet’.

 

However, the majority of Neo-Aramaic speakers in our sample, unlike the English, did not find the prototypical pair šaxina ‘cold’: qarira ‘cold’ committed in the following test frame in (92)-(95) that reveals the directionality of committedness in English. For the Neo-Aramaic speakers, bɪš šaxina‘more hot’ and bɪš qarira‘more cold’ do not mean hot or cold to a greater degree as intuited by the English, consider the comparison between the English oddity[6] in (96) and (97) and Neo- Aramaic normality in (98) and (99) . Even in the previous test frame, some speakers claimed that šaxina and qarira were still uncommitted.

 

(92)	A bɪš šaxinale mɪn B → B bɪš qarirale mɪn A.
	‘A is hotter than B’ → ‘B is colder than A’.

 

(93)	A bɪš qarirale mɪn B → B bɪš šaxinale mɪn A.
	‘A is colder than B’ → ‘B is hotter than A’.

 

(94)	A šaxinale bas bɪš qarirale mɪn B.
	‘A is hot, but it is colder than B’.

 

(95)	A qarirale bas bɪš šaxinale mɪn B.
	‘A is cold, but it is hotter than B’.

 

(96)	“?It’s hot, but it’s colder than yesterday” (Cruse 1986:207).

 

(97)	“?It’s cold, but it’s hotter than yesterday” (Cruse 1986:207).

 

(98)	ʔɪdju xɪmmale bas bɪš qarθala mɪn tɪmmal.

 

(99)	ʔɪdju qarθala bas bɪš xɪmmale mɪn tɪmmal.

 

In the same vein, bɪš ʧuʕa ‘ more smooth’ does not mean ʧuʕa ‘smooth’ to a greater degree, but implies having a high level of smoothness (see Table 3). This high level of the specified quality allows the speaker a lot of latitude in interpreting the adjective as uncommitted. The entailments in (100) and (101) hold:

 

(100)	A bɪš ʧuʕa mɪn B → B bɪš xɪrxɪsāna mɪn A.
	‘A is smoother than B’ → ‘B is rougher than A’.

 

But this entailment does not necessarily hold:

 

(101)	A bɪš xɪrxɪsāna mɪn B → B bɪš ʧuʕa mɪn A.
	‘A is rougher than B’ → ‘B is smoother than A’.

 

Therefore, the speakers ranked (102) as acceptable:

 

(102)	A bɪš ʧuʕale mɪn B bas tɪrwaθlehen xɪrxɪsānena.
	‘A is smoother than B but both are rough’.

 

But they considered (103) odd:

 

(103)	#A bɪš xɪrxɪsānale mɪn B bas tɪrwaθlehen ʧuʕena.
	‘A is rougher than B but both are smooth’.

 

Neo-Aramaic equipollent pairs share the quality of committedness, which meshes well with markedness. Both members of this group lose the ability to neutralize the opposition in questions because the adjectives in this category describe a built-in quality. As soon as the inquirer formulates his question, a presupposition crystallizes (104) and (105). However, both members retrieve the ability to neutralize the opposition in questions when this built-in quality is used to calibrate the inherency of the quantified noun as in (106)-(108). These are highly contextualized uses and most speakers were skeptical about them at the beginning but rated them acceptable after considering the context of (107). For some speakers, the existence of the quantified noun in such questions partially contributed to their superficial anomaly, because merchants, coffee, and hay are by default rich, bitter, and dry respectively. This equivocality was eliminated when they intuited that the question was not whether the quantified noun has the inherent property or not but whether it exceeds the average. In these cases of neutralizations, the unmarked member is operative on half scale (i.e. the unmarked member does not extend beyond the neutral interval) as shown in Diagram 2. To summarize, when both members are marked, they function on the whole scale; however, when they are coerced into an unmarked context, they become functional on half scale.

 

(104)	māqadale šaxina ʔāða laxma?
	‘How hot is this bread’?

 

(105)	māqadale ћaluja ʔāða ʧaj?
	‘How sweet is this tea’?

 

(106)	māqadale zangin ʔāða tažɪr?
	‘How rich is this merchant’?

 

(107)	Tuma: ʔāðɪ qehwe lagmašta māqadɪla marɪtta.
	Tuma: ‘This coffee is undrinkable because it is very bitter’.
	Behnan:qaj māqadala marɪta (ʔāðɪ qehwe)?
	Behnam: ‘Why, how bitter is this coffee’?

