Skip to navigation – Site map

HomeIssues16PapersA Case Study of -some and -able D...


A Case Study of -some and -able Derivatives in the OED3: Examining the Diachronic Output and Productivity of Two Competing Adjectival Suffixes

Chris A. Smith


In this exploratory study, we seek to compare two adjectival suffixes from a diachronic perspective: the native -some suffix and the imported Romance suffix -able. We aim to provide answers to these questions: in terms of competition, what evidence shows that -able can be viewed as a direct competitor to -some? Also, what other influences may have contributed to its decline (many other adjectival suffixes form competitors (-ful, -ish, -ly)? Other than increasing morphosemantic competition, can subjectification explain the successfulness of -able? We consider several hypotheses based on our data, explaining the shift in the landscape of adjectival suffixation and the apparently resulting decline of -some suffixation. Firstly, a semantic study of key words shows that Vable derivatives and Vsome derivatives differ in their semantic makeup, in that -able adjectives have a passive sense, whereas some adjectives in Vsome have an active sense (TENDENCY TO category, such as meddlesome [1615] “prone to meddling”), but are also compatible with a passive sense occurring (INTENDED FOR category) as in ticklesome “apt to be tickled”). This active passive alternation may have led to semantic instability, loss of transparency, and resulting loss of productivity (as suggested in the frequency- productivity chain proposed in Fernandez-Dominguez [2010: 202]). Secondly, a corpus study in multiple corpora (EHBO, COHA, Project Gutenberg OEC, COCA), as well as the OED data, both suggest that -some adjectives have a low frequency of usage over all periods of English. This low token frequency would have likely slowed propagation and therefore contributed to the decline in availability of the pattern. Finally, it is possible that -some declined due to direct pressure from -able. This hypothesis is however difficult to establish for multiple reasons: 1) blocking is a gradient phenomenon, rather than a cut-and-dried pressure; 2) the highly different frequency of usage of -some and -able in historical and contemporary corpora make it difficult to compare on a large scale; 3) other pressures exist, which haven’t been included in this study, such as other suffixations which may also have caused a chain reaction of adaptation within the language system. To test this, we conducted several case studies comparing active-oriented -some adjectives (in the TENDENCY TO category, such as meddlesome) and then passive adjectives (in the INTENDED FOR category, such as ticklesome) with Vable alternates. The conclusions reached were threefold. Overall, -some can be seen as semantically instable compared to -able: active-passive reinterpretation occurs in a number of -some adjectives (winsome, fearsome). On the other hand, -able has a more predictable semantic pattern, having mostly passive-oriented senses. Secondly, existing-some adjectives have a low token frequency as shown by extensive corpus searches, and this is verified in all periods, except for the 1850s where -some adjective formation increased out of a deliberate attempt to increase native suffixation. Finally, despite this lack of usage, -some has not become obsolete and opaque, and remains an active suffix. This begs the question of what register-specific contexts favour the use of -some adjectives.

Top of page

Full text


  • 1 The evidence of competition is based on the lexicographic study of 261 -some adjectives in the OED, (...)
  • 2 We use the form -able here rather than -ble (which includes the alternative spelling -ible, both or (...)
  • 3 Durkin [2016a: 397] “Although the OED stands alone in its breadth of coverage and chronological tim (...)
  • 4 Fernandez-Dominguez [2010: 195] explains that “[i]n essence, a word-formation process is said to be (...)

1This study offers a diachronic analysis of two competing1 adjectival suffixations in English. On the one hand, the native -some suffix and on the other, the borrowed -able suffix, which both form adjectives from verbs, and nouns. The approach is twofold, starting with a lexicographic study of Vsome and Vable2 derivatives from a diachronic perspective using the OED3 as source material3. The aim is to trace the diachronic evolution (from 1100 to 2000) of the morphosemantic patterns of formation of -some suffixation first, then -able suffixation. This lexicographic approach will then be tested in a second phase by carrying out a sample study of the usage of several Vsome and Vable alternations in three historical (EHBC English Historical Books Corpus, COHA Corpus of Historical American, PG Project Gutenberg) and two contemporary corpora (OEC Oxford English Corpus, and COCA Corpus of Contemporary American) using distributional semantics as a measure of semantic similarity of the adjectives in context. This study will consider what specificities of -some adjectives versus -able adjectives might explain the continued use of a small number of -some adjectives, and the impossibility of certain -able formations, despite the overall outperformance of -able adjectives. We seek to answer the following related questions:
1) Are there different diachronic outputs for
Vsome and Vable, how does this affect diachronic productivity4?
2) How does the Romance loan suffix -
able affect the morphosemantic pattern of a pre-existing native suffix like -some, in other words did -able phase out -some?
3) Even if
Vable can be seen as a direct competitor to Vsome (from the evidence provided by the OED data, as we shall show in Section 1), what is the role of the numerous other competing derivatives (-ful, -ish, -ly) which remain productive? The question is of course whether it is feasible to restrict a competition analysis to two suffixes. However, based on our data and methodology, for this preliminary test study, we will focus on two competing suffixes, the Anglo Saxon -some suffix and the Romance loan suffix -able.
4) Could the successfulness of -
able be attributed to its increasing modal value (ability readings and deontic readings, as explained in Kjellmer [1986], Di Sciullo [1997], Oltra-Massuel [2014]) and could this in turn have led to the decline of -some? In other words, is the low productivity of -some a result of the success of -able via blocking theory (see Bauer [2004: 136])? -able is well known for its modal behaviour which has led to a lot of attention and research (a very brief overview of which shall be provided in Section 2).

  • 5 Also, the pre-existing wieldy [1413] in the same sense.
  • 6 However, two questions arise: firstly, transparency is not necessarily “a necessary condition for p (...)

2A plausible hypothesis is that, semantically, Vable derivatives are compatible with a passive sense only (meaning the adjective related to a noun is perceived as the patient of the action), whereas -some derivatives can carry both active and passive senses. As will be observed in Section 1, -some adjectives have the ability to refer to an active participation in an event (TENDENCY TO, such as meddlesome [1615] ‘prone to meddling’), or to a passive participation as the patient or object affected by the event (INTENDED FOR such as ticklesome, or wieldsome). Vable counterparts do not always exist: active adjectives like meddlesome [1615] do not have passive -able counterparts like *meddlable but do have an -ing counterpart meddling [1529]; temptsome [1849] ‘Apt to tempt, tempting’, has tempting [1400], but not *temptable. More strikingly, the passive adjective wieldsome5 [1565] ‘Easily managed, controlled, or handled’ does not have an -able counterpart *wieldable. It seems the existence of both active/passive senses of Vsome adjectives ‘intended for V, prone to Ving, addicted to Ving’ may be the cause of the semantic instability and possible decline of Vsome adjectives due to loss of transparency6.

  • 7 However, it must be noted that the OED is a non-homogenous secondary source as explained in Allan (...)
  • 8 “The Oxford English Dictionary is currently in the course of its first ever comprehensive revision, (...)

3A second hypothesis is that the frequency of usage of Vsome adjectives remained low and was not able to ensure propagation of the pattern to a productive, i.e. actively available pattern (frequency of usage vs frequency of rare word forms). This hypothesis is supported by the rare, obsolete, archaic status of many -some adjectives recorded in the OED7; Dixon [2014: 255] also notes categorically that -some derivation is no longer productive “save perhaps in nonce and jocular coinings”. The lack of homogeneity over different periods (evidenced in the OED data available on the OED website) may also lead to an over-representation of rare historical words and an under-representation of new words due to the revision process of the OED3, detailed in Durkin [2011], and Durkin [2016a: 395-396]8. Equally problematic, as pointed out in Allan [2011: 25], is that “while attestations indicate currency of any sense of a lexeme at a particular time, lack of attestation cannot be taken to provide any evidence of lack of use.”

4A third hypothesis is that of blocking, suggesting that the co-existence of two competing suffixes is subject to a form of natural selection. Aronoff [1976: 43] defines blocking as “the non-occurrence of one form due to the simple existence of another”. However, as Bauer [2004: 136] explains, blocking is not a straight-and-fast rule as much as a tendency:

In other words, the pressures which cause and prevent the application of blocking are both gradient pressures, so that which wins out on any given occasion is determined by the relative pressures being applied in that case.

  • 9 According to the “loi de répartition” of synonyms suggested in Rafaelli [2012]; also see Renner [20 (...)

5Blocking or avoidance of synonymy cannot predict word forms correctly as synonyms do continue to exist, and variations between synonyms (stealer versus thief) continue to exist, potentially causing a chain-reaction of adaptation9.

6There is quite a large gap in interest between the two adjectival suffixes since there is an extensive literature on the -able suffix (Oltra-Massuel [2014], Müller et al. [2015], Schulte [2015]), whereas the -some suffix is comparatively under-studied (except for Dixon [2014: 254-257] who, while viewing that -some is no longer productive, contrasts the suffix -some with other active adjectival suffixes, including -ful, -ous, and -ly). The suffix -able is listed in Plag [2003] as an adjectival suffix, whereas -some is not even listed, likely to its modest perceived productivity in contemporary English as suggested in Marchand [1960]. This appears to be confirmed by the small number of -some derivatives listed in the OED, the high proportion of rare and archaic forms, and no attestation dates later than the early 1930s (chillsome [1927] in the sense “chilling, chilly”, shiversome “causing shivers’ [1930]), and curvesome [1933] “full of curves”).

7This paper is organised as follows: Section 1) gives the state of the art, and an overview of -some suffixation based on 261 -some adjectives listed in the OED, carries out a morphosemantic analysis and a study of the historical productivity. Section 2) then focuses on the rival loan suffix -able and carries out a case study of the morphosemantic overlap between the two suffixes. This leads to a suggestion that the potential perceived semantic instability of -some adjectives may have played a part in its waning productivity. Finally, Section 3) offers a distributional semantics analysis of the usage of synonymous -some/-able adjectives in several historical and contemporary corpora with the aim to test the factors at play in the obsolescence of a suffix.

1. -Some adjectives in the OED3: output trajectories, semantic classification and morphological typology

1.1. The question of how to measure diachronic productivity

  • 10 “The assumption underlies this belief that frequency commonly accompanies a word during its progres (...)
  • 11 As far as potential productivity is concerned Trips [2009: 33] underlines the issues of choosing an (...)

8Let us first consider the question of how to assess and measure diachronic productivity. Morphological productivity is the ability of a language to form new words based on a pattern, it is commonly hypothesized that frequency has an effect on productivity. Frequently occurring features have a higher chance of being significant and causing patterning: the repetition of a pattern, i.e. type frequency aids productivity as suggested in Plag et al. [1999], Bauer [2004], Trips [2009], Fernandez-Dominguez [2010: 202]10. There are actually 3 methods for measuring productivity: number of tokens (raw frequency), number of types (extent of use, i.e. repetition of the pattern), number of hapaxes (i.e. number of neologisms) as explained in Plag et al. [1999: 215-216]. Trips [2009: 31] distinguishes on the one hand realised productivity (type frequency), and on the other potential productivity (number of hapaxes for the number of tokens) which can give a measure of future productivity11.

9What is the most appropriate measure for diachronic study? As suggested earlier, frequency aids, and therefore precedes, productivity according to Trips [2009: 29], Fernandez-Dominguez [2010: 202] proposes this model shown in Figure 1 representing the chain of factors involved in the successful of a word form over time.

Figure 1. Factors affecting token frequency (from Fernandez-Dominguez [2010: 202])

Figure 1. Factors affecting token frequency (from Fernandez-Dominguez [2010: 202])
Since we aim to study past productivity using the OED as a diachronic source, using a raw frequency count will produce data that is relevant, i.e. the number of words that have existed over time (total output). Productivity can also be measured by the transparency of the pattern. Trips [2009: 31] after Hay & Baayen [2002] show that “there is a relationship between relative frequency, parsing and morphological productivity”, i.e. they are interrelated phenomena. A further issue appears when looking at a morphological pattern which is borrowed from a donor language. In the case of loan words, the output does not automatically correlate to internal formation if the morphological structure is not transparent. Bauer [2004: 144] addresses this problem specifically with reference to the number of -able adjectives in the OED, underlining the confusion between loan words and internal formation.

10To summarise, a straight count may capture forms which were in use before any word-formation process had arisen. This is the danger of confusing the process of word-formation (sometimes called Wortbildung in German) from the analysis of the results of word-formation (sometimes termed Wortgebildetheit – wordformedness or analysability) (see, for example, Dokulil [1968], Bauer [2004: 144]).

11So morphological productivity is not purely the raw number of words that exist since this can include external morphological patterns (like loans), however there is a relation between frequency and productivity, since a low frequency will entail a low productivity (as with -ter formations, such as laughter [Bauer 2004: 145]. The OED and its diachronic perspective also allow for a measure of productivity based on a timeline of new additions, which give an idea of how many new coinages created by that morphological process, as Bauer [2004: 156] writes:

The more productive a morphological process is, the more coinages that occur created by that morphological process in a given time period.

