7. Pronouns

7.5 Relative Pronouns

7.5.1 Relative clauses with expressed relative pronouns

In the material for the present study, the chief relative pronoun is what, and this is used with both human (also termed personal or animate) and non-human (also termed non-personal or inanimate) antecedents. [23] What is used in cases in which StE has who, whom, that, or which. The relative pronouns who, which and that play only a minor role in the relative system. Who occurs only with human antecedents, whereas that is attested with both human and non-human antecedents. Which is used mainly as a sentential relative, i.e., referring to the whole clause. The object relative pronoun whom and the possessive (also termed genitive) relative pronoun whose are not attested in the material for the present study. The non-occurrence of the form whom in Cambridgeshire dialect speech is consistent with Wright’s observation: “Whom is never used in dialects” (EDG 1905: §423); cf. the discussion of whom as an interrogative pronoun in 7.6. For relative as, see the discussion below. For the non-expression of a relative pronoun, see 7.5.2. [24]

The discussion here focuses on adnominal relative clauses, which are the central type of relative clause, although sentential and nominal relative clauses are also briefly touched upon. The antecedent of the adnominal relative clause forms part of the main clause (matrix clause), whereas the nominal relative clause ‘contains’ its antecedent (Quirk et al. 1979: 11.20), and in the sentential relative clause the antecedent is the whole clause. Most of the relative clauses that occur in the material for the present study are restrictive. [25] Restrictive relative clauses are closely linked with the antecedent and give essential information for the identification of the antecedent that they modify, whereas non-restrictive (also termed parenthetic or descriptive) relative clauses stand in loose relation to the antecedent and give additional, non-essential information about the antecedent they modify (e.g. Curme 1968: 167-168; Jespersen 1969: 357; Quirk et al. 1979: 13.8-13.14). Prototypically, the antecedent is either a simple noun phrase (NP) or the head noun with its various attributes.

Relative pronouns occur in subject function and in object function. Examples 7.38 (a-e) illustrate the relatives what, who and that in subject function.

In the infrequent cases where relative that is attested in the interviews carried out for the present study, it usually occurs with antecedents including one of the following items: only, first, last or an adjective in the superlative, as in example 7.38 (d). For a discussion of possible reasons for the scarcity of that, see below and cf. also the discussion of the zero relative in 7.5.2. Who refers to human antecedents, as it does in StE (e.g. Quirk 1957: 97-98). Who is not attested as a direct object; [26] it occurs in subject function and, very infrequently, as the object of a preposition. For the uses of the relatives who and what as the object of a preposition, see 7.5.3. In contrast with relative who, relative what is used as a direct object, as exemplified in 7.39 (a-b).

(In example 7.39 (b), the asterisk is used instead of a name.) As can be seen in examples 7.38 (a-b) and 7.39 (a-b), what refers to both a human and a non-human antecedent. What is the relative pronoun used by all informants in the present study. Even the speaker who occasionally uses the relatives who and/or that often changes to what in the course of a discussion, as shown by example 7.38 (c). A great variety of antecedents are referred to with what. In addition to persons, objects and things, the reference may be to nouns denoting animals (e.g. Have you seen that little colt lately what I bought off ye? (Rampton TR)), which are often referred to with masculine/feminine personal and possessive pronouns and thus might have been referred to with who/that. What is also used to refer to the antecedent all and to indefinite pronouns ending in -thing, as illustrated by examples 7.40 (a-b).

What may also be preceded by the pronoun them, the combination them what corresponding to StE those who, as in example 7.41 (a), and StE those which/those that, as in example 7.41 (b).

Similarly, that what stands for StE the one who, as in example 7.42 (a), and for StE that which, as in example 7.42 (b).

In StE, the latter sentence could also be And was it interesting what I said about father?

In non-restrictive relative clauses, what occurs with human and non-human antecedents in both subject and object function, as exemplified in 7.43 (a-b).

Example 7.43 (a) also illustrates the use of a redundant personal pronoun (i.e. he) after the relative clause. This common phenomenon is discussed in 7.1.1 and in 7.5.4.

In nominal relative clauses, illustrations of what include:

As in StE, what may function as a subject or object within the nominal relative clause, as in examples 7.44 (b) and (a), respectively. Another usage corresponding to that in StE is the use of what as a determiner (attributive what), as in ’Cause everybody went to these pumps after what water they had (Waterbeach BB).