 

(108)	māqadale wiša ʔāða qɪrša?
	‘How dry is this hay’?

 

 

It has been noted that this is not an across-the-board phenomenon, because the informants have failed to provide neutralized questions for some adjectives such as: pšila ‘cooked’: naja ‘uncooked’, swiʔa ‘full’: kpina ‘hungry’. They have also accepted (109)-(111) but found (112) odd.

 

(109)	māqadale kpina?
	‘How hungry is he’?

 

(110)	kpinale?
	‘Is he hungry’?

 

(111)	swiʔale?
	‘Is he full’?

 

(112)	#māqadale swiʔa?
	‘How full is he’?

 

The only counter example we have come across is pṣixa ‘happy’: muqhra ‘sad’. In spite of the fact that both members are committed, the positive member is unmarked (113) in questions whereas the negative member is marked (114).

 

(113)	māqadale pṣixa?
	‘How happy is he’?

 

(114)	māqadale mquhra?
	‘How sad is he’?

 

Table 3 Equipollent antonyms and committed members.

 

We have stated earlier that nominalizations of both members, in Neo-Aramaic, are morphologically derived as illustrated in Table 4. However, nominalizations of equipollent antonyms, unlike their polar counterparts, do not produce acceptable questions (115) and (116). A cogent argument that may explain this phenomenon is that equipollents do not have conventional units of measure. The measurable ones, šɪxnūθa ‘hotness’ and qararūθa ‘coldness’, were an exception to this rule. The former produced a completely acceptable but biased question, whereas the latter was less so (117)-(118). Apparently, heat has more detrimental effect than cold in this culture, as it is associated with hot things such as water, furnaces, and fever; therefore, šɪxnūθa has become more frequent and thus more acceptable. Furthermore, since they conceptualize temperature as incremental rather than decremental value, sentence (119) sounded odd to most of the subjects. When forced into a measure-phrase test (120), these adjectives compelled the speakers to take recourse to Arabic units (i.e. daraža ‘degree’).

 

(115)	#māqadala ћɪljuθɪd ʔāða ʧaj?
	‘What is the sweetness of this tea’?

 

(116)	#māqadala paxuθɪd ʔāða jxāla?
	‘What is the unsaltedness of this food’?

 

(117)	māqadala šɪxnuθɪd māja?
	‘What is the heat of the water’?

 

(118)	māqadala qararuθɪd ʔāni māja?
	‘How cold is this water’?

 

(119)	qararuθɪd māja tɪtʔe daražela.
	‘The coldness of the water is two degrees’.

 

(120)	šɪxnuθɪd māja tɪtʔe daražela.
	‘The hotness of the water is two degrees’.

 

The mid-interval terms are all committed and marked, and they do not have nominalizations except pajuxa ‘cool’. In (121), pajaxūθa ‘coolness’ does not yield an acceptable question because it is operative on the qararūθa ‘coldness’ scale.

 

(121)	#māqadala pajaxuθɪd maja?
	‘What is the coolness of the water’?

 

Table 4 Morphologically derived nominals for equipollent antonyms.

 

3.3 Overlapping (evaluative) antonyms

 

Overlapping antonyms in Table 5 express personal attitudes and judgements. One member is evaluatively positive and has more of the scaled property; the other is evaluatively negative and indicates a decrease in the property. The comparative of the E-positive member is impartial but the E-negative is committed. For this reason, (122) is acceptable but (123) is not. In other words, vain people can be modest but modest people cannot be vain- this explains the overlapping nature of this category (see Diagram 3).

 

(122)	Behnam bɪš makkixale mɪn Tuma bas tɪrwaθlehen rāmānena.
	‘Behnam is more modest than Tuma but both are vain’.

 

(123)	#Behnam bɪš rāmānale mɪn Tuma bas tɪrwaθlehen makkixena.
	‘Behnam is vainer than Tuma but both are modest’.

 

 

Two pairs deviate (see Table 5) from this norm due to their bidirectional uncommittedness (124) and (125). They behave like polars but they are categorized as overlapping antonyms. There are two reasons that may explain this: first, they have evaluative polarity because they depend on speaker’s evaluations and affective behavior; second, the unavailability of conventional measuring units makes them impervious to calibration.

