Directionality in affixation: the applicability of Marchand’s (1964) semantic criteria
Abstract
7
Full text
Directionality in affixation: the applicability of Marchand’s (1964) semantic criteria Alba E.Ruz (University of Granada)— Bożena Cetnarowska (University of Silesia in Katowice) Alba E.Ruz— Bożena Cetnarowska ABSTRACT This paper presents astudy of directionality in word-class changing affixation. Specifically, it addresses the various semantic criteria proposed in Marchand (1964) for directionality in conversion as applied to asample of affixation in English. The semantic criteria by Marchand have frequently been pointed out in the literature but have not been studied in detail in either conversion or affixation. The aim is to determine whether these criteria are applicable in overt affixation, where directionality does not seem to pose aproblem (because, unless back-formation is attested, the affix signals the directionality of the process), or whether the criteria fail to apply even to these cases. KEYWORDS affixation, conversion, directionality, lexical semantics, semantic criteria DOI https://doi.org/10.14712/18059635.2023.1.1 1 INTRODUCTION Directionality remains adifficult issue in many cases of English word formation, e.g., smilen / smilev (smilev > smilen or smilen > smilev), debonedadj (de- + bonedadj or debonev + -ed), or reconsiderationn (re- + considerationn or reconsiderv + -ation) (cf. Plag 2003: 2.4, on multiple affixation). In conversion, this is partly the result of the difficulty inherent in the process, at least under aview that conversion is aword-formation process involving several paronymic lexemes (see e.g., Cruse 1986: 130, 132, for adescription of paronymy). In both conversion and affixation, this is also partly due to the criteria used in the identification of directionality, and the level of application of the criteria (lexeme vs. sense) (cf. Plank 2010). This paper tests Marchand’s (1964) directionality criteria in asample of wordclass changing affixation in English, with the following aims: i. To study whether Marchand’s (1964) semantic criteria for directionality prove applicable in asample of word-class changing affixation in English. ii. To test how measurable the criteria are when considering senses, e.g., how true it is for abase to show fewer restrictions of usage than its derivatives. Testing Marchand’s criteria in asample of affixation in English is justified by the fact that conversion has often been described by analogy to other word-formation proOPEN ACCESS
8 LINGUISTICA PRAGENSIA 1/2023 cesses, mainly to affixation.1 However, word formation by affixation may or may not react to directionality tests similarly. This is also arelevant issue in the description of present-day English, especially because the (non-)applicability of the criteria brings forward various possibilities yet to be studied, each with its consequences for the description of English word formation today and, in particular, for the status of conversion: i. If the criteria prove applicable in affixation, it may be expected that they be applicable in conversion too, at least under aview that conversion and affixation are parallel processes in that they may have similar functionalities in language and produce semantically equivalent outputs. ii. If the criteria do not prove applicable in affixation, it may then be expected that these criteria would neither be applicable in conversion. The paper is structured as follows: Section 2 outlines the theoretical background for the study. Section 3 elaborates on the method employed for the extraction of the derivational paradigms and for the analysis of the criteria for directionality. Section 4 presents the results of the study. Section 5 discusses the results, and Section 6 offers the conclusions of the study. 2 DIRECTIONALITY IN ENGLISH WORD FORMATION Various criteria have been proposed in the literature for the identification of directionality in word formation in unclear cases. They include diachronic criteria relying on historical evidence and attestation dates to decide which lexeme2 is basic (as in Biese 1941). They also contain synchronic criteria with an emphasis on the use and semantics of the lexemes involved (as proposed by Marchand 1963, 1964). Since then, several studies have taken on the issue from various perspectives and in various theoretical frameworks (see e.g., Cetnarowska 1993, Štekauer 1996, Twardzisz 1997, Plag 2003, Balteiro 2007, Bram 2011, Lohmann 2017, on English, Don 2003, based on phonological and morphological properties in Dutch, Plank 2010, on German, but potentially extendable to other languages, or Tribout 2020, on French, among others). However, to the best of our knowledge, no solid method to identify directionality has been found and directionality remains adeadlock in many respects (cf. Bauer and Valera 2005: 11). This is because, interestingly, the criteria tested in the literature do not seem to agree, and have typically been found to contradict one 1 This study makes sense within aframework which considers affixation and conversion as parallel processes in the sense that both are asymmetrical and have similar functionalities. However, in order not to bias the analysis and because the aim of this study is to test the criteria for directionality in affixation, conversion is not included in the derivational paradigms, only word-class changing affixation is represented. 2 The term lexeme is used here in the traditional sense to refer to words assigned to one word class only, e.g., loven and lovev being two different lexemes (cf. Lyons 1977: 423–424). OPEN ACCESS
ALBA E.RUZ— BOżENA CETNAROWSKA 9 another, indicating opposite directions in many cases (cf. e.g., Bram 2011, for areview of the diachronic and synchronic criteria used in the literature and their degree of agreement in English). This fact supports the idea that directionality is not easily predictable, and that areliable method to tell directionality is yet to be found. Thus, considering that no experimental research has confirmed directionality convincingly and that there is little evidence on theoretical or empirical grounds to confidently claim adirection in unclear cases, this paper aims at testing amethod of assessing directionality by applying aseries of semantic criteria in asample of affixation in English, for which, initially, directionality does not seem to pose aproblem. Specifically, taking asynchronic approach, it is here believed that alook at the semantic complexity of the lexemes involved may help predict the direction in derivation, as it would be expected, e.g., that the semantically more complex lexeme in apair be derived from the semantically less complex one. The focus is on Marchand’s (1964) semantic (or ‘content’) criteria for directionality (see Section 3 for abrief description of the