Although the use of what as a relative has a long history, the dominant position of this pronoun in the material for the present study may be a relatively recent development. What has been attested in relative use since the 11th century, “but it is not very common” and “occurs mainly after antecedents of less definite character, like all and nothing” (Mustanoja 1960: 194); cf. examples 7.40 (a-b) above. In East Anglia (Norfolk and Suffolk), what has been used since at least the early 19th century. This use of what in this area was first noted by Forby in his Vocabulary of East Anglia (1970: 138-139), published posthumously in 1830. He observed that “[what] is certainly provincial, if it be not peculiar to us”. Forby’s example, The woman what came yesterday, illustrates the usage with a human antecedent. By the end of the 19th century, Wright (EDG 1905, §423) had established that “[w]hat can be used when it refers to persons as well as to inanimate objects in some of the north-midland counties and in nearly all the counties south of the north midlands”. Wright further observed that what was in general colloquial use and illustrated the usage of various counties, including Suffolk (The brave queen what fought along o’ them Romans [Fison Merry Suf. 1899, 30]), Norfolk (I’d lost the totty little screw what held it in [Emerson Marsh Leaves.1898, 125]) and Essex (A great, fine, strappin’ feller for a ’usbin’ what knows ’is own mind [Carr Cottage Flk. 1897, 31]).

According to Lowman’s Survey of Middle and South England, which was carried out in the early 20th century, both what and that were used in Suffolk, Norfolk and Essex, whereas in Cambridgeshire only that was used (cf. Viereck 1975: Map 207; Viereck 1980: 27).

The SED material, collected some twenty years earlier than the material for the present study, includes responses to two questions intended to elicit relative pronouns. In both of these questions, the linguistic context is a restrictive relative clause with a human antecedent in subject function. The responses to question III.3.7 (“If I didn’t know what a cowman is, you would tell me: He is the man ... looks after the cows.”) show the relatives as and what for Cambridgeshire and various other counties, including the neighbouring counties (clockwise from the north) of Lincolnshire, Norfolk, Essex, Hertfordshire, Bedfordshire, and Huntingdonshire. Suffolk relatives are what and at. For Norfolk, Suffolk and Essex, the zero relative (He is the man Ø looks after the cows) is also given. Responses to question IX.9.5 (“The woman next door says: The work in this garden is getting me down. You say: Well, get some help in. I know a man ... will do it for you.”) are the zero relative (I know a man Ø will do it for you) for northern Cambridgeshire and Lincolnshire, Norfolk, Suffolk, Essex, Hertfordshire, and Huntingdonshire, and the relative at for Essex. The Southern Cambridgeshire relative is who. [27] To sum up, the Cambridgeshire relatives recorded as responses to the SED questions III.3.7 and IX.9.5 are as, what, who and zero. However, the SED Cambridgeshire material is very limited; only two localities (Elsworth and Little Downham) were visited by the SED fieldworker. Moreover, the context of the relative pronoun in the SED is restricted to that of subject function in restrictive relative clauses with an antecedent with human reference. Thus, the picture provided of the Cambridgeshire relative pronoun system is based on very limited evidence. The pronoun as, which was used in East Anglia at the turn of the 19th century (e.g. Those as sleep. SH, Forby (1970 [1830]: 10) and was still common in East Anglia a hundred years later (Wright EDG 1905, §423), might not have been in frequent use in Cambridgeshire at the time when the SED material was collected in the 1950s, and by the 1970s its use had become almost extinct, the pronoun what having gained ground. [28] In the interviews recorded for the present study, no instances of as are attested, with the exception of I'll find you some as you know (Rampton HP), which could also be interpreted as ‘I’ll find you some, as you know I will’. Nor is the relative at attested in the material for the present study, although it is recorded in the SED responses (III.3.7 and IX.9.5) for Suffolk and Essex, the counties bordering Cambridgeshire. According to Mustanoja (1960: 191), relative at occurred side by side with that in the North, where it was common in the 14th and 15th centuries, but became rare after 1500. [29]

The tendency in Cambridgeshire may have been similar to that in Suffolk, where, according to Claxton (1968: 11), “‘What’ takes the place of the relative pronouns ‘that’, ‘which’ and ‘who’, e.g. The woman what live next door and The pigs what Oi bought”. Claxton does not mention relative as or at, perhaps because of their scarcity in Suffolk at the beginning of the 20th century.

Cambridgeshire dialect speech is not the only regional dialect in which relative what is frequent. For example, what has been recorded in the traditional rural dialects of Norfolk [30] and Suffolk. [31] According to Hughes & Trudgill (1979: 18), relative clauses with what are particularly common in non-standard urban dialects, compared to relative clauses with which, as, at, and zero. [32] However, the relative frequencies of the relatives what, who, which, as, at and zero vary to a great extent in different non-standard dialects. For instance, Shorrocks’ (1980: 554-563) study on the dialect of Farnworth and district, in the north of England, shows the main relative pronoun to be as, followed by zero, with the pronouns what, who and which being less frequent. Shorrocks also discusses constructions without a relative pronoun and the juxtaposition of two main clauses as an alternative to a relative structure. [33] Ihalainen’s (1980: 187-196) examination of interviews with traditional dialect speakers in Somerset, in the south-west of England, reveals that the most frequent device for introducing relative clauses is the non-expression of the pronoun, both in subject and in object function, followed by that, which, what, as and who (in descending order of frequency). Ihalainen also includes the categories RELATIVE + PERSONAL PRONOUN and PERSONAL PRONOUN as devices used to introduce relative clauses. [34] For constructions with no relative pronoun and other devices used for expressing relative relations in the material for this study, see 7.5.2 and 7.5.5.

