1This paper investigates the nature and lexical behaviour of phonesthemes, in order to evaluate their impact on the lexicon. The term “phonestheme” was coined by J. R. Firth in 1930, and refers to recurrent pairings of phoneme clusters and elements of meaning (Mompean et al. [2020: 516]). Sounds that convey meaning have also been named submorphemes (Bottineau [2008]) or sublexical markers (Philps [2002, 2012]). These meaning-carrying sounds are “emotionally expressive” (Marchand [1960: 313]).
2This notion of expressivity is already pointed out by Sapir [1929: 226], which compares the conventional “referential symbolism” to some “expressive symbolism” noticeable at the phonetic level. Sapir’s [1929: 233] “phonetic symbolism” is echoed by Jespersen [1922: 396-411], one of the first to talk about “sound symbolism”, defined as “a natural correspondence between sound and sense” [1922: 396]. Some early authors even mention comparable phenomena without naming them explicitly, as when Bloomfield [1914: 103] underlines the “phonetic similarity between the forms” in the words “flame, flare, flimmer, flash or in flash, crash, dash”. These similarities are also what Bolinger [1950: 128] calls the “magnetic attraction” between words that sound alike and share a similar meaning. The phonological structure of certain words has long been noticed to echo their meaning, as Firth [1930: 35-36] pointed out when considering the many well-known nursery rhymes and alliterations.
3Sound symbolism has shown an increase of interest for the last couple of decades (Hinton et al. [1994], Monneret [2020: 2]), although the exact reasons behind its existence are not fully understood yet. The status of phonesthemes is challenging since conventional theories in morphology and word-formation studies (Bolinger [1950], Blust [1988]) fail to accurately describe and interpret their nature and behaviour in context). For example, Martinet’s [1960] concept of “double articulation” as well as Hockett’s [1958] concept of “duality of patterning”, in which phonological elements are forbidden from carrying any meaning, are both binary approaches that do not adequately account for the semantic and formal complexity of phonesthemes (see Kwon & Round [2014: 3] on the theoretical issues induced by phonesthemes).
4On the other hand, many psycholinguistic (Hutchins [1998], Abelin [1999], Bergen [2004]) and corpus-based statistical studies (Drellishak [2006], Otis & Sagi [2008], Liu et al. [2018]) have yielded positive results, showing that phonesthemes do exist on a psychological level, and are statistically significant across the lexicon. Gonnerman et al. [2007] brought experimental evidence that semantic-phonological overlapping triggered a higher priming effect than morphological decomposition. What is still largely unclear are the precise conditions of the emergence of phonesthemes and their role in the evolution of the lexicon. An increasingly large body of evidence now shows that they have a regulation role in the organisation of the mental lexicon (Benczes [2019], Smith [2022b: 129-134], Tsur & Gafni [2022]). Phonesthemes are also thought to act as multimodal cues that shape the structure of the lexico-grammatical continuum (Smith [2022b: 130], see also Rhodes [1994: 287-288] on the synesthetic aspects of phonesthesia and the transmodal embedment of phonesthemes, and Nobile [2020] on the psychological and multimodal aspects of sound symbolism).
5Sound symbolism appears in a multitude of languages, and from different language families (see e.g. Blust [1988] on Austronesian languages, Abelin [1999] on Swedish, Hamano [1998] on Japanese, and Nobile [2020] for an overview of the evidence on sound symbolism). But English phonesthemes do not work like ideophones in Japanese or Korean, as they represent a less direct type of iconicity (Dingemanse et al. [2015]). While ideophones are marked forms that have distinct grammatical characteristics (Dingemanse [2012]), phonesthemic words in English do not feature the same level of divergence from the rest of the lexicon.
6Phonesthemes in English mostly appear as meaning-carrying consonant onsets. The words grumble, groan, grunt, grieve or grudge all relate to a form of complaint (Waugh [1994: 59]), which has led early observers to tag this recurrent /gr/ sequence as a phonestheme. Because of their hybrid status as transitory elements between phonology and morphosemantics, phonesthemes invite to further investigate phonological aspects of lexicology.
7Firth [1930: 47] includes phonesthemes in his own classification of lexical units, defining them primarily as instances of a “phonetic habit”, which is “an attunement, a setting of the central nervous system, which will be touched off by the appropriate phonetic stimulus.” Certain phonological sequences may have acquired meaning after being used by speakers over long periods of time. Firth quotes sl- words like “slack, slouch, slush, sludge, slime, slosh, slash, sloppy, slug, slut, sly” and many more as entrenching a “pejorative phonetic habit” [1930: 51]. This sound-sense association therefore seems to have emerged from and to be determined by usage: “But a group of words such as the above has a cumulative suggestive value that cannot be overlooked in any consideration of our habits of speech” (Firth [1930: 51]).
8Words with similar “phonetic links” are “brought together by alliterative and experimental analogy”, this notion of analogy being at the core of what Monneret [2020: 11] terms the “iconic function of analogy”. Firth [1930: 48-49] adds:
The general opinion is, however, that words, not phones or phonemes or phoneme systems, are the units of speech. But just as phones are grouped in phonemic habits, so “words” group themselves into families of linked words or related habits. Some words are lonelier than others.
9This paper adopts the cognitive view that lexical knowledge is drawn from memory representations and that the structure of language is based on sequential chunking (Bybee [2010: 76]). Phonesthemes are thus understood as phonological chunks that entrench a form-meaning pairing through usage. As Bybee [2007: 279] puts it:
The data examined provide support for the proposals that much of linguistic knowledge is procedural knowledge, that chunks of linguistic experience much larger than the analytic units of morphemes or even words are the usual units of storage and processing, that there is no real separation of lexicon from grammar, and that phonological alternations whose domain is larger than a word can indicate the size of processing units.
10The non-compositional properties of phonesthemes are thus interpreted as a consequence of their emergence and evolution. Phonological aspects of the lexicon are therefore not forbidden from acquiring meaning. The semantic content of phonesthemes, although difficult to tackle, must have emerged through usage and can only be described by observing how phonesthemic words are used in context (see Lenci [2008: 7] on the relation between use and meaning). Phonesthemes have also been shown to alter the evolution of the lexicon, by affecting semantic change (Smith [2022a: 277]). As such, phonesthemes show that phonological aspects of language affect, and are affected by, language use and evolution.
- 1 See the three meanings of gr- (mentioned by Waugh [1994]) in the next section of this paper.
11Even though phonesthemes might show several semantic components1, this paper holds the cognitive view that meaning is shaped through time by usage and does not contain any constant and static core meaning. Evidence drawn from the study of ideophones suggests that sound is of primary importance in the “degree of penetration of iconicity in the language” (see Flaksman [forthcoming] on the coining of iconic words): “there are no reports of languages with ideophones for textures or inner feelings which do not also have ideophones for sounds and movements.” (Dingemanse [2012: 663]) This multimodal aspect of phonesthemes will not be further inquired here, but our perspective is that language users’ ability to innovate is what entrenches meaning-form associations. It is still unclear whether the sound-to-sound mapping is what constitutes the primary entrenchment of phonesthemic motivation or not, and further studies would be required to investigate this issue. The fact that gr- exhibits both sound-based motivation (as in groan, grunt) as well as motivation based on something else than sound (as in grasp, grim) is an interesting characteristic of this phonestheme, although this paper does not focus on transmodality.
12The cognitive approach to language, combined with distributional semantics and a corpus-based methodology (see Fabre [2015] and Heylen & Bertels [2016]) motivate my choice to work from both lexicographic and corpus-based data.
13The present paper focuses on the emergence, evolution and lexical role of the phonestheme gr-, by applying a twofold method drawn from both lexicographic and corpus-based material. Waugh [1994: 59] distinguishes between three semantic fields concerning gr-:
(1) something unpleasant (grim, grisly, gritty, gruesome, gruff, grumpy); (2) complaint (grumble, groan, grunt, grieve, grudge, gripe and even disgruntled); (3) undesirable rubbing (grind, grate, grovel, grub).
14Swedish, as a Germanic language, seems to exhibit a comparable gr- sound symbolic element evoking a “bad mood”, as reported by Abelin [1999: 87]. Similarly, Drellishak [2006: 43] found out gr- to be associated with a “deep or complaining noise”. Drellishak’s [2006: 37-45] statistical study confirmed the phonesthemic status of gr-, alongside sl- “slide; careless”, gl- “light, vision”, fl- “motion, repeated or fluid” and many more.
15The key words “threatening noise, anger, grip” were found to be prominent in gr- words, as in growl (Abramova et al. [2013: 1698]). The word growl itself had previously been defined as an “onomatopoeic expression” (Crystal [2008: 234]). English words starting with gr- denote “animal cries” as in “growl, […] groan, grunt” (Wright [2012: 3-4]) and some gr- verbs appear to relate to an act of “eating noisily and greedily”, alongside words starting with /tʃ/ and /skr/ (Wright [2012: 6]).
16Another common meaning conveyed by gr- is that of “prehension” as in the words “grab, grapple, grip, grope, grub” (Philps [2008: 128]). The phonological sequence /gr/ has also been compared to /kr/ in crack, crash, creak or crunch (see Chadelat [2008] on the sublexical meaning of cr-). These two phonesthemes share a common /r/ sound, which is “frequently associated with noise” across languages, as reported by Tsur & Gafni [2022: 27]. They also indicate that “[r] serves as onomatopoeia for noise as in roar, rustle.” Although the question of the exact transmodal interface between sound and meaning will not be asked directly in this paper, it is worth mentioning that sounds have been proven to express different levels of emotion or affection (see Smith [2020b: 49-50] for a semantic analysis and multimodal interpretation of the articulatory and acoustic features of fl-, following a method proposed by Tsur & Gafni [2022]).
17All those observations are used as preliminary clues to test out. Some of them seem to be based on intuition, while others are backed up with statistical data or corpus-based material. My purpose is to assess how divergent my results will be, with the final aim of identifying key recurrent patterns that can explain the emergence of gr-.
18Since phonesthemes are at the interface between sounds and meaning, how can we use corpus-based data to investigate the interrelation between phonology and lexicology? In order to answer this broad question and to narrow down the scope of the study, I ask:
-
Since the phonesthemic status of gr- has been statistically confirmed, why are there different subsemantic fields conveyed by gr-? And how did those affect language use?
-
Each submeaning of gr- seems to feature a lot of synonyms or near synonyms (e.g. grab, grasp, grope, etc.). Why are there so many words that denote the same meaning and what are the implications? What is the role of these groups of form-meaning associations?
The methodology is twofold, combining a lexicographic analysis and a corpus analysis.
19The method was first laid out by Smith [2016] to study the phonestheme fl- (see also Smith [2020b]), and was subsequently applied to gr- words by Farina [2020] and to sw- words by Smith [2022a]. These studies have brought evidence of an internal lexical cohesion of phonesthemic words in the Oxford English Dictionary and an intrinsic organising role of phonesthemes. As Smith [2022a: 275-276] points out, the common ground between lexicography, cognitive linguistics and usage-based semantics needs to be further consolidated, as those methods are interrelated (see Allan [2001], Allan [2012] and Durkin [2016] on using the OED as lexicographic evidence). Those findings need to be further tested out through corpus-based evidence.
- 2 OED online subscription version, 3rd edition.
20Starting from the entry definitions searchable on the OED32, recurring key words that share extended semantic (synonymous, metonymic, metaphorical) properties are grouped together into semantic categories. Since the purpose of this paper is to test the internal semantic cohesion of gr- and not its overall representation in the lexicon, certain gr- words were discarded, including frequent words such as green or great, which do not contain enough recurring key words in their OED definitions.
21The semantic traits are good indicators of the semantic coherence of the phonestheme. Three types of analyses were conducted: 1) a general statistical overview, which showed which features were the most frequently attested, and which ones combined the most with other features; 2) an examination of semantic change, in order to tackle the convergence of senses towards a particular phonosemantic feature; and 3) an isolation of etymological roots in order to see if the existence of those phonosemantic features could be explained by etymology alone or if other processes were at stake.