criteria used in this study and their application). Another key aspect of this study is the adoption of Plank’s (2010) claim that directionality should be studied at the level of senses rather than at the level of lexemes, because several directions (and, thus, several bases and derivatives) may be found for the various senses which develop between apair of lexemes. Similarly, Marchand claims that “aword may be aderivative in one sense and not in another” (1964: 12), while Quirk et al. affirm that “conversion shows lexicalization having specific sense orientation, in that only aparticular sense of aword may be converted to another word class” (1985: 1529). Although these claims are widely accepted, few studies pay attention to senses when commenting on the issue of directionality (cf. Ševčíková 2021). This study tests the semantic criteria proposed by Marchand (1964) (specifically, semantic dependence, semantic pattern, semantic range, and restrictions of usage, hereafter SD, SP, SR, and RU, respectively) at the level of senses, while corpus data (range of registers in which alexeme is used and frequency of occurrence, hereafter Reg and Freq) is given for lexemes as awhole. 3 METHOD This section presents the method for the analysis of Marchand’s (1964) criteria in this paper. Specifically, Section 3.1 describes the decisions taken for the data sample selection and extraction, while Section 3.2 describes and exemplifies how the analysis of the criteria for directionality was carried out. 3.1 DATA SAMPLE SELECTION AND EXTRACTION The data sample in this study includes 30 underived bases: 10 nouns, 10 verbs, and 10 adjectives (see Table 1). These bases are the point of departure for the study because they were classified as simple in 40 European languages by participants in an international research project with afocus on crosslinguistic investigation into derivational networks (Projekt Monika, cf. Körtvélyessy et al. 2020, and specifically Popova 2020 on English). OPEN ACCESS
10 LINGUISTICA PRAGENSIA 1/2023 Körtvélyessy et al. (2020) draw adistinction between aderivational network, which is viewed as asystem of complex words (arranged into orders of derivation) grouped around asingle underived lexeme, and aderivational paradigm, which is defined as aset of first-order derivatives from agiven lexeme. Here, in contrast to the position taken by Körtvélyessy et al. (2020), the term derivational paradigm is used in awider sense to refer to a“[…] series of related morphological forms which share abase or base type”3 (Bauer 1997: 245, cf. also Bauer 1983), or to agroup of words sharing acommon root (Beecher 2004: 17; Fernandez-Dominguez et al. 2020: 4, among others). For each base, the derivational paradigms by word-class changing affixation were extracted after searches in both the British National Corpus (Davies 2004–, hereafter BNC) and the Oxford English Dictionary (hereafter OED). In order not to overlook any possible derivative, alist of word-class changing derivational affixes based on Quirk et al. (1985) and Stockwell & Minkova (2001) was used. After extraction of the derivatives, asample of 317 derived lexemes was obtained. It must be noted that not all the derivatives were found in both the OED and the BNC. Some derivatives appeared only in the OED (around 37%), other derivatives were found only in the BNC (around 4%), and, probably, certain senses or derivatives may not be attested in either source. Withal, acombination of these two sources is believed to build afairly complete picture of the paradigms for the 30 bases considered. Table 1 shows the total number of derivatives by affixation per base.4 Nouns nDerivatives Verbs nDerivatives Adjectives nDerivatives bone day dog eye fire louse name stone tooth water 16 7 17 11 10 9 14 15 20 20 burn cut dig drink give hold know pull sew throw 8 10 8 16 7 15 13 5 5 7 bad black long narrow new old straight thick thin warm 4 11 12 7 5 6 9 14 7 9 Sum 139 94 84 Table 1.Number of derivatives by affixation per paradigm base Once the paradigms were extracted, the criteria for directionality were applied, as described and exemplified in Section 3.2. 3 The term base is used here in the sense of aword-formation base, i.e., amotivating lexeme. For discussion on the vagueness of the term paradigm see Fernández-Alcaina & Čermák (2018: 70, and references therein). 4 The semantics of the lexemes for this study were last checked in the OED in October 2021. OPEN ACCESS
ALBA E.RUZ— BOżENA CETNAROWSKA 11 3.2 APPLICATION OF MARCHAND’S CRITERIA FOR DIRECTIONALITY The criteria used in the study are described and exemplified in this Section. Some criteria, following the idea that word semantics must be studied by considering the role of senses, are based on information from the OED, specifically the criteria of SD (Section 3.2.1), SP (Section 3.2.2), RU (Section 3.2.3), and SR (see Section 3.2.4). Other criteria, specifically the criteria of Reg (3.2.5) and Freq (3.2.6) are studied at the level of lexemes, based on data from the BNC. 3.2.1 SEMANTIC DEPENDENCE (SD) Semantic dependence (henceforth SD) is one of the criteria described by Marchand (1964: 12), who claimed that “[t]he word that for its analysis is dependent on the content of the other pair member is necessarily the derivative”, e.g., knifen > knifev ‘to Vwith aknife’. As exemplified in Table 2 for the paradigm of bonen,5 this criterion was measured by counting the senses6 for each derivative showing SD or not. BASE D1 D2 nSenses + SD – SD ? nSD bonen22 bonedadj 3 1, 2, 3 3 bonelessadj 31, 2 3 2 bonelessn* (1) 1 bonelesslyadv 1 1 1 bonelessnessn1 1 1 bonyadj/boneyadj 31, 2 3 2 bonyn1 1 1 bonyv1 1 1 bonilyadv - (1) 1 boninessn1 1 1 boningn41, 2, 3a 3b, 4 2.5 bonern41, 2a 2b, 3 4 1.5 boneishadj 21, 2 2 debonev1 1 1 debonedadj 2 1, 2 2 deboningn1 1 1 Total 30 24 Table 2.Semantic dependence in the paradigm of bonen (OED) (nSenses: number of senses that each lexeme takes in the OED; +SD: senses showing SD; –SD: senses not showing SD; nSD: total number of senses showing SD) 5 An asterisk (*) in this and other tables indicates that no specific senses are provided for alexeme which appears in the same entry in the OED as the lexeme that precedes it in the table. Ahyphen (-) in column nSenses means that the lexeme does not appear in the OED (but is attested in the BNC). Ageneral sense is counted in these cases. 6 Arabic-numbered senses were noted down for the analysis of the semantic criteria relying on the OED, and subsenses were considered too when relevant, as shown, e.g., in OPEN ACCESS