Contrary to the frequent use of relative what, the use of relative that is rare in the material for the present study. This scarcity of relative that may be related to the fact that the pronoun that is used in another function, that is, as an anaphoric pronoun in cases where StE has the personal pronoun it. The personal-pronoun use of that is particularly common in subject function in the initial position. For a more detailed discussion, see 7.1.1. The personal-pronoun use of that has been common in East Anglia since at least the late 19th century (Wright EDG 1905: 75). It is documented in Suffolk by Claxton (1968: 11), and in Norfolk, Suffolk and Essex in the SED. [35] (No evidence of the personal-pronoun use of that in Cambridgeshire is provided in the EDG or in the SED.) In Cambridgeshire dialect speech, the increasing personal-pronoun use of that may have caused its use as a relative pronoun to decrease. The rise of what as a relative pronoun compensated for this decreasing use of that. [36]

Like the relative pronouns that and who, the relative pronoun which is infrequent in the material for the present study. Which is used mainly as a sentential relative, that is, to refer to the whole of the preceding sentence. This type of relative clause, illustrated by example 7.45 (a), occurs in the speech of the youngest informant in particular. Similarly, in non-restrictive relative clauses, which mainly occurs in the speech of the youngest informant, although it is occasionally attested in the speech of other speakers as well. Which in subject function is illustrated by example 7.45 (b), and in object function by example 7.45 (c) (cf. the use of what in non-restrictive relative clauses and examples 7.43 (a-b) above).

Generally speaking, non-restrictive relative clauses are infrequent in the interviews carried out for the present study. Instead of non-restrictive relative clauses, speakers use co-ordinate and clauses and paratactic constructions. These are discussed in 7.5.5.

As mentioned above, the possessive (or genitive) relative whose is not attested in the material for the present study [37] (cf. the discussion of interrogative pronouns in 7.6). However, the form whose has been used as a relative “since earliest ME” (Mustanoja 1960: 200). According to the SED, it occurred in various dialects in the 1950s. For details, see Orton et al. 1978 (LAE Map M81). [38] The responses to the SED question which is intended to elicit the relative whose or an equivalent (IX.9.6; That’s the chap WHOSE uncle was drowned) give as his for northern and whose for southern Cambridgeshire. As his is also given for the neighbouring counties (clockwise from the north) of Lincolnshire, Suffolk, Essex, Hertfordshire, Bedfordshire and Huntingdonshire. [39] Earlier evidence drawn from Lowman’s Survey of Middle and South England, carried out in the 1930s, shows that his father for western Cambridgeshire and his father for eastern Cambridgeshire in the frame he’s a boy WHOSE father (cf. Viereck 1975: Map 208). That his father was also recorded for western Norfolk, and his father for eastern Norfolk and Essex. Going even further back in the history, Wright (EDG 1905: §423) records the construction as his (e.g. That’s the chap as his uncle was hanged) for numerous areas, and at his for Scotland (e.g. the man at his coat’s torn). Wright further claims that “as a rule the possessive relative cannot be expressed by a single word in dialects; instead a periphrasis or parenthetical sentence is substituted”. The periphrastic constructions that his father (Lowman) and as his uncle (the SED, the EDG) were not attested in the 1970s Cambridgeshire material. Nor were the constructions what his father, which was recorded for western Suffolk, or the construction that’s father (i.e. the reduced variant of that his father), which was recorded for eastern Suffolk (cf. Viereck 1975a: Map 208). For constructions used to express genitival relations, see the discussion in 7.5.5.

7.5.2 Non-expression of a relative pronoun

When discussing constructions without a relative pronoun, linguists often speak of omission, deletion or ellipsis of the relative pronoun, or contact clauses. Mustanoja (1960: 203-204; 206-208) presents a short overview of what he terms ‘Non-introduced Relative Clauses’ and points out that the term ellipsis is inaccurate. In OE there is “no ellipsis at all (i.e., nothing has been omitted that belongs to the organic structure of the statement), and that instances often interpreted as pronominal ‘ellipses’ are usually those of non-expression of the personal pronoun, not of the relative. … Non-introduced relative clauses make their appearance in the second half of the 14th century”. Thus, relative clauses without a relative pronoun have a long history, as also noted by Strang (1974: 142-143), who, in her discussion of contact clauses, writes that “it is important to realize that contact-clauses are ancient structures of independent origin, not just relatives with pronouns left out”. Similarly, Jespersen (1969: 360) describes contact clauses as “very old in the language” and as having been “for centuries extremely frequent in speech and in all kinds of literature, except in translations which tend to reproduce foreign idioms”.

