4.1 Overall distribution

Table 1 sets out the variants found in the different geographical varieties. The forms are illustrated in examples (7)–(13).

Table 1. Overall distribution of forms expressing perfect meaning in ICE

^a Only written material available

1. (7) The main impetus for economic growth has come from expansion in the tourism industry, rehabilitation of export in agriculture, recovery in the bauxite/alumina industry. . . (ICE-JAM:W2B-018#98:2)

2. (8) The versatile star of Sophie's Choice and Out of Africa reportedly beat out fellow Oscar winner Geena Davis for the role <O> music </O> // You just heard the one o'clock news (ICE-PHI:S2B-016#84-85:6:A)

3. (9) Give it a rest, Benjamin. They’re gone and the game's up. (ICE-GB:W2F-012#103-104:1)

4. (10) They just finish national population and housing census in Nigeria. (ICE-NIG:EX_25)

5. (11) You seen enough to make you concerned that you know that this is a real challenge for the company people. (ICE-JA:S1B-045#125:1:A)

6. (12) She don't come yet? (ICE-JA:W2F-015#11:1)

7. (13) I have saw Andy Lau concert organised by Chase Credit card (ICE-HK:W1B-005#143:12)

The grouping of the variants in table 1 is such that some of the labels include potentially ambiguous forms, as follows:

• – ‘PP’ includes forms such as There's a piece of plaster ’s come off the ceiling so they'll replaster the ceiling as well (ICE-GB:S1B-071 #262-263:1:C) where the highlighted ’s (in the zero relative clause) can be understood as the abbreviated form of is or has;

• – ‘preterite’ includes the forms finished, got, heard, said, thought and told (8), which can also be past participles;

• – ‘base form’ includes forms like come, as in example (4) above (this young history teacher who just come from Africa (ICE-JAM:W2A-019#68:1)), in which verbs or past participles can also be present (see section 3 on methodology);

• – ‘other’ includes examples of present tense used with perfect meaning (10) and combinations like have saw, which can also be interpreted as performance errors (for a discussion of the difference between errors and incipient changes, see Hundt Reference Hundt, Seoane and Suárez-Gómez2016).

The overall distribution of forms expressing perfect meaning set out in table 1 shows that the most common form in all varieties is the canonical present perfect construction. However, this form is more frequent in ICE-GB, with 86 per cent of cases, than in any other variety, where it ranges from only 56.1 per cent in ICE-JAM to 80 per cent in ICE-EA, a percentage slightly higher than that found for ICE-USA (79.6 per cent). One obvious reason for the high proportion of PP in ICE-EA is the diachronic gap between the corpus representing this L2 variety and the other L2s, as pointed out in section 3. Another potential reason for the frequent use of PP forms in ICE-EA is that African varieties are said to show a convergence with L1 varieties and ‘are still relatively closely oriented towards (written) standard BrE’ (Brato & Huber Reference Brato, Huber and Hickey2012: 181). ICE-IND and ICE-SL, which come fourth and fifth in the use of the PP, contain texts from the 1990s and texts written between 2003 and 2009, respectively. In the case of IndE there is agreement that it is more elaborate (Xiao Reference Xiao2009) and traditional than other varieties, because English here has been learnt through school, relying on an exonormative British model. Likewise, Sri Lankan English tends to pattern very much like Indian English, as if this has become a new epicentre or a source of influence for South-Asian varieties, Sri Lanka included (Hundt Reference Hundt, Schreier and Hundt2013: 191). Additionally, it must be borne in mind that the results from ICE-SL contain only written English, usually more conservative than spoken language (see section 4.3 on mode).