12Following Bauer [2004: 159], Schulte [2015], Smith [2016], [2018], the OED is viewed as the most appropriate source of data for morphological lexical study and can be used in conjunction with a corpus analysis case study as a second step. Despite the limitations related to the use of a dictionary versus a corpus, the use of the OED is motivated by the diachronic nature of the dictionary and the inclusion of obsolete and rare words (possible hapaxes) which provides insight to relevant past productivity.

13Consulting a corpus will allow rare, easily-interpretable, words and nonce words to be included, but the coverage is likely to be less complete than that of a dictionary – at least, unless the corpus is extremely large indeed, or the dictionary very small. Given this difference, the use of a dictionary as a starting point and the use of a (suitably large) corpus as a point of comparison could be a positive compromise.

1.2. Using the OED as data and method

  • 12 We shall also see that the behaviour of the suffix-some includes deadjectival derivation thereby fo (...)

14The present section deals with -some derivatives in the OED. The suffix -some originates from Old English with Germanic roots, according to the OED etymology, see also Jespersen [1942], and Marchand [1960]. The suffix initially forms denominal adjectives Nsome (first attested winsome [900] “pleasing or attractive, handsome, comely”), but we shall see that Vsome output considerably outperforms Nsome in the 19th century with no fewer than 85 new words between 1800-1900, i.e. one third of total output, although this performance may be short-lived12. As mentioned previously, the literature on -some derivation is scant. Jespersen [1942: 456] gives a description of the -some suffix in a section called “final batch of suffixes” which include -ship, -dom, and -hood. The suffix is described as being productive yet of somewhat infrequent use. He describes the suffix as denominal rather than deverbal (calling this rare) and producing the meaning “like, having the quality of”. In terms of productivity “the suffix has been productive during all periods, though comparatively little used.” Following Jespersen [1942: 456], Marchand [1960: 262-263] identifies -some to be an Old English (OE) suffix having the same stem as same, forming substantives, adjectives and verbs meaning “like, characterized by, apt to”. Marchand [1960] also notes that the derivatives were initially primarily denominal and no deverbal adjectives survived into the Middle English (ME) period. Only three have survived, winsome [900] from OE wynsam which is no longer analysable as win N ([OE] meaning “Joy, pleasure, delight, bliss; a source of joy, a delight”) + some, lovesome and longsome, which are given as being archaic and dialectal only. However, in contrast with Jespersen, Marchand does comment on the change of morphological patterning and between the 17th and 19th century output in terms of meaning, concluding the -some suffix has “lost its productivity for deadjectival coinages while the deverbal type has grown in importance, the derivatives meaning ‘apt to or apt to cause to…’”. Dixon [2014: 254-255] describes -some adjectives as having the sense ‘apt to’, and notes the “gradual” decline in frequency of use over time as well as the lack of productivity today.

15In this section, we aim to collect data to track patterns of output. The question we aim to answer: is -some no longer morphologically productive (actively available) in all patterns? If so, what explains the loss of productivity? How can historical productivity be measured, i.e. how frequent was the usage of -some neologisms (i.e. did they propagate into general usage)?

16Our methodology in this initial phase is as follows:
1) First, we extract -some derivatives in the electronic subscription version of the OED3 (261 words or lemmata);
2) We then analyse the senses of the adjectives using a semantic feature analysis, which has previously been applied to -age words in Smith [2018] and to fl- words in Smith [2016], [2019];
3) We then carry out a morphological structure analysis, based on the OED etymology and definitions;
4) We then correlate morphological pattern with the semantic feature analysis;

5) And finally, we compare timelines of morpho-semantic output.

1.3. Morphosemantic classification: base word classification and key word analysis

17An analysis of the dictionary meanings and key words provides the basis for a feature analysis of the 261 -some adjectives in the OED. Although the OED definitions are not systematized so we cannot expect to obtain immediately relevant categories (which will be developed later), the information can be used to form conceptual categories (see also Smith [2016] on phonesthemes, and Smith [2018] on -age words). Figure 2 shows the raw token frequency of key word categorisation in the 261 -some adjectives generated from the OED:

Figure 2. The semantic categories of -some derivatives

Figure 2. The semantic categories of -some derivatives

18The OED definitions of -some adjectives are analysed and key words in the definitions are identified: “given to”, “that has a predilection for”, “apt to”, “inclined to”; “characterised by”, “full of”, “inspiring”, “causing”, “having a tendency to”, “easy to”, “addicted to”, etc. These key terms are then used to build broader conceptual categories that regroup some of the variations found in the key words, in order to capture patterns. As Schulte [2015: 45] points out, the OED definitions and paraphrases are not perfectly systematized and standardized, and therefore exhibit some variation in the treatment of meanings. Despite this drawback, the OED senses remain useful in that they are the product of in-depth analysis of a corpus by a trained lexicographer. In addition, by regrouping key terms into conceptual categories, this limits the heterogeneity of the lexicographic material.

19Another problem lies in the inconsistency of the paraphrases provided in the OED. Similar words are sometimes described very differently, but it is doubtful where this signifies a marked difference in their semantics.
Take for example, meddlesome [1615] which is defined as “given to meddling or interfering; characterized by meddling”, or wranglesome [1817] “given to wrangling, quarrelsome”, or picklesome [1885] “inclined to mischief”, or perilsome [1593] “full of peril, dangerous”.

20Another issue of inconsistency or problematic classification in the OED can be gleaned from an adjective like eyesome [1587] “pleasing to the eye; attractive”. Eyesome is given as a Nsome formation which falls into the category “related to N”. However, the sense of eyesome is far more specific than would suggest this paraphrase. Instead, however that eyesome can be interpreted as Vsome (“intended to be eyed”) in order words as an involuntary object of an eventive structure. Although the occurrences provided in the OED range into 1950 and 2002, eyesome appears as rare, poetic or regional.

(1) 1950 Billboard 24 June 44/3 Cole’s partner, a honey-haired eyesome lass.

(2) 2002 M. Sangster in Lallans 60 57 The weemen wi hair up in net caps An skinklin aprons minded me o days gaun by... An those eesome ghaists. Were veesitin, the blin oors o Daith forgot.

21This recent usage of eyesome appears archaic or regional (“eyesome lass”). Rather than suggesting that -some remains a productive (in the sense creative, morphologically available) suffix, it seems to indicate that these are stored archaic forms that are used deliberately. However, this remains to be considered in this study. We turn to the question of how to measure diachronic productivity and potential (continued or prospective productivity) in Section 1.4., after analysing the morphological output and the diachronic timeline of output of -some adjectives.

Figure 3. A morphological typology of -some derivatives

Figure 3. A morphological typology of -some derivatives
  • 13 See Szymanek [2015: 155] on tautology in word formation, in particular affixal tautology: “morpholo (...)
  • 14 Jespersen [1942: 457] “In connexion with a numeral -sorne represents OE sum ‘some’ as used after a (...)

22The morphological makeup here in Figure 3 is based on the raw data from 261 adjectives: it shows Nsome outnumbers the other types at 44%, which Vsome is a second strongest morphological type at 26%. A remaining 30% of cases are either N/Vsome without clear distinction, or ADJsome (such as wholesome [1200] “promoting well-being, salutary, beneficial”, falsesome [1533] “untrue, deceitful”, or quietsome [1595] “quiet, that produces a feeling of peace”)13. The remaining OTHER category is represented by 20 -some forms, including 12 -some nouns from mostly numeral bases forming nouns referring to a collective group14, such as foursome, threesome, thirdsome, othersome, etc. These derivatives are of a different nature altogether as they form nouns of groups with the sense “group of” (so will be disregarded for the remainder of this study which focuses on adjectival affixation). The 8 remaining -some adjectives are not given as derivations in the OED, either because non-transparent (i-some [OE] “unanimous, agreed reconciled, at peace”) or because they stem from a deformation or alteration of a prior word, such as dubersome [1818] “dubious, doubtful, duberous” given as a “corruption of dubious”, hugesome [1555] “erroneous alteration of ugsome horrible dreadful, compare hugge” or halesome [1200] “alteration of wholesome. Finally, some have unknown origins as in puchersome [1400] “of uncertain meaning: perhaps difficult, troublesome, devilish”.

23Figure 4 below shows what happens when the morphological makeup is contrasted with the feature analysis:

Figure 4. Morphosemantic patterns of -some derivatives

Figure 4. Morphosemantic patterns of -some derivatives
  • 15 Regarding the existence of deadjectival adjective suffixation, “As far as motivations for innovatio (...)

24The preliminary conclusion is that there appears to be a strong correlation between key word features and base word type. This means that Nsome = full of N, Vsome = TENDENCY to V or INTENDED FOR V. The only area of overlap is CAUSING N which is shared by Nsome and Vsome alike. However, the question remains: how does Nsome evolve into Vsome? What motivates that change? What does this say about historical derivation? Is Vsome a natural extension of Nsome, and if so how? Other questions arise, although we won’t address this here; for instance, what is the correlation between the approximate feature of ADJsome derivation (which resembles -ish affixation) as in uglisome, darksome, wearisome15? As for the GROUP OF category, it is represented by a small number of nouns formed on numeral bases, which form a separate category given that this suffix -some is not related to the adjectival suffix; they will be disregarded.

1.4. Nsome and Vsome reanalysis

25Observation of the data collected suggests there is evidence that a N/V convergence may have taken place as early as 1450. Several examples of semantic shift in existing Nsome derivatives are indicative of a reanalysis from Nsome to Vsome formations. Eyesome has already been mentioned earlier, where there is some evidence to suggest that the morphosemantic pattern is readily interpretable as eventive rather than being a relational adjective “related to the eyes”.

  • 16 This reading of derivations as eventive might plausibly have given rise to the spike in Vsome forma (...)

26To continue to investigate this point, take the example of the adjective handsome [1440], which is categorised as Nsome. The sense of handsome is initially given as meaning “easy to handle”, which activates an eventive sense “easy to V” rather than a descriptive sense. In this context handsome seems synonymous with non-existent *handlesome. Similarly, feelsome 1450 also appears to activate an eventive sense, although the OED suggests that it may be a case of ADJsome derivation from the adjective feel signifying “good”16. Another instance of shift concerns the earliest listed adjective in the OED: winsome [OE] is derived from the OE noun win “joy, pleasure, delight” (which apparently became obsolete sometime after 1700 in Early Modern English). The classification of senses in the OED entry for winsome shows a shift between a Middle English sense (“pleasant, delightful, agreeable”, and “kindly, gracious, merciful”), to a Modern English sense [1677] “handsome, comely, of winning characters or manners”, and the regional sense “cheerful, joyous, gay” [1787]. The semantic shift in winsome from “joyful” to “attractive” appears to hinge on the reanalysis of the base as the verb win, from which a transparent interpretation becomes “intended to win over, apt to win over”, in other words, winning or attractive as it is explicitly suggested in extract (5). It is notable that a corpus search for winsome in COCA produces 361 occurrences, 457 in COHA, 840 in OEC, and 0 in EHBO. Winsome is repeatedly found as a premodifier to smile (24 occurrences in OEC out of 840, and 10 occurrences out of 457 in COHA, 17 out of 361 in COCA).

(3) Sara savored his winsome smile before returning to her customer. “I’m assuming you’re here to talk about planning a wedding, Mr. Porter. If you can wait, I’ll find you a planner as soon as I get these two settled. I have to watch them until my brother returns. As you’ve noticed, they take a lot of watching.” # “I did notice.” Cade Porter’s lips lifted in a grin. He squatted down and spoke quietly to the twins, showing them. COCA, 2009.

(4) upon only one were her brightest glances and her most winsome smiles lavished, and that was George Clayton, a young man from South Carolina, who was said to be very wealthy. COHA, Cousin Maud and Rosamond, 1860, Holmes.

(5) there’s a distinct, welcome absence of cuteness. Many child actors try to win us over with winsome smiles and charm; that’s not the case here. Meanwhile, Max Minghella does the best he can with a choppily written role. (Aaron would have benefited greatly from a handful of additional scenes.) Richard Gere and Juliette Binoche are professional, although Gere fares better here. OEC, Movie reviews, 2005.

27A study of the semantic shift in all Nsome adjectives would be of interest here (see Smith [2016] and [2019] for this approach to phonesthemes), but this won’t be the focus here, as we aim to consider the competition between Vable and Vsome adjectival derivation. However, what is of interest is that morphological productivity is not as straightforward as the evidence in the OED etymologies, since diachronic semantic shift appears to indicate a plausible shift from Nsome to Vsome which may have led to the sudden productive spike of Vsome between 1750 and 1850.