Constructions without a relative pronoun in subject function are considered ungrammatical or “very colloquial” in present-day (written) StE (Quirk 1968: 103; Quirk et al. 1979: 865). However, non-expression of a subject relative pronoun is not a recent phenomenon, but was already appearing in texts from the latter half of the 14th century (Mustanoja (1960: 205), although, as in OE, “many of the cases usually interpreted as non-expressions of the relative pronoun might equally well be taken as non-expressions of the personal pronoun”. Later, at the turn of the 17th century, “contact clauses were still extensively used where the ‘relative’ had subject function, as in Shakespeare’s I see a man here needs not live by shifts” (Strang 1974: 142-143). Like non-expression of the subject pronoun, non-expression of the object pronoun – accepted as a standard feature in Present-Day English – appears towards the end of the 14th century. However, it is much rarer than non-expression of the subject pronoun, although it gains ground towards the end of the ME period (Mustanoja 1960: 205). Thus, constructions without an object relative (e.g. The man I saw is your uncle) have long been common (Strang 1974: 68).

In dialects, constructions without a relative pronoun in subject and in object function were frequent between 1700 and 1900 (Wright EDG 1905: §423). In their discussion of relative pronouns, (Hughes & Trudgill 1979: 17-18) list constructions attested in non-standard English; among them is the construction “including omission of pronouns referring to the subject That was the man done it”. More recent evidence is provided by Shorrocks (1980: 556), who notes that, in the dialect of Farnworth and district, “zero-relatives occur in subject function as well as in object function”, and by Ihalainen (1980: 188), who observes that the zero-form, both subject and object, is the most frequent construction in his Somerset dialect sample. (cf. note [34]) For Cambridgeshire, evidence for this construction is provided by the SED responses to question IX.9.5 (“I know a man ... will do it for you.”) with the construction without a subject relative (I know a man Ø will do it for you). [40]

In the material for the present study, the majority of cases without a subject relative (marked by Ø) are attested in existential there sentences and in sentences with the VP have/(have/has) got, a context corresponding to that of existential there sentences. These cases usually have a human reference (including animals), as illustrated by examples 7.46 (a-c), but occasionally also a non-human reference, as in example 7.46 (d).

In example 7.46 (d), the context shows that the meaning is not ‘I made the first tractor come’ (cf. ‘I made him come’) but the form ‘come’ is used for StE ‘came’ (cf. examples 7.49 (b-c)).

In addition, the subject pronoun is not expressed in a few cases, such as the following:

Existential there sentences and (have/has) got sentences are also the contexts which favour the constructions without an object pronoun. Examples are given in 7.48 (a-b).

A few cases, such as those illustrated by examples 7.49 (a-c), may at first seem problematic.

In Cambridgeshire dialect speech, the third-person singular of the simple present is usually expressed without the -s/-es suffix, and a few verbs (e.g. come) have identical forms in the simple present and past. Considering the frequency of the constructions without a pronoun in this particular context, cases such as these are interpreted as cases without a pronoun, rather than, for instance, cases with the bare infinitive. Thus, the examples are interpreted as follows: She’s got a sister Ø lives …, we got a baker Ø comes … and There is a man Ø came….

Constructions without an object pronoun, although frequent in existential sentences, are mostly those in which the antecedent contains one of the items first, next, last or the superlative, as in the examples I think I can show you my first = envelope Ø I ever drawed off the Ouse Catchment Board (Willingham AA) and The coldest winter Ø I = I knew = remembered was = nineteen sixteen = end of nineteen sixteen beginning of nineteen = nineteen seventeen (Harston AS).

Existential there sentences and have/(have/has) got sentences may be particularly favourable for the non-expression of a relative. Shorrocks (1980: 556) claims that, in the dialect of Farnworth and district, constructions without a pronoun are regular after there is, there are and there were. Similarly, Ihalainen’s (1980: 187-196) study on Somerset dialect speech shows the frequent occurrence of the zero pronoun in existential sentences. “Sentences introduced by there is/are seem to be an environment where omission of the subject pronoun is particularly frequent”, as exemplified by There was a young barrister keep questioning en, you know (Isle Abbots HA) (Ihalainen 1980: 189). [41]

In the material for the present study, cleft sentences provide a further context for object pronoun non-expression, often of the type that’s all + PRONOUN + got, as in We got a pillar box. That’s all we’ve got (West Wickham CC).

For the non-expression of the relative pronoun as object of a preposition, see 7.5.3.

7.5.3 Relative clauses and preposition stranding

In the material for the present study, prepositions always stand at the end of relative clauses. This preposition stranding occurs in relative clauses in which a relative pronoun is expressed as well as clauses with the non-expression of a relative pronoun (marked with Ø), as illustrated by examples 7.50 (a-c).

As might be expected, preposition stranding is most frequent with the relatives what and zero, which are the two dominant relatives in the material for the present study. Preposition stranding with who is a marginal phenomenon, since relative what is also used with human referents. Which and that are not attested as prepositional objects.