The second most frequent form is the preterite, especially frequent in ICE-SIN, ICE-PHI and ICE-NIG, where it scores around 30 per cent of uses (I will return to preterite use in table 3). In fact, ICE-HK, ICE-NIG, ICE-SIN, ICE-PHI and ICE-JAM stand out as the varieties which show the highest rate of alternative forms to the PP. Within these, JamE deserves special attention, not only because it shows the highest frequency of non-PP forms (nearly 60 per cent), but also because it has the highest frequency of be periphrases, base forms and past participles (see table 1). According to Seoane (Reference Seoane2016a: 217) the use of be periphrases in JamE takes place predominantly with go, an intransitive verb of motion. In her opinion, this periphrastic structure is the result of diffusion from the input language, that is, an earlier variety of English in which the grammaticalization of have + past participle for the expression of perfect meaning had not yet taken place. The use of be with intransitive verbs of motion occurred more often in the period of colonization in the seventeenth century. In JamE, be also combines relatively frequently with the verb finish. The ambiguous status of the construction be finished is acknowledged in the literature: it is considered a ‘pseudo-passive’ construction with perfect meaning in the literature (see Quirk et al. Reference Quirk, Greenbaum, Leech and Svartvik1985 and also section 4.5 below) and also a ‘prefab’ (see footnote 4; Yerastov Reference Yerastov2015: 176). As for the use of base forms and past participles in JamE, they can be attributed to the influence of Jamaican Creole (JamC) (see, e.g., Christie Reference Christie2003; Deuber Reference Deuber2014; Seoane Reference Seoane2016a).

4.2 Type of perfect meaning

Classic accounts of the present perfect distinguish the following four semantic subtypes: resultative, experiential, recent past and persistent situation (Comrie Reference Comrie1976; see also Dahl Reference Dahl1985: ch. 5, Reference Dahl, Brown and Miller1999: 290–1; Dahl & Hedin Reference Dahl, Hedin and Dahl2000: 385–8; Miller Reference Miller2000: 327–31, Reference Miller, Fischer, Norde and Perridon2004: 230; Huddleston & Pullum Reference Huddleston and Pullum2002: 143–5). Following Miller (Reference Miller, Fischer, Norde and Perridon2004), I classified as resultative those examples which express an action having current relevance, such as (14), where the action of ‘giving advice to schools’ has current relevance for the topic. Examples like (15) were classified as experiential perfect, since they express an indefinite event (in (15) the speaker's experience at SRC) whose temporal frame leads to the present (the speaker's life). Regarding the expression of recent past, examples usually contain explicit temporal markers that express recency, such as the adverb recently in (16). Finally, examples containing persistent-situation typically describe recurrent events occurring in a period that leads up to the present. This is the case with (17), the recurrence emphasized by always.

1. (14) The Education Department is absolved since it has already given advice to schools to emphasise reading, listening and speaking in lower primary school (ICE-HK:W2B-006 #39:1)

2. (15) The only the only I have ever seen regularly at uh S R C is Max le Blond (ICE-SIN:S1A-082#197:1:A)

3. (16) Have you heard anything from Pam recently? (ICE-USA:W1B-004#56:1)

4. (17) Dominguez has corroborated what this newspaper has always said, and which the bankers and the government have been so hard put to cover up. (ICE-PHI:W2E-010#50:2)

In table 2 we can see that the canonical present perfect is favoured irrespective of the type of meaning expressed, but this formal variant is especially frequent in the expression of resultative meaning, which is the most common one and hence can be considered the perfect meaning par excellence in the corpus, and also in the expression of experiential meaning.Footnote ⁷ The same is the case with the expression of persistent situation perfect meaning, although the figures here are rather low. For the expression of recent past, however, the PP is in strong competition with alternative forms, especially the preterite, as clarified in table 3.