1.5. Historical output in the OED and evidence of shift from Nsome to Vsome

28The definitions of the 261 some adjectives in the OED3 are analysed into frequently occurring key words which are then grouped into conceptual categories (also see Smith [2016], Smith [2018] and Smith [2019] for a presentation of this methodology). The semantic categories based on dictionary definitions are to be viewed as an empirical tool to track semantic behaviour, and do not represent a suggested semantic core per se. Dictionary definitions do not claim to be systematized, however we posit that they can be a starting point for the analysis of semantic behaviour. Using the 10 larger conceptual features based on the OED definitions, 2 of these features appear to be most frequent. TENDENCY TO and FULL OF are by far the most frequent raw output, i.e. the total number of words falling under this category). The CAUSING feature appears to be somewhere in the middle, as is the RELATED TO feature. The remaining features fall behind in terms of raw frequency, as is clearly visible in Figure 5:

Figure 5. Raw output per conceptual feature

Figure 5. Raw output per conceptual feature

29Figure 6 shows the compared raw output for the two most frequent features (TENDENCY TO and FULL OF), and we have included CAUSING N in this category since this feature is not as infrequent as the remaining features. A clear spike is visible for the feature TENDENCY TO from the period starting 1750 reaching a peak around 1850 and sharp drop off after 1900. Note that the sense “moderately Adj” correlates with ADJsome derivatives (such as deepsome, roughsome, lithesome), and that the sense “group of N” correlates with numeral formation (twosome, threesome, etc.).

Figure 6. Compared raw output between the 3 main features

Figure 6. Compared raw output between the 3 main features
  • 17 From the base word lumber 1400 in the OED given as “Possibly two or more words may have coalesced. (...)

30A comparison between raw output of the three main features, and the three morphological types of -some derivation shows there is some clear overlap between semantic and morphological types (see Figure 6). Vsome adjectives are generally associated with TENDENCY TO feature: for instance, lumbersome17 1834 “cumbrous, unwieldy”, i.e. “with a tendency to lumber”, formed along a similar pattern as capersome 1852 “given or inclined to capering”, whereas Nsome are associated with FULL OF feature (such as joysome [1616], angersome [1650], griefsome [1635]).

31However, as suggested by a few cases of semantic shift (eyesome, handsome), evidence of reanalysis is also present in some of these Nsome derivatives: in addition to having the sense “full of” they can also be interpreted as “causing N” with an eventive reading. This evidence of potential shift from a static to an eventive reading is available among the 116 Nsome forms listed in the OED. Out of the 116 Nsome adjectives, 16 adjectives are compatible with a CAUSING (i.e. causing or inciting) sense: loathsome [1300], lustsome [1300], healthsome [1548], troublesome [1548], burdensome [1598], tulyiesome [1598], favoursome [1601], griefsome [1635], angersome [1650], frightsome [1689], hurtsome [1699], bothersome [1817], riddlesome [1843], clamoursome [1855], picklesome [1885], chillsome [1927]. This shift can be interpreted as a natural semantic shift based on metonymisation, but this natural shift may have led to an extension of the derivative to a Vsome base.

  • 18 The etymology of the base of frolic is unclear: the OED gives the adjective as being either Vsome o (...)

32Further evidence of shift can be seen in the number of derivatives with an either/or base. Out of the 261 -some adjectives, 26 are labelled as having an “either/or” base word, in otherwise a non-specific class of base word. The semantics are compatible with either a nominal or a verbal base. These blurry lines can be seen in a variety of derivatives given as Nsome or Vsome such as frolicsome18 [1699] “full of frolic, gay, merry, mirthful”, “causing frolic”, tripsome [1819] “characterized by tripping, nimble”, worrisome [1702] “apt to cause worry or distress, given to worrying”. These adjectives can all be interpreted from either a nominal or verbal root and are compatible with a stative or an eventive meaning, which justifies the need for context-individual research, as will be carried out here in the following sections. The episodicness of the characterisation seems to a defining factor.

Figure 7. Compared output of Vsome, Nsome and ADJsome

Figure 7. Compared output of Vsome, Nsome and ADJsome

33Figure 7 shows 2 spikes in Nsome ouput, and the single spike for Vsome overlapping with the increase in “tendency to” feature in Figure 2. This suggests that the output increase is in fact related to the performance of Vsome derivation from 1750 onwards, which suddenly outperformed the historical Nsome formation. However, this Vsome spike may be artificially emphasized due to uneven data distribution for this period. Culpeper & Clapham [1996] show there is an increase in the OED coverage of the period, which leads us to question the homogeneity of available data for different periods of the timeline of the English lexicon, see Culpeper & Clapham [1996: 202]:

The most serious problem encountered by any chronological study of the OED concerns the fact that more citations come from some periods than others. This is for a variety of reasons. Radically different amounts of material were available in different periods. Certain periods, notably the Elizabethan period, seem to have been favoured for their literary pre-eminence. Important medieval and Renaissance authors, such as Chaucer, Gower, and Shakespeare, are represented exhaustively (OED 1989: xlviii). Moreover, these periods are characterised by the ready availability of textual editions for scrutiny. And sometimes organisation of the readers of the OED was less than rigorous, so that certain periods suffered.

34Therefore, it is likely the spike in Vsome may be artificially boosted due to the data concentration in the 1850s. A second issue that affects the timeline results is the reliability of data provided in particular, since attestation dates which are well-known to be approximate (Trips [2009], Durkin [2016a]), which is highlighted specifically by Durkin [2016a: 404] regarding the extensive ongoing revisions to the OED. This revision has led to a number of antedatings as the OED revision process benefits from the increased availability of large diachronic databases with reliable annotations, see Durkin [2016a: 406]:

[W]hat is reported in historical dictionaries is based on analysis of the evidence available at time of publication of the dictionary entry, and may well be subject to review if and when further evidence comes to light. First dates of attestation are particularly subject to change, as new evidence becomes available, and as the dating of existing evidence is reconsidered. In particular, the increased availability of electronic text databases in recent years has swollen the flow of new data to a torrent.

35Although the OED is not unproblematic, the use of a diachronic corpus would produce the same if not more limitations; not all the vocabulary of English can be found in one place with completely homogeneous data for each period. The OED remains a valuable tool of diachronic study that can be complemented and combined with a corpus analysis. There are two related issues to resolve:
1) firstly, checking the actual frequency of usage of
Vsome adjectives in a diachronic corpus would test the propagation abilities of the pattern,
2) and considering what causes lie behind the productivity drop-off of
Vsome and Nsome from 1900s, i.e. the apparent total loss of productivity of the suffix -some.

36In what follows, we will be using corpus testing using a number of historical and contemporary English corpora to check for the usage and frequency of the adjectives under study: COHA (Corpus of Historical American), EHBC (English Historical Books Corpus), Project Gutenberg Corpus, OEC (Oxford English Corpus), and COCA (Corpus of Contemporary American) will be our reference corpora. For reference, EHBC contains over 800 million words of text (published words) and covers the older periods of English from 1473 to 1820 (Texts are from EEBO Phase I, ECCO and Readex’s Evans projects). COHA contains more than 400 million words of text and covers a later period of English from 1810s-2000s and the corpus is balanced by genre decade by decade. Project Gutenberg English contains 400 million words of text from all English e-books available in the Gutenberg database. As for the contemporary corpora, COCA contains more than one billion words of text and covers a contemporary period from 1990-2020. The second contemporary corpus, the OEC, contains 2 billion words of text covering the 21st century (mainly websites chosen in the way of presenting all types of English, from literary novels to everyday newspapers and the language of blogs and even social media, see also Brinton [2016: 204]).

1.6. Preliminary conclusions and morphological competition

37The analysis of the data retrieved from the OED confirmed two distinct trajectories for Vsome and Nsome formation, it revealed that the main conceptual features related to morphological structure. Two features outperform all others in terms of raw frequency: FULL OF N 67, TENDENCY TO V 75 which both correlate to Nsome formation and Vsome formation respectively. The remaining semantic categories such as INTENDED FOR V 26, RELATED TO N, GROUP OF N, CAUSING N have fewer than half as many members.

38This appears to consolidate the notion that two distinct patterns exist, one emerging historically as denominal Nsome, and the second emerging in the 1750s as deverbal Vsome. The emergence of Vsome does appear to be consistent with the semantic shift of Nsome into Vsome, suggesting that Vsome is in fact a natural extension of Nsome. Evidence of early eventive interpretation of Nsome (like eyesome, handsome, winsome), and the hybridity of a number of -some derivatives suggest the shift is naturally motivated in language change. In addition, the conceptual analysis of the secondary features of -some derivatives appear to derive, or at least be related, to primary features though metonymization; INTENDED FOR is related to TENDENCY TO via active/passive change, and CAUSING is related to FULL OF (via dynamic/ static change).

39If Nsome and Vsome are compatible patterns and follow consistent growth over time, how can the apparent lack of productivity after 1900 be explained? There are several plausible reasons for the obsolescence:
1) The sudden decrease in output may be related to unequal distribution of data across time periods in the
OED, as shown in Figure 8, based on data provided by the OED3 website:

Figure 8. Timeline of new words registered in the OED from 1000-2000

Figure 8. Timeline of new words registered in the OED from 1000-2000

402) A related plausible theory is that the OED registration process of recent new words is slow. There is evidence of fewer inclusions post 1900 and especially post 1950 (see also Simpson et al. [2004], Durkin [2016a] on the OED’s revision process). Lexicalisation is known to be a diachronic process (see Lipka et al. [2004: 6]), and the lexicalisation criteria in the OED are based on frequency and continued propagation of usage across text types and over a certain time period. Words attested historically tend to become fossilised, in the sense that they become unanalysable relics – of ME or even of OE (such as handsome).
3) Finally, the decrease in output could be a sign of the lessening productivity of a suffix that has been outperformed by a multitude of available suffixes forming alternatives, i.e. possible synonyms (-
ish, -ful, -y, ing, -able, etc.).
4) -
some may have waned due to semantic instability, i.e. lack of transparency. There are a number of studies investigating competing suffixes and diachronic productivity and competition: Baayen & Lieber [1991], Trips [2009], Lindsay & Aronoff [2013], Arndt-Lappe [2014], Schulte [2015], Esteban-Segura [2018]. Morphological productivity is found to be linked to the transparency of derivative as argued in Plag & Baayen [2009: 125]: “productive processes are semantically and phonologically transparent”. This means that semantic ambiguity may be a factor in facilitating obsolescence or productivity in that it delays processing ability (see Plag & Baayen [2009: 141]).

41In what follows we will focus on considering the motivations for the loss of productivity of -some derivation, in particular with the regard to the highly productive and well-studied suffix -able (see Aronoff [1976], Kjellmer [1986], Di Sciullo [1997], Schuwer [1999], Plag [2003], Bauer [2004], Oltra-Massuel [2014]). An observation of the OED data shows that a non-negligible number of -some adjectives are paraphrased with Vable adjectives, indicating a potential semantic overlap, i.e. synonymy. A search in the OED data shows that 42 of the -some adjectives (i.e. 42/261, 16%) have -able words in the dictionary paraphrase provided by the OED such as: batsome [1555] > battable (obs rare); favoursome [1601] > that is the object of favour (acceptable); harboursome [1596] > hospitable, lovesome [OE] > lovable; metesome [1674] > measurable; likesome [1565] > likeable, pennisome [1631] > profitable (lucrative); gainsome [1579] >profitable; ruesome [1833] >pitiable; handsome [1440] > admirable; relishsome [1593] > relishable. Of course, a caveat here is necessary, as this methodology is simply a shortcut to detecting synonymy in the first instance. The OED definitions show variability in treatment, are not systematized in that regard and can only serve as an approximate tool for semantic analysis (as previously stated). Many of the adjectives listed above are attested in Late Middle English and Early Modern English.

42In Section 2, I will now provide a brief overview of -able derivation, which has extensively been researched, before pursuing the study of suffix competition in Section 3.

2. -Able versus -some competition: evidence in the OED

2.1. -Able: a well-studied loan suffix with a modal meaning

43In stark contrast to the limited research on -some adjectives, there is very extensive interest in -able adjectival derivation, as illustrated by the number of publications devoted to -able or using -able as an instance of exemplary derivation: Jespersen [1942], Marchand [1960], Aronoff [1976], Kjellmer [1986], Di Sciullo [1997], Schuwer [1999], Plag [2003], Bauer [2004], Oltra-Massuel [2014].The suffix -able is frequently discussed as an example of morphological productivity, since it is widely known to be extremely productive to this day. This leads to studies of the constraints bearing on the productivity of -able derivation, as Baayen & Lieber [1991: 809] demonstrate:

The suffix -able, which forms adjectives from verbs, attaches only to verbs with an appropriate argument structure; potential bases for -able must have both an external and a direct internal argument (washable, *snorable). So among the productive affixes we must be able to distinguish different degrees of productivity. Baayen (1989) develops a number of statistical measures.

  • 19 Subsequently the suffix was extended to form denominal adjectives such as saleable adj. (first half (...)
  • 20 The issue remains whether Nable and Vable are the same pattern or not, and whether there is converg (...)