Preposition stranding may be a regular feature in dialect speech, as it is typical of colloquial speech in general. Shorrocks (1980: 558) writes that, in his study of the dialect of Farnworth and district, he has “not noted any cases at all in which a preposition precedes the relative pronoun”.

7.5.4 Extraposed relative clauses

In the material for the present study, a relative clause may be extraposed, that is, the relative clause may not immediately follow the antecedent to which it refers. The extraposed relative clause usually begins with what, which is consistent with the dominant position of what among the attested relative pronouns. Examples of extraposed relative clauses include:

What-clauses like these usually give additional information and thus resemble non-restrictive which-clauses in StE. However, extraposed relative clauses may also be restrictive, as in example 7.51 (e) ‘the man who used to guard them and (who) saw after them used to drive…’. In cases in which the additional information preceding the extraposed relative clause is expressed by a long and heavy structure, a personal pronoun is often added as a kind of restart of the main clause, as in example 7.51 (e). For more examples with a redundant personal pronoun, see 7.1.1.

Extraposed what-clauses are also a frequent phenomenon in traditional Somerset dialect (Ihalainen 1980: 192). The usage is illustrated by Ihalainen with the example They really do it for sport what catch in these rivers here (Wedmore LV). [42] The use of extraposed what-clauses is consistent with the general tendency of spontaneous dialect speech to proceed without a premeditated plan and particular structures in mind.

7.5.5 Alternatives to relative structures

As alternatives to relative clauses, Cambridgeshire dialect speakers use (i) and-coordination; (ii) two main clauses with a personal pronoun as a linking device; or (iii) they simply put one clause after another. The first alternative is exemplified in 7.52 (a-b).

Example 7.52 (a) can be interpreted as ‘he had an ordinary hammer, which got a groove in it’ and example 7.52 (b) as ‘there’s only about one in the village now, which is up the road’. [43]

The second alternative is illustrated by example 7.53.

This alternative is infrequent in the material for the present study. [44] In other non-standard dialects it may be more frequent. Trudgill (1974b: 41) includes it in his list of relative clauses and exemplifies it with He’s a man he likes his beer. This variant of alternative devices is also noted by Ihalainen (1980: 188) in his Somerset sample, and is illustrated with the example I was workin’ wi’ one man he used to kill pigs, too (Stocumber JC). Ihalainen explains: “Characteristic of this construction is that a personal pronoun is used as a linking device and the two clauses are simply strung together to form a semantic unit.” In his study of the dialect of Farnworth and district, Shorrocks (1980: 556) gives this construction with two main clauses juxtaposed as one of the alternatives to relative clauses. The example ‘An (and) I’ve started [to work] with a chap he come (came) out of Campbell Street’ is explained by Shorrocks as follows: “there are no pauses, and the second clause has the same intonation pattern that a relative clause would have.”

Clauses may follow each other without any linking word, as in example 7.54.

This example can be interpreted as ‘we had a butcher up here who lived across there’ or ‘we had a butcher up here who’d come and kill it for you’, with a clause ‘He lived across the road there’ inserted in the middle, to give extra information. Compare example 7.54 with the following example.

In this example, the antecedent is followed by a relative clause, which is followed in turn by clauses giving additional information, after which the main clause is restarted with a redundant personal pronoun.

As an alternative to whose-clauses, Cambridgeshire dialect speakers use a construction with a possessive pronoun, as in examples 7.56 (a-b).

The first example can be interpreted as ‘the young farmer, whose mother and father died, …’ and the second ‘*’s grandfather, whose name was *, …’. As mentioned in 7.5.1, the possessive relative whose is not attested in the material for the present study.

7.5.6 What in clauses introduced by than or as

In comparative clauses introduced by than or as, the nominal relative what is often inserted. [45] Illustrations such as the following represent this usage:

According to the OED (what, pron. 6), this is an old usage (e.g. I think I laughed heartier than what I do now (1818 SCOTT Hrt. Midl.)), which has survived (e.g. They’re all about the same age as what we are. 1966 P. WILLMOTT Adolescent Boys ii. 26), but is now considered dialectal or non-standard. The SED has no questions which try to elicit a whole comparative clause of the type He is younger than I am. Wright (EDD 1898-1905: 443) observes that what is used redundantly after like (e.g. She’s like what she is. Sc., w.Yks, Mid., War.), but does not mention the usage after than or as. However, in addition to its occurrence in Cambridgeshire dialect, this usage is noted in the dialect of Farnworth and district, and exemplified in ‘he is stronger tin (than) what I am’ (Shorrocks 1980: 560). [46]

My material also includes cases such as 7.58:

Originally, the adverb of time then and the connective particle than were the same word. In the literary language, than and then had become clearly differentiated in meaning and spelling by the 17th century. In dialects, then is still used after a comparative in the sense of than (OED, then). Thus, in example 7.58, the informant may still adhere to this old usage. Another possible interpretation of this example is that than is not expressed before what. The latter interpretation is supported by the fact that than is often shortened to ’an or ’n, as in example 7.57 (b), above.