Table 2. Distribution of present perfect vs non-present perfect forms according to type of perfect meaning

Table 3. Distribution of preterite forms expressing ‘recent past’ per variety

^a Only written material available

Turning to table 3, then, frequencies of the preterite for the expression of recent past are set out: the use of the preterite is higher than the global mean (43.7 per cent) in ICE-USA, an L1 variety, and in most L2s, especially in ICE-PHI, which shows the highest score of the preterite with this meaning (and in general, as shown in table 1). It is tempting to attribute the high incidence of the preterite in Philippines English (PhilE) to the fact that AmE was the colonizer variety and in AmE the preterite with perfect meaning is more frequent than in BrE, as can be seen both in the corpus itself and in previous studies (Hundt & Smith Reference Hundt and Smith2009); however, as we know, the cultural and linguistic influence of the US today is not restricted to the Philippines but extends to a wide range of territories around the world, triggering other types of language change in many other varieties (Leech et al. Reference Leech, Hundt, Mair and Smith2009: 258).

Falling well below the mean (43.7 per cent) we find ICE-GB and ICE-EA, the only variety which patterns closely to ICE-GB, at some considerable distance from the rest. We might recall that this was also the case when we examined the extent of PP use (see section 4.1), in which ICE-EA reflects ICE-GB closely; we also noted in this context the time gap between different ICE components.

From these results we can confirm eWAVE feature number 99 (Kortmann & Lunkenheimer Reference Kortmann and Lunkenheimer2013), which specifies ‘levelling of the difference between present perfect and simple past, with the simple past or preterite used for StE present perfect’.Footnote ⁸ This is what we observe in ICE, but only for the expression of recent past. Furthermore, the results from table 4 show that, when the selected form is different from have + past participle, the recent past is the meaning that relies most strongly (51.7 per cent) on the presence of explicit adverbial markers of time (mainly just, today and now) and that in most cases this alternative form is a preterite.

Table 4. Presence of adverbial markers according to meaning

4.3 Mode: spoken vs written

The distribution of PP vs non-PP per variety according to mode (spoken vs written language) is given in table 5. Footnote ⁹

Table 5. Distribution of PP and non-PP forms according to variety and mode

^a Only written material available

The findings in table 5 show that, in five out of the eight varieties where both spoken and written language are available, the present perfect is more frequently attested in the written than in the spoken mode. In these five varieties, therefore, non-PP forms tend to occur in the spoken mode, which might perhaps be expected, given that innovative alternatives of expression often appear in spontaneous spoken interactions but are filtered out in written discourse.

The strongest contrast between spoken and written modes is observed most notably in ICE-PHI, but also in the two African varieties and in ICE-SIN. In fact, of all those varieties in which the PP is more frequent in the written mode, it is PhilE where the contrast is most pronounced, with 62.3 per cent of PP forms in spoken English, compared to 76.2 per cent in written English.Footnote ¹⁰ An explanation for this might be found in table 3, where we can see that PhilE is the variety with the highest proportion of preterite (thus non-PP) forms. As for the African varieties, there is again a fairly pronounced gap between written English and spoken English in ICE-EA (86.0 as against 76.8 per cent), which could be interpreted as a reflection of the early stage of development of the varieties it comprises, from Kenya, at stage 3 Schneider's (Reference Schneider2007) Dynamic Model, and from Tanzania, very early at stage 3. We might see in this evidence for the fact that adherence to the kind of linguistic norms taught in schools is reflected primarily in written English, and less so in the spoken mode. However, ICE-SIN is one of the most advanced varieties (stage 5 in Schneider's model) yet it also shows a similar gap between use of the PP in spoken versus written English (73.4 vs 63.5 per cent).

The three exceptions to the high frequency of non-PP forms in spoken material are ICE-GB, an L1 variety where non-PP forms are scarce in general, and ICE-IND, a very elaborate variety which shows a high incidence of present perfect forms irrespective of mode. Once more, the fact that English in India is learnt at school, with British English as the exonormative model, explains the homogeneity of the variety, where no clear differences between spoken and written texts can be observed in the expression of perfect meaning. In ICE-JAM, the pervasiveness of non-PP forms is not exclusive of spoken texts but is generalized to written text types too, which testifies to the fact that the influence of JamC is strong and predominant in a wide range of text types across both modes.