12 LINGUISTICA PRAGENSIA 1/2023 Table 2 exemplifies the analysis of SD for the paradigm of bonen. The numbers in columns +SD, –SD, and ? correspond to the specific Arabic numbers of the senses for the lexemes, as they appear in the OED. For the paradigm in Table 2, it can be seen that out of the total number of senses found for all the derivatives by affixation from bonen (n = 30), 24 senses seem to show SD (e.g., bonelessadj, sense 1 ‘Having no bones; lacking bones’, or bonedadj, 7 sense 2 ‘Provided with bone or bones’). When synonyms were used for the description of the lexemes, the entries for the synonyms were checked and if they referred to the original bases, the derivatives were analyzed as showing SD towards the base too. While most senses seem to satisfy the criterion in this paradigm, senses 2b and 3 for the derivative bonern are marked as not showing SD: (1) 2b. Acow of moderate to poor quality or condition whose meat is typically used for low-grade beef products. Frequently attributive. (2) 3. slang (chiefly North American). Amistake, ablunder; frequently (and in earliest use) Sport (originally and chiefly Baseball) apoor decision or tactical error, esp. one that causes one’s team to lose agame. Frequently in to pull aboner: to make amistake. Cf. bonehead n.1b. In the case of sense 2b (1) even if arelation may be perceived by speakers, the base is not mentioned in the definition in the dictionary and, thus, aSD relation cannot be confidently claimed. Sense 3 (2), on the other hand, seems to have emerged specifically for the derivative, as no related sense is found in the base. Also, senses classified as unclear regarding whether they show SD to the base or not are listed under column “?”. For example, sense 3 below for bonelessadj (3) or sense 4 for bonern (4) because no clear reference to bonen is made. (3) bonelessadj 3. figurative. Lacking substance, solidity, or strength; (of aperson) having little strength of character or willpower; lacking ‘backbone’. (4) bonern 4. slang (originally U.S.). An erection of the penis. Hence figurative: astrong attraction to or state of excitement about something specified. Another case classified as unclear is sense 3 for bonyadj (5), because even if it makes reference to sense 12 in bonen (6), the latter (i.e., sense 12) also makes reference to the former (i.e., sense 3) of the adjective: Tables 2–4. For instance, for bonern in Table 2, nSD (the sum of the number of senses showing dependence to the base bonen) is 1.5 because subsenses were counted as 0.5, and under our analysis, just sense 1 and subsense 2a show SD. 7 This formation can be viewed as adeverbal adjective, ultimately referred to abase noun and resulting from atwo-step derivation (i.e., N > V> ADJ), or, as noted by an anonymous reviewer, as adenominal adjective (N > ADJ, similarly to the adjectives haired and talented). OPEN ACCESS
ALBA E.RUZ— BOżENA CETNAROWSKA 13 (5) bonyadj 3. U.S. Mining. Of coal: containing aconsiderable amount of slate or shale. Cf. bone n.1 12. (6) bonen1 12. Mining. Slaty or shaly material embedded in coal seams; coal containing such material. Cf. bony adj. 3. 3.2.2 SEMANTIC PATTERN (SP) For the criterion of SP, Marchand (1964: 15) stated that “[c]ertain words have characteristic meanings which mark them as derivatives”, e.g., fatherv ‘to act as afather’. This criterion was measured here by calculating the number of senses for each lexeme showing aSP typical of aderivative (i.e., nSP), as in Table 3 for the paradigm of bonen.8 BASE D1 D2 nSenses –SP/? nSP bonen22 bonedadj 3 3 bonelessadj 3 3 bonelessn* (1) 1 bonelesslyadv 1 1 bonelessnessn1 1 bony*adj 3 3 bonyn1 1 0 bonyv1 1 bonilyadv - (1) 1 boninessn1 1 boningn4 3b 3.5 bonern4†1, 2b, 3, 4 0.5 boneishadj 2 2 debonev1 1 debonedadj 2 2 deboningn1 1 Total 30 25 Table 3.Semantic pattern in the paradigm of bonen (OED) Table 3 exemplifies the analysis of the criterion of SP for the paradigm of bonen, for which the number of senses for each lexeme showing aSP typical of aderivative was noted down. Specifically for the paradigm of bonen, alarge number of senses (n = 25) were found to take aSP typical of aderivative from the nominal base or the related base in D1. In some cases, no SP signaling the assumed derivative character of the senses is used, as in bonyn (see example (5)). The related sense, bonen (sense 12 in example (6)), is defined using the same wording as the definition for bonyn, thus, 8 Marchand (1969) lists aseries of paraphrases for each affix which were particularly useful in the analysis of this criterion. OPEN ACCESS
14 LINGUISTICA PRAGENSIA 1/2023 no SP typical of aderivative is found, but the two lexemes can be used to refer to the material. 3.2.3 RESTRICTION OF USAGE (RU) Following Marchand, this study expects derivatives to show more RU than their bases. Marchand (1964: 13) claimed that “[i]f one word has asmaller range of usage than its pair member, it must be considered the derivative”, listing various possibilities for RU (1969: 13–14). More specifically: i) for one of the words not to be generally accepted while the other is commonly used; ii) for aword to be restricted to certain forms as one of the word classes while it is not restricted as the other; iii) for aword to be used as half serious or semifacetious; or iv) to take aliterary or poetic use. Marchand (1964) also mentions frequency in relation to RU, in the sense that if aword is less commonly used than the other in apair, then, this may be expressed in terms of frequency, being the less commonly used lexeme less frequently used too. In this study, however, Freq is analyzed, not as part of RU, but as aseparate, though clearly not unrelated, criterion. In this study, the criterion of RU was measured by counting the total number of senses showing any restrictions in the OED. This criterion is divided into several types, as shown in Table 4 for the paradigm of bonen: i. RU1: not commonly accepted senses because, e.g., the sense is marked as rare, archaic or obsolete in the OED; it is restricted to aspecific group of speakers or alanguage variety (e.g., Australian, Scottish, American English, etc.); it is specific of aregister (e.g., slang and dialectal senses), or used in aspecific discipline (e.g., Mining, Astrology, etc.). ii. RU2: senses the use of which is restricted to aspecific form as one word class, not taking all of the inflected or grammatical forms typical of its category, while the use of the other pair member seems to be unrestricted, e.g., verbs that may be used only in one of their forms such as the participle -ing form, nouns appearing only in plural form, etc. iii. RU3: half serious, semi-facetious senses, and those described as humorous senses in the OED. iv. RU4: not colloquial, i.e., hyperbolic, literary or poetic senses in the OED. Figurative and extended senses were initially listed down within this category too (in gray font in column RU4, Table 4), but these were later quantified as restrictions in our analysis only in some cases, while in others they were rather considered as special uses for the lexemes, typically widening the SR or scope of the derivatives. In other words, if asense of aderivative contrasts with the base-related sense in that it can only be used figuratively, then it would be analyzed as restricted, but if the derivative sense covers the use of the base sense and additionally it can take afigurative interpretation, then this cannot count as restricted because it would rather widen the possibility of use of the derivative sense. OPEN ACCESS