22The purpose of this lexicographic study is to evaluate how phonesthemic senses influence each other across the OED by identifying recurring features in the definitions. Key words are used as indicators of semantic cohesion inside the OED. This method is based on the assumption that lexicographers must have left indications of semantic evolution in the definitions, indirectly influenced by the corpus database.
23Headwords starting with the initial consonant cluster gr- were selected by hand from OED definitions (See Smith & Farina [forthcoming] for a detailed account of the selection procedure). Only monomorphemes were kept for the analysis: morphemes which cannot be further divided into smaller regular compositional elements. For example, only grey was kept, as opposed to greyback, grey-backed, greybeard, grey-blue, grey-faced, grey-haired and so on. However, certain historical morphemes that are no longer productive in Modern English were kept: -le (as in gripple or griddle), certain -er endings (as in grater, grover), as well as -el(l) (as in grummel or gravell) and -et (as in grummet). Following Smith [2016, 2020b], homonyms were treated as different entries, so that each sense could be analysed individually, with the final aim of examining recurring semantic features. Contrary to what Drellishak [2006] postulates, the shared historical roots were not excluded, because of the need to examine the divergences from a same root as well as processes of blending and potential phonosemantic change. The sample corpus is composed of a total of 219 monomorphemes, which correspond to 834 senses, among which 510 are non-obsolete (61%), i.e. still in use nowadays.
24Table 1 shows how the key word process was conducted for grunt v., and how key words serve as a basis for the construction of conceptual categories. Every semantic trait is named after its most recurring key words. Each word of the dataset can correspond to several traits, so that traits combine.
Table 1. Key word collection from gr- monomorphemes in the OED
Entry word
|
Sense
|
OED definition
|
Date
|
Use
|
Etymology
|
Key words
|
grunt
|
v. (1a)
|
Of a hog: To utter its characteristic low gruff sound. Also of other animals and of persons (with conscious allusion to the pig): To utter a sound resembling this.
|
725
|
In use
|
Old English grunnęttan (= Old High German, modern German grunzen), frequentative of grunian (compare Middle High German grunnen) to grunt, an echoic formation parallel with Latin grunnīre.
|
utter, low, sound
|
grunt
|
v. (1b)
|
To groan.
|
1340
|
Obsolete
|
|
groan
|
grunt
|
v. (2a)
|
To utter a similar sound, expressive of discontent, dissent, effort, fatigue, etc.; to grumble, murmur.
|
1325
|
In use
|
|
utter, sound, discontent, grumble, murmur
|
grunt
|
v. (3a)
|
To grind (the teeth).
|
1400
|
Obsolete
|
|
grind, teeth
|
grunt
|
v. (3b)
|
To grind with the teeth; cf. grind v.1, grint v.
|
1426
|
Obsolete
|
|
grind, teeth
|
grunt
|
n. (1)
|
The characteristic low gruff sound made by a hog; a similar sound uttered by other animals.
|
1615
|
In use
|
< grunt v.
|
low, sound, utter
|
grunt
|
n. (2a)
|
A similar sound, uttered by a human being; sometimes expressive of approbation, or the opposite. †In early use, a groan.
|
1553
|
In use
|
|
sound, utter
|
grunt
|
n. (2b)
|
An infantry soldier.
|
1969
|
U.S. slang
|
|
-
|
grunt
|
n. (3a)
|
A name for American fishes of the genus Hæmulon and allied species (as Orthopristis chrysopterus). So called from the noise they make when taken.
|
1705
|
In use
|
Apparently not connected with Dutch gront, grunt, which is a shortened form of grondel grundel n., and denotes a different fish (Cyprinus gobio).
|
-
|
grunt
|
n. (3b)
|
An English fish, ?the perch.
|
1851
|
In use
|
|
-
|
The attestation dates allow us to determine how many words acquire the trait through time, thus giving an evolutive indication. The etymological information given by the OED, combined with the raw frequencies of the traits across the sample and the attestation dates of the different senses allowed us to keep track of the semantic changes that occurred.
25Table 2 displays all the categories that emerged through the protocol. Key words that were not recurrent enough were naturally discarded as the process went on. After a quantitative comparison of the categories, four of them appeared to stand out: 1) words expressing bad humour or negative emotions (grump, grunch, grutch, grouse, grisly), represented by the weep/complain/anger (WCA) category; 2) words relating to unpleasant sensations (grit, grind, grate), represented by the harsh/rough/scrape (HRS) category; 3) unpleasant or deep sounds (groan, growl, grunt, graunch, gruntle), represented by the sound/voice/utter (SVU) category; and finally 4) the notion of prehension (grip, grope, grab, grasp), represented by the hold/clutch/touch (HCT) category.
26These phonosemantic features confirm what has been observed in the literature about the submeaning of gr-. The major traits were the most widely represented across the sample and were also the ones with best internal cohesion, meaning that their key words were semantically very close to each other.
Table 2. Key words grouped into 8 conceptual features for gr- monomorphemes in the OED
Feature abbr.
|
Conceptual categories
|
Example words
|
Number of senses
|
Proportion within the sample (%)
|
WCA
|
weep/complain/anger
|
grouch, grudge, grump, grumble, grutch, grouse, grieve, grauly, grisly
|
330
|
32.5
|
HCT
|
hold/clutch/touch
|
grab, grabble, grapple, grasp, grip, grope
|
137
|
13.5
|
HRS
|
harsh/rough/scrape
|
grash, grate, grind, grist, grint, graze
|
121
|
11.9
|
SVU
|
sound/voice/utter
|
graunch, groan, growl, grunt, gruntle
|
114
|
11.2
|
LDD
|
low/deep/dig
|
grave, groove, ground
|
97
|
9.6
|
PSF
|
plant/soil/feed
|
graft, grape, grass, graze, groot, grove, grub, gruel
|
95
|
9.4
|
LTB
|
large/thick/bulk
|
grand, great, gross, gruff
|
72
|
7.1
|
STB
|
snout/teeth/body
|
groin, grume, graisle, gralloch, grebe
|
48
|
4.7
|
The less quantitatively dominant categories are not as discriminating, and most of them are in the same order of magnitude. The minor categories plant/soil/feed, large/thick/bulk and snout/teeth/body are much less semantically coherent and recurring. They probably correspond to false positives and were therefore discarded from the rest of the study (see Drellishak [2006: 21] on statistical filters and discarding false positives).
27Features that attract a lot of senses and influence the evolution of monomorphemes tend to appear in combination with other features inside the definitions. These attractive features must have played a role in the entrenchment of the phonestheme in the mental lexicon. Combination patterns of co-emergent features indicate the lexical productivity of features, while post-attested features indicate whether senses converge or diverge from each other. They indicate a phonesthemic convergence. Table 3 shows the number of features each category combines with.
Table 3. Combinations of features in the OED.
Feature
|
Combination 0
|
Combination 1
|
Combination 2
|
Total
|
sound/voice/utter
|
37
|
32.5%
|
58
|
50.9%
|
19
|
16.7%
|
77
|
67.5%
|
harsh/rough/scrape
|
63
|
52.1%
|
41
|
33.9%
|
17
|
14.0%
|
58
|
47.9%
|
low/deep/dig
|
68
|
70.1%
|
19
|
19.6%
|
10
|
10.3%
|
29
|
29.9%
|
weep/complain/anger
|
234
|
70.9%
|
80
|
24.2%
|
16
|
4.8%
|
96
|
29.1%
|
hold/clutch/touch
|
116
|
84.7%
|
19
|
13.9%
|
2
|
1.5%
|
21
|
15.3%
|
The “combination 0” column shows the number of senses that only carry one feature and do not combine with any of the other features. The “combination 1” column shows the number of senses whose feature combines with 1 other feature. The “combination 2” column shows occurrences of category that combine with 2 other features. Figure 1 renders those data into a bar chart.
Figure 1. Combinations of features in the OED.
28The categories SVU (sound/voice/utter) and HRS (harsh/rough/scrape) are the ones that combine the most, which means that words which refer to these features also tend to refer a lot to other features. On the other hand, HCT (hold/ clutch/touch) combines very rarely, on top of not being well attested overall. SVU is not amongst the most attested features but it combines with a lot of other ones This shows that the notion of sound is inherent to a lot of gr- words, whatever their category.
29The SVU feature exhibits a lot of semantic change. Most gr- words that relate to sound are particularly polysemous and a lot of them show a semantic shift towards other categories, as the combinatory data pointed out. Semantic shift from etymological meaning to another phonosemantic feature is interpreted as evidence of a phonesthemic impact on the lexicon. Table 4 shows that the adjective gruff has shifted from an original meaning dating back to Old English hréof “rough, scabby”, relating to the HRS feature, to the OED sense adj. (3b) “Of the voice and speech, implying the utterance of hoarse or guttural sounds”, relating to the SVU category.
Table 4. Senses of gruff adj. [1533] in the OED.
Entry word
|
Sense
|
OED definition
|
Date
|
Use
|
Semantic features
|
gruff
|
adj. (1a)
|
Coarse, coarse-grained; containing coarse or rough particles.
|
1533
|
Obsolete exc.
Scottish and technical.
|
large/thick/bulk (LTB) harsh/rough/scrape (HRS)
|
gruff
|
adj. (1b)
|
Of immaterial things: Rude, gross, unpolished. Also said of a guess = ‘rough’.
|
1681
|
Scottish
|
harsh/rough/scrape (HRS)
|
gruff
|
adj. (2)
|
Of a surface: Rough, rugged.
|
1697
|
Obsolete. Rare.
|
harsh/rough/scrape (HRS)
|
gruff
|
adj. (3a)
|
Rough, surly, or sour in aspect or manner; said also of appearances.
|
1691
|
In use
|
harsh/rough/scrape (HRS) weep/complain/anger (WCA)
|
gruff
|
adj. (3b)
|
Of the voice and speech, implying the utterance of hoarse or guttural sounds.
|
1712
|
In use
|
sound/voice/utter (SVU) harsh/rough/scrape (HRS)
|
gruff
|
adj. (3c)
|
Quasi-adv.
|
1841
|
In use
|
sound/voice/utter (SVU) harsh/rough/scrape (HRS)
|
gruff
|
n. (1a)
|
In pharmacy, the coarse residue, which will not pass through the sieve in pulverization.
|
1853
|
Pharmacology
|
large/thick/bulk (LTB)
|
gruff
|
n. (2)
|
A quarrel, ‘tiff’.
|
1857
|
? U.S; regional
|
weep/complain/anger (WCA)
|
30The lexicographic analysis of gr- has shown the following points (See Smith & Farina [forthcoming] for a comparison between fl-, sw- and gr-):
-
The terms used by lexicographers vary, but all in all 11% of the sample words are dialectal words, regional peculiarities or slang words, 12% are said to be rare, archaic or specific and 39% of all the senses are obsolete;
-
The 3 major features, WCA, HRS and SVU, account for 56% of all the senses (294 senses, out of 522);
-
In general, high level of polysemy can be observed across the sample;
-
Homonymy tends to be rather common: see e.g. the nouns graft n.1 [1871], graft n.2 [1644], graft n.3 [1620], graft n.4 [slang; 1853] and graft n.5 [U.S.; 1865], as well as the verbs graft v.1 [1483], graft v.2 [dialect; 1823], graft v.3 [slang; 1859] and graft v.4 [U.S.; 1859];
-
The semantic features do not all behave in the same way. Words that belong to the categories WCA, HRS and SVU combine a lot with other features, they show a high degree of semantic variation and they also seem to acquire new features through semantic change. They also show a lot of post-attested features, i.e. words that did not originally contain them, but came to acquire those semantic elements overtime;
-
On the other hand of the spectrum, the HCT category seems to undergo very little semantic change, if any at all, and words that trigger this category do not acquire new features overtime.
31The previous section showed that 1) the semantic features SVU, WCA and HRS display more semantic change than the feature HCT and that 2) SVU, WCA and HRS show evidence of phonesthemic convergence, as opposed to HCT.
