In this section, the distribution of individual relativizers in restrictive and non-re-strictive relative clauses will be considered. According to Quirk et al. (1985: 1245), restrictive relative clauses are overall more frequent than non-restrictive relative clauses (see also e.g., Biber et al. 1999; Hundt and Denison 2013). As discussed ear-lier, in dialectal speech non-restrictive relative clauses are less frequently used than restrictive relative clauses (see, e.g., Peitsara 2002; Herrmann 2003).
As table 7 indicates, restrictive relative clauses are more frequent than non-restrictive relative clauses in news reporting, too. In BrE the distribution between restrictive and non-restrictive relative clauses is nearly even, while in LSSE the dif-ference in the frequency of these two types of relative clauses is the most explicit.
This variation between BrE and LSSE proves statistically significant (Fisher’s exact test: p = 0.0222). The percentages do not change notably when adverbial relatives are excluded from the investigation (see table 3 in Appendix A). However, the dominance of restrictive relatives is somewhat stronger.
Table 7. Distribution of restrictive and non-restrictive relativizers including adverbial relatives where, when and why
HISSE LSSE IrE BrE
R % NR % R % NR % R % NR % R % NR %
That 12.5 1.1 15.6 1.2 7.3 1.0 11.5 2.7
Zero 15.6 0.0 15.2 0.0 10.0 0.0 11.1 0.0
Which 10.6 22.8 10.0 18.7 14.9 17.8 10.6 22.4
Who 10.8 15.1 12.5 14.2 20.3 17.1 13.3 12.0
Whom 0.0 0.3 0.1 0.3 0.0 0.5 0.2 0.2
Whose 0.3 1.5 0.7 0.7 0.0 2.4 0.2 2.5
Where 2.9 3.0 1.9 4.6 1.0 3.7 1.4 5.2
When 1.1 2.0 1.2 2.6 1.7 2.2 2.3 4.2
Why 0.2 0.0 0.5 0.0 0.0 0.0 0.2 0.0
Total 54.1 45.9 57.6 42.4 55.3 44.7 50.9 49.1
As for the distribution of restrictive relative clauses, that and zero are used clearly more frequently in SSE news than which and who. In IrE the wh-relatives have the upper hand and who is especially frequent, whereas in BrE the distribution of these four relativizers is rather equal. Amongst the non-restrictive relativizers, the wh-rel-ativizers which and who dominate in all the varieties: which takes the most relativizer positions across the varieties.
Interestingly, from a diachronic perspective the Scots/SSE situation has not changed since the Early Scots period, when in written Scots that was the most com-mon relativizer in restrictive and quhilk (i.e. which) in non-restrictive relative clauses (Caldwell 1974). This point will be further elaborated in chapter 6.
As figure 3 and table 7 indicate, the wh-relatives and non-restrictive relative claus-es are frequent in SSE unlike in spoken Scots and other spoken English varieticlaus-es (see, e.g., Macafee 1983; King 1997; Beal and Corrigan 2002; Tagliamonte 2005, Gut 2011).
The overall frequency of restrictive vs. non-restrictive relative clauses in Scottish speech is 79.3 vs. 20.7 (Herrmann 2003: 107), while in news writing the frequencies are 55.0 and 45.0 in HISSE and 60.6 and 39.4 in LSSE, respectively. The current results are congruent with Romaine’s (1980: 236) findings for Scottish newspapers regarding the distribution of the wh-relatives vs. that. These findings show that the frequency of non-restrictive relative clauses is clearly higher in news in comparison to speech (cf. Biber et al. 1999). Also, the fact that the wh-relatives, especially which, are more frequent in non-restrictive than restrictive relative clauses in news seems to increase the overall number of non-restrictive relative clauses. This applies to all the varie-ties under study, not only SSE. In a sociolinguistic study in Ayr, western Scotland, Macaulay (1991) observes that the wh-relatives and non-restrictive relative clauses are more frequent in middle than lower class speech, a finding which supports the current results and is reflected in the frequency of these features in written news, which are produced by educated writers, i.e. middle class informants.