ALBA E.RUZ— BOżENA CETNAROWSKA 15 BASE D1 D2 nS RU1 RU2 RU3 RU4 nRU bonen22 Hist./obs./rare: 3, 10, 18, 19b, 20, 21 Specific: 5, 8b, 12 Slang: > 11, 14 Pl.: 5a, 7, 9, 16 (+coll.), 17 Mass n.: 8a Figurative: 1c, 4b, 6, 9 15 bonedadj 3Specific: 2 1 (also fig.) 1 bonelessadj 3Figurative: 3 1 bonelessn* (1) 1 bonelesslyadv 1Figurative: 1 1 bonelessnessn1 0 bony*adj 3Specific: 3 (U.S. Mining.) 1 bonyn1Specific: 1 (U.S. Mining.) 1 bonyv1Obsolete (+ nonce-word): 1 1 bonilyadv - (1) 0 boninessn1 0 boningn4Specific: 1, 2, 3 Slang: 4 – Pl. 4 bonern4Slang: 1, 3, 4 3 boneishadj 2Obsolete: 1 1 debonev1 0 debonedadj 2Specific, rare: 1 1 deboningn1Specific: 1 1 Total 30 Total nRU (Der) 17 Table 4.Restrictions of usage in the paradigm of bonen (OED) Table 4 exemplifies the analysis of RU for the paradigm of bonen. As can be seen, this criterion is divided into various types (columns RU1–RU4), but these are later quantified as awhole for each lexeme (column nRU), which allows for abase-derivative comparison of RU. The numbers in columns RU1–RU4 are the specific Arabic numbers that the senses presenting the restrictions specified in each column take in the OED. Specifically for the paradigm from bonen, it can be seen that many senses for the derivatives show restrictions of various types (n=17), however, it is to be noted that the base itself has alarge number of senses which do show restrictions in use too (n=15). 3.2.4 SEMANTIC RANGE (SR) Regarding the criterion of SR, Marchand (1964: 14) claimed that “[o]f two homophonous words exhibiting similar sets of semantic features the one with the smaller field of reference is the derivative”, e.g., convertv > convertn ‘one who has been converted to areligion/belief’. This criterion is measured qualitatively in this study for the lexOPEN ACCESS
22 LINGUISTICA PRAGENSIA 1/2023 4.1 SEMANTIC DEPENDENCE (SD) The quantitative results of the analysis of the criterion of SD are presented in Tables 10, 11, and 12 in the appendix (A.1), for the nominal, adjectival and verbal bases, respectively. The tables present the number of senses showing SD towards the base out of the total number of senses for the derivatives by order of derivation,12 based on information from the OED, and only for those derivatives recorded in the dictionary. From the results in Tables 10–12, the criterion of SD applied at the level of sense seems to give fairly clear indications as to the directionality of the lexemes in the paradigms studied, i.e., most of the senses for the derivatives in D1, D2, or D3 seem to show SD towards their bases. This is more obvious for derivatives in D2 or D3, which seem to take fewer senses, these being more specific and typically using the semantics of the base more directly in their description. Still, it can be seen that not all senses in the derivatives do show SD towards the base. Some examples of senses not showing SD to the base are presented in examples (7–9): (7) bonern 2b. Attributive. Cow of poor quality whose meat is used for low-quality beef products. (8) daylessadj †1. Without redress or reward; without adesired result or effect. Obsolete. (9) warmingn 2. Athrashing, trouncing. Also figurative. In examples (7–9), the senses listed do not appear to be related to or derived from any of the senses listed for their bases in the OED. Our interpretation is that, for those specific senses, there is no SD towards the bases, even if arelation may be perceived by acommunity of speakers in some cases, e.g., in example (9), for which sense 2 seems to have emerged through metonymy, i.e., awarmingn, which is the result of ‘athrashing, trouncing’ is used to name the action. Other senses for the same lexemes in the examples are directly related to the senses in the base and show SD, e.g., sense 1a for warmingn ‘the action of making warm; the state of becoming warm’.13 4.2 SEMANTIC PATTERN (SP) The quantitative results of the analysis of the criterion of SP are presented in Tables 13, 14, and 15 in the appendix (A.2), for the nominal, adjectival and verbal bases, respectively. Based on the definitions in the OED, the tables present the number of senses showing SP out of the total number of senses for each derivative by order of derivation. The tables present the data only for the derivatives recorded in the OED. 12 SD for derivatives in D2 is measured in relation to their bases in D1. 13 Because the focus of this study is on overt affixes only and converted lexemes are not included, derivatives taking one derivational affix are always compared to the bases of the paradigms, but these may be derived, or at least some of their senses may be more closely related to the converted counterparts of the bases, as seems to be the case in warmingn (its senses being more closely related to warmv). This is common for derivatives in -ing and -ed in the nominal and adjectival bases. OPEN ACCESS
ALBA E.RUZ— BOżENA CETNAROWSKA 23 Tables 13–15 show that the criterion of SP also seems to indicate directionality when considering senses, i.e., most senses for the derivatives in D1, D2, or D3 are described in the OED by using aSP typical of aderivative, although results differ for each paradigm, and cases are found too where this criterion is not fulfilled. Examples (10) and (11) are senses which do not clearly show aSP typical of aderivative: (10) cuttingn † 2.An intersection; also asection. Obsolete. (11) narrowlyadv 4. †a. Barely, scarcely. Obsolete. rare. Still, overall results show that in D1 most senses seem to take aSP typical of aderivative, and that senses in D2 and D3 do so even more strongly. It seems that the more complex alexeme is in terms of the number of affixes it takes, the more derivativelike its senses tend to be (or at least be described as such). 