32Those preliminary results drawn from the OED need to be tested through a corpus analysis. Analysing the collocational behaviour of gr- words in context will give the frequency of use and the nature of the texts they appear in. This will show which collocates may have had an influence on the semantic change of gr- words. The initial hypothesis is that a few gr- words have played a key role in entrenching the form-meaning association between /gr/ and the submeanings pointed out by the key word-based lexicographic results. The postulate is that the more widely used a word is, the more impact it has on the lexicon, based on the view that “lexical storage is highly affected by language use” (Bybee [2007: 280]), and that lexical use is a form of procedural knowledge (Bybee [2007: 282]). As Benczes [2019: 17] puts it: “the use of forms and patterns has an effect on their mental representation; those that are used more often will have stronger representations and will also be accessed more easily.”
33In addition, a smaller number of occurrences of a word in a usage-based corpus means it is more difficult to assess the level of prototypicality of the contexts. That being said, frequency was not the only factor taken into account. It is also possible that the phonesthemic entrenchment might be restricted to certain semantic domains, and not to the whole lexicon. As Kilgarriff [2004] points out, different word senses may be the most common senses of different lexical domains. The overall frequency of a word is not necessarily a good indication of its role inside its specific lexical domain.
34The selected sample words are: groan, grunt, grudge, grip and grasp. These are all initially verbs, although they also convert to nouns. Three of them denote sounds, while the other two refer to an act of prehension. None of them are obsolete or dialectal words. The fact that the first three on the one hand, and the last two on the other, are so semantically close raises the question of synonymy. My hypothesis is that these two groups of words should display divergent nodes of synonyms in the corpus, and their lexical behaviour in context is expected to show some degree of variation. Table 3 shows the number of occurrences of these selected gr- words across the corpus, both for verbs and nouns.
Table 5. Frequencies of the selected gr- words in the OEC
Gr- word
|
Verb
|
Noun
|
Total
|
groan
|
9,294
|
3,973
|
13,267
|
grunt
|
3,656
|
3,782
|
7,438
|
grudge
|
542
|
4,702
|
5,244
|
grip
|
17,929
|
37,709
|
55,638
|
grasp
|
25,194
|
14,023
|
39,217
|
35This handful of gr- words acts as a gateway to tracking down the semantic and distributional behaviour of the phonestheme gr-. As Lenci [2008: 1] puts it: “the statistical distribution of words in context plays a key role in characterizing their semantic behavior”. The corpus analysis of gr-words will thus help determine the semantic content of gr-.
36The lexicographic data have established that among the four sets of major semantic features, three of them entertain close semantic bonds (both metaphorical and metonymic) and extensively combine with each other. Therefore, is this statement also confirmed by corpus data? If so, what can explain such a coherence? Is there evidence of phonesthemic clustering within the collexeme network of our gr- word sample?
37I chose a corpus that could be searched by Sketch Engine and that was large enough to allow for statistically valid interpretations. The Oxford English Corpus was chosen because of its large inventory of two billion words and the broad variety of texts it contains. It encompasses both UK and US English contexts, as well as contexts from Australia, New Zealand, the Caribbean, Canada, India, Singapore, and South Africa3. The OEC spans over a short period of time, from 2000 to 2006, which makes it a micro-diachronic investigating tool. The dataset is mostly drawn from the Internet and processed to make it searchable by the Penn Treebank tag set (Culpeper [2009: 66]).
38The OEC was treated via the distributional online tool Sketch Engine with the aim of tracking the grammatical and collocational behaviour of the selected gr- words. The large amount of textual material provided by the OEC combined with the distributional proximity analysis tools offered by Sketch Engine were judged profitable to my investigation.
39The collocation scores as well as syntactic templates searchable on Sketch Engine allow to see how gr- words are used in context and if some semantic aspects are triggered over others according to the context (see Kilgarriff et al. [2004] and Kilgarriff et al. [2012] on using Sketch Engine as linguistic evidence). A protocol was established to treat the data in a profitable manner.
40In order to tackle the semantic and distributional behaviour of my sample of gr- words, two functions of Sketch Engine are used:
-
The Thesaurus presents a list of the most frequent and most similar lexicogrammatical collexemes. Note that the terminology can be confusing since Sketch Engine uses the term “synonym” to refer to the semantic network-induced collexemes;
-
The Word Sketch function relies on lexicogrammatical parsing, which allows to choose the most frequent or most similar collocates to be looked at in more details. As mentioned further below, the parsing algorithm shows numerous mistakes.
41A number of constraints had to be respected:
-
Only the 20 first typical collexemes generated by Sketch Engine via the Thesaurus were looked at, since the typicality scores become insignificant as one scrolls down the list for too long;
-
Collexemes that show a similarity of less than 0.1% were discarded, as well as those that have a score smaller than 1;
-
For each of the 5 gr- words, only a limited number of collexemes were looked at in more details. This number varies depending on the reliability of the tagging and parsing established by Sketch Engine, and on the potential for semantic change. But in general, at least 2 collocates for each syntactic category were selected to be further investigated;
-
Because the scope of this study had to be constricted, grammatical collexemes were not taken into account;
-
Proper names were also discarded, for the sake of clarity and time, and because onomastics is a special case that needs extra care with its own questions that will not be addressed here;
-
When possible, I tried to match the context of use of gr- words to the OED definitions in order to assess the reliability of the OED and also to determine what contextual elements might have triggered this particular sense.
42However, Sketch Engine corpus probing has shown a number of tagging and parsing errors. These difficulties are addressed, and a number of methodological solutions are proposed:
-
Most web links gathered in the OEC date back to the early 2000s and are no longer active;
-
Some syntactic functions are not well identified by the algorithm. The wrong attributions were not discarded straight away, because although the quantitative data must be at least partly wrong (typically the general frequencies and proximity scores), the contexts themselves might still be valuable;
-
In general, the more occurrences of a collocate, the fewer tagging and parsing errors, which is also why the size of the corpus matters. In the end, the syntactic parsing category that proved to be the most reliable was the and/or group. Most of these collocates were exploitable as they contained much less attribution errors than the rest;
-
It is common to encounter copies of the same original work treated as distinct sources, which skews the quantitative data. Sometimes the sentences are not exactly the same, but the core text is very similar;
-
There were many more mistakes with nouns than verbs, thus the final data sheet contained more verbal contexts than nominal ones.
43Tables 6 and 7 show the Thesaurus data for groan. Table 6 shows the most similar collexemes of the verb groan, while Table 7 shows the most similar collexemes of the noun groan. Figures 2 and 3 show those same collexemes as a visual representation. The listed collexemes are based on the similarity of collocational behaviour and their overall frequency. The closer to groan on the collexeme space, the higher the collocation similarity, while the bigger the circle, the more frequent the collexeme across the OEC. Groan is the most frequent of the five gr- words selected. Both the verb and noun collexemes relate to a sound production of some sort, mostly human (sigh, mumble) or animal (howl, roar, hiss) in origin. In most cases the subject of such verbs can be either human or animal. Four gr- words are among the collexemes, displaying relatively high similarity scores, compared to what can be found elsewhere. Some of the collexemes also carry a phonesthemic part on top of gr- (sm- in smirk, bl- in blush, sn- in snort, wh- in whisper, shr- in shriek, squ- in squeal, scr- in scream) or seem to be onomatopoeic or mimetic in origin (chuckle, giggle, mumble, grunt, growl, howl, hiss). There is even a case of rhyming with the word moan. More generally, a lot of these words have expressive meanings (giggle, exclaim). A lot of these collexemes also convey a negative connotation or feeling (wince, chuckle, frown, wail, sob). Some of them also depict facial expressions (frown, grimace), or a distorted smile (grin, smirk). This suggests that groan tends to appear in multimodal contexts, in combination with other means of communication and expression. The element -le is also present in three of them (chuckle, giggle, mumble). Most collexemes correspond to the semantic category sound/voice/utter, and a few to the category weep/complain/anger (sigh, wail, sob).
Table 6. Frequency and similarity scores of the collexemes of groan v.
Collocate
|
Frequency
|
Similarity
|
sigh
|
33,402
|
0.343
|
gasp
|
12,221
|
0.340
|
moan
|
11,317
|
0.298
|
chuckle
|
11,899
|
0.276
|
wince
|
5,437
|
0.274
|
giggle
|
12,251
|
0.264
|
shudder
|
7,267
|
0.260
|
exclaim
|
17,017
|
0.254
|
frown
|
13,602
|
0.251
|
growl
|
6,568
|
0.249
|
mutter
|
16,916
|
0.248
|
grunt
|
3,656
|
0.244
|
mumble
|
8,640
|
0.239
|
grin
|
24,045
|
0.237
|
smirk
|
7,223
|
0.236
|
blush
|
10,153
|
0.235
|
snort
|
5,573
|
0.228
|
hiss
|
6,893
|
0.227
|
grimace
|
3,015
|
0.217
|
whisper
|
32,378
|
0.215
|
Figure 2. Thesaurus representation of groan v.
44A substantial amount of the noun collexemes are verbs misrecognised by the algorithm. Virtually, all the collexemes presented as verbs, on the other hand, seem to be real verbs. Only sigh, gasp and moan are close semantic relatives of groan, which indicates that groan probably has a key role to play in the mental lexicon, potentially as a foundation word (see Smith [2022a: 284]).
Table 7. Frequency and similarity scores of the collexemes of groan n.
Collexeme
|
Frequency
|
Similarity
|
moan
|
3,034
|
0.358
|
gasp
|
7,134
|
0.335
|
sigh
|
20,593
|
0.329
|
shriek
|
2,661
|
0.293
|
howl
|
3,359
|
0.277
|
grunt
|
3,782
|
0.263
|
growl
|
2,396
|
0.261
|
wail
|
2,326
|
0.254
|
squeal
|
1,947
|
0.250
|
chuckle
|
5,036
|
0.245
|
scream
|
13,931
|
0.239
|
giggle
|
5,653
|
0.236
|
yelp
|
1,134
|
0.234
|
snort
|
1,268
|
0.226
|
sob
|
3,966
|
0.226
|
shout
|
8,616
|
0.223
|
yell
|
2,690
|
0.204
|
roar
|
7,229
|
0.204
|
hiss
|
3,093
|
0.192
|
yawn
|
3,167
|
0.191
|
Figure 3. Thesaurus representation of groan n.
45The lexicogrammatical positions provided by the Word Sketch show that the and/or category contains the most frequent and the most similar collexemes. Moan is the most frequent as well as the most similar collexeme here, as it was already in the Thesaurus. The X* in N category shows the source of groaning. Most convey a negative connotation (pain, agony, despair) and only a few have a positive connotation (pleasure).