44The OED suggests that the non-transparent word forms compatible with Vable and Nable due to the dual nature of the base words led to the extension of a Vable pattern to a Nable pattern. For instance, reasonable [1325] is a borrowing from French that is potentially reanalysed from Vable (raisonner in the donor language) to Nable in English with the sense “full of reason”. Historically, -able stems from borrowing of French and Latin forms, with the formation extending to English bases fairly quickly19 (from the 1350s according to Jespersen [1942: 398]). It is also suggested by the OED that reinterpretation of French and Latin loans encouraged the analogy, due to the transparency of the borrowed base words. Three factors appear to have contributed to the success of -able derivation:
1) Firstly, the importance of morphological reanalysis of -
able words appears as a major factor in the extension of -able to English word formation according to Jespersen [1942: 398]: “able was treated as a living suffix, mainly because of form-association with the adjective -able”.
2) Secondly, the high productivity coupled with the transparent interpretation of -
able derivation probably led to it becoming highly regular in Middle English where “it is possible to form adjs in -able from practically any verb” (Jespersen [1942: 400]).
3) Finally, a third factor has impacted -
able derivatives, and that is the blurriness of the base word leading to the extension of the derivation from Vable to Nable (Jespersen [1942: 402-403]). In this regard, it can be noted that -able derivation appears to follow a similar pattern to -some derivation, in that it is compatible with both denominal and deverbal formation20.

45The literature on -able traditionally describes two morphosemantic types of -able adjectives. Kjellmer [1986: 12] argues that there are two semantic categories of -ble (this notation includes both -able and -ible spellings, which are viewed as a single pattern) adjectives in English, those that are neutral and objective in their eventuality (possibility, “pure potentiality”), and those that express subjective opinion (suitability, likelihood). Kjellmer’s [1986: 26] corpus experiment of -ble words confirm that the suffixed -ble adjectives fall into two interpretations, and that frequency of use is a factor in triggering the more subjective sense of the adjective. However, -able appears to show a high degree of semantic regularity over time:

[F]requent words are semantically more complex than infrequent words, at least as far as their polysemous character is concernedbut -ble words are particularly interesting in the semantic regularity of their development. [Kjellmer 1986: 26]

  • 21 « Le suffixe de ces adjectifs attributifs est, comme pour les adjectifs épithètes, porteur de plusi (...)

46Plag [2003: 94] argues that the two Vable have a core sense that draws a comparison with Vee derivatives in that both cases there is a reference to the non-volitional participation in an event. Oltra-Massuel [2014: 17] distinguishes a high -able and low -able meaning. The terms high and low are based on transparency and decomposability: one type of -ble adjective is “idiosyncratic and lexicalized”, whereas the other is “regular and transparent’. As far as whether Vable and Nable are the same pattern, both Di Sciullo [1997: 89-90] and Oltra-Massuel [2014: 303] share the position that it is a case of extension of Vsome to Nsome. For Di Sciullo, Vable and Nable bases denote a transitory property, which is not limited to a particular grammatical category, whereas Oltra-Massuel views both roots to be of an eventive nature “The suffix -ble does not attach to verbs, it needs an eventive (verbalized) root that expresses a transition and that can meet the two requirements imposed by -ble.” Overall, there is consensus in that there are two readings of Vable adjectives, which either refers to a potential or an actualized participation in an event (via euphemistic usage). The distinction in senses can also be correlated to the distribution of the adjective, as either a predicative or premodifier use of. If Vable is used as a premodifier the sense is potential, if Vable is used as a predicative the sense tends to be deontic (Schuwer [1999: 17]). Schuwer [1999: 21] suggests the deontic sense is more widespread21:

  • 22 [To sum up, when the adjective is used predicatively, the prediction is based on the properties of (...)

Pour résumer, lorsque l’adjectif est en position attribut, la prédiction se fonde sur les propriétés de B, évaluées dans la perspective de leur impact : l’événement est perçu comme une conséquence (hautement probable) des propriétés de B. Tandis que lorsque l’adjectif est prénominal, c’est à partir de ses propres connaissances que l’énonciateur déclare la haute probabilité du procès. [Schuwer 1999: 17]22

47To summarise, contrary to the native Anglo-Saxon suffix -some, -able originates from extensive borrowing from French before becoming an internal formation process, thereby potentially competing with other pre-existing lexical forms and existing productive patterns. The process of internalisation of the loan suffix follows three stages: i) the extension to English verb bases; ii) the extension to nominal bases; iii) the continued ability of the pattern to produce new words that remain in use (i.e. active continued productivity). This overview and semantic makeup of Vable indicates a degree of similarity with -some adjectives from two perspectives: 1) the dual nature of the root, and 2) the eventive characterisation of the derivation.

48However, to our knowledge, with the exception of Dixon [2014], these similarities have not been commented on in any prior literature, possibly because the suffix -some is generally disregarded because it is viewed as a non-productive suffix. As we have shown in Section 1, -some suffixation is compatible with eventive readings of the root and also appears to focus on potential realisation of causation effects/ resultative effects. This is evident in uses such as worrisome, bothersome, which remain in frequent usage today: “which is likely to cause bother”, which actually causes bother. The key feature analysis showed that this eventive reading is particularly active from 1800-1900 with a peak in the 1850s.

49This allows us to now move on to the issue of the productivity of -able, and to considering the causes behind the obsolescence of one derivation and the continued success and productivity of the other.

2.2. The OED output for -able adjectives

50As previously mentioned, -able adjectives far outnumber -some adjectives in the OED, which in addition show a large number of obsolete or regional word forms. The output according to the OED3 data shows -able adjectives total 3700 entries (although this includes a few false hits with homonym able). -able is therefore considerably more productive than -some adjectival by over tenfold.

Figure 9. -able output per century 1200-1999 based on origin

Figure 9. -able output per century 1200-1999 based on origin
  • 23 One of the many dating inconsistencies already noted, as the OED does not always have access to fir (...)

51Figure 9 represents the output of -able adjectives based on loan words versus internal formations when -able becomes productive in English. An origin search in the OED provides out of a total of 3700, 300 French origin -able word forms, and 400 Latin origin -able headwords, and 2600 English origin words. The diachronic distribution is shown in the figure above. As we know, the data provided in the OED is not entirely reliable especially regarding origins. For instance, the first “English language origin” adjective provided delitable (obs.) 1290 turns out to be a loan from French, and the other adjective unstable 1225 (dated before stable 127523) is also actually of French or Latin origin. The number of direct loans from a European language (i.e. French) are numbered at 700 only, which is likely an underestimation. The first loan adjective listed is changeable [1275] (from French).

52Of course, the nature of productivity of loan morphology poses a different issue, that is internal productivity versus external productivity. The raw number of -able derivatives is not a reliable measure of diachronic internal productivity, however the measure of non-borrowed -able adjectives can be interpreted as a measure of internal productivity. Figure 8 shows that English formations remain consistently numerous and outnumber loans as early as 1300, taking an increasing lead from 1700 where the number of loans from Latin and French become extremely rare.

53The importing of derivational morphology is a well-known phenomenon, specifically for English which has borrowed a large number of French and Latin suffixes (such as the -age suffix, which remains an active suffix in English today, see Smith [2018]). Gardani et al. [2015: 13] explain that derivational morphology is more susceptible to borrowing than inflectional:

The general consensus about this claim rests ultimately in the abundance of derivational borrowings in the most studied language of the world, English, dating back to the time when (Middle) English extensively borrowed from French.

54As for the borrowing of full words, it precedes morphological borrowing. Research has shown that core meanings are more resistant to borrowing, as Zenner et al. [2014: 74-75] state: “A traditional claim in contact linguistics holds that core vocabulary is highly resistant to borrowing [...]”. The question of a loan word then outperforming a native word is addressed in Durkin [2014], [2016a]: Durkin [2016a: 393] provides some evidence of positive outcomes for borrowed words, such as carry/bear, cry/weep, soil/earth, ground/dirt. What factors induce increased frequency of one form over the over, change or obsolescence? Kay & Allan [2016: 225-22] argue that semantic clash or semantic instability, in other words, homonymy and polysemy, are at issue: this is illustrated through the example of farm in the sense “cleanse, empty, purge”, which disappeared due to pressure from the frequent core word farmer.

Examining farm and farmer leads to a possible example of homonymic clash, where two words of different origins happen to have the same form and are semantically incompatible. As with polysemic clash, homonymic clash can lead to obsolescence or change of meaning the homonymy between farm and farmer leads to reanalysis and problematic ambiguity.”

55Thus, there is evidence of borrowed words becoming mainstream and overshadowing loan words due to pressures on the system. So, although -able is the imported suffix, it quickly became an internal suffix which then rapidly overshadowed -some affixation in terms of productivity. The overall number of registered -able adjectives outnumber -some derivatives by no less than ten to one. The question remains: what multiple influences may have led to this overwhelming change? As pointed by Durkin [2011], it is very difficult to determine and identify all of the “complex interplay” of influences which altered the continually evolving ecosystem (to use the term in Renner [2020]), leading to such striking changes. In our attempt to identify some of these influences, the following step is a key word comparison of -some and -able derivation using the OED definitions.

2.3. Verifying the key words of - able adjectives in the OED

56Not only would a manual semantic feature analysis of -able adjectives in the OED be too time-consuming given there are 3700 entries to analyse, but it would also create a large discrepancy in data between -some and -able adjectives. So as to maintain a balance of data between the lower frequency and higher frequency derivatives, an automatic search was carried out based on a preliminary analysis using the OED search system of 3700 -able adjectives listed. The search for semantic patterns is based on the knowledge acquired from the extensive literature on -able suffixation. The key words are modal auxiliaries and modal adjectives.

Figure 10. Key features of Vable adjectives in the OED

Figure 10. Key features of Vable adjectives in the OED
  • 24 This descriptive typology has the advantage of being detailed and empirically-based rather than for (...)

57Figure 10 shows the raw number of occurrences of each feature in -able adjectives in the OED. These semantic features are of course by no means exhaustive, nor do they appear in competition with one another, rather the key words in *able adjective definitions tend to co-occur. They are listed here with the total frequency of use of each key word: capable (1000), can (700), may (1200), likely (81), liable (222), deserving (57), intended (9), suitable (166), worthy (78). The key words search is based on prior literature regarding the semantic structure of -able derivatives, as noted in Section 3.1. notably with Oltra-Massuel’s [2014] semantic typology24: active (agreeable, perishable ‘that V’), passive (modifiable, realizable ‘that can be Ved’); denoting possibility (modifiable, realizable ‘that can be Ved’); expressing some evaluative judgment (admirable, enviable ‘that should be Ved’), or being causative (horrible, terrible ‘that causes N’).

2.4. Overlap of -able and -some: Causing/ tendency features: testing passive/ active alternation

58From the analysis of the trajectory of -some adjectives, and a preliminary overview of -able adjectives, it is apparent that the output rates vary vastly. In terms of semantic makeup, it is possible to suggest that both suffixations lead to a partial semantic similarity based on eventive readings. Vsome (and some technically Nable) adjectives correlate with a resultative meaning, “causing x”, “intended for V”. On the other hand, Vable adjectives correlate with an eventive reading of eventuality “capable of/ likely to”. We can conclude that -some appears more compatible with an active reading and -able with a passive reading. The question therefore arises: does active-passive account solely for unacceptable Vables?

  • 25 As any corpus researcher knows, one of the main constraints is the choice of corpus which is liable (...)

59This question deserves investigation here, so I aim to analyse some alternations of Vsome adjectives in comparison with potential Vable counterparts. In order to achieve this, I cannot rely on quantitative data gleaned from the OED. Attempting to determine what percentage of Vsome adjectives have an explicit Vable counterpart in the OED would not be quantitatively reliable since the OED definitions do not have a systematic nature. Since a quantitative approach is not feasible, a closer look at a few individual word studies will provide some preliminary exploratory data. From the Vsome adjectives in our data set, most -some adjectives entering in competition with Vable tend to be classified in the active category “Causing N” (inciting N, creating N) or “inclined to”, with an eventive or resultative sense. Of the 74 some adjectives in the TENDENCY category, no fewer than 61 -some adjectives have a verbal or N/V base. Of these adjectives, we will examine a few cases of -some adjectives with and without -able counterparts, using three historical corpora (COHA, EHBC and Project Gutenberg Corpus) and two contemporary corpora (COCA and OEC) for comparison25. The value of word histories in linguistic observation is recognized in Durkin [2011: 103-104]:

The contributions [in Allan & Robinson 2011] by Allan and by Kerremans et al. show the value of close attention to the detail of individual word histories, in conjunction with an understanding of how these individual word histories interact with the larger structures of the lexicon of the language. All three contributions remind us of the complex interplay of factors involved in any instance of lexical change.

  • 26 Durkin [2016b: 252] on treating words in historical dictionaries: “However, words are units in comp (...)
  • 27 Also see Geeraerts [2010: 76]: “Rather than being the opposite of a more traditional hermeneutic ap (...)

60Word histories offer a valuable insight into complex sets of influences that help gather data setting a precedent for theorising of semantic change (see also Durkin [2016b: 25226]). Word histories also help to focus on the historical and individual contextualisation of diachronic change (see Geeraerts et al. [2012: 11]), allowing for future hypothesis testing based on larger sets of data27. The selection of a sample of Vsome adjectives for study here was determined by two requirements. The first was the need to compare Vsome adjectives with a verifiable explicit Vable counterpart with Vsome adjectives that do not have a Vable counterpart. Secondly, the selection was determined by frequencies of usage in the corpora: it was necessary to use words for which occurrences were available across the historical corpora. Given the relative infrequency of many -some adjectives, this selection was necessary and justified by the need to study occurrences. 9 Vsome adjectives from the active TENDENCY TO category will be compared with the corresponding Vable derivatives.