4.3 RESTRICTIONS OF USAGE (RU) The quantitative results of the analysis of the criterion of RU are presented in Tables16, 17, and 18 in the appendix (A.3), for the nominal, adjectival and verbal bases, respectively. Following the information in the OED, this criterion was measured by calculating the number of senses showing RU out of the total number of senses for each lexeme. The tables offer results for the senses of the derivatives which appear in the OED. Also, they present the restrictions for the base and derivatives in each paradigm, by order of derivation, to stress the fact that not only derivatives show RU, but senses in the bases do so too. From Tables 16–18, it can be observed that the picture changes from paradigm to paradigm, but for paradigms in which derivatives show aconsiderably high percentage of restricted senses, the percentage of restricted senses seems to be typically high for the bases of the paradigms too. Example (12) illustrates how restricted senses in the derivatives are typically related to senses which are also restricted in the base. Here, sense 2e for the derivative pullingn, marked as ‘British slang’ in the OED, is directly related to sense 12a for the base pullv, also indicated as ‘British slang’: (12) pullingn 2. e. British slang. The action of picking up asexual partner. pullv 12. a. transitive. British slang. To pick up (apartner), esp. for sexual intercourse; to seduce. 4.4 SEMANTIC RANGE (SR) The qualitative results of the analysis of the criterion of SR are given in Section 4.4.1 and results of acomparison of the number of senses of each lexeme in the OED are provided in Section 4.4.2. 4.4.1 QUALITATIVE ANALYSIS OF SR The qualitative results of the analysis of the criterion of SR are presented in Tables 19, 20, and 21 in the appendix (A.4.1), for the nominal, adjectival and verbal bases, respectively. The tables present the number of derivatives showing asimilar (≈), slightly OPEN ACCESS
24 LINGUISTICA PRAGENSIA 1/2023 narrower or very similar (≲), narrower (<), or wider (>) SR than that of their base by order of derivation, based on information from the OED. Column ?/na OED lists unclear cases and those not attested in the OED. The criterion of SR was measured qualitatively partly because it is difficult to make any claims when the number of senses for bases and derivatives in the OED is low, and partly because of inconsistencies in the organization of senses in the OED, e.g., senses which appear as main senses in the base may appear as subsenses in the derivative, or the other way around. Also, some derivatives are defined as taking asense ‘in all of the senses of the base’ or ‘in various senses’, e.g.: (13) eyelessnessn2 ‘The state or condition of being eyeless (in various senses)’. The results of the analysis reveal that most derivatives in our sample seem to take anarrower SR, or their SR seems to be slightly narrower or very similar to that of the base, but with the semantic change associated to the word-class category change. Few lexemes show asimilar or awider SR, and overall, it seems that taking avery similar or slightly narrower SR is more typical the more formally complex alexeme is. 4.4.2 NUMBER OF SENSES In relation to the criterion of SR, the number of senses in the OED for each lexeme was considered to see if, as expected, the bases do take more senses than their derivatives, these being more specific or covering asmaller SR. Tables 22–24 in the appendix (A.4.2) present the number of derivatives taking, as expected, alower number of senses in use than their bases (>), or in contrast to the expectation for derivatives, taking the same (=) or ahigher (<) number of senses than their bases for each lexeme in the nominal, adjectival, and verbal paradigms considered, by order of derivation. As the study of this criterion is based on the representation of lexemes in alexicographic source, the tables also provide the number of derivatives for which no information on the senses was supplied in the OED (in column No info, by order of derivation). Tables 22–24 show that, for most cases in this study, derivatives seem to take alower number of senses in the OED, independently of the order of derivation. For most paradigms, apicture like the one in Figure 1 below for the paradigm of blackadj is obtained. In the paradigm of blackadj, the derivatives in D1 seem to take amuch more reduced number of senses than the base, and derivatives in D2 seem to take fewer senses than their bases in D1, e.g., blackenv (D1) has ahigher number of senses in the OED (n = 4) than its derivatives in D2, blackenedadj (n = 2), blackeningn (n = 2), blackeningadj (n = 1), or blackenern (n = 1). As Tables 22–24 show, where this is not true, the number of senses is in most cases equivalent, and there are only afew exceptions in which the derivatives take ahigher number of senses than the base. 4.5 RANGE OF REGISTERS (REG) Tables 25–27 in the appendix (A.5) present the quantitative results of the analysis of Reg of the lexemes, i.e., they provide the number of derivatives for which Reg in the OPEN ACCESS
ALBA E.RUZ— BOżENA CETNAROWSKA 25 BNC is lower (<), equivalent (=), or higher (>) as compared to the Reg of the bases in the nominal, adjectival, and verbal paradigms considered, by order of derivation. Tables 25–27 show that in our sample, even though there are cases for which the Reg of derivatives is more restricted than that of their base, there are many examples of derivatives used in the same number of registers as their base too. Thus, alook at the number of registers in which each base and derivative appears in the BNC does not seem to prove as useful or conclusive for the purpose of this study, at least when the lexemes in the paradigms are as frequent as the ones in this study. 4.6 FREQUENCY OF OCCURRENCE (FREQ) Tables 28–30 in the appendix (A.6) present the quantitative results of the analysis of the lexemes’ Freq, i.e., the number of derivatives for which the normalized frequency of occurrence in the BNC is lower (<) or in contrast higher (>) to that of the bases for the lexemes in the nominal, adjectival, and verbal paradigms, by order of derivation. Data is presented for derivatives in the corpus, and the percentage for the derivatives not represented in the BNC by order of derivation is given too (Not attested D1/D2/D3). Tables 28–30 show that the Freq of the lexemes proves applicable in the sense that most derivatives in the BNC have alower frequency than their base. The paradigms in this study present themselves as cases in which bases are more frequent than their derivatives in D1, and at the same time, bases in D1 are typically more frequent than their related derivatives in D2, and so on. This is exemplified, e.g., by the paradigm of bonen in Table 8 (Section 3.2.6). Just afew examples were encountered where Freq of one of the derivatives was higher than that of their base, e.g., lousyadj in Table 31: Figure 1.Number of senses for each lemma in the paradigm (by word-class changing affixation) of blackadj 0 2 4 6 8 10 12 14 16 blacK_ADJ blackness_N blackly_ADV blacking_N blacken_V blackened_ADJ blackening_N b lackening_ADJ blackener_N blacked_ADJ blacker_N blacky_N OPEN ACCESS