Table 8. Collexemes for groan v. in the OEC (Word Sketch)
And/or
|
Freq.
|
Score
|
N subj V*
|
Freq.
|
Score
|
X* in N
|
Freq.
|
Score
|
moan
|
316
|
11.4
|
shelf
|
31
|
8.4
|
travail
|
11
|
9.3
|
creak
|
84
|
10.5
|
creation
|
33
|
7.8
|
pain
|
115
|
9.3
|
grunt
|
77
|
10.1
|
table
|
29
|
7.0
|
frustration
|
38
|
8.8
|
roll
|
61
|
8.8
|
earth
|
12
|
6.1
|
agony
|
|
|
wince
|
12
|
7.6
|
[...]
|
|
|
annoyance
|
|
|
gasp
|
16
|
7.6
|
|
|
|
exasperation
|
|
|
grumble
|
11
|
7.5
|
|
|
|
ecstasy
|
|
|
sigh
|
31
|
7.4
|
|
|
|
sympathy
|
|
|
slump
|
9
|
7.4
|
|
|
|
despair
|
|
|
complain
|
16
|
7.4
|
|
|
|
disgust
|
|
|
weep
|
15
|
7.3
|
|
|
|
pleasure
|
|
|
shriek
|
10
|
7.2
|
|
|
|
[...]
|
|
|
writhe
|
9
|
7.2
|
|
|
|
|
|
|
shudder
|
9
|
7.2
|
|
|
|
|
|
|
strain
|
8
|
7.0
|
|
|
|
|
|
|
hiss
|
9
|
6.9
|
|
|
|
|
|
|
whimper
|
7
|
6.9
|
|
|
|
|
|
|
[...]
|
|
|
|
|
|
|
|
|
gripe
|
6
|
6.9
|
|
|
|
|
|
|
[...]
|
|
|
|
|
|
|
|
|
grimace
|
6
|
6.7
|
|
|
|
|
|
|
[...]
|
|
|
|
|
|
|
|
|
mumble
|
6
|
6.6
|
|
|
|
|
|
|
[...]
|
|
|
|
|
|
|
|
|
grieve
|
5
|
6.3
|
|
|
|
|
|
|
[...]
|
|
|
|
|
|
|
|
|
mutter
|
5
|
6.3
|
|
|
|
|
|
|
bleed
|
6
|
6.2
|
|
|
|
|
|
|
crack
|
7
|
6.2
|
|
|
|
|
|
|
growl
|
5
|
6.1
|
|
|
|
|
|
|
cry
|
17
|
6.1
|
|
|
|
|
|
|
[...]
|
|
|
|
|
|
|
|
|
Table 9. Collexemes for groan n. in the OEC (Word Sketch)
And/or
|
Freq.
|
Score
|
X mod N*
|
Freq.
|
Score
|
V obj N*
|
Freq.
|
Score
|
moan
|
114
|
11.0
|
audible
|
35
|
8.2
|
stifle
|
43
|
7.5
|
grunt
|
83
|
10.4
|
strangle
|
8
|
7.0
|
elicit
|
46
|
7.0
|
creak
|
38
|
10.0
|
muffle
|
12
|
6.9
|
emit
|
33
|
6.8
|
sigh
|
61
|
9.7
|
guttural
|
8
|
6.9
|
utter
|
17
|
6.0
|
shriek
|
27
|
9.1
|
pained
|
8
|
6.7
|
suppress
|
30
|
5.4
|
cry
|
39
|
8.1
|
exasperated
|
8
|
6.5
|
smother
|
4
|
5.2
|
gasp
|
10
|
7.7
|
unearthly
|
6
|
6.5
|
hear
|
213
|
4.7
|
scream
|
22
|
7.6
|
smother
|
5
|
6.4
|
repress
|
4
|
4.7
|
sob
|
8
|
7.5
|
[...]
|
|
|
swallow
|
4
|
3.5
|
grumble
|
6
|
7.4
|
loud
|
66
|
6.4
|
prompt
|
14
|
3.4
|
grimace
|
6
|
7.2
|
frustrated
|
20
|
6.4
|
induce
|
9
|
3.0
|
squeak
|
7
|
7.2
|
orgasmic
|
5
|
6.2
|
provoke
|
5
|
2.6
|
cheer
|
16
|
7.1
|
agonized
|
4
|
6.1
|
draw
|
22
|
2.1
|
gripe
|
5
|
7.0
|
disgusted
|
4
|
5.9
|
ignore
|
5
|
1.0
|
boo
|
6
|
7.0
|
stifle
|
7
|
5.9
|
|
|
|
guffaw
|
4
|
6.8
|
[...]
|
|
|
|
|
|
giggle
|
7
|
6.8
|
anguished
|
4
|
5.4
|
|
|
|
yell
|
6
|
6.8
|
ache
|
5
|
5.3
|
|
|
|
laugh
|
17
|
6.8
|
mournful
|
4
|
5.3
|
|
|
|
squeal
|
5
|
6.7
|
disappointed
|
4
|
5.0
|
|
|
|
yawn
|
4
|
6.7
|
faint
|
11
|
4.9
|
|
|
|
lamentation
|
4
|
6.7
|
[...]
|
|
|
|
|
|
rattle
|
5
|
6.6
|
agonizing
|
4
|
4.8
|
|
|
|
chuckle
|
4
|
6.5
|
[...]
|
|
|
|
|
|
curse
|
7
|
6.3
|
slight
|
25
|
4.0
|
|
|
|
whisper
|
5
|
6.3
|
hollow
|
4
|
3.8
|
|
|
|
shout
|
5
|
6.0
|
soft
|
29
|
3.7
|
|
|
|
tear
|
17
|
5.7
|
[...]
|
|
|
|
|
|
crack
|
7
|
5.6
|
deep
|
31
|
2.7
|
|
|
|
laughter
|
9
|
5.5
|
[...]
|
|
|
|
|
|
complaint
|
8
|
5.1
|
low
|
47
|
1.6
|
|
|
|
click
|
4
|
4.7
|
|
|
|
|
|
|
noise
|
5
|
3.4
|
|
|
|
|
|
|
[...]
|
|
|
|
|
|
|
|
|
sound
|
7
|
3.1
|
|
|
|
|
|
|
ground
|
4
|
2.2
|
|
|
|
|
|
|
[...]
|
|
|
|
|
|
|
|
|
46The nature of the contexts of use of groan greatly depends on the collocates. In general, phonesthemic and onomatopoeic collocates tend to trigger contexts that contain a large proportion of phonesthemic and onomatopoeic words. They also tend to contain more expressive marks, like capital letters, expression points, insults and repetitions, although it is not systematic. They also tend to appear in dialogues and conversations. Collocates starting with cr- are particularly prolific in onomatopoeic contexts, like in (1). Intensifying adverbs also seem to trigger more phonesthemic and onomatopoeic contexts, as in (2). The OEC does not always provide the source from which the context is quoted, and the web links are dead for most of them, but they are still included here when indicated.
(1) As the sound of the thunder rolled away, they all heard a noise of creaking and groaning coming from outside the house. Rushing to the windows, they saw the kaldis tree – split and blackened where it had been struck – sway for a moment and then crash to the ground. Oh, no! cried Dunelm in anguish. (source unknown)
(2) As I walked towards Josh’s motionless body, glass cracked and wood crunched. Tile crumbled from the walls and sparks erupted from falling lights. The floor was shimmering with the metal from instruments. A soft wind blew my hair and rain moistened my bloody face. I glared down at the back of Josh’s head. A maroon stream of blood was slithering down the back of his it, dyeing the color of his hair deep red. My lips trembled as he groaned loudly. “Y-yeah.” he sputtered. I smiled widely. “W-W-What’s going on?” Josh said with a hint of terror in his tone. “It must be Josh.” I thought to myself. Another piece of weight was added. I assumed it was the kid. The car door slammed shut from the wind pressure, crushing my already pulsing leg in the door. Another gust of wind ripped it open, deafening my screams of pain. (source: http://www.fictionpress.com/read.php?storyid=1754021&chapter=1)
47Tables 10 and 11 show the most frequent collexemes of grunt v. and grunt n., respectively. The same goes with Figures 4 and 5. The collexeme network of grunt is very similar to that of groan, since they share the same collexemes, and in roughly the same order, from most to least similar. They both exhibit the same gr- collexemes, with the addition of grumble here, and the absence of grin. Once again, there are words that depict facial expressions (grimace, scowl, smirk). A substantial amount of collexemes contain the long vowel /i:/ (squeal, shriek, squeak, scream) or the short vowel /ɪ/ (grimace, whimper, hiss, giggle). Although high acoustic frequencies and closed vowels have been statistically correlated with smallness (cf. teeny, wee, itsy-bitsy) (Ohala [1994: 335-336]), which has also been confirmed by experimental evidence (Hoshi et al. [2019]), these collexemes do not follow this pattern here.
Table 10. Frequency and similarity scores of the collexemes of grunt v.
Collexeme
|
Frequency
|
Similarity
|
groan
|
9,294
|
0.244
|
growl
|
6,568
|
0.239
|
moan
|
11,317
|
0.215
|
mumble
|
8,640
|
0.208
|
snort
|
5,573
|
0.207
|
gasp
|
12,221
|
0.200
|
grimace
|
3,015
|
0.199
|
whimper
|
2,233
|
0.198
|
squeal
|
4,103
|
0.185
|
shriek
|
5,252
|
0.184
|
hiss
|
6,893
|
0.184
|
pant
|
3,817
|
0.182
|
chuckle
|
11,899
|
0.177
|
mutter
|
16,916
|
0.168
|
grumble
|
6,685
|
0.168
|
murmur
|
7,821
|
0.168
|
sigh
|
33,402
|
0.168
|
scowl
|
3,501
|
0.167
|
smirk
|
7,223
|
0.166
|
yelp
|
1,671
|
0.166
|
Figure 4. Thesaurus representation of grunt v.
Table 11. Frequency and similarity scores of the collexemes of grunt n.
Collexeme
|
Frequency
|
Similarity
|
groan
|
3,973
|
0.263
|
growl
|
2,396
|
0.253
|
gasp
|
7,134
|
0.210
|
moan
|
3,034
|
0.206
|
sigh
|
20,593
|
0.206
|
shriek
|
2,661
|
0.198
|
squeal
|
1,947
|
0.196
|
yelp
|
1,134
|
0.194
|
howl
|
3,359
|
0.190
|
snort
|
1,268
|
0.183
|
shout
|
8,616
|
0.173
|
giggle
|
5,653
|
0.167
|
exclamation
|
3,560
|
0.166
|
murmur
|
3,472
|
0.165
|
scream
|
13,931
|
0.162
|
yell
|
2,690
|
0.155
|
wail
|
2,326
|
0.153
|
chuckle
|
5,036
|
0.151
|
squeak
|
1,576
|
0.151
|
whoop
|
2,424
|
0.149
|
Figure 5. Thesaurus representation of grunt n.
48The collocation profile of grunt is very similar to that of groan. Once again, the and/or category contains the most frequent and the most similar collexemes. Groan is the most similar collexeme of the category. The X* in N category, although less coherent, is very similar to that of groan. These observations show that these two words are distributionally close. There are a few verbs of body movement (heave, nod, shrug, slump), suggesting that grunt also has a multimodal behaviour.
Table 12. Collexemes for grunt v. in the OEC (Word Sketch)
And/or
|
Freq.
|
Score
|
V* obj N
|
Freq.
|
Score
|
X* in N
|
Freq.
|
Score
|
groan
|
77
|
10.1
|
acknowledg- ment
|
5
|
6.0
|
annoyance
|
5
|
8.1
|
sweat
|
36
|
9.3
|
pig
|
10
|
5.8
|
pain
|
28
|
7.3
|
growl
|
16
|
8.5
|
greeting
|
6
|
5.7
|
reply
|
22
|
7.2
|
moan
|
28
|
8.3
|
hello
|
4
|
5.0
|
satisfaction
|
6
|
7.2
|
snort
|
12
|
8.3
|
reply
|
8
|
4.6
|
frustration
|
10
|
7.0
|
strain
|
10
|
8.1
|
approval
|
9
|
3.5
|
surprise
|
9
|
6.8
|
heave
|
7
|
7.8
|
response
|
5
|
1.2
|
disgust
|
6
|
6.3
|
wince
|
7
|
7.7
|
|
|
|
agreement
|
13
|
5.7
|
grumble
|
7
|
7.6
|
|
|
|
response
|
36
|
4.7
|
squeak
|
6
|
7.6
|
|
|
|
return
|
5
|
2.8
|
wheeze
|
7
|
7.6
|
|
|
|
|
|
|
squeal
|
7
|
7.6
|
|
|
|
|
|
|
howl
|
7
|
7.4
|
|
|
|
|
|
|
mumble
|
6
|
7.4
|
|
|
|
|
|
|
scowl
|
5
|
7.3
|
|
|
|
|
|
|
swear
|
11
|
7.2
|
|
|
|
|
|
|
grimace
|
4
|
7.0
|
|
|
|
|
|
|
point
|
12
|
7.0
|
|
|
|
|
|
|
struggle
|
8
|
6.9
|
|
|
|
|
|
|
snarl
|
4
|
6.8
|
|
|
|
|
|
|
sniff
|
4
|
6.7
|
|
|
|
|
|
|
shriek
|
4
|
6.6
|
|
|
|
|
|
|
gasp
|
5
|
6.5
|
|
|
|
|
|
|
curse
|
5
|
6.3
|
|
|
|
|
|
|
cough
|
6
|
6.2
|
|
|
|
|
|
|
grind
|
4
|
6.0
|
|
|
|
|
|
|
lay
|
4
|
6.0
|
|
|
|
|
|
|
scream
|
16
|
6.0
|
|
|
|
|
|
|
puff
|
4
|
5.9
|
|
|
|
|
|
|
shout
|
8
|
5.7
|
|
|
|
|
|
|
nod
|
8
|
5.3
|
|
|
|
|
|
|
sigh
|
5
|
5.0
|
|
|
|
|
|
|
[...]
