The English Verb Phrase: Findings

Select findings for ‘The changing verb phrase in present-day British English’.

1. Modal Verbs: MUST, HAVE TO and HAVE GOT TO

Research questions

  • Is there evidence that the core modals in Present-Day English are decreasing?
  • If MUST is found to be decreasing can this be related to an increase in “rival” semi-modal forms?
  • Is there support for the idea that core modals are becoming monosemous?

Summary

There is a statistically significant decline in the frequency of MUST and a statistically significant rise in the frequency of HAVE TO.

HAVE GOT TO has decreased over the thirty year period (this is not statistically significant).

Table 1. Frequencies of MUST, HAVE GOT TO and HAVE TO in DCPSE (’s’ = significant at p<0.01).
(Semi-) modalLLC frequencyICE-GB frequencyChange in frequency
 rawper 100,000 wordsrawper 100,000 words%chi-square score
MUST42792.0117240.82-55.6436.29 s
HAVE GOT TO18740.3015637.02-8.143.10
HAVE TO18840.5122553.40+31.8231.94 s
TOTAL802172.82553131.24-24.0671.32 s

When viewed as proportions of total MUST, root uses remain constant (around 39%) and epistemic uses show a slight increase; there is no evidence that MUST is becoming monosemous.

Table 2. Distribution of semantic types of MUST in DCPSE (N=frequency per 100,000 words).
Source corpusEpistemicRootPerformativeAmbiguousTotal
 N%N%N%N%N
LLC43.9647.7836.6339.819.4810.301.942.1192.01
ICE-GB21.1251.7416.1439.532.375.811.192.9140.82
TOTAL65.0848.9952.7739.7311.858.923.132.35132.83

Root

In root contexts, HAVE TO is more frequent than MUST even in the 1960s corpus (LLC). By the 1990s, HAVE TO is more than twice as frequent as MUST, suggesting that HAVE TO is taking over some of the uses of root MUST.

Table 3. Frequencies of root MUST, HAVE GOT TO and HAVE TO in DCPSE (’s’ = significant at p<0.01).
(Semi-) modalLLC frequencyICE-GB frequencyChange in frequency
 rawper 100,000 wordsrawper 100,000 words%chi-square score
MUST17437.497016.61-55.6923.61 s
HAVE GOT TO18439.6515135.84-9.610.12
HAVE TO18539.8620849.36+23.8312.32 s
TOTAL543117429101.81-12.9836.05 s

Epistemic

Epistemic uses of HAVE TO and HAVE GOT TO are rare throughout the thirty year period.

Table 4. Frequencies of epistemic MUST, HAVE GOT TO and HAVE TO in DCPSE.
(Semi-) modalLLC frequencyICE-GB frequencyChange in frequency
 rawper 100,000 wordsrawper 100,000 words%chi-square score
MUST20644.399121.60-51.340.19
HAVE GOT TO20.4340.95+120.933.36
HAVE TO20.4330.71+65.121.83
TOTAL21045.259823.26-48.605.03 

These results were presented at ICEHL 15, Munich, and ISLE 1, Freiburg, in 2008. 

2. The Progressive

Research questions

  • Is there evidence that the progressive is increasing in present-day spoken English as is said to be the case for written English?

Summary

There is a statistically significant increase in the frequency of use of progressive measured as a proportion of progressivisable verb phrases over the period covered by the LLC (1958-1977) and ICE-GB (1990-92).

Table 5: the rise of the progressive year by year in DCPSE.
Yearprogressiveerrornot progressiveTotal
195842 (2.39%)±0.72%1,7131,755
195941 (5.27%)±1.57%737778
1960112 (4.93%)±0.89%2,1592,271
1961187 (4.01%)±0.56%4,4714,658
196320 (3.90%)±1.68%493513
196482 (3.63%)±0.77%2,1762,258
1965108 (3.93%)±0.73%2,6432,751
1966123 (3.76%)±0.65%3,1453,268
196789 (3.83%)±0.78%2,2352,324
196999 (3.19%)±0.62%3,0013,100
197080 (5.03%)±1.08%1,5091,589
1971220 (3.83%)±0.50%5,5235,743
197287 (2.85%)±0.59%2,9653,052
197337 (5.21%)±1.63%673710
1974279 (3.90%)±0.45%6,8747,153
1975561 (4.01%)±0.32%13,44014,001
1976196 (3.61%)±0.50%5,2305,426
197736 (3.93%)±1.26%881917
1990261 (4.79%)±0.57%5,1935,454
19912,193 (5.59%)±0.23%37,00939,202
1992698 (6.21%)±0.45%10,54111,239

This can be viewed more clearly in the graph below.