Thus, whereas at first glance the general picture seems coherent, with a tendency for the PP to predominate in the written mode, other factors must be at play to account for differences between varieties. These differences are likely to be related to the composition of the written and spoken components of the corpus, that is, the different text types and their linguistic conventions. Such differences in the composition of the corpora can at times distort the comparison of the spoken versus written modes (Mair Reference Mair and Collins2015: 141; Seoane Reference Seoane2016b).

4.4 Register: formal vs informal

Another variable relevant to us here is register (formal vs informal), determined by the different text types available. In spite of the caveats described in section 3 (methodology) regarding the ICE data, this section provides an attempt at comparing registers. In order to determine registers (as defined by Biber & Conrad Reference Biber and Conrad2009; see section 1), I selected specific text types of the spoken and written modes and classified them as formal and informal according to the following parameters: type of audience, communicative purposes and production circumstances.

Two balanced subgroups of 260,000 words each were created. Subgroup A represents informal register and includes the text types private conversation (spoken mode) and personal letters (written mode). Broadly speaking, the audiences in both text types are similar in that they are close to the speaker/writer and there is shared knowledge between them. These text types also share communicative purposes, because there are non-official, informative and conversational-like texts. Finally, they are produced under similar conditions: with relative spontaneity and a low degree of sophistication, little planning or editing, and unlikely to involve revision. Regarding subgroup B, which represents a more formal register, this contains the ICE categories scripted monologue (spoken mode) and academic writing, instructional writing and reportage (written mode). These share the following characteristics: (i) the type of audience is similar;Footnote ¹¹ (ii) they are informational; (iii) they are carefully planned; and (iii) they are edited, revised and produced under similar circumstances.

One unexpected finding from the comparison set out in table 6 concerns the raw numbers, which show that the distribution of perfect meaning (independently of the form it takes) varies widely across varieties: if we take ICE-HK, for example, we see that it expresses perfect meaning 774 times in the selected informal registers, whereas in ICE-GB only 369 examples in the same registers are recorded. The contrast in formal registers is not so pronounced; nevertheless, 328 such examples from ICE-EA contrast with 149 in ICE-GB. There is clearly room here for a more fine-grained, qualitative study in order to reveal why these differences are observed in these text types across varieties. Since we are comparing exactly the same text types in all varieties, such differences between varieties are unexpected and may have to do with the fact that the selection and interpretation of text types is culturally dependent. As Hundt (Reference Hundt, Biber and Reppen2015) shows, a student essay in ICE-PHI, for example, is not linguistically similar to a student essay in ICE-GB. So comparing text types with the same label (e.g. ‘social letters’) does not necessarily entail comparing similar text types, after all.

Table 6. Distribution of PP vs non-PP forms across a selection of formal and informal registers

Regarding the results of register, table 6 shows that subgroup A, representing informal registers, tends to exhibit a higher proportion of non-PPs than subgroup B, which contains formal registers. The only exception is ICE-IND, which shows a more balanced distribution of PP forms across registers, as it also did across modes (see section 4.3). As argued above, the homogeneity exhibited by ICE IND might be a reflection of the status of English in India, a co-official language learnt at school with prescriptive exonormative rules which speakers might be expected to apply in a broad range of contexts, which may turn Indian English into a monostylistic variety.Footnote ¹² The frequent use of a periphrastic structure might also be fostered by the existence of a periphrastic structure to express perfect in Hindi.Footnote ¹³ The remaining varieties have more than 30 per cent of non-PPs. As to the distribution of subgroup B, representing formal register, there is a much higher incidence of PP forms.

These results must be interpreted with caution, since registers are heterogenous due to cultural differences and different principles adhered to in the compilation of corpora (e.g. Hundt Reference Hundt, Biber and Reppen2015: 384). However, the findings here do show that register seems to be a strong predictor for variation in the realm of perfect meaning, and appears to override mode (see Seoane Reference Seoane2016b).