|
|
|
|
|
|
|
|
|
Table 13. Collexemes for grunt n. in the OEC (Word Sketch)
And/or
|
Freq.
|
Score
|
X mod N*
|
Freq.
|
Score
|
V obj N*
|
Freq.
|
Score
|
groan
|
83
|
10.4
|
guttural
|
13
|
7.5
|
utter
|
12
|
5.5
|
moan
|
24
|
8.8
|
annoy
|
12
|
7.4
|
emit
|
13
|
5.5
|
squeal
|
19
|
8.7
|
grunt
|
10
|
7.3
|
elicit
|
5
|
3.8
|
growl
|
14
|
8.3
|
non-committal
|
8
|
7
|
hear
|
68
|
3.1
|
snapper
|
13
|
8.1
|
blackberry
|
7
|
6.7
|
lack
|
5
|
1.0
|
sigh
|
19
|
8.1
|
blue-striped
|
5
|
6.7
|
manage
|
6
|
1.0
|
squeak
|
9
|
7.7
|
muffle
|
10
|
6.6
|
|
|
|
gasp
|
8
|
7.4
|
irritated
|
7
|
6.5
|
|
|
|
howl
|
7
|
7.3
|
satisfied
|
10
|
6.5
|
|
|
|
scream
|
17
|
7.3
|
audible
|
11
|
6.4
|
|
|
|
croak
|
5
|
7.2
|
mono-syllabic
|
5
|
6.4
|
|
|
|
whimper
|
5
|
7.2
|
Cherub
|
4
|
6.3
|
|
|
|
snort
|
5
|
7.2
|
unintelligible
|
5
|
6.2
|
|
|
|
nod
|
12
|
7.1
|
animalistic
|
4
|
6.0
|
|
|
|
shrug
|
6
|
7.1
|
low-end
|
8
|
5.9
|
|
|
|
mono-syllable
|
4
|
7.0
|
inarticulate
|
4
|
5.9
|
|
|
|
hoot
|
5
|
7.0
|
frustrated
|
11
|
5.5
|
|
|
|
grind
|
6
|
7.0
|
appreciative
|
5
|
5.4
|
|
|
|
twitch
|
5
|
6.9
|
loud
|
26
|
5.1
|
|
|
|
sniff
|
4
|
6.8
|
infantry
|
7
|
4.9
|
|
|
|
yelp
|
4
|
6.7
|
pig
|
8
|
4.8
|
|
|
|
snarl
|
4
|
6.6
|
low-level
|
6
|
4.7
|
|
|
|
yell
|
5
|
6.6
|
gaming
|
11
|
4.4
|
|
|
|
roar
|
5
|
6.5
|
monkey
|
4
|
4.4
|
|
|
|
slump
|
5
|
6.5
|
occasional
|
30
|
4.3
|
|
|
|
shriek
|
4
|
6.4
|
Marine
|
10
|
4.1
|
|
|
|
hiss
|
5
|
6.2
|
angry
|
12
|
3.7
|
|
|
|
cry
|
10
|
6.2
|
sheer
|
13
|
3.4
|
|
|
|
grunt
|
4
|
6.1
|
soft
|
19
|
3.1
|
|
|
|
shout
|
5
|
6.0
|
army
|
9
|
3.0
|
|
|
|
gesture
|
12
|
5.9
|
level
|
7
|
2.8
|
|
|
|
glance
|
4
|
5.9
|
mighty
|
4
|
2.8
|
|
|
|
whistle
|
8
|
5.6
|
graphics
|
4
|
2.8
|
|
|
|
jack
|
4
|
5.6
|
army
|
9
|
2.5
|
|
|
|
cough
|
6
|
5.4
|
processing
|
4
|
2.5
|
|
|
|
laugh
|
5
|
5.0
|
slight
|
8
|
2.3
|
|
|
|
sweat
|
6
|
4.9
|
engine
|
4
|
2.2
|
|
|
|
breathing
|
4
|
4.9
|
intellectual
|
10
|
2.1
|
|
|
|
click
|
4
|
4.7
|
enemy
|
4
|
2.1
|
|
|
|
crash
|
4
|
4.5
|
odd
|
7
|
2.0
|
|
|
|
noise
|
5
|
3.4
|
marine
|
4
|
1.9
|
|
|
|
call
|
4
|
2.6
|
happy
|
6
|
1.4
|
|
|
|
sound
|
5
|
2.6
|
animal
|
6
|
1.3
|
|
|
|
word
|
5
|
1.9
|
|
|
|
|
|
|
[...]
|
|
|
|
|
|
|
|
|
music
|
6
|
1.0
|
|
|
|
|
|
|
[...]
|
|
|
|
|
|
|
|
|
49Some collexemes (acknowledgment, reply, hello, gesture) show a high level of interaction across the contexts, as well as the expression of emotions, especially in the X* in N group (annoyance, satisfaction, frustration). Those emotions are predominantly negative and occur in arguments and disputes. In most cases, the word grunt does not appear directly inside the dialogue, but near it, as in (3). In this case, grunt is used as a reporting verb and serves to convey the disapproving tone of the subject.
(3) “Anyone checked the basement for pods?” Jake mumbled. “I beg your pardon, sir?” Eight inquired quizzically. “We have no basement and therefore, no pods.” “It was a joke, Eight.” Jake replied. Jake grunted an acknowledgment and continued his inspection of the eggs. (source unknown)
50Overall, contexts including grunt are very similar to the ones including groan. Grunt appears extensively in collocation with groan and moan, and those contexts tend to feature a lot of other phonesthemic or iconic collocates, as in (4). Once again, the intensifier loudly also triggers phonesthemic collocates, as well as expressive markers (exclamative points and question marks), as in (5).
(4) It was the raw, infectious beat of tracks like I Feel Good that truly revealed his art. To witness a James Brown concert was to see “performance” taken to its extreme. His gravelly voice extended to include every conceivable screech, scream, grunt and moan; he wore eye make-up, his trousers were tight, his pompadoured hair could appear to defy gravity; he would shuffle and shimmy and perform flying splits. (source: https://www.telegraph.co.uk/news/obituaries/1537815/James-Brown.html)
(5) Dimitri was relieved that he had not noticed how long it had taken him to whip up something from scratch. “It’s pasta sir,” He replied. I made it you bloody git! How could he think that it was frozen food!? Dimitri cringed away from the raging monster and quickly stumbled towards the door. He heard his father grunt loudly and mutter. He grabbed his leather jacket from the stand in the corridor and quickly exited the house. He yawned and hauled himself up onto the bus. It was so much effort. He groaned and leaning his head on the seat in front of him, turning away from his friend. (source: https://www.fictionpress.com/read.php?storyid=1646777&chapter=7)
51Contexts are often conversational, even though a lot of those conversations are fictional. Using phonesthemes seems to produce expressive contexts, rather than purely descriptive and informational ones.
52As for the previous case study words, Tables 14 and 15, as well as Figures 6 and 7, show the Thesaurus results for grudge, both verb and noun. The situation is very different than for the first two words, as this time there are hardly any words that are phonesthemic or iconic in any way, apart from grievance, which also conveys a negative meaning. The frequencies are much more irregular and chaotic than with the collexemes of groan and grunt, and the similarity scores are lower. Some of the collexemes of grudge v. are paradigmatic forms (repining, forbode, missing), indicating that not all the forms of each word have been successfully parsed by Sketch Engine. This issue may have distorted the figures, but not as to show such discrepancies in frequencies and such low scores. Some of the collexemes are in fact not even verbs at all (naïve, jar, gall), which may be partly explained by the fact that a lot of instances of grudge v. in the OEC correspond to the set phrase “to hold a grudge against […]”.
53However, the collexemes virtually all show some level of negative meaning or connotation. Some of them are synonyms, especially nouns (animosity, resentment, enmity, rancor), but their similarity scores are low. Most of them match the OED-drawn category weep/complain/anger. A large number of collexemes cling around grudge, which suggests that grudge does not play the same lexical role as groan and grunt do. It probably does not hold a special place in the mental lexicon of native speakers as groan and grunt do.
Table 14. Frequency and similarity scores of the collexemes of grudge v.
Collexeme
|
Frequency
|
Similarity
|
naïve
|
1,204
|
0.098
|
condescend
|
2,373
|
0.097
|
repining
|
28
|
0.094
|
begrudge
|
2,075
|
0.088
|
underwhelm
|
2,208
|
0.087
|
forebode
|
88
|
0.082
|
nauseate
|
1,410
|
0.079
|
cloy
|
1,347
|
0.074
|
disbelieve
|
1,888
|
0.066
|
convolute
|
1,205
|
0.065
|
pixelate
|
580
|
0.063
|
jar
|
3,322
|
0.057
|
gall
|
2,269
|
0.056
|
missing
|
35
|
0.055
|
discomfort
|
726
|
0.055
|
dispirit
|
1,595
|
0.054
|
garble
|
1,342
|
0.054
|
oversimplify
|
1,709
|
0.054
|
misguide
|
4,249
|
0.051
|
abash
|
431
|
0.051
|
Figure 6. Thesaurus representation of grudge v.
Table 15. Frequency and similarity scores of the collexemes of grudge n.
Collexeme
|
Frequency
|
Similarity
|
vendetta
|
2,326
|
0.214
|
grievance
|
13,492
|
0.153
|
animosity
|
4,916
|
0.151
|
resentment
|
14,817
|
0.128
|
enmity
|
2,963
|
0.127
|
animus
|
1,171
|
0.110
|
jealousy
|
9,087
|
0.109
|
bitterness
|
8,647
|
0.107
|
malice
|
4,620
|
0.104
|
rancor
|
1,573
|
0.101
|
tirade
|
3,764
|
0.099
|
revenge
|
26,940
|
0.095
|
antipathy
|
3,006
|
0.089
|
hate
|
20,596
|
0.088
|
hurt
|
17,553
|
0.086
|
prejudice
|
30,075
|
0.086
|
antagonism
|
4,614
|
0.085
|
distrust
|
6,488
|
0.085
|
mistrust
|
4,044
|
0.082
|
envy
|
9,228
|
0.082
|
Figure 7. Thesaurus representation of grudge n.
54The and/or category shows more semantic cohesion than the others. All collexemes in this category relate to a form of negative emotion. Overall, there are much more nouns than verbs. The Word Sketch announces 4,702 nouns for 542 verbs, and a lot of the alleged verbs are nouns. The amount of collexemes that feature a high similarity score is rather low, especially compared to what was observed for groan and grunt. This means that grudge does not have very close semantic relatives, which is confirmed by the Thesaurus data. The Word Sketch gives a more qualitative picture and shows that only a restricted meaning of grudge is observable in the OEC.