Notes

[23] Terminology varies in different studies. For instance, Shorrocks’ (1999: 98) terms are human and non-human antecedents. Human and non-human antecedents are also referred to as personal and non-personal antecedents (e.g. Quirk et al. 1985). Cheshire et al. (1993: 68-70) use the terms animate and inanimate.

[24] In the interviews carried out for the present study, the most frequent relative pronoun is what (64%); next in frequency comes the zero relative (29%); then come the other relatives: who (2.6%), which (2.6%), that (1.3%) and as (0.2%) (Vasko 2005: 73). (As occurs in an ambiguous example I'll find you some as you know (Rampton HP), which can also be interpreted as ‘I’ll find you some, as you know I will’.) Instances in which the ‘antecedent’ of as is such or the same are excluded, since I consider these to be comparative rather than relative.

The Cambridgeshire percentages above, based on the total of 476 relative tokens, are somewhat different from those given by Peitsara (2002a: 180) in her study “Relativizers in the Suffolk Dialect”. Comparing the relatives in Suffolk and Cambridgeshire, Peitsara gives the following percentages for Cambridgeshire (for Suffolk in brackets): what 53.1% (17.7%), zero 38.3% (37.4%), who 3.4% (6.2%), which 3.4% (18.2%), that 1.6% (19.2%) and as 0.2% (1.3%). The differences in Cambridgeshire percentages are partly due to a different sum total of relative tokens. Peitsara’s percentages are based on a total of 444 tokens for the Cambridgeshire data (and a total of 390 tokens for the Suffolk data). This total does not include nominal relative clauses. Another reason for the differences is the fact that, as Peitsara (2002a: 167) says, “studying relativizers in dialect speech involves both general and dialect-specific problems. It is not always easy to recognize a relative construction in speech where the borderlines between dependent and non-dependent structures are often obscured”. The data sets for both Cambridgeshire and Suffolk consist of traditional rural dialect material collected in the 1970s.

[25] This finding corroborates Miller & Weinert’s (1998: 349) statement that two constructions of relative clauses, namely non-restrictive relative clauses and infinitival relative clauses are very rare in spontaneous speech. This scarcity is also noted in various other studies, e.g. Miller’s (1993: 112) study on Scottish English and Harris’ (1993: 139-186) Grammar of Irish English. The infrequency of non-restrictive relative clauses may be a reason why they are excluded in some studies, e.g. Cheshire et al.’s (1993: 68-70) survey on grammatical features in British urban centres and Tagliamonte’s (2008: 107) study on relative markers based on evidence from the Roots Archive.

[26] The same observation is made of the Suffolk dialect by Peitsara (2002a: 171).

[27] For a more detailed discussion of the relative pronouns recorded in the SED responses, see Upton et al. (1994: 489), which arranges responses according to the occurrence of features in individual counties. See also Orton et al.’s Linguistic atlas of England (LAE) (1978), which presents selected SED questions as maps on which isoglosses are drawn, and Viereck’s Computer developed linguistic atlas of England (1991), which lists individual SED responses and marks them with symbols on maps.

[28] While relative as is described as “in general use” in East Anglia in Wright’s EDG (1905: §423), it became very sporadic after 1900. In the Lowman’s Survey, relative as is no longer recorded for East Anglia, and later on, it apparently retreated even further west than Cambridgeshire (Herrmann 2003: 75-77). In traditional East Anglian dialect data from Suffolk and Eastern Cambridgeshire (i.e. Soham), no instance of as was encountered. This finding by Herrmann is based on data drawn from a sub-corpus of the Freiburg Corpus of English Dialects (FRED). The majority of the interviews in FRED are documents from oral history projects recorded between the 1970s and 1990s (Herrmann 2003: 20-21).

[29] Relative at, recorded in the SED responses (III.3.7 and IX.9.5) for Suffolk and Essex, may be a weak variant of that, since the initial sound is often dropped in weakly stressed words (e.g. they > ’ey), although historically relative at is not a reduced form of that. My view of at as a weakened form of that and not a separate relative is shared by Poussa (1996: 529; 2001a: 244-247), Peitsara (2002a: 169) and Herrmann (2003: 69).

[30] Poussa (1994a, 1994b, 1999b, 2001a, 2002b, 2006) discusses Norfolk relatives in two data sets: Francis’ SED data (from the end of the 1950s) from southern Norfolk and the data collected (circa 1990) by Poussa herself in north-west Norfolk from speakers of similar background to the SED informants (‘Hickling data’). Poussa (2001a: 247) concludes that the traditional dialectal relative marker in Norfolk is what. Poussa (1994b, 2002b) also found Francis’ Norfolk data “peculiarly what-full and that-less” (cf. the similar finding in the material for the present Cambridgeshire study). One reason for the rise of what in Norfolk, according to Poussa (2006: 323), is the loss of that in subject position, which Poussa takes to be a phonological question, similar to the that>at development in Yorkshire.