61The OED data show that -some adjectives are compatible with TENDENCY TO sense in an active sense: 74 -some adjectives are compatible with the feature of “tendency to/ given to/ inclined” with a more active sense. The OED gives formulations such as “apt to, given to, addicted to, prone to”: worrisome [1702] “apt to cause worry, given to worrying”: both passive and active sense; venturesome [1661] “disposed to venture”; toilsome “hard-working, given to hard work; temptsome [1849] “apt to tempting”; growsome [1579] “apt to grow”.

62I will begin with a comparison of three Vsome adjectives: the first has no Vable counterpart (meddlesome 1615), the second temptsome [1849] does have a Vable counterpart (with an opposite sense) and finally tricksome [1815] does not have an attested Vable counterpart.

Table 3. Vsome adjectives in the TENDENCY TO category and their competition/ counterparts

Vsome adjective with the active sense TENDENCY TO

Synonyms provided by the OED

Potential -able counterpart passive sense?

meddlesome “given to meddling”



temptsome “apt to tempt, tempting’



tricksome “playful, frolicsome”

playful, frolicsome


wieldsome “easily managed, controlled, or handled”



savoursome “having a pleasant or savoury taste or smell; (also) able to be savoured”,

No alternatives

savourable (?)

laughsome “inclined to laughter; mirthful”



frightsome “causing fright; frightening, frightful’

Later sense “frightened, fearful”

frightful, frightening

frightened/ fearful

*frightable/ *fearable

fearsome “Fear-inspiring; frightful, dreadful”.

Later sense “? erroneous. timid, apprehensive, frightened.”

frightful, dreadful


*frightable/ *fearable

awesome “Arousing or inspiring awe”

Later “In weakened use: staggering, prodigious, huge; remarkable, striking; (in negative contexts) challenging, daunting; (in positive contexts) notably good or impressive”.

No alternatives



63The adjective meddlesome [1615] has an active sense described in the OED as “Given to meddling or interfering; characterized by meddling”, i.e. falls into the category TENDENCY TO. The hypothetical *meddlable would have a passive sense, “that can be meddled, or should be meddled with” and is not acceptable nor attested. The adjective meddling [1529] “That meddles; interfering” according to the OED definition is a better synonym but lacks the typicality aspect “given to”. Meddlesome is one of the more frequently used remaining -some adjectives in contemporary English. A corpus frequency search for meddlesome in both historical and contemporary corpora provides the following tally: COCA 278 occurrences, EHBC 7, PG 331, OEC 400, BNC 11.

64On the other hand, the adjective temptsome [1849] is given as rare and defined as “Apt to tempt, tempting”, with an active sense TENDENCY TO in direct competition with the pre-existing adjective tempting [1400]. The -able counterpart temptable 1628 has the opposite passive sense “That may be tempted; liable or open to temptation”. Temptable has 5 occurrences in COHA, 8 occurrences in EHBC and temptsome has no occurrences in COHA, 0 in EHBC, 0 in OEC. Similarly, tricksome [1648] has an active participant reading “given to playing tricks”. The OED provides three distinct senses for tricksome, with usage examples providing adjectives bearing on a human referent, [1815] “playful, frolicsome”, which refers to human beings or behaviours (gait, graces), and [1820] a use referring to musical abilities. Trickable is not listed in the OED (nor in Merriam-Webster), although the word is plausible with the sense “apt to be tricked; capable of being tricked”. A search in COCA found 2 occurrences (BNC 0, COHA 0, OEC 0) both taken from a US TV series as shown in (6):

(6) Blukic: It’s like they don’t want us to test our Tachyon drive. Ben: Blukic! Driba! Is that you? Blukic: Who else could it be, Albedo? - Driba: Some randomly selected Galvans who have elected to bring you food? Ben: I-It’s me... Ben! Albedo switched places with me! You’ve got to get me out of here! Driba: That’s exactly the sort of thing Albedo would say to trick us! Blukic: And we are not trickable.

65The adjective wieldsome [1565] is given as rare and obsolete and has a passive sense “Easily managed, controlled, or handled”. This infrequency is verified by a corpus search in EHBC and COHA: wieldsome: COHA 0, wieldable 0; wieldsome EHBC 1, wieldable 1. Wieldsome appears synonymous with wieldy [1413] in the sense “Easily handled, controlled, or used; manageable, esp. in size or number.”, although the semantic trajectory of wieldy shows dramatic semantic shift, notably with the opposite sense of “unwieldy, Difficult to control, manage, or use” appearing 1588. This semantic development into the oppositive meaning is consistent with the idea that -some may be a semantic unstable affix (possibly due to lack of transparency, the word no longer tending to be analysed). On the other hand -able is analysed separately, and even confused with the adjective -able which has probably ensured its survival and accelerated its propagation (see Dixon [2014]).

66The adjective wieldable [1688] is attested in the OED, although much later, in the passive sense capable of being wielded; easily handled, used, or manipulated.”

67Another example of a passive adjective is savoursome [1595] “having a pleasant or savoury taste or smell; (also) able to be savoured”, which is in direct competition with savourable [1485] “capable of being savoured”. However, this adjective is given as Nsome in the OED, despite the passive sense, which points to a probable reanalysis as Vsome. If we compare this pair with another synonymous pair flavoursome and flavourful, it becomes apparent that the latter pair is exclusively denominal “full of N”, without a Vable counterpart *flavourable. This example indicates that the denominal to deverbal shift in Nsome adjectives might be a factor in the obsolescence of -some adjectival derivation. The data collected don’t take into account the apparently frequent semantic shift in -some adjectives from denominal to deverbal, and from active “inclined to” to passive “capable of being Ved”.

68Take the adjective laughsome [1612] “inclined to laughter; mirthful”; [1798] “that causes laughter, amusing”. The second sense is attested in the OED as late as 1944 and 2009.

(7) 1944 Joplin (Missouri) Globe 27 Feb. 2/3 Others among the 16 all-star acts include the laughsome mimicry of Stan Greenspan.

(8) 2009 Independent (Nexis) 9 Mar. (Extra section) 4 Pitch-perfect pastiche of America’s dominant musical styles: immediately recognisable, and instantly laughsome.

  • 28 Note a lot of negative -some adjectives of evaluation have remained in use: troublesome, wearisome, (...)

69If we compare laughsome with laughable [1600], laughable has a passive sense only: “able to be laughed at; amusing”. The semantic change visible in this adjective is that it takes on a stronger evaluative judgment28 “Now chiefly: ludicrous, absurd.” Laughable remains in frequent usage in current contemporary English, as evidenced by a corpus search in OEC, COCA.

  • 29 Labelled ‘erroneous’ in the OED definition for this sense of fearsome.
  • 30 Also see Robinson [2010] for a study of the adjective awesome from a cognitive sociolinguistic pers (...)
  • 31 There is potentially an iconic motivation of this semantic shift and heightened success of the adje (...)

70Fearsome [1768] is an interesting example of a hybrid N/Vsome suffixation which is given as meaning “fear-inspiring”, therefore in the category “causing N / inspiring N”. It follows a frequent pattern of adjectives viewed as hybrid which are compatible with both a static sense “full of N” and a dynamic resultative sense “causing X”. A second sense is attested in 1863 as meaning the opposite, i.e. “timid, frightened”. This second incongruous sense29 is compatible with the reading “capable of being frightened”, which represents a kind of analogical confusion between the verbs fear and frighten. It is possible that this misreading may be a symptom of the semantic instability of some adjective suffixation. It is notable that frightsome [1689] follows the same pattern of shift from the sense “causing fright”, to a secondary sense of fearful in 1827. On the other hand, a synonymous adjective awesome30 [1578] has an initial sense of “causing awe, reverential fear”, but then undergoes a different semantic shift, a weakening of its use in first a negative context, and then a positive context (attested from 1916)31. This shift could be perceived as a potential case of subjectification insofar as Traugott [2016: 385] underlines that “pejoration (and some types of amelioration) of lexical meanings can be seen to arise via subjectification and to be used subjectively”.

71The adjective awesome is so widespread in usage that it will be of no surprise that Google n-gram shown in Figure 10 is indicative of the increased frequency of usage of the adjective awesome. This tendency is in very stark contrast with the rarified usage of many remaining derivatives, such as winsome and eyesome:

Figure 10. Google Books n-grams for awesome / winsome / eyesome

Figure 10. Google Books n-grams for awesome / winsome / eyesome

72Of course, using Google Books for diachronic research specifically may appear surprising given its lack of representativity in terms of corpus balance, as pointed out in Davies & Chapman [2016: 147] who wonder at the unexpected accuracy of its results in comparison with the historical COHA. The conclusion is that size is paramount over “modelling”: if a corpus is large enough, sufficient variety will be present.

Proportion and balance from type to type and from decade to decade are also important considerations. Yet Google Books has disregarded these principles of representativeness. The creators of this collection simply scanned everything available in several large university libraries. Unlike COHA, there was no attempt to sample from multiple genres or to balance the selection across groups and decades. And yet Google Books provides data on lexical change (as measured by lexical frequency) that is very similar to that of COHA, which is a well-designed corpus. How can this be?
The answer may be simpler than we think. The concept of representativeness says that we should accurately “model” the entire target population of texts in the “real world”. But what if you have, in effect, the entire target population at your disposal, or at least a sufficiently large percentage of it? In this case, modeling is not as important. The variety of text-types will be taken care of by a sample that is large enough to catch that variety. And this is precisely what Google Books has done.

73After this investigation into the diachronic emergence of Vsome and Vable alternations based on adjectives classified in the active TENDENCY TO category, we will now follow the same approach at Vsome and Vable competition for adjectives belonging to the passive INTENDED FOR category.

2.5. Semantic shift and active/passive instability of -some adjectives

74Out of the OED data collected regarding the 261 -some adjectives, most belong to the TENDENCY TO category with an active sense. There are a much smaller number of -some adjectives given as having a passive sense INTENDED FOR. We count 9 Vsome adjectives which are classified as “intended for”: wieldsome, ticklesome, relishsome, furthersome, metesome, clipsome, tewsome, cuddlesome, hugsome.

75If we consider e.g. the case of ticklesome, there are two distinct senses provided in the OED that correspond to diachronic shift. Ticklesome [1585] is defined as “That tends to tickle; difficult, critical, delicate, precarious, ticklish. Now dialect.”, then [1844] “Easily tickled; tickly; ticklish; suitable or fitted for tickling or laughter”, as a consequence of an active-passive shift. As for relishome [1593], the adjective is defined as “relishable; tasty, appetizing”, in other words, it appears a direct competitor for the co-emerging relishable [1605] (“capable of being relished; appetizing, enjoyable” according to the OED). From the OED contexts of use provided, relishsome appears to be still in use although rare, with an example from 2004.

(9) 2004 K. Hulme Stonefish 7 A relishsome mix, that would be perfect with bacon chunks.

76A corpus search in both historical and contemporary corpora indicates that relishsome is extremely infrequent, as shown by the frequency scores in each corpus: COHA 0, COCA 0, EHBC 1, PG 0, OEC 0. The counterpart relishable is also relatively infrequent although the frequency scores are higher; with 5 occurrences in COHA, 1 in COCA, 8 in EHBC, 13 in PG and 11 in OEC.

77The infrequent use of -some derivatives suggests a lack of propagation of -some, which in turn may infer a lesser cognitive availability of the pattern, i.e. an available derivational pattern that is inactive. The question is whether -some adjectives have the ability to provide data for a productive series, as raised by Trips [2009: 28]. The answer is clearly yes. -Some derivation had the ability to produce a productive series given the replication of the Vsome pattern in the 1750s. This means that a replication pattern did not give rise to sufficient frequency of use in order for the suffixation to continue to be productive. Still, contrary to other archaic suffixes (-th, -dom), it can be argued that -some is not obsolete and retains the ability to trigger an analogical patterning given the right conditions. In other words the low propagation of -some adjectives may result from increasing specialization of its usage (for instance in terms of register).

3. -Some and -able competition: corpus testing using collexeme analysis

78This section will be devoted to corpus testing of competition between Vable and Vsome adjectives as suggested in Durkin [2016a: 394] who calls for testing of competition between lexemes in a corpus. Given that our data set is limited given the infrequency of Vsome adjectives, as explained previously, our methodology will focus on a few case studies. We will combine both qualitative manual analysis, and quantitative collocation analysis. Collexeme analysis provides a quantitative methodology for the comparison, based on a selection of alternating Vsome and Vable forms.