Table 16. Collexemes for grudge v. in the OEC (Word Sketch)
N subj V*
|
Freq.
|
Score
|
X* in N
|
Freq.
|
Score
|
V* obj N
|
Freq.
|
Score
|
respect
|
3
|
6.5
|
praise
|
5
|
5.9
|
expense
|
2
|
8.4
|
parliament
|
2
|
3.3
|
quarter
|
2
|
2.3
|
prepared-ness
|
2
|
6.6
|
sound
|
2
|
2.8
|
support
|
3
|
1.6
|
admiration
|
4
|
4.8
|
Moore
|
2
|
2.6
|
|
|
|
acknowledg-ment
|
2
|
4.7
|
part
|
2
|
2.6
|
|
|
|
respect
|
12
|
3.6
|
|
|
|
|
|
|
acceptance
|
3
|
3.5
|
|
|
|
|
|
|
resignation
|
2
|
2.9
|
|
|
|
|
|
|
rival
|
2
|
2.7
|
|
|
|
|
|
|
sacrifice
|
2
|
2.5
|
|
|
|
|
|
|
recognition
|
3
|
2.3
|
|
|
|
|
|
|
match
|
6
|
2.2
|
|
|
|
|
|
|
tribute
|
2
|
1.6
|
Table 17. Collexemes for grudge n. in the OEC (Word Sketch)
And/or
|
Freq.
|
Score
|
X mod N*
|
Freq.
|
Score
|
V obj N*
|
Freq.
|
Score
|
vendetta
|
6
|
8.0
|
long-held
|
10
|
6.6
|
harbor
|
134
|
8.3
|
grievance
|
16
|
7.9
|
deep-seated
|
5
|
5.5
|
nurse
|
101
|
8
|
animosity
|
9
|
7.7
|
long-standing
|
29
|
5.5
|
bear
|
445
|
7.5
|
feud
|
8
|
7.6
|
bear
|
12
|
5.2
|
hold
|
993
|
6.1
|
resentment
|
14
|
6.6
|
petty
|
14
|
5
|
settle
|
36
|
5.2
|
rivalry
|
6
|
6.0
|
long-running
|
6
|
4.5
|
nurture
|
10
|
4.8
|
hurt
|
6
|
5.9
|
personal
|
170
|
4
|
carry
|
90
|
3.7
|
alliance
|
8
|
5.5
|
[...]
|
|
|
forget
|
11
|
2.7
|
jealousy
|
5
|
5.5
|
last
|
8
|
3.7
|
overcome
|
5
|
1.8
|
hate
|
5
|
5.1
|
bitter
|
9
|
3.2
|
have
|
489
|
1.5
|
hatred
|
8
|
4.8
|
lifelong
|
5
|
2.9
|
pursue
|
6
|
1.3
|
[...]
|
|
|
ancient
|
17
|
2.6
|
|
|
|
prejudice
|
5
|
4.3
|
hidden
|
5
|
2.4
|
|
|
|
[...]
|
|
|
old
|
112
|
2.3
|
|
|
|
|
|
|
deadly
|
5
|
2.1
|
|
|
|
|
|
|
deep
|
20
|
2
|
|
|
|
|
|
|
historical
|
13
|
1.8
|
|
|
|
|
|
|
tremendous
|
5
|
1.7
|
|
|
|
|
|
|
secret
|
7
|
1.6
|
|
|
|
|
|
|
serious
|
17
|
1.1
|
|
|
|
|
|
|
[...]
|
|
|
|
|
|
55Most uses of grudge in the OEC contexts correspond to the two senses defined in the OED: 1) “To envy (a person). Also intransitive. To be envious.” [sense: v. (3); date: 1577]; and 2) “Ill-will or resentment due to some special cause, as a personal injury, the superiority of an opponent or rival, or the like.” [sense: n. (3a); date: 1477]. But instances of the verb are very rare. The nominal use can be found in (6). The sense “To murmur; to utter complaints murmuringly; to grumble, complain; to be discontented or dissatisfied.” [v. (1a); 1440] can hardly be found across the corpus. It is found in (7), which evidently is a copy of a much older text, although both the web link and the original source are unspecified.
(6) The killers of Meredith Kercher had no obvious motive for murder and displayed no “animosity or grudge” towards the British student, an Italian court has concluded. Meredith Kercher’s murder was “without planning, without any animosity or grudge against the victim” ruled the judges. (source: https://www.telegraph.co.uk/news/worldnews/europe/italy/7369255/Meredith-Kerchers-killers-had-no-grudge-against-her-rule-judges.html)
(7) What he himself spake on his death-bed for example to other, I thought not less to pretermit. Who hearing that he should die, and that there was no remedy, murmured and grudged, wherefore he should die, having so much riches, saying, that if the whole realm would save his life, he was able either by policy to get it, or by riches to buy it; adding and saying, moreover, “Fie,” quoth he, “Will not death be hired? Will money do nothing? [...]” (source unknown)
56Overall, the contexts in which grudge appears are less expressive than for groan and grunt. The nature of the contexts is different, as there are fewer dialogues, and more descriptive and political contexts, as in (8), historical content as in (9), or sport commentating as in (10).
(8) Yorkshire in fine old style. But hacks imprisoned on the Tory battlebus for the last three weeks or so were grudging in their praise. Britain’s first General Election of the new millennium pressed kitchens, pubs and even a fish and chip shop into service as polling stations. (source: https://www.yorkshirepost.co.uk/viewarticle2.aspx?sectionid=55&articleid=192022)
(9) In the end Melville won but the monument was not erected on the first site proposed, Calton Hill, but in its present position in the square, 16 years after work began. If that sounds grudging, the Americans took 100 years to complete the national monument to General Washington. (source: https://www.findarticles.com/p/articles/mi_qn4156/is_20001029/ai_n13955308)
(10) Moving to Jaguar last year, he accounted for 17 of the 18 points carded by the Big Cat. Impressive in a sport where everything is relative and respect can be grudging. After early-season struggles, the French GP last weekend was a better race for Jaguar. (source: https://www.scotsman.com/sport)
57Grip is the first of the selected sample words to refer to the hold/clutch/touch category. Some of its collexemes relate to diseases or health issues (plague, paralyse, cripple), and more generally, to the semantic field of affliction (trouble, afflict, ravage, hurt). Surprisingly, the intuitive “prehension” meaning that was expected does not prevail in the verb list (only clutch, touch and seize, displaying a rather low score). The noun list, however, shows a number of words relating to prehension (hold, grasp, handle, touch), which strongly resonate with the key words collected from the OED. Once again, a substantial amount of these are verbs misinterpreted as nouns, but the semantic cohesion of the lexicography-based features is partly confirmed by these corpus data.
58Once again, a few phonesthemic collexemes are in the semantic network (grab, cripple, wrack, crush), among which some are onomatopoeic (wrack, crush). The only close collexeme of grip is hold, which happens to be the most frequent key word of the conceptual category hold/clutch/touch in the OED. The word grip must therefore occupy a key place in the mental lexicon, as groan and grunt do, although its overall collocation behaviour is very different.
59However, a specific submeaning of grip, associated with the lexical field of disease, affliction and infirmity seems to have taken over in usage. It has stepped over most of the other submeanings of grip, namely that of prehension. The prehension meaning is defined as following in the OED: “To grasp or seize firmly or tightly with the hand; to seize with the mouth, claw, beak or other prehensile organ” [sense: verb 1 (1a); date: 950]. Another sense that originated from this prehension meaning developed from semantic shift: “Transferred. Said of a disease” [sense: v.1 (1c); date: 1818].
60Following Flaksman’s model of the iconic treadmill [2017: 18], which states that iconic words gradually become less expressive, it is possible that an originally iconic meaning of gr- associated with prehension gradually lost its motivated sense overtime, to the benefit of other submeanings. This model postulates that motivated forms usually acquire new collocation abilities, which leads to the development of polysemy (Flaksman [2017: 28]), but it has rarely been acknowledged in corpus studies. The collocational profile of grip is an example of a semantic restriction that led to the de-iconisation of gr-, and therefore constitutes a case of phonesthemic divergence.
Table 18. Frequency and similarity scores of the collexemes of grip v.
Collexeme
|
Frequency
|
Similarity
|
plague
|
16,773
|
0.246
|
haunt
|
19,534
|
0.236
|
overwhelm
|
25,012
|
0.227
|
rock
|
28,285
|
0.223
|
beset
|
5,172
|
0.222
|
paralyze
|
12,625
|
0.213
|
trouble
|
31,237
|
0.210
|
touch
|
110,652
|
0.206
|
engulf
|
9,055
|
0.206
|
grasp
|
25,194
|
0.202
|
afflict
|
9,643
|
0.201
|
clutch
|
11,605
|
0.201
|
devastate
|
42,598
|
0.198
|
ravage
|
6,595
|
0.197
|
grab
|
85,225
|
0.197
|
overtake
|
16,050
|
0.182
|
overshadow
|
11,131
|
0.182
|
disturb
|
26,461
|
0.179
|
ruin
|
35,618
|
0.179
|
hurt
|
107,724
|
0.178
|
cripple
|
10,770
|
0.175
|
wrack
|
2,264
|
0.168
|
crush
|
37,605
|
0.167
|
shatter
|
23,005
|
0.166
|
seize
|
60,278
|
0.162
|
Figure 8. Thesaurus representation of grip v.
Table 19. Frequency and similarity scores of the collexemes of grip n.
Collexeme
|
Frequency
|
Similarity
|
hold
|
49,439
|
0.223
|
grasp
|
14,023
|
0.181
|
handle
|
21,990
|
0.179
|
focus
|
149,508
|
0.160
|
touch
|
99,792
|
0.159
|
emphasis
|
73,828
|
0.158
|
impact
|
233,318
|
0.157
|
influence
|
169,428
|
0.152
|
burden
|
57,293
|
0.145
|
perspective
|
124,901
|
0.144
|
impression
|
77,293
|
0.141
|
lock
|
26,721
|
0.140
|
stance
|
40,474
|
0.138
|
reflection
|
51,546
|
0.138
|
restriction
|
69,899
|
0.137
|
weight
|
177,967
|
0.137
|
control
|
484,375
|
0.136
|
frame
|
88,376
|
0.136
|
combination
|
124,043
|
0.136
|
mark
|
104,165
|
0.135
|
Figure 9. Thesaurus representation of grip n.
61Many more Word Sketch categories were relevant this time. Not all the categories are presented because it would be too long, and only the most similar collexemes are presented. For grip v., the Word Sketch shows that the N subj V* category is the construction pattern where the metaphorical sense of affliction emerged. Most subjects of grip are nouns denoting negative emotions (fear, anxiety, terror, horror) or some form of physical or mental illness (crisis, paralysis, hysteria, paranoia). The and/or category contains verbs of motion and manual action, but contains very few collexemes overall. The V* obj N category, on the other hand, contains the object nouns that can be gripped. A lot of them are physical objects (wheel, handle, hilt) and body parts (wrist, arm, throat), but some also correspond to a metaphorical sense (nation, imagination, country). Overall, this shows that the prehension meaning of grip is not as productive as the metaphorical sense.
Table 20. Collexemes for grip v. in the OEC (Word Sketch)
And/or
|
Freq.
|
Score
|
V* obj N
|
Freq.
|
Score
|
N subj V*
|
Freq.
|
Score
|
grip
|
10
|
9.2
|
wheel
|
136
|
7.7
|
fear
|
237
|
9.2
|
squeeze
|
8
|
7.8
|
handle
|
89
|
7.4
|
panic
|
90
|
9.0
|
pull
|
8
|
6
|
shoulder
|
128
|
7.4
|
fever
|
83
|
9.0
|
move
|
20
|
5.8
|
nation
|
243
|
7.2
|
hand
|
318
|
8.4
|
hold
|
15
|
5.8
|
imagination
|
94
|
7.1
|
tension
|
54
|
7.5
|
|
|
|
railing
|
40
|
7.1
|
anxiety
|
30
|
7.4
|
|
|
|
wrist
|
43
|
6.6
|
crisis
|
74
|
7.2
|
|
|
|
armrest
|
26
|
6.6
|
finger
|
44
|
7
|
|
|
|
hilt
|
26
|
6.6
|
paralysis
|
14
|
6.9
|
|
|
|
handlebar
|
26
|
6.6
|
terror
|
22
|
6.9
|
|
|
|
edge
|
90
|
6.5
|
panic
|
11
|
6.6
|
|
|
|
arm
|
215
|
6.5
|
urge
|
12
|
6.5
|
|
|
|
cock
|
31
|
6.4
|
hysteria
|
11
|
6.5
|
|
|
|
sword
|
48
|
6.3
|
chaos
|
14
|
6.4
|
|
|
|
rail
|
30
|
6.3
|
drought
|
14
|
6.4
|
|
|
|
country
|
366
|
6.1
|
tension
|
9
|
6.3
|
|
|
|
[...]
|
|
|
unrest
|
11
|
6.3
|
|
|
|
forearm
|
19
|
5.9
|
paranoia
|
10
|
6.3
|
|
|
|
chest
|
27
|
5.9
|
madness
|
10
|
6.2
|
|
|
|
throat
|
32
|
5.9
|
insurgency
|
11
|
6.2
|
|
|
|
[...]
|
|
|
delusion
|
9
|
6.2
|
|
|
|
hand
|
250
|
5.4
|
recession
|
16
|
6.2
|
|
|
|
heart
|
83
|
5.4
|
excitement
|
11
|
6.1
|
|
|
|
[...]
|
|
|
famine
|
9
|
6.1
|
|
|
|
chin
|
15
|
5.3
|
horror
|
10
|
6
|
|
|
|
hip
|
15
|
5.2
|
excitement
|
7
|
6
|
|
|
|
doorknob
|
10
|
5.2
|
uncertainty
|
12
|
5.9
|
|
|
|
stomach
|
17
|
5.1
|
[...]
|
|
|
|
|
|
[...]
|
|
|
grief
|
8
|
5.8
|
|
|
|
elbow
|
14
|
5.1
|
[...]
|
|
|
|
|
|
[...]
|
|
|
|
|
|
|
|
|
thigh
|
10
|
5
|
|
|
|
|
|
|
[...]
|
|
|
|
|
|
|
|
|
flesh
|
10
|
4.5
|
|
|
|
|
|
|
[...]
|
|
|
|
|
|
|
|
|
ankle
|
9
|
4.4
|
|
|
|
|
|
|
[...]
|
|
|
|
|
|
|
|
|
knee
|
12
|
4.4
|
|
|
|
|
|
|
[...]
|
|
|
|
|
|
62For grip n., the and/or category essentially contains technical terms and industry-related objects (traction, downforce, frame, polymer, shaft). The V obj N* and N* subj V categories mainly contain verbs relating to manual work, although some metaphorical senses can also be observed.