26 LINGUISTICA PRAGENSIA 1/2023 Lexeme Freq NF lousen192 1.99 lousyadj 221 2.29 Table 31.Frequency of occurrence of lousen and lousyadj (BNC) The reason for the derivative to show ahigher Freq than the base may be identified if the semantics of the lexemes are considered, as in Table 31.In this case, the base lousen is used mainly to refer to the insect, and sometimes with an extended or transferred use to refer to human beings with some characteristics associated with the insect. On the other hand, lousyadj denotes in one sense the property ‘full of lice’ (with subsenses), which is the most literal sense, and it also takes afigurative or extended sense used to characterize people or things in various related ways associated with the noun ‘dirty, filthy, obscene. Also as ageneral term of abuse: Mean, scurvy, sorry, vile, contemptible. Also, inferior, poor, bad; ill; in low health or spirits’. This would explain the higher Freq of the derivative, as it seems more likely that one may use the characterization of the adjective and describe someone or something as lousy in any of the senses of the adjective than to use the base to refer to the insect, or in its extended/transferred sense to refer to someone as alouse. 5 DISCUSSION The results of testing the criteria for directionality in affixation were presented in Section 4, and adiscussion follows on the issues if semantic criteria for directionality are applicable in affixation and if directionality is indeed predictable on the basis of semantic analysis, and to which extent directionality criteria are applicable at the level of sense or lexeme. To help the reader follow the discussion below, results are examined by criterion, too. Based on the results of this study, Section 5.1 focuses on the applicability of the criterion of semantic dependence (SD), Section 5.2 on semantic pattern (SP), Section 5.3 on restrictions of usage (RU), and Section 5.4 on semantic range (SR), all based on data from the OED. Finally, Sections 5.5 and 5.6 focus on the range of registers (Reg) and on the frequency of occurrence of the lexemes in the BNC (Freq), respectively. 5.1 SEMANTIC DEPENDENCE (SD) The results in Section 4.1 show that the criterion of SD is applicable at the level of sense and it is useful in predicting the direction in affixation. It indicates the expected directionality, from the base to the formally more complex lexeme, for alarge part of the senses in the derivatives of the paradigms in this study and even more strongly for second and third order derivatives. This is probably because the more complex the derivative, the fewer senses it seems to take, these being typically more specific and with atendency to depend on the meaning of the base more directly. Even if not all senses in the derivatives do show SD towards the base, and OPEN ACCESS
ALBA E.RUZ— BOżENA CETNAROWSKA 27 the degree to which SD indicates abase>derivative direction varies for each paradigm, and for each derivative, overall, the criterion is fulfilled for most senses in the derivatives and thus considered useful here for the study of directionality. Asystematic assessment of instances in which aSD towards the base is not identified, or where various directions or patterns can be identified may provide further information in this regard. 5.2 SEMANTIC PATTERN (SP) The results in Section 4.2 show that the criterion of SP is applicable at the level of sense and is useful in predicting adirection in affixation too, indicating the assumed base>derivative direction for alarge part of the senses in the derivatives of the paradigms in this study. This is manifested even more strongly the more complex aderivative is, similarly to the results of the criterion of SD. Again, it must be noted that not all senses in the derivatives show aSP typical of aderivative, and it would be interesting to study if this occurs precisely because different directions may be involved in affixation. 5.3 RESTRICTIONS OF USAGE (RU) The results in Section 4.3 indicate that the criterion of RU is applicable in affixation only for specific senses. It was shown that while it is true that many of the senses for the derivatives are restricted, this cannot be taken as evidence of directionality before adetailed analysis of the senses of the base and its derivatives is carried out. This is because senses in the bases may be restricted, and these restrictions will most probably be passed to the related derivative senses, independently of whether the sense is basic in the simpler or in the more complex lexeme. Only the identification of related senses which appear as restricted in one of the lexemes in apair and as unrestricted in the other one would prove useful for the study of directionality. Thus, RU is applicable and relevant only at the level of sense, probably not offering conclusive results for an analysis of lexemes as awhole. 5.4 SEMANTIC RANGE (SR): QUALITATIVE AND QUANTITATIVE ANALYSIS OF SENSES The results from aqualitative analysis of SR in Section 4.4.1 seem to confirm the conclusion that the senses in the derivatives typically show either anearly similar SR or asignificantly lower SR than the base but not including all the senses in the base. Few or no cases were found in which the derivative SR was higher or similar to that of the base. Thus, SR seems applicable for the study of directionality by senses in affixation. This method, however, proved to be time-consuming and not as easily measurable. Also, the use of semantic information from the OED complicated the analysis in terms of directionality because it includes obsolete senses,14 as well as rare, specific or restricted senses. 14 The quantitative study of SR presents the results of considering only senses in use because the focus of this study is synchronic, but adiachronic study of directionality requires other wise. OPEN ACCESS
28 LINGUISTICA PRAGENSIA 1/2023 Acomparison of the number of senses in the OED for lexemes in apair (see Section 4.4.2) showed that the percentage of lexemes in D1 with alower number of senses than their base was higher than, e.g., lexemes in D2 as compared to their base in D1, which were more frequently found to display an equivalent number of senses. The difference in the number of senses was also found to be lower the more complex the derivative in terms of the number of affixes. Although acomparison of the number of senses alone, without any further considerations, is definitely less time-consuming than the qualitative analysis carried out, it does not prove useful in identifying directionality in affixation. Also, it presents various challenges in its applicability at the level of sense, among them the fact that i) inconsistencies in the representation of senses in the OED are found, ii) the extent to which adifference in the number of senses between apair, be it wider or narrower, may serve to indicate adifference in the SR of the lexemes is unclear. It is here thus believed that SR for specific pairs may be more accurately described following aqualitative analysis as in Section 4.4.1. However, such an analysis would prove time-consuming for the analysis of large quantities of data. 5.5 RANGE OF REGISTERS (REG) Reg may also be taken as an indication of the SR of alexeme for specific cases. It is undeniably linked to the frequency and to the polysemanticity of lexemes, i.e., the general expectation is that the higher the number of senses of alexeme or the wider its SR, the wider its Freq and Reg will be too. The results in Section 4.5 show that at least for lexemes as frequent as the ones in this study, Reg does not prove to be as useful in deciding on the directionality between two lexemes. This is because many bases and derivatives in our paradigm are well-known and their use is widespread, and thus, many base-derivative pairs are found to appear in all the registers in the BNC. Also, results seem to vary from paradigm to paradigm, and by order of derivation. While it may seem to prove more useful when considering more complex derivatives, it is unclear whether this may just be related to alow Freq of the lexemes or whether it really indicates that the Reg of those derivatives is narrower.15 To which extent amore in-depth classification into registers or by senses may offer results for the directionality in affixation remains unclear. 5.6 FREQUENCY OF OCCURRENCE (FREQ) The results in Section 4.6 show that astudy of the Freq of lexemes is applicable in affixation and seems to indicate the assumed directionality for most base-derivative pairs, the bases being typically more frequent than their derivatives in D1, D2, and D3. While it seems to offer results at the level of lexeme, it is here believed that considering Freq by senses would be useful in identifying directionality between lexemes in unclear cases, and in identifying pairs for which various directions of derivation may 15 If alexeme appears just acouple of times in the BNC, it will naturally appear in alow number of registers, which may not mean that it cannot be used in other registers, but may just be due to alack of representation of that specific lexeme. OPEN ACCESS
ALBA E.RUZ— BOżENA CETNAROWSKA 29 exist. Otherwise, reaching more concrete results regarding Freq (and Reg), or identifying patterns of directionality in this respect will not be possible. Also, the extent to which adifference in the Freq between apair may be relevant or not is unclear and should be defined before undertaking amore accurate analysis of directionality in terms of Freq in unclear cases. 6 CONCLUSION This paper aimed to investigate Marchand’s (1964) directionality criteria in asample of word-class changing affixation in English in order to test if the semantic criteria prove applicable outside conversion, and to determine how feasible it is to apply Marchand’s semantic criteria at the level of sense and not of lexeme. An analysis of the criteria seems to indicate the following: i. For most cases in our sample, both the criteria of SD and SP seem to indicate correctly abase>derivative direction of derivation. However, as has been shown above, this is not true for every sense, and some variation is found for the individual paradigms. Independently of the reasons why the criteria may not be satisfied in some cases, the same is expected to happen for other wordformation processes. ii. The criterion of RU proves useful only at the level of sense on abase-derivative comparison, and only for specific cases. iii. Most of the derivatives seem to take anarrower SR, but this is best measured qualitatively partly because of inconsistencies in the dictionary sense organization which make aquantitative analysis difficult, the number of senses not always offering relevant results. Aqualitative analysis, however, proves timeconsuming. iv. Reg may be areliable diagnostic if studied by senses, and perhaps only if amore fine-grained division of registers is considered. The investigation of the number of registers per se does not always lead to convincing conclusions. v. Freq of the derivatives has been found to be typically lower than that of their bases for the paradigms studied. It must be noted, however, that Freq may serve as auseful diagnostic if studied by senses. Moreover, what should be determined first is to what extent and from which point adifference in Freq can be considered relevant. This is especially important for unclear cases. It should be emphasized again that the criteria have been tested here with asample of affixation in which the bases of the paradigms are typically much more frequent than the derivatives in the paradigm, and show ahigher degree of polysemy too. Determining adirection between two lexemes, however, will be achallenge for pairsof lexemes which are both polysemous and which show, e.g., asimilar SR, number of senses, Reg, and Freq. This is where an analysis by senses would prove most beneficial. OPEN ACCESS
30 LINGUISTICA PRAGENSIA 1/2023 Also, it should be noted that the semantic criteria have synchronic relevance, in the sense that directionality is not rigidly determined, and may vary with time for the different senses which may emerge for apair of paronymic lexemes. Overall, this study serves to confirm that an analysis by senses, although timeconsuming, is desirable for polysemous pairs of lexemes (in line with Plank 2010). The results reinforce the idea that an analysis by senses may offer more accurate results regarding directionality, the semantic development of lexemes and the patterns found between pairs and within derivational paradigms. ABBREVIATIONS BNC British National Corpus OED Oxford English Dictionary N Noun V Verb Der Derivatives DP Derivational Paradigm ADJ Adjective SD Semantic dependence SP Semantic pattern SR Semantic range RU Restrictions of usage Reg Range of registers Freq Frequency of occurrence NF Normalized frequency of occurrence D1 First-order derivation D2 Second-order derivation Acknowledgements We would like to express our gratitude to Prof. Salvador Valera for his helpful comments on an earlier version of this article and to Cristina Fernández Alcaina and Cristina Lara Clares for their suggestions. We would also like to thank the editors and the two anonymous reviewers for their feedback. This publication has been supported by the project PID2020-119851GB-I00 MCIN/ AEI/10.13039/501100011033, funded by the Spanish State Research Agency (AEI) and the Ministry of Science and Innovation (MCIN). The research activities were also co-financed by the funds granted under the Research Excellence Initiative of the University of Silesia in Katowice. OPEN ACCESS