3.1. Hatesome / lovesome & lovable / hateable

79The adjectives hatesome and lovesome both have -able counterparts, however a frequency analysis shows that their trajectories are not symmetrical. The compared frequency of the synonym pairs indicates a divergence in usage, particularly for hatesome and hateable, which remain extremely rarified in multiple corpora, whereas lovesome / lovable have fared a little better (see Table 4). A possible explanation is the highly frequent use of the adjective hateful which remains the most successful derivatives of hate across all corpora, including the historical COHA, EHBC, and PG.

Table 4. Compared frequency of usage of synonym pairs




































  • 32 It must be said however that the distinction between nouns and verbs decreased diachronically in th (...)

80Hatesome [1382] appears to have a denominal structure (but is also compatible with Vable reading32), and a sense corresponding to “that arouses feelings of hate”, “causing hate”, similar to hateful [1382] which is provided in the definition. The adjective is compatible with a patient sense: likely to induce a feeling in someone. This sense relies on evaluative assessment, but also subjective interpretation. It resembles the idea of hateable (Vable according to the OED) [1425] “deserving of hatred; that deserves to be hated or greatly disliked; odious”. This reading of hateable focuses more on an evaluative judgement. Hateable appears then to be deverbal whereas hatesome is given as denominal, although the senses of these adjectives appear comparable: is one more eventive than the other? Hatesome is given as rare although a context is provided for 2005: the context appears to be a historical text, suggesting the usage is deliberately archaic (for a novel based on 18th century Scotland) in (10):

(10) L. C. Higgs Fair is Rose (2005) lvii. 365 She must cast all the blame upon herself. Not on Jamie, not even on her hatesome father.

81However, a corpus search for hatesome shows no hits in EHBC, attesting to a possible lack of generalised usage, and this is replicated in most other corpora; in addition, hatesome is also absent from the English Dialect Dictionary.

82The antonym lovesome [OE]33 follows a similar semantic pattern activating the causing N sense, or the full of N sense. Contrary to hatesome which is synonymous with the far more frequent hateful, lovesome is synonymous with lovable (referring to the quality of something “that induces the feeling of love in someone” or that deserves / is worthy of being loved). The pair lovesome [OE] and hatesome [1382] underlines the asymmetry of the -able derivation and the semantic instability of -some derivatives. To prove this instability, let us now consider the semantic shift of lovesome, as given in the OED:

1) Sense [OE] “worthy of love, that inspires love”: sense 1) has a passive reading as lovable, which can be loved, although it focuses on the evaluative judgement of worthiness.
2) Sense [OE] “friendly, affectionate”: sense 2) has an active reading “which/who gives loves”, which is given as being rare and obsolete. It appears simultaneous to sense 1) the passive reading.
3) Sense [1175] “inspiring love through beauty, beautiful”: sense 3) is metonymical (cause for effect) and an extension of sense 1), based on the idea that beauty is worthy of love.
4) Sense [1575] is another active sense “showing love”, similar to sense 2, although this time love is to be interpreted in the amorous sense.

83Lovesome has many occurrences in the OED citations ranging historically from OE (13) to 2007 (12), or 2004 (14), showing that usage of lovesome is still occasional, despite the absence of occurrences in large corpora. Not only that, but each of the senses of lovesome is illustrated with examples ranging historically up until 2000s, with the possible exception of the weakened sense “affectionate” as shown in (17):

(11)1883 J. Ingelow in Longman’s Mag. Sept.533 While lovesome and moansome thereon spake and falter’d the dove to the dove.

(12) 2007 Washington Times (Nexis) 8 Mar. a18 A car terrorizes and attacks a lovesome pink piggy bank.

(13) St. Juliana (Bodl.) 115 Ihesu crist þet ich on leue, & luuie as leoflukest & lufsumest lauerd.

(14) 2004 Press (Christchurch, N.Z.) (Nexis) 9 Oct. 20 A redcurrant laden with shiny red berries is a lovesome feast for the eye.

(15) 1901H. C. Welch Anselm iii. 48 This increasing influence was due to the happy lovesome temper which plays through his letters.

3.2. Semantic behaviour loathsome / hateful

84A number of derived adjectives compete for the expression of negation evaluation of something or someone; loathsome [1389] is defined as “exciting disgust or loathing” and falls within CAUSE N category. Hatesome [1382] is diachronically co-emergent “that arouses or provokes feelings of hatred; hateful, odious, detestable”. Detestable [1477] is later-emerging loan from French with the sense “To be detested; intensely hateful or odious; execrable, abominable”. The frequency counts of these synonymous adjectives were tested in COHA and in EHBC, showing that hatesome is very rare, in comparison with the highly frequent loathsome, hateful, with the frequency of detestable being considerably less in COHA than in EHBC.

  • COHA: loathsome 1105, hatesome 0/ hateful 2007, detestable 794

  • EHBC: loathsome 2619, hatesome 0/ hateful 4081, detestable 4198

  • 34 See Glynn [2010: 27] on the importance of quantitative and qualitative analysis: “It is not the fre (...)

85In order to investigate the variation in semantic behaviour, we will now turn to a distributional semantic analysis, which provides a list of significant collocates according to the distribution of the collexeme in the corpus EHBC. Distributional semantics and collexeme analysis is a well-established method in semantics and lexicography for operationalising meaning34. Sketch Engine provides a tool allowing for a compared collocate analysis, which is well equipped for testing the semantic similarity of use of two lexemes. The results are shown in Table 5. Collocation scores in the two final columns are based on the LogDice measure of co-occurrence, which aims to provide a reading of the likelihood of the collocation in the corpus: the higher the score the less likely the collocation, and therefore the more significant the co-occurrence will be.

Table 5. Compared AND/OR collocates of hateful and loathsome

Raw count with hateful

Raw count with loathsome

Score with hateful

Score with loathsome
























































86The area in green represents the significant collocates of hateful, in yellow the significant collocates of loathsome, and in the central area the shared significant collocates of both adjectives, which represent the potential overlap in usage of hateful and loathsome.

87A further collocate analysis in Table 6 shows the collocates in subject position, with the adjective in predicative usage:

Table 6. Compared subject collocates of hateful and loathsome

Raw count with hateful

Raw count with loathsome

Score with hateful

Score with loathsome
























8.3 approx

















88The area of overlap of collexeme in subject position is quite varied, although the noun sin is manifestly one of the most significant for both adjectives, with a slight preference for hateful (this sin is hateful) rather than loathsome.

89The collocate comparison tends to confirm the synonymy and shared usage. The most differentiated collocate pattern occurs in the premodifier category. It appears only hateful is more liable to take an adverbial premodifier of qualitative degree (infinitely, peculiarly, justly) as in (16) and (17), whereas loathsome does not. The shared significant premodifiers are comparatives (as, so and rather/than), and the adverbs very and altogether as in (18) and (19):

(16) 1746 Edwards Jonathan. And not only so, but it convinces the Soul of some thing further concerning Sin, that it saw nothing of, while only under legal Convictions; and that is the infinitely hateful Nature of Sin, and it’s Dreadfulness upon that Account. EHBC

(17) 1775. Cooper Robert. O ſinners, permit me to uſe, with a little variation, language which, according to its common acceptation, is juſtly hateful, but in this caſe proper and emphatical: Let me in reſpect of God charge you to non-reſiſtance, and the moſt unlimited and chearful obedience. EHBC.

(18) 1680. Thomas Strafford. That Atkinson afterwards fell into decay, and was Imprisoned; and the Prison being very loathsome, the Bishop wrote unto him, this Deponent, and sent him a Lease, under the Hand and Seal of him the said Bishop, and the Incumbent, with a Label for his the Deponents hand, and desired him to seal it for 40 s. a year to another, that Atkinson might pay his Debts, and stock himself with Cattle. EHBC.

(19)1656 Boccalini Traiano (translation), his Majesty holds him to be but an impetuous, proud, impertinent fellow, a capricious wit of the first head, one that means well, but judges ill, and one who is all zeal, crusted up in imprudency; which qualities are very hateful to Apollo, who thinks it very ill done to bestow publike imployments upon such companions, as ought onely to be conferred upon men civilly behaved, EHBC.

90The occurrences of very hateful show a remarkable tendency for hateful to be followed by a prepositional complement to, whereas loathsome does not. This tendency is confirmed in the collocate search for prepositional complements, suggesting this may be a syntactic specificity of hateful.

3.3. Laughsome / laughable

91Laughsome is virtually non-existent in all corpora as shown in Table 7. The only occurrence found in Google Books refers to William Barnes [1952], in a passage which sees Barnes [1952: 28], initially published in 1863 under the title A Grammar and Glossary of the Dorset Dialect, argues for the revival of Anglo-Saxon -some derivatives, which he assumes have fallen into obsolescence outside of his Dorset dialect, and which can serve as a guide for improvement of clarity:

Our useful adjectives ending in some, German sam, as quarrelsome, noisome, equivalent to the Latin ones in ax-loqu-ax, given to talking, or bundus, -vagabundus, given to wandering, naming the state of a noun likely or given to do an action, would have been well taken into the national speech from any dialect in which they might be found, instead of those borrowed from the Latin; as heedsome, attentive; winsome, likely to win or captivate; lovesome, disposed to love; blithesome, disposed to be blithe; fadesome, laughsome, runsome (as mercury), meltsome (as butter or lead). Winning and loving are bad substitutes for winsome and lovesome, since winsome does not mean actually winning one, but likely to win one; and lovesome is not amans, but amasius.

Table 7. Compared frequency of laughsome / laughable

Raw occurrence in each corpus





Google Books

















92The OED however provides occurrences of laughsome as late as 2009 in (20) in the sense “causing laughter”, although the sense “mirthful” appears rare, one of the later examples is provided in (21) in 1884:

(20) Independent (Nexis) 9 Mar. (Extra section) 4 Pitch-perfect pastiche of America’s dominant musical styles: immediately recognisable, and instantly laughsome.

(21) 1884 G. Allen Philistia I. iv. 113 Fly away, sweet little frolicsome, laughsome creature.

3.4. Irksome, worrisome, fatiguesome, boresome: “heavy with X”

93From observation of the 261 adjectives on our data set, a number of -some adjectives appear to fall into a similar semantic category of negative evaluative adjectives referring to emotional reactions: take irksome, worrisome, fatiguesome, boresome, bothersome. Additionally, the number of -some adjectives still in use referring to negative evaluations may potentially represent a patterning, in other words a form of convergence. We hypothesize here if this could lead to reanalysis of -some adjectives as meaning “weighed down by V”, although more evidence and analysis would be required to test this hypothesis. Table 8 shows the raw frequencies (i.e. tokens) of 8 -some adjectives across 2 historical corpora and 2 contemporary corpora (of different sizes and different makeup, which reduces the impact of any comparison):

Table 8. Frequencies of usage of a set of synonyms of tiresome





irksome 1435





tiresome 1513





wearisome 1460 (adj+some)





bothersome 1817





worrisome 1702





troublesome 1548





burdensome 1578





cumbersome 1487





94What can be observed however, despite the shortcomings of such a comparison, is that several adjectives remain in frequent use, thereby contrasting with the many rare infrequent -some adjectives which compete with high frequency -able adjectives (such as lovable / lovesome). The adjective cumbersome is particularly interesting as it shows the increasing frequency of use of one of the older adjectives dating back to 1487.

Figure 10. Google Books n-grams for cumbersome / troublesome

Figure 10. Google Books n-grams for cumbersome / troublesome

95The cumbersome / troublesome Google n-grams (in Figure 10 above) shows a rising frequency pattern as opposed to a falling pattern for the adjective troublesome. This increased usage may potentially be a result of other motivational factors, in particular through reanalysis with lumber/lumbering, i.e. “heavy”. It is also clear that none of these adjectives compete directly with -able counterparts, or even -ful adjectives (*botherful, *botherable, *troubleful, *troubleable). Further study would of course be required in order to test the hypothesis of reanalysis and to test the rising pattern of usage of certain -some adjectives.

4. Conclusion: motivation in suffix competition, obsolescence, productivity and outlook

96The suffixes -able and -some therefore share certain morphological and semantic features: first, they can be used as both deverbal and denominal derivatives, and secondly they can form adjectives with the sense “capable of being Ved”, “intended for V”, placing them in direct competition with one another. The issue of synonymy or interchangeability is addressed in Guimier [1985: 164] who underlines that no two word-forms are completely identical:

[T]he two suffixes -ic and -ical show no difference of meaning in some cases: geometric(al), strategic(al); in other cases, they are not interchangeable (economic problems but an economical housekeeper). This means that the opposition -ic / -ical is meaningful, even though, owing to the semantic nature of some adjectives, it may be neutralized. On a more theoretical level, it has already been said that there are no semantically empty signs in language.

