4. Competing types of passives, analysis and theoretical background
5.2 Passives in the Strathy corpus: Using a broader database
The data from C19 and S-sub yielded some low-frequency results, particularly for some less frequent verbs and more generally for the pronominal passives. For the sake of greater statistical confidence in this area, data from the full Strathy corpus (S-full below; 45.4 million words) will be added and discussed here. First, the absolute fre- quencies of the lexical passives of each verb are provided in Table 10. The tallies follow the same guidelines as those for C19 and S-sub above. Instances of ‘This email has been sent to you by [sender]’ from CBC News Online texts were filtered out, being automati- cally attached to mails published there.
Table 10. Lexical passives in S-full Verb First pass. Sec. pass.
total Prep
pass. –E Prep
pass. +E Prep
pass. total Verb
total % first pass.
assign 114 0 152 24 176 290 39.3
bring 1 0 163 35 198 199 0.5
deny 109 0 6 2 8 117 93.2
give 1343 0 264 41 305 1648 81.5
grant 122 0 11 2 13 135 90.4
offer 171 0 43 7 50 221 77.4
sell 10 0 58 6 64 74 13.5
send 33 0 288 25 313 346 9.5
serve 9 0 4 1 5 14 64.3
show 46 0 17 4 21 67 68.7
teach 45 0 4 2 6 51 88.2
total 2003 0 1010 149 1159 3162 63.3
Free ebooks ==> www.ebook777.com
Passives of so-called ‘ditransitives’ in nineteenth century and present-day Canadian English 167 As with S-sub, first direct passives are the clearly dominant type, and this is again largely due to the frequencies for give. The relative shares of direct and prepositional passives vary considerably for each verb, reflecting their different semantics and indi- vidual properties. In the case of offer, for instance, the participant following to is usually human or at least animate, whereas send has no such restriction and boasts many loca- tive themes that increase the number of prepositional passives noticeably.8
With 1648 occurrences, give accounts for about two thirds of all lexical passives for the verbs examined in this study. After deny, grant and teach it has the fourth highest percentage of first passives. If give were discarded from the lexical passives in Table 10, the total share of first passives in the last column would drop from 63.3% to 43.6%. The high share of first passives also owes much to frequent collocations such as give sb the opportunity/power /option/right to do sth, which typically occur only in first passives and allow no dative alternation. Such collocations of course favour a higher percent- age of direct passives. At the extreme points of the scale we find deny with the highest preference for first passives and bring with the lowest. Both verbs will be explored in Section 5.3. Second lexical passives did not occur at all among the verbs investigated; a pronominal realisation of the beneficiary seems to be required here.
Table 11 lists the pronominal passives (defined as for C19/S-sub) found in S-full.
In line with the S-sub data (Tables 7 and 8) first pronominal passives are rare. All first passives found here thus end in it/them as in … they’re taught it right from the time they’re born (Strathy, 2002, Fifth Estate [TV], spoken).
Huddleston & Pullum’s note on second passives that “[a]cceptability is greater when we have a personal pronoun + by phrase” (2002: 249) was tested for give, offer and send in S-full. It could not be corroborated for the present Canadian data: with these verbs, only three second passives were followed by a by-phrase, whereas in 15 instances no agent was specified.
Table 12 provides a summary of lexical and pronominal passives in the three cor- pora. While the percentage of pronominal passives in S-full is 5.9%, and thus clearly higher than in S-sub (3.3%), it is still much lower than the 32.1% found for C19. The reversal of the clear majority of prepositional over first direct lexical passives in C19 found in S-sub is confirmed by the data from the full Strathy corpus. The differences here between C19 and S-sub/S-full are statistically highly significant as shown in column 4.
8. Rohdenburg (2009b: 304) comments that AmE permits more verbs in the ‘double-object construction’ than BrE and that that there is “no doubt that, generally speaking, the preposi- tional variant, which was rare in Early Modern English, has dramatically increased its range of application over the last four centuries.” (ibid. 2009b: 308). Such variation does not, however, shed light on the relative strength of the first and the prepositional passive typical of the give group but rather applies to the (generally rare) second direct passive and its prepositional equivalent as in Questions were asked (of) him.
www.ebook777.com
Free ebooks ==> www.ebook777.com
168 Matthias L.G. Meyer
Table 11. Pronominal passives in S-full Verb First
pass. Sec.
pass.
–E
Sec.
pass.
+E
Sec.
pass.
total
Prep pass.
–E
Prep pass.
+E
Prep pass.
total
Verb
total % Sec.
pass.*
assign 0 2 0 2 3 0 3 5 40
bring 0 0 0 0 0 5 5 0
deny 2 3 4 7 3 1 4 15 63.6
give 3 4 10 14 38 37 75 92 15.7
grant 0 1 2 3 3 1 4 7 42.9
offer 2 1 1 2 10 4 14 18 12.5
sell 0 0 0 0 5 0 5 5 0
send 0 1 1 2 22 9 31 33 6.1
serve 0 0 0 0 0 1 1 1 0
show 0 0 0 0 6 2 8 8 0
teach 2 1 0 1 3 3 6 9 14.3
total 9 13 18 31 93 63 156 196 16.6
* This percentage relates second passives to prepositional passives only; first passives are ignored here.