 

(124)	Behnam bɪš qšiṭale mɪn Tuma bas tɪrwaθlehen ðaʕifna.
	‘Behnam is fatter than Tuma but both are thin’.

 

(125)	Behnam bɪš ðaʕifle mɪn Tuma bas tɪrwaθlehen qšiṭena.
	‘Behnam is thinner than Tuma but both are fat’.

 

The E-positive member is unmarked and neutralizes the opposition in questions, whereas the E- negative member is marked and yields biased questions. It is worth mentioning that we have found that how-questions yielded anomalous constructions as in (126)-(128), except for the deviant pair where ʔagran ‘expensive’ and qšiṭa ‘fat’ yielded normal but biased how-questions. In spite of the fact that these two adjectives are evaluative, they can be calibrated because ‘currency’ and ‘pounds’ can respectively modulate ʔagran and qšiṭa and normalize the questions in (129) and (130).

 

(126)	# māqadale tarri?
	‘How fresh is it’?

 

(127)	#māqadale šenāja?
	‘How friendly is he’?

 

(128)	#māqadale žwanqa?
	‘How young is he’?

 

(129)	māqadale ʔagran?
	‘How expensive is it’?

 

(130)	māqadale qšiṭa?
	‘How fat is he’?

 

Neo-Aramaic speakers have shown a consistent tendency in preferring yes/no questions in this category. The unmarked member yields neutral questions (131) and (132) and it is operative on the whole scale (see Diagram 3 above), whereas the marked member yields a biased question (133) and is operative on the whole scale (i.e. šenāja ‘friendly’: kuvi‘unfriendly’ scale). Admittedly, it was hard to engineer examples such as (134) and (135) to neutralize the opposition, but the inherently negative noun mpalta ‘a fall’ and inherently expensive gardāna ‘a golden necklace’ facilitated the process. In (134), pis ‘bad’ is operative on half scale and thus unmarked.

 

Obviously, the positive unmarked members collocate oddly with the inherently negative nouns as in (136). Succinctly, the positive unmarked member is operative on the whole scale, whereas the negative marked member operates on the whole scale but restricted to half scale only when associated with inherently negative noun where it becomes unmarked.

 

(131)	tarrile?
	‘Is it fresh’?

 

(132)	makkixale?
	‘Is he modest’?

 

(133)	kuvile?
	‘Is he unfriendly’?

 

(134)	māqadaja pis mpalteh?
	‘How bad was his fall’?

 

(135)	māqadaju ʔagran gardānah?
	‘How expensive was her (golden) necklace’?

 

(136)	#maqadaja rande mpalteh?
	‘How good was his fall’?

 

Table 5 Overlapping antonyms and committed members.

 

Morphological nominalizations (see Table 6) in this group did not have the same effect as the previous categories did. In other words, our informants have flagged the following as odd:

 

(137)	#māqadala šɪðanūθeh?
	‘How much is his madness’?

 

(138)	#māqadala ramūθeh?
	‘How much is his vanity’?

 

However, we attempted to find out whether these nominalizations are neutral or biased outside the how-question frame. Nominalizations of both the positive and negative members yielded a biased statement.

 

(139)	ћkili ʕan makkixūθa dɪšwāwe.
	‘Tell me about the neighbors’ modesty’.

 

(140)	la kɪmmɪlleli ʕan ʔarzanūθɪd šuqa.
	‘He did not tell me about the cheapness of the market’.

 

(141)	qreli xa kθāwa ʕan (šɪmnūθa) šɪmna.
	‘I read a book on fatness’.

 

Table 6 Morphologically derived nominals for overlapping antonyms.

 

4. Gradable complementaries

 

Neo-Aramaic is not rich in gradable complementaries. This category combines two distinct features and thus allows two interpretations. First, an antonymic one, which takes for granted the existence of the scaled property and demonstrates the ratio of its existence; second, a complementary interpretation traces and verifies the existence of a property. It is important to state that context is an intrinsic factor in establishing this category. A few candidates have been spotted, namely naðifa: šɪxtāna ‘clean: dirty’ and ṣɪpja: hɪršāna ‘pure: impure’. These candidates have already established their antonymic status after passing the criteria of markedness and committedness. naðifa and ṣɪpja are evaluatively positive and denote absence of the scaled property. As a result of being unmarked, they yield neutralized questions. That said, we will provide a piece of evidence to confirm their complementary standing. naðifa and ṣɪpja denote absence of dirt and impurities respectively; therefore, Neo-Aramaic speakers, unlike their English peers (see Cruse and Togia 1995 for more details on English), found (142) and (143) anomalous, which endorses the complementary reading. A characteristic feature of a complementary reading is the absence of a midinterval - things can be either clean or dirty.