97The decrease in -some output is likely a sign of the lessening productivity of a suffix that has been outperformed by a multitude of available suffixes forming alternatives, i.e. possible synonyms (-ish, -ful, -y, ing, -able, etc.). -Some may have waned due to semantic instability insofar as “productive processes are semantically and phonologically transparent” (Plag & Baayen [2009: 125]). We can safely say that low productivity is directly correlated to the loss of transparency of -some adjectives: the loss of transparency itself may be caused by the loss of the base word, or by change of meaning – especially compared to the higher frequency and transparency of -able. Lexicalisation and fossilisation are also viewed to be an inhibitor to productivity: fossilised words continue to exist, forming a paradigm of no longer parsable word forms: length, width, health, depth, girth (see Anshen & Aronoff [1997: 9]). Nonce-words fall into the non-transparent category and therefore do not contribute to diachronic productivity; they conform to an existing pattern but are stored without the pattern, i.e. retrieved as complete words rather than affixed words (see Mattiello [2017: 25-26]). Arndt-Lappe [2014: 540] concludes that analogical reasoning is at the heart of the preferedness of suffix productivity:

It was shown that the productivity profiles that we see in the contemporary data are the result of an ongoing and consistent development throughout that period, which is characterised by a constant increase of the productivity of -ity in the domains in which it occurs. This is accompanied by a corresponding decrease of the productivity of -ness in those domains. The diachronic facts therefore provide further evidence that competition between -ity and -ness in language use involves analogical reasoning.

98Overall, the semantic instability (active passive changes as in fearsome), the active-passive reinterpretation of -some adjectives (winsome) suggest that -some has a complex semantic behaviour. -Able has surpassed -some derivation however the overlap is only partial since -able forms mostly passive-oriented adjectives. The low frequency of usage of many -some adjectives tend to show that the suffixation has waned after a sudden creativity peak in the 1850s, likely boosted by a deliberate effort to preserve Anglo-Saxon suffixation over borrowed suffixes (as exemplified in Barnes [1863]). However, it is of interest that despite the decrease in usage, -some adjectives remain in use, both as frequent lexicalised word forms which may be opaque (handsome), and frequent lexicalised word forms that appear transparent (tiresome, burdensome, cumbersome). -Some continues to exist as a slightly archaic suffix (eyesome); nevertheless, a revival of -some derivatives may not be out of the question. -Some remains in a state of dormant productivity, as contrary to other archaic suffixes (-th, -dom), -some has not become obsolete and opaque. Jespersen’s remark [1942: 456] remains true: “the suffix has been productive during all periods, though comparatively little used”, echoed by Dixon’s [2014: 255] conclusion that -some “always had a fairly low frequency”.

  • 35 Traugott [1989: 51]: “Pragmatic strengthening and relevance as I use the terms largely concern stra (...)

99Another line of enquiry concerning the success of Vable adjectives may lead us to consider the subjectification hypothesis outlined by Traugott [1989]. Is Vable compatible with a higher degree of modal interpretation, which would then be a case of increasing subjectification, increased pragmatic strengthening35, such as the shift from deontic to epistemic and to evidentiality as described in Traugott [1989: 32])?

Although not all linguists would include evidentials among epistemics, since they regard evidentials as markers of the speaker’s information source and epistemics as markers of the speaker’s state of knowledge or belief, Lyons defines epistemology as ‘concerned with the nature and source of knowledge’ (1977: 793), which suggests that epistemics and evidentials are related linguistically.

100If this subjectification theory holds true for this word formation, we should expect -able adjectives to take on epistemic senses (although we have no evidence of this so far). However, as Traugott underlines, subjectification shift isn’t easy to identify and doesn’t necessarily involve epistemic shift. Traugott [2016: 389] provides three major tendencies of semantic change, which are external description shifts to internal evaluation; meanings becoming increasingly textual and pragmatic; and meanings evolving towards increasing interpersonal hearer/speaker interactions. However, qualitative analysis is required taking into account several contextual factors:

(a) Subjectivity is a gradient less-or-more phenomenon; so is subjectification. Therefore, an analysis must be sensitive to micro-differences both synchronically and over time. (b) Because subjectification is highly dependent on linguistic context and interlocutors’ objectives, all putative examples of subjectification must be evaluated in extended textual contexts. (c) Because most subjectivized items are polysemous and polyfunctional, a paraphrase must be found that will distinguish the subjectivized from the less subjectivized meaning.

101In order to assess the validity of a subjectification hypothesis, further large-scale qualitative analysis of Vable / Vsome competition is required. Furthermore, as underlined in Fernandez-Dominguez [2017: 112] the difficulty of assessing competition “seems to largely involve exploiting partial or incomplete datasets, insofar as it requires tracking down not only existing lexemes (i.e. the prevailing ones), but also derivatives that do not succeed in competition, and which often fall into disuse”. Given the limited frequency of usage of -some adjectives, the task of studying suffixal competition cannot be merely quantitative but requires sufficient comparable data as to allow for a reliable comparison of semantic and distributional behaviours. The study has also shown that despite a low frequency of usage -some adjectives have not been totally phased out and remain in use, admittedly mostly with a “jocular” or “archaic” usage. The study of the effect of register and text type will also possibly bring to the fore relevant results on variation in the usage of derivatives based on language situations.

102We believe this investigation has shown the relevance of studying diachronic productivity using both the OED and historical corpora to test and question the trajectories of morphological suffixation, and the motivations behind increasing or decreasing usage.

Top of page


Allan Kathryn, 2011, “Using OED data as evidence for researching semantic change”, in Allan Kathryn & Robinson Justyna (Eds.), Current methods in historical semantics, Berlin / Boston; De Gruyter, 17- 40.

Arndt-Lappe Sabine, 2014, “Analogy in suffix rivalry: the case of English -ity and -ness”, English Language and Linguistics, 18, 497-548.

Anshen Frank & Aronoff Mark, 1997, “Morphology in real time”, in Booij Geert & van Marle Yaap (Eds.), Yearbook of Morphology 1996, Dordrecht: Springer, 9-12.

Baayen Harald & Lieber Rochelle, 1991, “Productivity and English derivation: a corpus-based study”, Linguistics 29, 801-843.

Bauer Laurie, 2004 (2001), Morphological productivity, Cambridge: Cambridge University Press.

Culpeper Jonathan & Clapham Phoebe, 1996, “The borrowing of Classical and Romance words into English”, International Journal of Corpus Linguistics 1:2., 199-218.

Di Sciullo Anna Maria, 1997, “Selection and derivational affixes”, in Dressler Wolfgang U., Prinzhorn Martin, & Rennison John R. (Eds.), Advances in Morphology, Berlin, New York: Mouton De Gruyter, 79–95.

Davies Mark & Chapman Don, 2016, “The effect of representativeness and size in historical corpora: An empirical study of changes in lexical frequency”, in Chapman Don, Moore Colette & Wilcox Miranda (Eds.), Studies in the History of the English Language VII (Generalizing vs. Particularizing Methodologies in Historical Linguistic Analysis), 131-150.

Dixon R. M. W., 2014, Making new words: Morphological derivation in English, Oxford: Oxford University Press.

Durkin Philip, 2011, “Commentary; Data and uses”, in Robinson Kathryn & Robinson Justyna A. (Eds.), Current methods in historical semantics, Berlin / Boston: De Gruyter, 97-105.

Durkin Philip, 2014, Borrowed words: a History of Loanwords in English, Oxford: Oxford University Press.

Durkin Philip, 2016a, “The OED and HTOED as tools in practical research: a test case examining the impact of loanwords on areas of the core lexicon”, in Kytö Merja & Pahta Paivi (Eds.), The Cambridge Handbook of English Historical Linguistics, Cambridge: Cambridge University Press, 390-406.

Durkin Philip, 2016b, “Etymology, word history and homonymy”, in Durkin Philip (Ed.), The Oxford Handbook of Lexicography, Oxford: Oxford University Press, 236-252.

Esteban-Segura Laura, 2018, “A study of noun-deriving suffixes in competition in Middle English”, Miscelánea: a Journal of English and American studies 57, 59-77.

Fernandez-Dominguez Jesus, 2010, “Productivity vs lexicalization: frequency-based hypotheses on word formation”, Poznań Studies in Contemporary Linguistics 46(2), 193–219.

Fernandez-Dominguez Jesus, 2017, “Methodological and procedural issues in the quantification of morphological competition”, in Santana-Lario Juan & Valera-Hernández Salvador (Eds.), Competing Patterns in English Affixation, Bern: Peter Lang, 67-117.

Gardani Francesco, Arkadiev Peter & Amiridze Nino, 2015, “Borrowed morphology. An overview”, in Gardani Francesco, Arkadiev Peter & Amiridze Nino (Eds.), Borrowed Morphology, Berlin / Boston / Munich: De Gruyter, 1- 23.

Geeraerts Dirk, 2010, “The Doctor and The Semantician”, in Dylan Glynn & Kirstin Fischer (Eds.), Quantitative Methods in Cognitive Semantics, Berlin / Boston: De Gruyter, 63-78.

Geeraerts Dirk, Gevaert Caroline & Speelman Dirk, 2012, “How anger rose: Hypothesis testing in diachronic semantics”, in Allan Kathryn & Robinson Justyna (Eds.), Berlin / Boston; De Gruyter, 109-132.

Guimier Claude, 1985, “On the origin of the suffix -ly”, in Fisiak Jacek (Ed.), Historical Semantics- Historical Word formation, Amsterdam: De Gruyter, 155-170.

Jespersen Otto, 1942. A modern English Grammar on Historical Principles, Part 6: ‘Morphology’, Copenhagen: Munksgaard.

Kay Christian & Allan Kathryn, 2016, “Change in the English lexicon”, in Kytö Merja & Pahta Paivi (Eds.), The Cambridge Handbook of English Historical Linguistics, Cambridge: Cambridge University Press, 220-236.

Kjellmer Göran 1986, “Legible but not readable: On the semantics of English adjectives in -ble”, Studio Neophilologica 1, 11-38.

Kratzer Angelika, 1981, “The notional category of modality”, in Eikmeyer Hans-Jurgen & Rieser Hannes (Eds.), Words Worlds, and Contexts, New Approaches in Word Semantics, Berlin: De Gruyter, 38-74.

Lipka Leonhard, Handl Susanne & Falkner Wolfgang, “Lexicalization & Institutionalization: The State of the Art in 2004”, Skase Journal of Theoretical linguistics, 2-19.

Lyons Christopher, 1977, Semantics, Vol 1, Cambridge: Cambridge University Press.

Marchand Hans, 1960, The Categories and Types of Present-Day English Word-Formation. München: C. H. Beck.

Mattiello Elisa, 2017, Analogy in Word formation. A study of English neologisms and occasionalisms, Berlin / Boston: De Gruyter.

Oltra-Massuel Isabel, 2014, Deverbal Adjectives at the Interface. A Crosslinguistic Investigation into the Morphology, Syntax and Semantics of -ble, Berlin / Boston: Mouton De Gruyter.

Plag Ingo, Dalton-Puffer Christiane & Baayen Harald, 1999, “Morphological productivity across speech and writing”, English language and linguistics, (3), 209–228.

Plag Ingo, 2003, Word Formation in English, Cambridge: Cambridge University Press.

Plag Ingo & Baayen Harald, 2009, “Suffix Ordering and Morphological Processing”, Language, Vol. 85, No. 1, 109-152.

Rafaelli Ida, 2012, « La loi de répartition dans le cadre de la linguistique cognitive », in Berlan Françoise & Berthomieu Gérard (Eds.), La synonymie, Paris, Presses de l’Université Paris-Sorbonne, 35-50.

Rainer Franz, Dressler Wolfgang U., Gardani Francesco & Luschützky Hans Christian, 2014, “Morphology and meaning – an overview”, in Rainer Franz, Gardani Francesco, Luschützky Hans Christian & Dressler Wolfgang U. (Eds.), Morphology and meaning, Amsterdam & Philadelphia: John Benjamins, 3-46.

Renner Vincent, 2020, “An ecosystem view of English word-formation”, The Mental Lexicon, 15(1), 4-20.

Renouf Antoinette, 2019, “Big Data: Opportunities and Challenges for English Corpus Linguistics”, in Suhr Carla, Nevalainen Terttu & Taavitsainen Irma (Eds.), From Data to Evidence in English Language Research, Leiden / Boston: Brill, 29-65.

Robinson Justyna, 2010, “Awesome insights into semantic variation”, in Geeraerts Dirk & Taylor John R. (Eds.), Advances in cognitive sociolinguistics, Berlin / New York: De Gruyter 85-110.

Schulte Marion, 2015, The semantics of derivational morphology. A synchronic diachronic investigation of the suffixes -ery and -age in English, Tübingen: Narr Verlag.

Simpson John, Weiner Edmund & Durkin Philip, 2004, “The Oxford English Dictionary today”, Transactions of the Philological Society, Volume 102:3, 335–381.

Schuwer Martine, 1999, « Le suffixe adjectival –ble : Impacts sémantique et syntaxique de la dérivation », ASp [En ligne], 23-26 | 1999, mis en ligne le 05 mars 2011, consulté le 25 juin 2020,

Smith Chris A., 2016, “Tracking semantic change in fl- monomorphemes in the Oxford English Dictionary”, Journal of Historical Linguistics, Vol. 6:2, 165–200.