Table 21. Collexemes for grip n. in the OEC (Word Sketch)
And/or
|
Freq.
|
Score
|
V obj N*
|
Freq.
|
Score
|
N* subj V
|
Freq.
|
Score
|
beavertail
|
35
|
8.9
|
tighten
|
1,157
|
10.6
|
loosen
|
46
|
9.6
|
traction
|
30
|
8.2
|
loosen
|
594
|
10.0
|
tighten
|
101
|
9.5
|
pistol
|
49
|
8
|
relax
|
174
|
8.0
|
lessen
|
7
|
6.8
|
stance
|
39
|
8
|
consolidate
|
115
|
7.2
|
weaken
|
10
|
5.7
|
grip
|
36
|
8.0
|
strengthen
|
227
|
7
|
slip
|
17
|
5.2
|
downforce
|
18
|
7.9
|
lose
|
1,452
|
6.8
|
fit
|
11
|
4.3
|
forearm
|
23
|
7.6
|
loose
|
57
|
6.7
|
allow
|
12
|
2.2
|
grin
|
16
|
7.0
|
relinquish
|
58
|
6.6
|
feel
|
17
|
2.1
|
frame
|
42
|
6.7
|
maintain
|
499
|
6.3
|
hold
|
16
|
1.9
|
strap
|
13
|
6.6
|
weaken
|
70
|
6.3
|
fail
|
8
|
1.3
|
barrel
|
17
|
6.4
|
release
|
364
|
6.2
|
remain
|
13
|
1.2
|
trigger
|
11
|
6.4
|
get
|
2,705
|
5.9
|
keep
|
7
|
1.2
|
polymer
|
9
|
6.2
|
exert
|
47
|
5.7
|
cause
|
8
|
1
|
posture
|
12
|
6.2
|
retain
|
134
|
5.6
|
|
|
|
shaft
|
9
|
6.1
|
break
|
218
|
5.5
|
|
|
|
sight
|
20
|
6
|
keep
|
620
|
5.4
|
|
|
|
handle
|
10
|
6
|
cement
|
22
|
5.2
|
|
|
|
finish
|
15
|
5.9
|
adjust
|
43
|
5.2
|
|
|
|
blade
|
12
|
5.9
|
slacken
|
15
|
5.1
|
|
|
|
lever
|
7
|
5.9
|
[...]
|
|
|
|
|
|
hammer
|
10
|
5.8
|
feel
|
89
|
3.9
|
|
|
|
screw
|
8
|
5.8
|
[...]
|
|
|
|
|
|
handling
|
17
|
5.8
|
twist
|
9
|
3.8
|
|
|
|
steel
|
22
|
5.4
|
[...]
|
|
|
|
|
|
feel
|
9
|
5.3
|
slip
|
9
|
3.7
|
|
|
|
swing
|
8
|
5.3
|
[...]
|
|
|
|
|
|
hand
|
47
|
5
|
grasp
|
9
|
3.4
|
|
|
|
peak
|
11
|
4.9
|
[...]
|
|
|
|
|
|
balance
|
18
|
4.9
|
threaten
|
18
|
3.2
|
|
|
|
leather
|
7
|
4.9
|
[...]
|
|
|
|
|
|
tape
|
10
|
4.5
|
|
|
|
|
|
|
[...]
|
|
|
|
|
|
|
|
|
arm
|
12
|
3.5
|
|
|
|
|
|
|
[...]
|
|
|
|
|
|
|
|
|
body
|
9
|
2.1
|
|
|
|
|
|
|
[...]
|
|
|
|
|
|
|
|
|
63Uses of grip in context can be literal, as in (11) and (12), or metaphorical, as in (13) and (14). Both subjects of grip v. in (13) and (14) are negative emotions: grief and fear. Grip does not appear in conversational contexts as groan and grunt tend to do, but rather serves to describe movements (11 and 12) or emotions (13 and 14).
(11) It looked like it had once been bright and cheerful, but now darkness - evil - had taken over. The shutters on the window - like those on old Victorian houses, hung in awkward positions. The windows were cracked and some were even shattered, as if some neighborhood pranksters had thrown rocks at the window. “Well it’s almost time for my break,” the taxi driver told them, his knuckles white and his hands gripping the steering wheel as if it was going to run away. “Queer fellow,” Cougre said grinning. (source: http://www.fictionpress.com/read.php?storyid=1453281&chapter=3)
(12) “Your life is going to depend on these training sessions, so I suggest you cut the sarcastic teenage wise cracks and pay attention. I’m dead serious, girly. These vampires don’t fool around, and if you don’t know how to defend your self you can just kiss your sorry ass goodbye.” Inside Azerith’s house, Rika was slowly being set down on the couch while tightly gripping her shoulder and screaming as loudly as she possibly could every few seconds. (source: http://www.fictionpress.com/read.php?storyid=521768&chapter=9)
(13) On Tuesday morning, a 35-year-old air traffic controller told Congress what it was like to be on the other end of the pilot’s calm radio exchanges. The controller, Patrick Harten, 35, of Long Beach, N.Y., told a Congressional panel in Washington that raw moments of shock and grief have gripped him since the controlled crash of US Airways Flight 1549 on Jan. 15, which was caused by a double-engine failure following a bird strike. (source: http://www.nytimes.com/2009/02/25/nyregion/25crash.html?scp=1212&sq=&st=nyt)
(14) Leaving what? A stump? And what did one cut through? Bone, tendon? Or did one find the joint as in a chicken? I would sooner be dead, thought Alison. But I know why. It is because I am a vain, a wicked woman, who thinks too much of this world, and of her own body. I am not humble, I cannot face old age, I cannot face ugliness and decay. She shivered. She was afraid. A large, a terrible fear gripped her. Of death? No, not of death. To die in one’s sleep, to fade away, seemed easy enough, as a prospect. Of mutilation? Why, of mutilation? Nobody had ever threatened her with it, so why, so unnaturally, so wastefully, at least so prematurely, face it, fear it? (source: http://awopbopaloobop.blogspot.com/2008_10_01_archive.html)
64Grasp is characterised by only two gr- collexemes (grab v. and grip n.) and even fewer phonesthemic collexemes than grip, and no onomatopoeic or mimetic collexemes at all. Grasp and grip do not share the same close semantic relationship as groan and grunt do, contrary to my expectations.
- 4 To what extent the word grasp might have known a similar fate as the Latin com-prehend, which origi (...)
65The word grasp features a specific collexeme network, which is oriented towards the semantic field of comprehension (understand, grab, reach), mental ability (master, imagine, reflection) and intellectual perception (articulate, discern, sense) as well as interpersonal communication (appreciate, embrace, manipulate, ignore). These words are instances of a submeaning of grasp that can be summed up by the following OED definition: “To lay hold of with the mind; to become completely cognizant of or acquainted with; to comprehend” [sense: verb (6); date: 1680]. It is possible that this specific sense of grasp, as was the specific sense of grip relating to illness, may have become more prevalent at some point in time and overcome the other senses. An in-depth analysis across a diachronic corpus would explain how and when this sense became dominant, and to what extent it is. The fact that only the word comprehend is in the closer zone of the mapping representation reveals the overrepresentation of this submeaning4.
Table 22. Frequency and similarity scores of the collexemes of grasp v.
Collexeme
|
Frequency
|
Similarity
|
comprehend
|
12,966
|
0.366
|
appreciate
|
115,264
|
0.311
|
embrace
|
65,931
|
0.292
|
realize
|
317,138
|
0.288
|
master
|
17,982
|
0.282
|
articulate
|
25,011
|
0.269
|
understand
|
526,263
|
0.267
|
grab
|
85,225
|
0.263
|
discern
|
10,797
|
0.261
|
sense
|
27,887
|
0.253
|
clarify
|
33,187
|
0.246
|
shape
|
47,041
|
0.243
|
possess
|
68,950
|
0.241
|
ignore
|
152,260
|
0.240
|
manipulate
|
31,233
|
0.237
|
fulfill
|
59,444
|
0.237
|
explore
|
130,469
|
0.237
|
imagine
|
186,852
|
0.236
|
notice
|
173,656
|
0.235
|
transcend
|
14,184
|
0.232
|
Figure 10. Thesaurus representation of grasp v.
Table 23. Frequency and similarity scores of the collexemes of grasp n.
Collexeme
|
Frequency
|
Similarity
|
mastery
|
8,356
|
0.236
|
overview
|
18,566
|
0.212
|
appreciation
|
32,304
|
0.191
|
comprehension
|
7,465
|
0.188
|
grip
|
37,709
|
0.181
|
understanding
|
163,835
|
0.176
|
breadth
|
12,053
|
0.167
|
conception
|
29,221
|
0.167
|
reach
|
33,601
|
0.164
|
embrace
|
10,072
|
0.163
|
awareness
|
74,545
|
0.161
|
critique
|
26,623
|
0.157
|
impression
|
77,293
|
0.155
|
ignorance
|
26,632
|
0.152
|
essence
|
41,377
|
0.150
|
perception
|
69,215
|
0.150
|
realization
|
24,225
|
0.149
|
characterization
|
15,908
|
0.147
|
reflection
|
51,546
|
0.147
|
complexity
|
44,871
|
0.144
|
Figure 11. Thesaurus representation of grasp n.
66The Word Sketch function provides evidence of specific patterns of use of the word grasp in context, both for verbal and nominal uses. The and/or function confirms the Thesaurus data and shows a dominant use within the semantic field of comprehension and intellectual interactions (comprehend, understand, explain). The same goes for the other functions, although the semantic field of prehension is more present, especially with a few prehensive or prehensible body parts (hand, finger, claw, arm, wrist).