Table 12. Totals of first and prepositional lexical passives compared in three corpora First lex.
pass. Prep.
pass. Chi-square Give first lex. p. Give
prep. p. Chi-square
S-sub abs. 61 26 C19 vs S-sub: 42 3 C19 vs S-sub:
S-sub % 70.1 29.9 56.4 (p < 0.001) 93.3 6.7 57.78 (p < 0.001)
C19 abs. 25 105 7 41
C19 % 19.2 80.8 14.6 85.4
S-full abs. 2003 1228 C19 vs S-full: 1343 305 C19 vs S-full:
S-full % 63.3 36.7 95.49 (p < 0.001) 76.3 23.7 128.58 (p < 0.001)
Give, the most frequent verb in the present study, has a major share in the new dominance of the first passive, and in S-full its preference for this construction (76.3%) is more pronounced than average (63.3%), though not quite as high as in S-sub, where it is overwhelming (93.3%: see Table 12). The change in preference from a preposi- tional give passive to a first direct passive is highly significant, as the chi-square values in the last column show.
The S-full share of pronominal passives amounts to 5.9% and is significantly higher than in S-sub, but the relative percentage is still less than one fifth of its counterpart in C19. The magnitude of this difference is at first surprising but may be explained by looking at the lexical passives, where we observe a corresponding shift from a low share
Free ebooks ==> www.ebook777.com
Passives of so-called ‘ditransitives’ in nineteenth century and present-day Canadian English 169
of first direct passives (19.2% of the total of first and prepositional passives) in C19, to a majority use of such first passives both in S-sub (70.1%) and in S-full (63.3%).
Let us now return to the question of the influence of extraction (E) on passivisa- tion as outlined in Section 4.4. Relative clauses turned out to be the most common cases of extraction co-occurring with the passives investigated. They typically have a pronominal beneficiary as illustrated in (8) and (9) above. Also, a browse through active give-predicates in S-full showed that extraction (the book (that) you gave me) is generally far more common than second passives (The book was given me). Extractions were also found to occur with lexical realisations of NP1 as shown in (17):
(17) a. the time period that you gave the municipalities to come into … (S-full, 2000–2001, Walkerton Water Tragedy Inquiry, spoken) b. the help they gave the soldiers during the Italian Campaign.
(S-full, 2004, CBC Magazine, news) This is noteworthy because such actives with extraction contrast sharply with the total absence of second lexical passives (*The time period was given the municipalities) in both S-sub and S-full. Thus, with the verbs investigated, only exceptionally can pas- sivisation separate a theme from its associated lexical beneficiary, while extraction can do so easily, as shown in (17). This reduces the question of whether extraction might render second passives more acceptable in CanE to cases with pronominal beneficia- ries. It has already been noted that for C19 and S-sub, no significant support for this hypothesis could be detected (see Table 8) but this might be attributed to the modest size of the corpora. S-full contains 31 pronominal second passives of the verbs inves- tigated, as opposed to 156 pronominal prepositional passives and 1159 instances of lexical prepositional passives. But even with the generally higher frequencies provided by a larger corpus and the greater marginality of second passives in Present-day CanE (compared to C19), extraction did not turn out to be significantly more common with them compared to their prepositional counterparts. A look at S-full frequencies of passives with and without extraction as provided in Table 11 shows us that indeed a majority of eighteen second passives with E (58.1%) corresponds to a minority of thirteen (41.9%) without E. In the corresponding prepositional construction only a Table 13. Summary data for S-full
First
pass. Second pass.
–E
Sec.
pass.
+E
Sec.
p. total Prep pass.
–E
Prep pass.
+E
Prep pass.
total
Verb total
Lexical 2003 0 0 0 1010 149 1159 3162
Pronominal 9 13 18 31 93 63 156 196
% extraction in
pron. passives 41.9 58.1 59.6 40.4
www.ebook777.com
Free ebooks ==> www.ebook777.com
170 Matthias L.G. Meyer
minority of 63 instances (40.4%) show extraction while 93 (59.6%) do not. The fact that extraction applies to a majority of second passives but only to a minority of prepo- sitional passives appears to suggest that extraction does favour the former, but despite the larger database the difference is once again not statistically significant (chi-square 3.29, p < 0.06). Statistical significance might, however, be obtained in a larger dataset.
The low absolute frequencies of the CanE pronominal passives motivated a com- parison with the AmE COCA data, whose 450 million words yielded much higher fre- quencies of individual verbs. Table 14 compares the relative shares of second passives (sth is given him) and prepositional passives (sth is given to him) in AmE and CanE.
Table 14. A comparison of second and prepositional pronominal passives in COCA and the Strathy corpus
Verb COCA
first pass.
COCA sec.
pass.
COCA prep pass.
COCA
% direct pass.
S-full sec.
pass.
S-full prep pass.
