Menu
Home
Log in / Register
 
Home arrow Language & Literature arrow Extending the Scope of Construction Grammar
Source

Conventionalization in the BPOC: evidence from the Corpus of Contemporary American English

Commenting on the BPOC, as well as on Vup a storm (e.g. {cook / curse / dance / party / sing / sweat / talk / tweet /...} up a storm), Jackendoff (2002a: 86) writes that the “the choice of verb seems totally open, whereas by contrast, the choice of NP is totally fixed”. Elsewhere I have shown that in actual language usage, there is quite a bit of variation in the choice of NP: every possible alternative word for ass and butt can make its appearance in the pattern, including behind, fanny, rear, tail, tush and any other term one could find in a good thesaurus. As for the verbs used in this pattern, Goldberg and Jackendoff (2004: 560) are more careful when they write that “the choice of verb is quite broad”. To examine how broad, I have conducted a corpus study, using the Corpus of Contemporary American English (COCA, Davies 2008-). Ten search strings were used, each time consisting of an open position (for a verb), followed by a possessive pronoun, a body part noun known to occur in the pattern (including a couple of the less frequently used alternatives for ass and butt) and the particle off or out. Table 1 shows the twenty most frequently used verbs in this pattern, with token frequencies for each postverbal sequence.

Table 1: Token frequencies for the twenty most frequently used verbs and ten common postverbal sequences appearing in the BPOC in the Corpus of Contemporary American English.

one’s oss off

one’s heart out

one’s head off

one’s eyes out

one's butt off

one’s tail off

one’s guts out

one’s lungs our

one’s balls off

one’s tush off

work

1

69

43

1

199

cry

4

80

2

1

104

scream

2

2

44

3

13

64

laugh

25

30

1

56

bawl

1

2

4

39

1

47

sing

1

43

1

1

46

freeze

24

8

1

8

1

41

play

2

16

5

3

1

27

puke

19

19

dance

8

8

1

1

18

run

4

5

3

12

bark

1

10

11

sob

*

2

11

fight

5

2

2

9

talk

8

1

9

sweat

a

1

1

4

8

cough

1

6

7

tie

3

4

7

yell

5

2

7

cheer

1

1

3

5

hapaxes

22

M

в

11

4

9

5

2

1

87

total tokens

162

145

136

129

100

47

43

31

14

1

807

hapaxes per tokens

14%

10%

10%

5%

11%

9%

21%

16%

14%

100%

11%

Grey-shade legend

tokens

1

2

<3

г 5

к 8

к 11

к 17 к 26 к 38

к 57

hapaxes/

tokens

(in%)

0

1-3

4-6

7-9

10-12

13-15

16-18

19-21 22-24 25-27

к 28

Note: the bottom row gives the hapax/token ratio for each post-verbal sequence, thus providing an indication of its productivity. Darker shades are used for easy identification of high token frequencies or high number of hapaxes as well as high hapax/token ratios (cf. legend).

Table 1 contains detailed information about the distribution of individual verbs and post-verbal sequences in the BPOC. It allows us to see (i) for each frequently used verb which post-verbal intensifier it takes and (ii) for each post-verbal sequence which high-frequency verbs may precede it and with how many verb tokens it is used. Verbs and post-verbal sequences are ordered by frequency from top to bottom and from left to right, respectively. Thus, work is the most frequently used verb in the BPOC and it is followed mainly by one’s butt off, one’s ass off, one’s tail off and one’s heart out in the corpus. Conversely, the

“subpattern” of the BPOC with the post-verbal sequence one’s ass off is used most frequently: it has the highest number of verb tokens (165), for which work (66 tokens), freeze (27 tokens) and laugh (25 tokens) are mainly responsible.

For each subpattern, Table 1 also gives the number of hapax legomena, or hapaxes, verbs which combine with that postverbal sequence only once in the corpus. The more hapaxes appear in the open slot of a pattern, the more lexical items may in fact be expected to be used on a one-off basis in this pattern outside the corpus. That is why hapaxes have been used as a measure of productivity of a pattern. One such measure, proposed by Baayen (1989) for morphological processes but perfectly applicable for above-word-level patterns as well, is formulated as the proportion of hapaxes in a corpus to the total number of words in the corpus used in that pattern. Of all verb tokens used in the investigated subpatterns of the BPOC in COCA, it appears that 11% occur in a particular subpattern only once. So, on average, every tenth verb in a subpattern will be a hapax. This can be taken as evidence that the BPOC is indeed a “pocket of productivity” within the more general caused-motion construction or the yet more general resultative construction.[1] The subpattern [V one’s butt off] is a representative one within the BPOC: 11 of the 100 occurrences of this subpattern contain a verb which is used just once in it, for instance dance, sweat and (not shown in the table) bike, coach, compete and lip-sync. If we ignore [V one’s tush off], which is found only once in the corpus (with the verb freeze) and which accordingly has full productivity, the subpattern [V one’s guts out] has the highest degree of productivity, with one fifth of the verb tokens found in it being hapaxes.