 

(142)	#lele la naðifa wala šɪxtana.
	‘It is neither clean nor dirty’.

 

(143)	#lele la ṣɪpja wala hɪršana.
	‘It is neither pure nor impure’.

 

The context of the conversational stages in (144) reinforces the complementary interpretation of naðifa.

 

(144)	a.	lele naðifa dex ʔaxlɪnne.
		‘It is not clean, how can I eat it’?

 

	b.	taqriban naðifale.
		‘It is almost clean’.

 

	c.	dāha naðifale.
		‘Now it is clean’

 

5. Nongradable complementaries

 

Many linguists, if not all, have agreed to characterize some adjectives, such as dead, alive, atomic, pregnant, etc. as non-gradable because they resist intensification, modification and the use in comparative form, such as (145) and (146):

 

(145)	# Jack is very dead.
(146)	# This bomb is more atomic than this one.

 

They claimed that such adjectives do not have a mid-interval, unlike the well-established adjectives in the literature; gradable adjectives can be represented on a continuum, such as hot, warm, cool, cold.

 

However, we can still find instances of those adjectives in our daily interactions where they can be intensified or used in the comparative form. It is noteworthy that in (147), (148) and (149) the bolded adjectives are still directly correlated with their intrinsic semantic features and they have not acquired any external or superficial implications unlike those in (150)-(155), which are in fact metaphors.

 

(147)	a.	You said that Jack was half dead and was taken to hospital yesterday, but he was very alive when I met him today.
	b.	Yeah, yesterday he was barely alive.
(148)	Catherine is more pregnant than Sarah is. She is pregnant with quads whereas Sarah is just pregnant with her second baby.[7]

(149)	A very atomic bomb was built in 1950s. (One that causes more destruction than a less atomic bomb).
(150)	The situation is pregnant with danger for the future (=full of risk)
(151)	That’s a pregnant decision (=fruitful).
(152)	Chomsky has a mind pregnant in ideas (=creative).
(153)	Plants never grow in dead soil (=barren).
(154)	The volcano is dead (=dormant).
(155)	The killer was dead to her plea for mercy (=unresponsive).

 

It has been argued that such sentences are rendered correct because the pragmatic component of the language has converted their inherent lexical properties into other context-dependent ones. In other words, we are not grading the adjectives themselves, but their connotations or secondary implications (Lyons 1977: 278). Therefore, Lyons (1977), Lehrer (1982) and many others have restricted the antonymic relation between dead and alive to a postulate claiming that dead applies only to things or humans that were once alive:

 

(156)	X is dead ↔︎ X is not alive.
(157)	X is not dead ↔︎ X is alive.

 

This analysis is untenable as lexical items gain their semantic quality from their appropriate linguistic context and not from occurring in isolation. In (147) we are still referring to the biological status of the person in question (cf.149-155). It is also clear that such terms neither allow an exhaustive set of modifiers nor do they permit comparative and superlative forms; therefore, it is semantically impossible to have the following:

 

(158)	#Tuma bɪš miθale mɪn Behnan. (Metaphorically acceptable)
	‘Tom is deader than John’.

 

(159)	#Tuma bxa ga miθale. (Metaphorically acceptable)
	‘Tuma is extremely dead’.

 

 

Neo-Aramaic speakers did not provide a strong evidence to refute the aforementioned state of affairs. All the respondents rated sentences like (147), (148) and (149) unacceptable; however, 45% of them accepted miθa ‘dead’ but just 15% found bɪxajʔe ‘alive’ normal in similar scenarios. Their bias towards miθa stems from their conceptualization of this word. To them miθa is split up into two halves: one is lifeless and the other has a half-life (see Diagram 4). It is a well-motivated analysis, as bɪxajʔe is Q-positive and E-positive, whereas miθa is imbued with negative attributes and evaluations. When the context was dexterously manipulated, Neo-Aramaic speakers have, unexpectedly, rated (160) acceptable- the complementary (default) reading was overshadowed by the antonymic one.