Charting the rise in spoken progressive use in English using DCPSE.
Figure 1. Charting the rise in spoken progressive use in English using DCPSE.

These results were presented at the symposium on Current Change in the English Verb Phrase at ICLCE3, London, July 2009. 

3. The Perfect

Research questions

  • Is there evidence of recent change in the overall frequency of the perfect construction in spoken British English?
  • Is the same pattern of change seen for the different tense forms of the perfect (present, past, and non-finite)?

Summary

Overall, there is a statistically significant decline in the frequency of the perfect construction from LLC (1960s-1970s) to ICE-GB (1990s).

The different tense forms of the perfect show different patterns of change:

  • The present perfect (the most frequent form) shows no significant change in frequency.
  • However, the past perfect and infinitive perfect both show statistically significant declines in frequency.
Table 6. Frequencies of perfect auxiliary tense categories in DCPSE (’s’ = significant at p<0.05).
Tense categoryLLCICE-GBChange in frequency
 rawpmwrawpmw%A: χ² (words)B: χ² (perfect)
present3,5728,020.173,3438,277.17+3.20%1.7221.71 s
past8351,874.824841,198.37-36.08%62.40 s41.74 s
infinitive6521,463.934131,022.58-30.15%32.90 s19.71 s
-ing participle78175.1358143.61-18.00%1.310.46
Total5,13711,534.054,29810,641.72-7.74%15.18 s 

The table shows two distinct series of chi-square tests. Column A compares the distribution of each term with the total number of words. Column B compares each term relative to the trend of the overall set of perfect auxiliaries. For example, Column A shows that the slight percentage increase of the present perfect is not considered significant compared with the number of words, while Column B shows that it does differ significantly from the overall pattern.

The normalised (pmw) percentage change over time from LLC to ICE-GB subcorpora can be plotted in the form of a bar chart, shown in Figure 2 below. The error bars represent p<0.05 confidence intervals.

Changes in pmw frequencies (Table 1 '%' column) with error bars for p&lt;0.05.
Figure 2. Changes in pmw frequencies (Table 1 ‘%’ column) with error bars for p<0.05.
Note that the data in Table 1 exclude instances of the ‘semi-modal’ HAVE got [to] (as in a lot of work has got to be done on it), but include instances of the combination HAVE + got which occur with an NP object and express a stative meaning (e.g. he’s got two kids ‘he has two kids’). These stative examples involve an idiom which is historically derived from a perfect construction. They are frequent in the present perfect data and should be taken into account in a detailed analysis of that category. However, they do not affect the findings summarised above: if the data are recalculated to exclude such examples, the overall pattern remains similar.

The Perfect Infinitive

A more detailed investigation was made of the perfect infinitive. A great majority of examples of the perfect infinitive in the corpus (88%) occur following a modal auxiliary, while the remainder occur in contexts following the infinitival marker to. Modal auxiliaries have themselves declined in frequency in our data. This led us to pose a further question.

Research question:

  • Can the declining frequency of the perfect infinitive simply be attributed to the declining frequency of the modal auxiliary as a potential context of occurrence?

Summary of findings:

  • When potential modal contexts and to-contexts are taken as the baseline, it is shown that the proportions of perfect infinitives fall within both types of context. Therefore the perfect infinitive shows trends of decline that are independent of the decline in modal auxiliaries.
Table 2a. Changes in the proportion of perfect infinitives in modal contexts in the LLC and ICE-GB components of DCPSE.
modalperfect infinitiveno perfect infinitiveTotalSummary
LLC561 (7.37%) 7,0507,611% change = −21.90
(c.i. ±11.20%)
ICE-GB371 (5.76%) 6,0746,445
Total932 (6.63%)13,12414,0562×2 χ² = 14.69 s
Table 2b. Changes in the proportion of perfect infinitives in ‘to’ contexts in the LLC and ICE-GB components of DCPSE.
toperfect infinitiveno perfect infinitiveTotalSummary
LLC 87 (1.33%) 6,4476,534% change = −51.83
(c.i. ±25.86%)
ICE-GB 40 (0.64%) 6,1976,237
Total127 (0.99%)12,64412,7712×2 χ² = 15.44 s

Findings

The English VP