Smith Chris A., 2018,Where do new words like boobage, flamage, ownage come from? Tracking the history of ‑age words from 1100 to 2000 in the OED3, Lexis [Online], 12 | 2018, Online since 14 December 2018:

Smith Chris A., 2019. « Approche cognitive diachronique de l’émergence du phonesthème fl- : réanalyse phonosymbolique et transmodalité dans le Oxford English Dictionary », Signifiances (Signifying), 3(1), 36-62.

Smith Chris A, 2021 (forthcoming), “How is stickage different from sticking: A study of the semantic behaviour of V-age and V-ing nominalisations (on monomorphemic bases)”, in Mignot Élise (Ed.), Nouns, between syntax and semantics, Berlin / Boston: De Gruyter.

Szymanek Bogdan, 2015, “Remarks on tautology in word formation”, in Bauer Laurie, Körtvélyessy Lívia & Štekauer Pavol (Eds.), Semantics of Complex Words, Heidelberg New York Dordrecht London: Springer, 143-161.

Traugott Elizabeth C, 1989, “On the rise of epistemic meanings in English: an example of subjectification in semantic change”, Language 65, 31-55.

Traugott Elizabeth C, 2016, “Identifying micro-changes in a particular linguistic change-type: the case of subjectification”, in Kytö Merja & Pahta Paivi (Eds.), The Cambridge Handbook of English Historical Linguistics, Cambridge: Cambridge University Press, 376-389.

Trips Carola, 2009, Lexical Semantics and Diachronic Morphology. The Development of -hood, -dom and -ship in the History of English, Tübingen: Max Niemeyer Verlag.

Zenner Elise, Speelman Dirk & Geeraerts Dirk, 2014. “Core vocabulary, borrowability and entrenchment: A usage-based onomasiological approach”, Diachronica 31:1, 74-105.

Electronic Resources: dictionaries and corpora (last accessed 14/10/2020)

COCA, Davies Mark

COHA, Davies Mark,

English Dialect Dictionary, Universität Innsbrück, based of Joseph Wright’s English Dialect Dictionary (1898-1905),

English Historical Books Corpus, via Sketch Engine.

Merriam Webster

Oxford English Corpus, Oxford: Oxford University Press. Accessed via Sketch Engine.

Oxford English Dictionary Online, 3rd ed. OED3. Oxford: Oxford University Press,

Project Gutenberg English, accessed via Sketch Engine.

Sketch Engine, Kilgarriff Adam,

Top of page


1 The evidence of competition is based on the lexicographic study of 261 -some adjectives in the OED, as this study will show. Based on OED data, 42 out of 261 -some adjectives have -able adjectives in the dictionary paraphrase provided by the OED (i.e. 42/261, 16%).

2 We use the form -able here rather than -ble (which includes the alternative spelling -ible, both originating from the same Latin source -abilis or -ibilis).

3 Durkin [2016a: 397] “Although the OED stands alone in its breadth of coverage and chronological time depth, there are a number of historical dictionaries which deal with specific periods and/or varieties of English”.

4 Fernandez-Dominguez [2010: 195] explains that “[i]n essence, a word-formation process is said to be productive if it has the potential for speakers to operate it in an unconscious and repetitive way for the rule-governed production of an indefinite number of words”.

5 Also, the pre-existing wieldy [1413] in the same sense.

6 However, two questions arise: firstly, transparency is not necessarily “a necessary condition for productivity”, (see Rainer et al [2014: 8]). Secondly, this hypothesis means that a more restricted affix survives over a more unrestricted affix, thereby contradiction some claims of grammaticalization theory. Many thanks to one of the reviewers for pointing this out.

7 However, it must be noted that the OED is a non-homogenous secondary source as explained in Allan & Robinson [2011: 4]: “[…] like any secondary source the OED cannot be taken at face value, but needs to be used critically, particularly because it is made up of entries from different periods that belong to different editions.”.

8 “The Oxford English Dictionary is currently in the course of its first ever comprehensive revision, in the course of which all aspects of each entry (in some cases first published as far back as the 1880s) are being reconsidered and where appropriate rewritten in the light of new evidence. The changes are often extensive, frequently involving new earliest (and later) examples, and detailed reconsideration of definitions and etymologies, and sometimes involving such features as splitting (or more rarely merging) of senses where this better reflects the available evidence, reconsideration of the spelling of lemmata as part of comprehensive reassessment of historical spelling evidence for each lemma, or sometimes reassignment of evidence to a different dictionary entry.”

9 According to the “loi de répartition” of synonyms suggested in Rafaelli [2012]; also see Renner [2020] on the notion of ecosystem of suffixes.

10 “The assumption underlies this belief that frequency commonly accompanies a word during its progress from one stage onto another, so that coinages will tend to be infrequent, institutionalized words will be more frequent and lexicalized units even more frequent, because they have been around for a longer time.”

11 As far as potential productivity is concerned Trips [2009: 33] underlines the issues of choosing an appropriate corpus for diachronic study of potential productivity. “What we expect is that among the ten hapaxes we will find more neologisms than among words that occur with a higher frequency than one. This is borne out in a study by Plag (2003) for the formations with -able where it is shown that the number of non-listed words with that suffix is high among hapaxes, and that therefore hapaxes can be used to measure productivity. As long as the corpora investigated are large the proportion of neologisms among the hapax legomena increases. However, in small corpora the proportion of neologisms among the hapaxes will be small, and it is likely that in most cases hapaxes are rare words in the language and not newly coined ones. This point is problematic for diachronic corpora since they are relatively small (see discussion below).”

12 We shall also see that the behaviour of the suffix-some includes deadjectival derivation thereby forming adjectives from a seemingly synonymous adjective base, such as late / latsome [OE], dark / darksome [1530], bright / brightsome [1548], murk / murksome [1590], or weird / weirdsome [1885].

13 See Szymanek [2015: 155] on tautology in word formation, in particular affixal tautology: “morphological tautology presupposes synonymy between the base and the derivative – a special kind of synonymy where the two words share not only the same meaning but also a fragment of their form, i.e. the base”.

14 Jespersen [1942: 457] “In connexion with a numeral -sorne represents OE sum ‘some’ as used after a numeral in the gen. pl, e.g. Beowulf 207 fiftena sum ‘as one of fifteen’. The modern expressions twosome (Se.), foursome are especially used for a game for two, etc.”

15 Regarding the existence of deadjectival adjective suffixation, “As far as motivations for innovation go, neologising does not principally fill a gap (also see Sylvester et al. (forthcoming) on lexical innovation), but rather can provide an alternative in a specific register, and in particular slang.” (Smith [2021, forthcoming]).

16 This reading of derivations as eventive might plausibly have given rise to the spike in Vsome formation in the 1750s. However, the question remains : why did Vsome die out after 1900?

17 From the base word lumber 1400 in the OED given as “Possibly two or more words may have coalesced. Middle English lomere may have been a frequentative formation on lome adj […]. The word, however, may be partly of direct imitative formation in English.”

18 The etymology of the base of frolic is unclear: the OED gives the adjective as being either Vsome or Nsome, but a search for the verb frolic then suggests it originates from the adjective frolic, itself a loan from Germanic origins “<Dutch vrolijk (in Kilian vrolick), = Old Saxon *frôlîc (whence frôlîco adverb), Old High German frôlîch (Middle High German vrôlich, vrœlic, modern German fröhlich); < Middle Dutch vrô = Old High German frô (Middle High German vrô, modern German froh) glad, joyous”.

19 Subsequently the suffix was extended to form denominal adjectives such as saleable adj. (first half of the 16th cent.), marriageable adj. (second half of the 16th cent.), carriageable adj. (early 18th cent.), etc. This extension was probably encouraged by the numerous cases where a formally identical noun existed alongside a verb which was the base of a formation in -able (or where all three had been borrowed < French: compare changeable adj. beside change v. and change n., or debatable adj. beside debate v.1 and debate n.1) (OED).

20 The issue remains whether Nable and Vable are the same pattern or not, and whether there is convergence.

21 « Le suffixe de ces adjectifs attributifs est, comme pour les adjectifs épithètes, porteur de plusieurs modalisations possibles, en fonction du sémantisme de la base. Là encore, la modalisation à valeur radicale est dominante » Schuwer [1999: 21]. [The suffix of these predicative adjectives, as with premodifying adjectives, can carry several modal interpretations, depending on the meaning of the base word. In this cas also, the root modal interpretation tends to be predominant.]

22 [To sum up, when the adjective is used predicatively, the prediction is based on the properties of N, assesses from the perspective of their impact: the event is perceived as a (highly probable) consequence of the properties of B. Instead, when the adjective is used as a premodifier, it’s the personal knowledge of the speaker that posits the high probability of the event or process.]

23 One of the many dating inconsistencies already noted, as the OED does not always have access to first date of usage (see Culpeper & Clapham [1996], Durkin [2016b]).

24 This descriptive typology has the advantage of being detailed and empirically-based rather than formal, although it does however show an overlap of the passive and possibility senses.

25 As any corpus researcher knows, one of the main constraints is the choice of corpus which is liable to affect results. It is well-known that size, representativeness and homogeneity affect results. For diachronic research specifically, Renouf [2019: 62] underlines the benefits of larger-size corpora in a view to studying lexical change and diachronic productivity: “In the form of very large text corpora, structured and with the current level of search and analytical software, big data brings obvious benefits to corpus linguistics over smaller corpora, in terms of the increased amount of information across the lexicon, which allows for a finer-grained analysis and understanding of the language. Very large corpora with a diachronic dimension give access to language innovation and change across ever greater stretches of time. These corpora also afford the corpus linguist in theory-based disciplines the opportunity to review and modify existing theories in the light of data”.

26 Durkin [2016b: 252] on treating words in historical dictionaries: “However, words are units in complex systems, showing complex patterns of mutual influence, and historical dictionaries have to find flexible ways of identifying, reflecting, and documenting at least the most important of these patterns; even ‘external’ influence from words in other languages will often occur during the history of a word, and not only at the initial point of origin… Additionally, the origin of many (perhaps most) words may be better conceptualized not as radiation emanating from the Big Bang of a single point of origin, but as the gradual coming together of multiple similar but distinct innovations each contributing to the emergence and growing establishment of a new lexical item.”

27 Also see Geeraerts [2010: 76]: “Rather than being the opposite of a more traditional hermeneutic approach, an empirical approach to antics is the completion and consummation of it.”

28 Note a lot of negative -some adjectives of evaluation have remained in use: troublesome, wearisome, worrisome, caresome, grievesome, troublesome, burdensome.

29 Labelled ‘erroneous’ in the OED definition for this sense of fearsome.

30 Also see Robinson [2010] for a study of the adjective awesome from a cognitive sociolinguistic perspective.

31 There is potentially an iconic motivation of this semantic shift and heightened success of the adjective awesome, but this is speculative at this stage.

32 It must be said however that the distinction between nouns and verbs decreased diachronically in the history of English, due to the loss of inflectional markers. In this light, it is not surprising that many of the examples are ambiguous – see also the work of Eitelmann, University of Mainz. Many thanks to one of the reviewers for their comments regarding the historical causes.

33 Listed in the English Dialect Dictionary:

34 See Glynn [2010: 27] on the importance of quantitative and qualitative analysis: “It is not the frequency per se of linguistic features that is of interest, but what this says about usage, the relative association of forms and meanings in context.”

35 Traugott [1989: 51]: “Pragmatic strengthening and relevance as I use the terms largely concern strategic negotiation of speaker-hearer interaction and, in that connection, articulation of speaker attitude.”

Top of page

List of illustrations

Title Figure 1. Factors affecting token frequency (from Fernandez-Dominguez [2010: 202])
File image/png, 11k
Title Figure 2. The semantic categories of -some derivatives
File image/png, 26k
Title Figure 3. A morphological typology of -some derivatives
File image/png, 64k
Title Figure 4. Morphosemantic patterns of -some derivatives
File image/png, 31k
Title Figure 5. Raw output per conceptual feature
File image/png, 34k
Title Figure 6. Compared raw output between the 3 main features
File image/png, 39k
Title Figure 7. Compared output of Vsome, Nsome and ADJsome
File image/png, 34k
Title Figure 8. Timeline of new words registered in the OED from 1000-2000
File image/png, 33k
Title Figure 9. -able output per century 1200-1999 based on origin
File image/png, 30k
Title Figure 10. Key features of Vable adjectives in the OED
File image/png, 16k
Title Figure 10. Google Books n-grams for awesome / winsome / eyesome
File image/png, 84k
Title Figure 10. Google Books n-grams for cumbersome / troublesome
File image/png, 67k
Top of page


Electronic reference

Chris A. Smith, A Case Study of -some and -able Derivatives in the OED3: Examining the Diachronic Output and Productivity of Two Competing Adjectival SuffixesLexis [Online], 16 | 2020, Online since 17 December 2020, connection on 18 July 2024. URL:; DOI:

Top of page

About the author

Chris A. Smith

Université de Caen Normandie – CRISCO EA4255

By this author

Top of page



The text only may be used under licence CC BY-NC-ND 4.0. All other elements (illustrations, imported files) are “All rights reserved”, unless otherwise stated.

Top of page
Search OpenEdition Search

You will be redirected to OpenEdition Search