The distribution of syntactic functions in relative clauses
As pointed out above, previous studies on relativization indicate that relativizers typi-cally function as the subject of the relative clause (see, e.g., Biber et al. 1999; Geisler 2002; Peitsara 2002; Tagliamonte et al. 2005). The results of the present study concur with these earlier findings as illustrated by tables 8 and 9.
Table 8. Distribution of syntactic functions in restrictive and non-restrictive relative clauses including the adverbial relativizers where, when and why
Subject Object Adverbial Predicative
complement
Total
Variety % % % % %
HISSE 65.6 17.9 16.3 0.2 100.0
LSSE 65.3 16.6 17.6 0.5 100.0
IrE 74.1 12.0 13.7 0.2 100.0
BrE 65.8 14.3 19.9 0.0 100.0
Tables 8 and 9 show that relativizers occur most frequently in subject function. The object and adverbial functions are also common and in close contest with each other.
BrE is the only variety in which there is discernible variation between the object and adverbial functions, to the advantage of the latter. Adverbial relativizers are more frequent by 5.6 percentage points, while in the rest of the varieties there is only mi-nor variation in the distribution between the adverbial and object functions. Relative clauses in predicative complement function are rare in news.
According to Biber et al. (1999: 621–2), relativizers in subject function are easier to process compared to other syntactic functions. Therefore it is the most frequently relativized syntactic function. When in subject function, the relativizer occurs in its standard position in the clause, and the clause elements follow the standard SVO order. This function is frequent in informational written registers, such as news, because the relative clause provides new information and identifies the antecedent (Biber et al. 1999: 622).
Table 9. Distribution of syntactic functions in restrictive and non-restrictive relative clauses excluding the adverbial relativizers where, when and why
Subject Object Adverbial Predicative
complement
Total
Variety % % % % %
HISSE 72.3 19.7 7.7 0.2 100.0
LSSE 73.3 18.7 7.5 0.5 100.0
IrE 81.0 13.1 5.6 0.3 100.0
BrE 75.8 16.4 7.8 0.0 100.0
When the adverbial relativizers where, when and why are excluded from the analy-sis (table 9), the frequency of subject function rises sharply, approximately between seven to ten percentage points.
Prepositional complementation
As discussed in chapter 3, prepositional complementation is realized either by pied-piping or stranding. This is well illustrated by table 10.
Table 10. Prepositional complementation excluding the adverbial relativizers where, when and why
Prepositional
comple-mentation Pied-piping Stranding
% of all RCs % of prepositional RCs
HISSE 7.3 59.3 40.7
LSSE 6.1 50.0 50.0
IrE 5.9 63.6 36.4
BrE 8.9 94.1 5.9
Prepositional complementation of relative clauses is a regular but sporadic feature in news, and occurs with a similar frequency across the studied varieties. Pied-piping is especially frequent in BrE, which in this respect differs clearly from all the other varieties. In HISSE and IrE approximately 40 per cent of the instances are stranded, whereas in LSSE prepositions are pied-piped and stranded with equal frequency.
Relative clauses with the adverbial relativizers where, when and why are excluded from table 10, because their prepositional complementation is not usually realized by pied-piping or stranding. In the current data there is only one occurrence of where with a pied-piped preposition (in HISSE). Instead, the preposition is usually “con-tained” by the antecedent in adverbial relative clauses (see chapter 3).
The total figures of prepositional complementation are in line with Van den Eynden’s (1996) findings on relativization in British newspapers. She is, however, able to show that register affects the choice of preposition placement: in quality news-papers stranding (9%) is less frequent than in tabloids (28%). Similarly, NigE, JamE, PhilE and SingE news follow the StE “guidelines” and pied-piping is their preferred option (Gut and Coronel 2012). In fact, pied-piping occurs in more than 65 per cent of the instances in all these varieties, a percentage which is higher than in SSE news.
Pied-piping is most frequently applied in JamE news, in over 90 per cent of the in-stances, which is very close to the BrE figures presented here.
5.1.2 Types of antecedent