 

(160)	lele la miθa wla bɪxajʔe.
	‘He is neither dead nor alive’.

 

A closer look at the adjectives in (146)-(148) would reveal that their inherent lexical meaning has not shifted due to the context and they are still interpreted as carrying their main sense. They can be considered non-prototypical type of gradable antonyms like vigorous/ feeble or a constrained sub-type of gradable adjectives. However, this gradability has been repudiated as being the side effect of a pragmatic interpretation or the context that plugs in an alien implication triggering a deviation from the inherent lexical properties of this item. This departure from the kernel to the peripheral semantic of the items in question can be exemplified by (149)-(154).

 

6. Conclusion

 

Some adjectives are not used in their absolute sense, their interpretation may vary from culture to culture, and from one context to another, and from one speaker to another. Thus, the average height of a Dutch male is 6ft and 1/2’’ whereas in India the average is 5 ft and 3’’. That said, a relatively tall Indian male would be relatively short compared to his Dutch peer and, similarly, a short Dutch male would be relatively taller than his Indian peer would. However, such facts do not affect polars as they operate on a unidirectional scale. On the other hand, such influence is clearly noticed in equipollents and a good example is the temperature scale. In Iraq, for example the average high temperature in summer is 114° F whereas in Ontario/Canada the average is 80°F. Again, what is considered summer for a Canadian, would be spring season for a Kuwaiti. This kind of shift within the same semantic scale depends on the context in toto. This kind of conceptualization has turned English prototypical equipollents hot and cold into an atypical case. Neo-Aramaic speakers apply a kind of psychological grading with a reference to a norm built on comparison. They base their analysis on a relative norm while the English apply a logical grading that is contingent on a more standardized norm (see Sapir 1944 for more on English grading). Any natural language speakers will employ various lexical items to get their communicative content across. This employment does not transpire in rigid isolation and fixed conceptual framework but rather in a variety of contexts and divergent mental images, which shape and contribute to the semantic structure of adjectives.

 

Polar antonyms can be considered a prototypical category; both members are uncommitted and the unmarked member yields a neutralized question. Nominalizations of the unmarked member have produced neutralized questions. On the other hand, the frequent structural gaps and asymmetrical morphological derivations (even for the unmarked member) explain the biased or anomalous questions formulated by morphologically derived nominalizations.

 

The committedness and markedness of equipollents generate biased questions. One pair turned out to be an exception (i.e. happy: sad); both members are committed, yet the positive unmarked member yields a neutralized question. Moreover, at least one counterexample (hot: cold) has been provided where both members are uncommitted but they are still marked. In this category, nominalizations do not yield neutralized questions. Markedness ceases to exist due to the inherent attributes of the quantified noun which collocates well with the adjective as in tažɪr zangin ‘rich merchant’ qehwe marɪtta ‘bitter coffee’ qɪrša wiša ‘dry hay’.

 

Evaluative antonyms have shown a strong correlation between markedness and committedness.

 

The E-positive member is uncommitted and consequently yields impartial questions whereas the negative is committed and does not neutralize the opposition. The evaluative nature of this category precludes how-questions but approves of the straight interpretation of yes/no question. Besides, two pairs in this category pattern in conformity with polars (i.e. both members are uncommitted and the Q-positive member neutralizes the opposition).

 

Inherency (i.e. context) can neutralize the opposition whether the member is committed ‘pis ‘ or ‘ʔagran’ uncommitted. However, nominalizations of both the positive and negative members yield anomalous questions but acceptable biased statements.

 

In spite of their considerably small number, gradable complementaries provide a compelling piece of evidence in support of the syncategorematic nature of adjective. This category consists of two genuine overlapping antonyms, which readily take on a complementary interpretation when their context is subject to modification.

 

Nongradable complementaries have also provided a strong piece of evidence that context is a powerful tool which can reshape the semantic features of adjectives. miθa ‘dead’ is applicable to organisms which have had life at some point; therefore, it is logically acceptable for life to diminish gradually as death advances, which makes the use of ‘half’ quite appropriate. On the contrary, bɪxajʔe ‘alive’ does not enjoy this privilege, thus the use of ‘half’ is prohibited. We predict that this semantic property, which is pertinent to death, can be universally motivated.