ALBA E.RUZ— BOżENA CETNAROWSKA 31 REFERENCES Balteiro, I. (2007) The Directionality of Conversion in English: ADia-synchronic Study. Berlin: Peter Lang. Bauer, L. (1983) English Word-Formation. Cambridge: Cambridge University Press. Bauer, L. (1997) Derivational paradigms. In: Booij, G. and J. van Marle (eds) Yearbook of Morphology 1996, 243–256. Dordrecht: Springer. Bauer, L. and S.Valera (2005) Conversion or zero-derivation: an introduction. In: Bauer,L. and S.Valera (eds) Approaches to Conversion / Zero-Derivation, 7–15. Münster: Waxmann. Beecher, H. (2004) Derivational paradigm in word formation. Unpublished research paper. Available at: citeseerx.ist.psu.edu/viewdoc/ download?doi=10.1.1.94.9071&rep=rep1& type=pdf [last accessed 14August 2021]. Biese, Y.M. (1941) Origin and development of conversions in English. Helsinki: Annales Academiae Scientiarum Fennicae B XLV/2. Bram, B. (2011) Major Total Conversion in English: The Question of Directionality. PhD Thesis, Victoria University of Wellington. Cetnarowska, B. (1993) The Syntax, Semantics and Derivation of Bare Nominalisations in English. Katowice: Uniwersytet Śląski. Cruse, A. (1986) Lexical Semantics. Cambridge: Cambridge University Press. Davies, M. (2004–) British National Corpus (from Oxford University Press). Available at https:// www.english-corpora.org/bnc/ [last accessed 30October 2021]. Don, J. (2003) Anote on conversion in Dutch and German. In: Cornips, L. and P.Fikkert (eds) Linguistics in the Netherlands, 33–43. Amsterdam: John Benjamins. https://doi. org/10.1075/avt.20.07don Fernández-Alcaina, C. and J.Čermák (2018) Derivational paradigms and competition in English: adiachronic study on competing causative verbs and their derivatives. SKASE Journal of Theoretical Linguistics 15/3, 69–97. http://www.skase.sk/Volumes/JTL38/pdf_ doc/04.pdf Fernández-Domínguez, J., A.Bagasheva and C.Lara-Clares (eds) (2020) Paradigmatic Relations in Word Formation. Leiden: Brill. https://doi.org/10.1163/9789004433410 Körtvélyessy, L., A.Bagasheva and P.Štekauer (eds) (2020) Derivational Networks Across Languages. Berlin: De Gruyter Mouton. https:// doi.org/10.1515/9783110686630 Lohmann, A. (2017) Phonological properties of word classes and directionality in conversion. Word Structure 10/2, 204–234. https://doi. org/10.3366/word.2017.0108 Lyons, J. (1977) Semantics. Cambridge: Cambridge University Press. Marchand, H. (1963) On aquestion of contrary analysis with derivationally connected but morphologically uncharacterized words. English Studies 44, 176–187. Marchand, H. (1964) Aset of criteria for the establishing of derivational relationship between words unmarked by derivational morphemes. Indogermanische Forschungen 69, 10–19. Marchand, H. (1969) The Categories and Types of Present-Day English Word-Formation. ASynchronic—Diachronic Approach, 2nd edn, x–xxvii, 1–545. München: Verlag C.H.Beck. Oxford English Dictionary (OED). Oxford: Oxford University Press. Available at https://oed.com/ [last accessed 30October 2021]. Plag, I. (2003) Word-Formation in English. Cambridge: Cambridge University Press. Plank, F. (2010) Variable direction in zeroderivation and the unity of polysemous lexical items. Word Structure 3/1, 82–97. https://doi. org/10.3366/E1750124510000498 Popova, G. (2020) Derivational networks in English. In: Körtvélyessy, L., A.Bagasheva and P.Štekauer (eds) Derivational Networks Across Languages, 147–156. Berlin: De Gruyter Mouton. https://doi. org/10.1515/9783110686630-015 Quirk, R., S.Greenbaum, G.Leech and J.Svartvik (1985) AComprehensive Grammar of the English Language. London: Longman. OPEN ACCESS
38 LINGUISTICA PRAGENSIA 1/2023 Table 20.Semantic range of the derivatives from the adjectival bases by order of derivation (OED) Abrief explanation of the data in Table 19 may help the reader understand Tables 19–21. Table 19 shows that, in the paradigm from bonen, all the derivatives found in first-order derivation (D1) present anarrower SR (<: 10/10) than the base of the paradigm, while out of the seven derivatives in second-order derivation (D2), three have anarrower semantic range (<: 3/7), one derivative exhibits aslightly narrower or very similar semantic range (≲: 1/7), and two derivatives show asimilar semantic range (≈: 2/7) to their base in D1. Another derivative does not appear in the OED (na: 1/7), and thus no comparison of its SR is made. For reasons of space, categories or types in which no derivatives were found are not represented in Tables 19–30, e.g., in Table 19, none of the derivatives in D2 were classified as having awider SR (>) than their bases in D1, thus, this option is not portrayed in the table. OPEN ACCESS
ALBA E.RUZ— BOżENA CETNAROWSKA 39 Table 21.Semantic range of the derivatives from the verbal bases by order of derivation (OED) Table 22.Number of senses for the derivatives from the nominal bases by order of derivation (OED) A.4.2 NUMBER OF SENSES OPEN ACCESS
40 LINGUISTICA PRAGENSIA 1/2023 Table 23.Number of senses for the derivatives from the adjectival bases by order of derivation (OED) Table 24.Number of senses for the derivatives from the verbal bases by order of derivation (OED) No info columns in Tables 22–24 provide the number of derivatives not recorded in the OED or recorded but for which no senses are specified. It can be observed, e.g., that in Table 22 all the ten derivatives found in D1 for the base bonen present alower number of senses than the base (Base > D1: 10/10), while of the seven derivatives found in D2, four have alower number of senses than their base (D1 > D2: 4/7), two derivatives exhibit asimilar number of senses (D1 = D2: 2/7), while the OED provides no description for one of the derivatives (No info: 1/7), therefore no comparison is made. OPEN ACCESS
ALBA E.RUZ— BOżENA CETNAROWSKA 41 A.5 RANGE OF REGISTERS (REG) Table 25.Range of registers for the derivatives of the nominal bases by order of derivation (BNC) In Table 25 it can be seen, e.g., that for the base bonen, three out of the ten derivatives found in D1 are not attested in the BNC (under Not attested D1 column in gray) therefore no comparison can initially be made, while five out of the six derivatives attested show asmaller Reg than the base (Base > D1: 5/6), and one derivative shows asimilar Reg as the base, i.e., it appears in the same number of registers in the BNC (Base= D1: 1/6). Similarly, for the six derivatives found in D2, all derivatives attested in the BNC show asmaller Reg than their base (D1 > D2: 3/3), while for the other three not attested derivatives no comparison is made. OPEN ACCESS
42 LINGUISTICA PRAGENSIA 1/2023 Table 26.Range of registers for the derivatives of the adjectival bases by order of derivation (BNC) Table 27.Range of registers for the derivatives of the verbal bases by order of derivation (BNC) OPEN ACCESS
ALBA E.RUZ— BOżENA CETNAROWSKA 43 Table 28.Normalized frequency of occurrence for the derivatives of the nominal bases by order of derivation (BNC) Table 29.Normalized frequency of occurrence for the derivatives of the adjectival bases by order of derivation (BNC) A.6 FREQUENCY OF OCCURRENCE (FREQ) OPEN ACCESS
44 LINGUISTICA PRAGENSIA 1/2023 Table 30.Normalized frequency of occurrence for the derivatives of the verbal bases by order of derivation (BNC) OPEN ACCESS