Table 24. Collexemes for grasp v. in the OEC (Word Sketch)
And/or
|
Freq.
|
Score
|
N subj V*
|
Freq.
|
Score
|
V* obj N
|
Freq.
|
Score
|
comprehend
|
12
|
8.0
|
hand
|
166
|
7.5
|
nettle
|
418
|
9.9
|
reach
|
20
|
7.7
|
mind
|
84
|
7
|
concept
|
533
|
8.2
|
master
|
6
|
7.1
|
intellect
|
11
|
6.7
|
handle
|
161
|
7.9
|
hold
|
35
|
7.0
|
finger
|
22
|
6.1
|
significance
|
242
|
7.8
|
articulate
|
8
|
6.9
|
claw
|
7
|
6.1
|
meaning
|
300
|
7.4
|
appreciate
|
19
|
6.7
|
understanding
|
7
|
5.5
|
essence
|
95
|
7.2
|
understand
|
54
|
6.4
|
philosopher
|
6
|
5
|
enormity
|
61
|
7.1
|
pull
|
9
|
6.1
|
reason
|
9
|
5
|
wrist
|
73
|
6.9
|
express
|
8
|
6.1
|
arm
|
11
|
4.9
|
reality
|
179
|
6.9
|
manipulate
|
9
|
6.0
|
viewer
|
13
|
4.8
|
implication
|
133
|
6.8
|
embrace
|
6
|
5.9
|
listener
|
6
|
4.8
|
nature
|
179
|
6.7
|
reject
|
8
|
5.7
|
public
|
25
|
4.8
|
magnitude
|
55
|
6.7
|
explain
|
10
|
5.6
|
sense
|
7
|
4.6
|
opportunity
|
528
|
6.6
|
experience
|
6
|
5.5
|
reader
|
25
|
4.5
|
nuance
|
47
|
6.5
|
accept
|
9
|
4.8
|
student
|
65
|
4.3
|
complexity
|
68
|
6.5
|
find
|
8
|
3.9
|
audience
|
13
|
4.2
|
dumbbell
|
39
|
6.5
|
see
|
14
|
3.9
|
being
|
7
|
4.1
|
shoulder
|
78
|
6.4
|
control
|
6
|
3.6
|
brain
|
6
|
4
|
basic
|
48
|
6.4
|
use
|
8
|
3.5
|
opponent
|
8
|
4
|
importance
|
162
|
6.4
|
try
|
17
|
3.1
|
male
|
7
|
3.9
|
extent
|
76
|
6.3
|
|
|
|
thought
|
6
|
3.7
|
fact
|
273
|
6.2
|
|
|
|
class
|
7
|
3.4
|
hand
|
442
|
6.2
|
|
|
|
[…]
|
|
|
truth
|
150
|
6.1
|
|
|
|
|
|
|
gravity
|
35
|
6.0
|
|
|
|
|
|
|
subtlety
|
31
|
6.0
|
|
|
|
|
|
|
bar
|
79
|
6.0
|
|
|
|
|
|
|
seriousness
|
33
|
6.0
|
|
|
|
|
|
|
hold
|
77
|
5.9
|
|
|
|
|
|
|
[…]
|
34
|
5.9
|
Table 25. Collexemes for grasp n. in the OEC (Word Sketch)
And/or
|
Freq.
|
Score
|
X mod N*
|
Freq.
|
Score
|
V obj N*
|
Freq.
|
Score
|
reach
|
9
|
6.3
|
firm
|
619
|
8.8
|
elude
|
79
|
8.3
|
understanding
|
18
|
4.5
|
intuitive
|
100
|
8.1
|
exceed
|
138
|
6.6
|
insight
|
6
|
4.1
|
tenuous
|
66
|
7.9
|
loosen
|
23
|
6.3
|
knowledge
|
17
|
3.5
|
shaky
|
44
|
6.9
|
escape
|
106
|
6.3
|
mind
|
10
|
3.3
|
instinctive
|
32
|
6.8
|
evade
|
24
|
5.9
|
sense
|
7
|
3.1
|
rudimentary
|
31
|
6.6
|
demonstrate
|
129
|
5.6
|
view
|
6
|
3
|
feeble
|
20
|
6
|
tighten
|
16
|
5
|
intelligence
|
6
|
2.8
|
thorough
|
48
|
6
|
display
|
67
|
4.8
|
memory
|
6
|
2.8
|
encyclopedic
|
14
|
5.9
|
slip
|
11
|
4.8
|
skill
|
8
|
2.1
|
sure
|
41
|
5.7
|
possess
|
26
|
4.2
|
experience
|
7
|
1.8
|
masterful
|
15
|
5.7
|
relax
|
6
|
3.9
|
|
|
|
keen
|
36
|
5.5
|
lack
|
33
|
3.7
|
|
|
|
well
|
255
|
5.4
|
have
|
2,209
|
3.6
|
|
|
|
solid
|
88
|
5
|
retain
|
26
|
3.4
|
|
|
|
masterly
|
7
|
5
|
exhibit
|
13
|
3.2
|
|
|
|
uncertain
|
17
|
4.9
|
get
|
413
|
3.2
|
|
|
|
tenacious
|
7
|
4.9
|
acquire
|
19
|
3.1
|
|
|
|
vague
|
17
|
4.8
|
show
|
139
|
3
|
|
|
|
innate
|
11
|
4.8
|
gain
|
37
|
3
|
|
|
|
sophisticated
|
36
|
4.7
|
imply
|
6
|
2.9
|
|
|
|
astute
|
9
|
4.7
|
extend
|
22
|
2.8
|
|
|
|
imaginative
|
13
|
4.7
|
maintain
|
35
|
2.6
|
|
|
|
frail
|
7
|
4.7
|
release
|
27
|
2.5
|
|
|
|
articulate
|
7
|
4.7
|
strengthen
|
8
|
2.4
|
|
|
|
uncanny
|
8
|
4.6
|
lose
|
60
|
2.3
|
|
|
|
conceptual
|
16
|
4.6
|
reveal
|
17
|
2.1
|
|
|
|
limp
|
6
|
4.6
|
question
|
7
|
2
|
|
|
|
flimsy
|
7
|
4.6
|
indicate
|
11
|
1.9
|
|
|
|
ready
|
11
|
4.6
|
obtain
|
16
|
1.8
|
|
|
|
icy
|
10
|
4.5
|
require
|
28
|
1.3
|
|
|
|
inadequate
|
17
|
4.5
|
combine
|
6
|
1.3
|
|
|
|
speculative
|
11
|
4.5
|
need
|
32
|
1.2
|
|
|
|
weak
|
43
|
4.5
|
avoid
|
11
|
1.2
|
|
|
|
confident
|
9
|
4.3
|
reflect
|
9
|
1.2
|
|
|
|
fragile
|
12
|
4.3
|
keep
|
33
|
1.2
|
|
|
|
intellectual
|
46
|
4.3
|
achieve
|
11
|
1.1
|
|
|
|
slender
|
7
|
4.3
|
develop
|
22
|
1.1
|
|
|
|
imperfect
|
6
|
4.3
|
suggest
|
8
|
1
|
|
|
|
[…]
|
|
|
|
|
|
|
|
|
greedy
|
6
|
4.1
|
|
|
|
|
|
|
[…]
|
|
|
|
|
|
67These data show lexical subpatterns observable in discourse that all trigger the same conceptual feature of grasp, either used in its primary physical sense or in its figurative sense. A closer analysis of the specific contexts shows that the literal sense tends to trigger more phonesthemic clustering, in (15), than the figurative sense, as in (16). Again, this indicates that the prehension submeaning of gr- might have seen a de-iconisation process.
(15) The loud crash of shattering glass interrupted him, and a hand grasping a long dagger snaked into the backseat of the car. Alex and Alastair screamed in one voice and both jumped to the opposite side of the car. The hand blindly threw the dagger and managed to slice the boy’s upper left arm. Alex, who was cowering away from the broken glass, heard him scream in pain and agony as the shadow crept off into the night. (source: http://www.fictionpress.com/read.php?storyid=1157429&chapter=4)
(16) It is not possible for ordinary aspirant to comprehend and grasp the essence of scriptures or religious philosophy (e.g. Vedanta, Yoga, The Gita) by merely reading the books or listening to talks. (source: http://www.boloji.com/hinduism/046.htm]
68The aim of this paper was to tackle the semantic characteristics and collocation behaviour of gr- words in usage, by combining two methods: 1) a lexicographic analysis based on data drawn from the OED and 2) a corpus-based analysis in the OEC, via the distributional tool Sketch Engine. The article set to answer two main questions: 1) how do the different subsemantic fields conveyed by gr- affect language use? and 2) how do meaning-carrying phonological elements behave in context?
69The lexicographic approach showed that the phonestheme gr- is associated with several submeanings and not with a single easily identifiable meaning. Four main form-meaning associations are induced by gr-: 1) vocal activity, both human and animal (groan, grunt); 2) vocally expressed negative emotions and bad humour (grudge, grieve, grump); 3) unpleasant sensations (grate, grind); and 4) prehension activity, with both a physical, manual meaning, and an abstract metaphorical meaning (grasp, grab, grip). The first three traits show a high level of semantic change, feature combination and phonesthemic convergence. On the other hand, the last trait was much more semantically inert.
70The corpus study showed that these semantic categories differ in usage. The set of case study words pertaining to the SVU and WCA categories tend to show comparable collexeme networks and share some collocational properties. In particular, the corpus analysis showed that groan and grunt shared a semantic space. The fact that these two words are so close semantically and distributionally raises the question of their relevance in the mental lexicon as near-synonyms. They also appear in contexts of a similar nature, mainly dialogues, and they combine with other phonosemantic forms, as well as expressive markers. A lot of these contexts also convey negative emotions. On the contrary, the corpus analysis showed that my set of case study words pertaining to the HCT category (grip and grasp) may have undergone phonesthemic divergence. De-iconisation processes might be at stake.
71These different submeanings of the phonestheme gr- are thought to play a different role in different lexical domains. In fact, the analysis in the OEC showed that gr- words relating to the HCT semantic category have much more specific uses than the other gr- words. These uses are mainly metaphorical and restricted to specific semantic fields (the field of comprehension for grasp and the field of illness for grip). They also tend to be unique in their synonym network, but because they are more distributionally isolated that groan and grunt, they probably hold a special place in their specific domain.
72Groan and grunt show much more phonesthemic clustering than grudge, grip and grasp. Comparing the first three case words, it is possible that groan acts as an in-between category, along the continuum that goes from the raw onomatopoeic grunt to the more conventionalised grudge. From this perspective, grudge is to be viewed as a partially de-iconised word that might have been onomatopoeic in origin but that, overtime, specialised in restricted senses due to semantic change. It is not entirely clear yet how this could play out from a diachronic perspective. Flaksman [forthcoming] suggests that some iconic and imitative words may lose their “original (sound- or articulatory-gesture related) meaning” overtime, although the precise conditions of this potential semantic loss are still unclear.
73This paper showed that the semantic behaviour of phonesthemic words depends on their surrounding collocates, and the nature of the context they appear in. Some semantic aspects of gr- words are triggered by certain lexical fields or multimodal cues. I observed a clustering of phonesthemic words and expressive markers in context. The different modes of expression that a text features play a key role in triggering a more or less iconic or phonesthemic tone.
74Some questions remain unanswered:
-
How semantically diverse is gr- compared to other phonesthemes? Only a broad-scale comparative study will allow to answer this question, but Smith & Farina [forthcoming] has already shown that fl-, sw- and gr- behave differently from one another.
-
The contexts of usage of grudge, grip and grasp are very specific. This raises the question of the typicality of the contexts. Can the results drawn from such specific uses be generalised to the whole lexicon? Are the results of this study only imputable to the behaviour of gr- words, or do they primarily depend on the nature and composition of the corpus?
-
The exact processes at stake behind the transmodality of gr- are still largely unknown. In particular, the relation between submeanings of gr- pertaining to sound imitation (see groan and grunt) and submeanings relating to visual representation (see grip and grasp) is not fully understood yet (see Nobile [2020] for an insight on transmodality within the scope of language motivation and iconicity).
-
To what extent can my statistical data be conclusive? The corpus study dataset contained only five words, which calls for a more broad-scale study. My aim is to pursue this study with an in-depth and fine-grained context analysis, which is much more time-consuming. Analysing the pragmatic aspects on a broader scale will give a more detailed picture. Sketch Engine is a powerful tool, but it also shows some limitations. The same is true for the OEC.
75My conclusions are to be understood in the scope of my dataset, and I still need to apply my method to other corpora. The EHBC, searchable via Sketch Engine, as well as the COHA, are good candidates for a diachronic study, which is the next step towards understanding the emergence and evolution of phonesthemes. The method exposed in this paper can be replicated to other phonesthemes (see Smith [2016; 2022a]), as well as more conventional lexical units (see Smith [2020a] on -some and -able derivatives). I aim to pursue these investigations by analysing other phonesthemic and phonosemantic formations in a usage-based approach.