4.5 Lexical verbs

The last variable analysed in detail is that of the forms used by the different lexical verbs. The analysis here sought to identify any distributional preferences.

Table 7 shows that the verbs finish, get, say, tell and think express perfect meaning in more than 35 per cent of cases by means of a non-PP form. Following a more detailed examination of the preferences of these lexical verbs and specific non-PP forms, table 7 shows that get, say, tell and think (i.e. all but finish) select the preterite as the most frequent non-PP form, the four verbs together representing 56.5 per cent of all occurrences of the preterite in the corpus (1,178/2,085). say is found very frequently in broadcast news and press news reports (of the ten verbs, say is in fact the most frequent verb in such texts, with 33.5 and 25.1 per cent respectively). Both of these text types encompass a style known as journalese, characteristic of the popular media, and often feature the structure X said today, as in (18):

1. (18) Hindi to give the latest tally and important results, official sources said here today. (ICE-IND:W2C-005#60:5)

Table 7. Distribution of forms by lexical verb

^a Category ‘other’ excluded

The same applies to tell, frequently used in the same text types. Both verbs are also recurrent in the construction As I told/said + before/earlier, as in (19).

1. (19) Now as I told you before Daswani would not accept Indian English as a variety of British <,> British English (ICE-IND:S2A-047#93:1:A)

finish, on the other hand, the only regular verb in the sample, is the most frequent example of a verb used in the base form (see example (10), repeated here for convenience as (20)) and also occurs frequently in the be periphrastic construction to express perfect meaning (21).

1. (20) They just finish national population and housing census in Nigeria. (ICE-NIG:EX_25)

2. (21) Results of another 7,000 applications will be ready in the first week of next month and those of the remaining applications will be announced as soon as the processing is finished. (ICE-HK:W2C-007#83:6)

The case of the base form might be the result of simplification of final consonant clusters. In fact, 82 per cent of the occurrences of finish in the base form are found in ICE-HK, ICE-SIN and ICE-JAM. This phenomenon has already been reported for Asian varieties with a Chinese substrate (Gut Reference Gut2009; Suárez-Gómez & Seoane Reference Suárez-Gómez, Seoane, Bech and Andersen2013) and for ICE-JAM, under the influence of Jamaican Creole (Sand Reference Sand1999: 120; Seoane Reference Seoane2016a: 205). As already mentioned, the use of finish in the be periphrasis (21) to express perfect meaning is also documented in the literature, and is referred to by Quirk et al. (Reference Quirk, Greenbaum, Leech and Svartvik1985) as a ‘pseudo-passive’ construction, or considered a ‘prefab’ by Yerastov (Reference Yerastov2015: 176; see also footnote 4). It is also important to bear in mind that verbs meaning ‘finish’ are a frequent source of grammaticalized perfect markers across the world's languages (Kortmann & Schneider Reference Kortmann, Schneider, Heine and Narrog2011: 270, 273).

1. (22) Give it a rest, Benjamin. They’re gone and the game's up. (ICE-GB:W2F-012#103-104:1)

2. (23) I'm kind of pressured now because my boss gone for like three days and the other <indig> bredrin</indig> that work with me him gone out too for (ICE-JAM:S1A-037#18:1:A)

Another very frequent verb choosing the be periphrasis is go (finish and go represent 86 per cent, 146/165 examples of be perfect) (see (9), repeated here as (22)). This combination (be gone) is reminiscent of the use of be + past participle of mutative intransitive verbs for the expression of perfect meaning in the history of English before the grammaticalization of the have perfect from the end of Early Modern English onwards (Rissanen 1999: 215). Connected to this construction is the use of go in the past participle (23) (81/107, 75.7 per cent) as a consequence of copula deletion, 61 per cent of them in ICE-JAM, reflecting the influence of Jamaican Creole (see table 1, and Christie Reference Christie2003; Deuber Reference Deuber2014; Seoane Reference Seoane2016a, among others).