S-full % direct
pass.
assign 0 2 63 3.1 2 3 40.0
bring 0 9 485 1.8 0 5 0.0
deny 1 82 17 82.8 7 4 63.6
give 4 84 622 11.9 14 75 15.7
grant 0 18 16 52.9 3 4 42.9
offer 3 29 101 22.3 2 14 12.5
sell 1 2 30 6.3 0 5 0.0
send 0 7 166 4.0 2 31 6.1
serve 0 2 12 14.3 0 1 0.0
show 0 7 38 15.6 0 8 0.0
teach 2 11 24 31.4 1 6 14.3
total 11 253 1574 13.8 31 156 16.6
The frequencies of first pronominal passives (John was given them) are additionally given for COCA here, but the percentages in column 5 are based on the competition between second and prepositional passives alone. Abstracting away from individual low-frequency values in CanE, we can see that the overall shares of second passives are low, but not insignificant, in both varieties (16.6% of all second and prepositional pronominal passives in CanE, and 13.8% in AmE). Note that the higher overall share of second direct passives in CanE (as compared to AmE) is not statistically significant (chi-square 1.04, p > 0.05). In line with the global behaviour of the verbs listed, give strongly favours the prepositional construction. Interestingly the strong preference for
‘be assigned to – pronoun’ in AmE is not reflected in CanE. On the whole, however, the
Free ebooks ==> www.ebook777.com
Passives of so-called ‘ditransitives’ in nineteenth century and present-day Canadian English 171 share of second passives in CanE seems roughly similar and individual discrepancies can be attributed to the low frequencies in Strathy and to the different weightings of the genres contained.
5.3 The verbs bring and deny
Among the verbs investigated, bring stands out as having only one occurrence of the first passive given in (18), while being frequent (198 occurrences) in the prepositional construction:
(18) “I’m brought a veritable tumbler tull [= full] of scotch,” wrote Peter.
(S-full, 1994, Daniel Gawthrop, Affirmation, non-fiction)
Since this single first passive amounts to only 0.5% of the total number of pronominal passives of bring, this might signal that it is close to ungrammatical in CanE. However, it seems far more acceptable than *I’m bought a bottle of scotch. Therefore bring illus- trates a borderline case in English grammar: it differs from the give group in largely resisting the formation of first passives but it also does not quite match verbs like buy or write, whose first passive sounds worse and which typically show a benefactive instead of a dative alternation (buy a book for Mary vs give/bring a book to Mary).
A search in COCA using the pattern ‘is/are/was/were/be/been brought a/the’ yielded eighteen first passives of bring, as opposed to 937 occurrences of the prepositional pattern ‘be brought to a/the noun’. While first passives in COCA amount to only 1.9%
of the total of the two sets, attestations such as (19) stem from various text types and cannot be easily ignored.
(19) When seated, you are brought a basket of warm, freshly baked country breads … (COCA, 1990, San Francisco Chronicle, news) Thus bring might, by way of analogy, adapt to the give group and eventually become part of this category, which was described by Visser as expanding since the 16th cen- tury (see Section 2 above). In doing so, it would be moving away from the fairly sub- stantial group of non-passivisable verbs more prototypically represented by buy, build or make in constructions such as (20):9
9. Basilico (2008: 731ff) refers to verbs such as write, make, buy or build as “benefactive double object construc tions” and distinguishes them systematically from the “dative double object construction” found with give verbs. Huddleston & Pullum (2002: 295) see different semantic roles behind the first complement of make in He made me an offer of $100 and He made me a cake. The first me, they argue, denotes the recipient and the second the beneficiary.
One may, but need not, go along with this. For speakers who accept the passive of the former but not of the latter, the former is analysed here as ‘make – O – CE’ and the latter as ‘make – BC – CE’. See Section 4.3 for details.
www.ebook777.com
Free ebooks ==> www.ebook777.com
172 Matthias L.G. Meyer
(20) … my husband made me a frame, an A-frame, … (S-full, 2004, Canadian Geographer, academic)
Kirchner’s (1936: 207) list of non-established verb constructions includes passives of find such as the Assyrians were at last found a home that seem comparable to bring. Bring thus appears to be indicative of a group of verbs that are grammatically ambivalent between the older benefactive-intransitive pattern ‘V – BC – CE’ to which buy/write conform and the newer extensive-transitive (passivisable) pattern ‘V – O – CE’ typical of the give-group. At present, however, bring is still much closer to the older pattern.
While first passives of bring are rare, deny has the highest percentage of first pas- sives in S-full. However this cannot easily be attributed to its hostility to the prepo- sitional construction. Siemund (2013: 221–2) lists deny Beth the ice-cream as a case in point showing that verbs of deprivation only allow the direct construction. In his example this is indeed the preferred construction because the recipient (Beth) is shorter than the theme. With heavy NPs, to seems normal and Herbst et al. (2004, s.v.
‘deny’) do not mark the prepositional pattern of deny as rare. The to-construction with heavy PPs is also attested in S-full, as shown in (21):
(21) “BCE denies any liability to any creditor of Teleglobe … ”
(S-full, 2002, CBC Magazine, news) Altogether eight lexical and four pronominal prepositional passives in S-full show that this construction is not uncommon. It is useful for topicalising the theme and suitably replaces marginal second passives in CanE.