In a comparison of her own (Francis’ SED and Hickling) data, my data, as evidenced in Ojanen 1982, and Pasanen’s Suffolk data (Peitsara 2002a), Poussa (2002b: 1-23) concludes that northern Norfolk and Cambridgeshire are similar in demonstrating a high proportion of what and a paucity of that, but that in Suffolk the frequencies of what and that are roughly equal. Thus, as Poussa (2002b: 17-18) expresses it, “The slight difference between the two parts of Norfolk fits in very neatly, geolinguistically speaking, with the findings of what in rural Cambridgeshire and Suffolk in the 1970s”. Poussa’s comparison of the findings from Ojanen 1982 and those based on the Francis corpus further reveal that Cambridgeshire and Norfolk have hardly any attestations of which, whereas in the Hickling data which was attested in the interviews with a few elderly individuals, although the dominant form was what.

[31] In Suffolk “that, what and which seem to be about equal in general frequency”, as appears as a relic form in a few instances, and the predominant relative is zero (Peitsara 2002a: 176). For the percentages of each Suffolk relative, see note [24]. Evidence in Peitsara (2002a) is drawn from the Suffolk sub-corpus (the Pasanen corpus) of the Helsinki Corpus of British English Dialects (HD) http://www.helsinki.fi/varieng/CoRD/corpora/Dialects/suffolk.html. For a more detailed discussion of relative pronouns in Suffolk, see Peitsara (2002a). See also the earlier discussion in Kekäläinen (Peitsara) (1985), which is based on the analysis of relative pronouns in interviews with nine Suffolk speakers, included later in the HD.

[32] Hughes & Trudgill’s observation is confirmed by Cheshire et al. (1989: 198; 1993: 68), who, on the basis of responses to a questionnaire sent to various urban schools, report relative what as a feature of all the urban centres of Britain today. Cheshire et al. further note that relative what is reported “as frequently in the North of the country as in the South, and that as and at as relatives are reported very infrequently”. Relative what is also reported by Beal (1993: 207-208) in Tyneside English. Somewhat contrary to Hughes & Trudgill’s, Cheshire et al’s and Beal’s observations about relative what, Edwards & Weltens’ (1985: 116) survey of research on non-standard English found that what appeared to be used less frequently than other non-standard relative pronouns such as as, and that it did not seem to occur at all in the north of England. This difference in findings may be due to the relatively limited material available for Edwards & Weltens’ survey in the early 1980s. For the north of England, cf. note [33]. For relative pronouns in Suffolk, see notes [24] and [31], for Norfolk, see note [30].

[33] Relative pronouns in the dialect of Farnworth and district (with new material added) are also discussed by Shorrocks (1999: 98) in his Grammar of the Dialect of the Bolton Area. The use of relative pronouns in the north of England is also discussed by Beal (1993:187-214) in her Grammar of Tyneside and Northumbrian English, by Beal & Corrigan (2002: 33-56.) for Tyneside and Northumbria, on the basis of the Newcastle Electronic Corpus of Tyneside English (NECTE), and by Beal & Corrigan (2005b: 211-229) for Newcastle and Sheffield. Tagliamonte (2008: 107-128) examines the use of several morphosyntactic features, including relative clauses, in Maryport, Cumbria. Tagliamonte’s examination, which draws evidence from the Roots Archive (Tagliamonte, Smith & Lawrence 2005b), is not limited to the north of England; the examination also covers the use of relative pronouns in Southwest Scotland (Cummock) and Northern Ireland (Cullybackey and Portavogie). Tagliamonte’s earlier study, “Variation and Change in the British Relative Marker System” (Tagliamonte 2002: 147-165), compares recent sociolinguistic variation and change as seen in a large corpus from York with smaller dialect samples from Maryport (in Cumbria, north-western England), Wheatley Hill (in County Durham, north-eastern England), Ayrshire (in Scotland) and the rural areas of Devon and Somerset (in south-western England). In particular, Tagliamonte studied the patterning of who, that and zero, and found an “unexpected paucity of WH forms” in all areas, including the south-western communities. All the peripheral communities favoured that and zero subjects more than York. For the use of relative pronouns in Scottish English, see also Miller (1993: 99-139), and for a discussion of relative clauses in spoken English, including Scottish English, see Miller & Weinert (1998: 346-351). For the use of relative pronouns in Somerset, see Ihalainen (1980) and Peitsara (2006). For Somerset, see also note [34].