At the same time, it can also be appreciated that there is a high degree of conventionality within this productive area of English grammar. For example, while [V one’s guts out] is indeed quite productive, there is one verb, puke, which accounts for almost half of its occurrences in the corpus; besides, no other intensifying post-verbal sequence appears to be used with that verb, which tells us that puke one’s guts out is a strong collocation. Note, furthermore, that only one in twenty of the verb tokens used in the subpattern [V one’s eyes out] are hapaxes. This subpattern is used practically exclusively with cry and bawl.

Of course, there is some motivation why puke combines with guts and cry and bawl with eyes, rather than vice versa. The motivation is, of course, that if one pukes, contents of one’s guts are ejected, and that if one cries, tears flow from one’s eyes. Similarly, [V one’s lungs out] naturally attracts verbs denoting an activity of forceful air expulsion, such as scream, cough, cheer and yell. In each of these cases, the metonymical construal (container for contained), may be supported by a conceptual metaphor according to which an intense activity can be represented as an excessive change of location, in casu the exhaustion or detachment of a body part (Mateu and Espinal 2007; Espinal and Mateu 2010). An utterance such as John cried his eyes out may therefore be paraphrased most accurately not just as ‘John cried intensely’ (cf. Jackendoff 1997a, b, 2002a, b) but as ‘John cried so much that his eyes almost came out’, as represented in Figure 3.

Pictorial representation of a hyperbolic reading of John cried his eyes out, from Mateu and Espinal (2007)

Figure 3: Pictorial representation of a hyperbolic reading of John cried his eyes out, from Mateu and Espinal (2007)

The existence of a hyperbolic reading, in which the literal interpretation of a BPOC utterance is conceived of as depicting the actual event in an exaggerated (and therefore realistically unfeasible way), is compatible with Kudo’s (2011) analysis mentioned in section 3.2 above. Crucial to that analysis is that BPOC utterances could in principle receive a literal interpretation, referring to an actual event in the real world. The encyclopaedic relatedness of the verb’s meaning and the denotation of the head noun helps explain why certain combinations have become strong collocations while other combinations (e.g. cough one’s butt off) have not. Most of these conventionalized sequences are therefore not “idioms of decoding” in Makkai’s (1972: 47) sense: their meaning could easily be guessed on the basis of what we know about the words used, the meaning of the resultative pattern, the function of a possessive pronoun etc. I would like to claim that they are still “idioms of encoding” (Makkai 1972: 57), however, since a speaker can’t be expected to realize that cry one’s eyes out is a conventional figure of speech in English without having learned that fact. In support of this, note again that in Dutch, whose grammar also provides its speakers the opportunity to make use of the resultative pattern, one doesn’t say *Jan huilde z’n ogen uit, the direct translation of John cried his eyes out.

There are some further surprising observations to be made on the basis of Table 1. First, note that while the subpattern [V one’s head off], with its hapax/ token proportion of 9%, is about as productive as the BPOC pattern in general, it occurs with the verb work only once in the corpus, although this verb accounts for almost one fourth of all verb tokens in the BPOC. This suggests that the use of work one’s head off may be blocked by the frequent occurrence of e.g. work one’s butt off and work one’s ass off. Second, while ass and butt are synonyms, there are 25 occurrences of laugh one’s ass off but no occurrences of laugh one’s butt off in the corpus.[2] (Remarkably, occurrences of laugh one’s ass off are not blocked by the common use of one’s head off as a post-verbal intensifier with this verb.)

In sum, what the corpus data reveal is (i) that subpatterns of the BPOC show varying degrees of productivity; (ii) that motivations based on the physical involvement of a certain body part and the nature of the activity expressed by the verb exist but that these motivations do nothing to diminish the conventionality of some of the high-frequency combinations; and (iii) that some verbs occur with a range of post-verbal intensifying sequences but may still exhibit unexpected “gaps” in this range, apparently due to blocking effects - though exactly when a blocking effect occurs is itself an unpredictable matter. In short, the usage patterns found in COCA suggest that native speakers of English make use of a great deal of knowledge about conventional combinations of individual verbs and particular post-verbal sequences. Native speakers apparently also know which combinations are to be avoided, even though they could have been perfectly possible given full productivity of the BPOC. Apparently, not anything goes.

  • [1] By comparison, the suffixes -able, -ful (‘measure’), -ful (‘property’), -ize, -ness and -wise, allof them said to be productive, have lower hapax/token ratios (Plag et al. 1999). While this may be an indirect indication of how productive the BPOC is, I am for now veryhesitant to compare morphological and phrasal constructions with respect to this measure,especially across different corpora.
  • [2] Near-rhyme (laugh - ass) could be a factor in the conventionalization of this combination.Reporting findings based on dictionaries of idioms, Boers and Stengers (2008) note that almostone fifth of the English idiom repertoire contains some sort of “catchy” sound pattern, involvingalliteration and/or assonance.
 
Source
Found a mistake? Please highlight the word and press Shift + Enter  
< Prev   CONTENTS   Next >
 
Subjects
Accounting
Business & Finance
Communication
Computer Science
Economics
Education
Engineering
Environment
Geography
Health
History
Language & Literature
Law
Management
Marketing
Mathematics
Political science
Philosophy
Psychology
Religion
Sociology
Travel