 

 

References

 

Aristotle. 350 B.C. Categories. Translated by E.M. Edghill. The internet Classic Archive <http://classics.mit.edu/Aristotle/categories.3.3.html>.

Bolinger, D. 1977. Neutrality, norm, and bias. Bloomington: IUlC.

Cohen, Eran. 2012. The syntax of Neo-Aramaic: The Jewish dialect of Zakho. NJ: Gorgias Press.

Coghill, E. and Deutscher, G. 2002. The origin of Ergativity in Sumerian, and the ‘inversion’ in pronominal agreement: a historical explanation based on Neo-Aramaic parallels. Orientalia 71: 267-290.

Cruse, David. 1976. Three classes of antonymy in English. Lingua 38: 281-91.

Cruse, David. 1980. Antonymy and gradable complementaries. In Kastovsky, D. (Ed.), Perspektive der lexikalischen Semantik (pp.14-25). Bonn: Bouvier Verlag Herbert Grundmann.

Cruse, David. 1986. Lexical semantics. Cambridge: Cambridge University Press.

Cruse, David. 1992. Antonymy revisited: Some thoughts on the relationship between words and concepts. In Adrienne Lehrer and Eva Feder K. (eds.) Frames, fields and contrasts: New essays in semantics and lexical organization (pp.389- 306). Hillsdale, NJ: Erlbaum.

Cruse, David and Togia, Pagona. 1995. Towards a cognitive model of antonymy. Lexicology 1: 113-141.

Doran, Edith and Khan, Geoffrey. 2012. The typology of morphological ergativity in Neo- Aramaic. Lingua 122: 225-240.

Hoberman, Robert. 1988. The history of the Modern Aramaic pronouns and pronominal suffixes. Journal of the American Oriental Society 108: 557–575.

Hoberman, Robert. 1989. The syntax and semantics of verb morphology in Modern Aramaic: A Jewish dialect of Iraqi Kurdistan. New Haven, CT: American Oriental Society.

Khan, Geoffrey. 1999. A grammar of Neo-Aramaic: The dialect of the Jews of Arbel. Leiden: Brill.

Khan, Geoffrey. 2002. The Neo-Aramaic dialect of Qaraqosh. Leiden: Brill.

Khan, Geoffrey. 2004. The Jewish Neo-Aramaic Dialect of Sulemaniyya and Halabja. Leiden: Brill.

Khan, Geoffrey. 2008. The Neo-Aramaic dialect of Barwar: Grammar, Vol. 1. Leiden: Brill.

Khan, Geoffrey. 2009. The Neo-Aramaic dialect of Sanandaj. NJ: Gorgias Press.

Klooster, W.G. 1972. The structure of underlying measure phrase sentences. Dordrecht: Reidel.

Krotkoff, Georg. 1982. A Neo-Aramaic dialect of Kurdistan: Texts, grammar, and vocabulary. New Haven, Connecticut: American Oriental Society

Lehrer, A. & Lehrer, K. 1982. Antonymy. Linguistics and philosophy 5: 483-501.

Lehrer, A. 1985. Markedness and antonymy. Journal of linguistics 21: 397-421.

Ljung, M. 1974. Some remarks on antonymy. Language 50: 74-88.

Lyons, John. 1963. Structural semantics. Cambridge: Cambridge University Press.

Lyons, John. 1968. Introduction to theoretical linguistics. Cambridge: Cambridge University Press.

Lyons, John. 1977. Semantics. Cambridge: Cambridge University Press.

Mutzafi, Hezy. 2004. The Jewish Neo-Aramaic dialect of koy Sanjaq. Wiesbaden: Otto Harrassowitz.

Mutzafi, Hezy. 2008. The Jewish Neo-Aramaic dialect of Betanure. Wiesbaden: Otto Harrassowitz.

Napiorkowska, Lidia. 1989. A grammar of the Christian Neo-Aramaic dialect of Diyana- Zariwaw. Leiden: Brill.

Plato. 360 B.C. Phaedo. Translated by Jowett, Benjamin. The internet Classic Archive <http://classics. mit.edu/Plato/phaedo. html>.

Sapir, Edward. 1944. Grading, a study in semantics. Philosophy of science 11: 93-116.

Zimmer, K.E. 1964. Affixal negation in English and other languages: An investigation of restricted productivity. Monograph Supplement 5 to Word, 20