[34] Ihalainen’s (1980: 187-196) study is based on the interview data he collected in Somerset in the early 1970s. These data are included in the Helsinki Corpus of British English Dialects (HD) as part of a larger Somerset dataset. For the Somerset sub-corpus, see http://www.helsinki.fi/varieng/CoRD/corpora/Dialects/somerset.html. Peitsara (2006: 269-280) considers Ihalainen’s larger Somerset Corpus, showing that the more extensive data both corroborates Ihalainen’s findings from the smaller corpus (which is not surprising since, as an expert on Somerset dialect speech, Ihalainen chose the best representatives for his study) and ‘corrects’ Ihalainen’s conclusion that relative that only occurs as the subject or object of preposition, never as a direct object. In addition to proving that Ihalainen’s statement on the non-occurrence of that as a direct object was too categorical, Peitsara (2006: 272) found an increased proportion of that, making it the most frequent relative, where in Ihalainen’s study this was zero; the proportions of which and as were also found to be reduced. Thus, the frequencies in Peitsara (2006) are (in descending order of frequency): that, zero, what, which, who and as.

Ihalainen’s (1980) conclusions are based on interviews with seven informants in seven villages, whereas the Somerset sub-corpus of the HD, with 23 informants from 15 villages, made it possible for Peitsara (2006) to investigate the data in greater detail. This larger corpus also revealed phenomena that could not be detected in a small corpus, and, thus, showed the value of larger datasets.

[35] More recent evidence comes from Peitsara (2002a: 169) for Suffolk and from Poussa (Poussa 1997, 2002b) for Norfolk. The frequent use of the pronoun that in East Anglia is further noted by Trudgill & Chambers (1991c: 8; also, e.g. Trudgill 1995: 136-137). Trudgill & Chambers claim that “in East Anglian dialects of English, it occurs only as an object pronoun, with third-person neuter singular subjects being indicated by that”. They illustrate this usage with the examples That’s raining. I don’t like it – that’s no good. Trudgill (2001a: 2) claims that the personal-pronoun use of that is still “certainly found in Norwich, Ipswich and Colchester”.

[36] According to Poussa (2002b: 18-19, 21), “the wholesale replacement of that by what in Cambridgeshire and north Norfolk might be accounted for by the coincidence of the loss of relic as and the phonetic wear of the remaining that, at the time when that was being replaced by it in the personal pronoun system”. Poussa’s conclusion is based on evidence drawn from two data sets: the Francis SED data (the end of the 1950s) from southern Norfolk and the data collected (1990-1991) by Poussa herself in north-west Norfolk (‘Hickling data’) from speakers of similar background to the SED informants. Poussa further suggests that what appeared in the Cambridgeshire-Norfolk area towards the end of the 18th century.

[37] Generally speaking, relative whose is not frequent in non-standard dialects. For instance, in his study on the dialect of the Bolton area, Shorrocks (1999: 56) found attestations of the construction as his (e.g. ‘[the man] as his (whose) uncle were (was) drowned last week’), which is also the construction recorded for Cambridgeshire in the SED material (IX.9.6). However, whose (a relative determiner in Shorrocks’ terminology) was also attested. In her discussion of relativizers in the Suffolk dialect, Peitsara (2002a: 171) claims that “case and gender distinctions do not seem to be important to the informants. … The only possessive instance in the corpus is periphrastic: the **s lived there an’ Mr ** now = one WHO HIS son’s in Oxen near Bury (Lavenham D)”. Peitsara draws evidence from the Suffolk sub-corpus of the Helsinki Corpus of British English Dialects, http://www.helsinki.fi/varieng/CoRD/corpora/Dialects/suffolk.html. The data for this sub-corpus were transcribed by Peitsara. On the basis of the evidence in Wright’s EDG, Poussa (2002b: 3) concludes that whose may have spread during the first half of the 20th century, “mainly at the expense of the periphrastic expressions as his and what his”. The spreading of whose may have been slow, since whose is not frequently attested or is not attested at all in a number of dialects, including those of Cambridgeshire and Suffolk.

[38] Statistics for the SED (X.9.6) whose map can be found in Viereck 1991 (CDLAE 1 Map S10).

[39] For a more detailed discussion of geographical distribution, see Upton et al. (1994: 489-490).

[40] The frequent occurrence of the zero relative has been noted in several regional dialects. In their survey of research on non-standard dialects, Edwards & Weltens (1985: 116) note that the relative pronouns who, whose and which are far less common in regional dialects than in StE, and whom is completely absent. Instead, that is used in some non-restrictive relative clauses, and the relative pronoun is “commonly omitted in subject positions”. In Suffolk, the zero relative is found in more than one third of the total of instances and used by all informants (Peitsara 2002a: 167-180). This observation is based on evidence drawn from the Suffolk sub-corpus of the Helsinki Corpus of British English Dialects (HD). Poussa’s (1994b: 180; 2001a: 246-252) studies of Norfolk relatives, which draw on evidence from the Francis Corpus collected for the SED and Poussa’s Hickling Corpus (1990-1991), indicate that the zero relative is common, competing with what in frequency. The other relatives, that, which and who, are rather marginal in frequency. Poussa (2001a: 250) comments that, in Norfolk, zero must represent deletions of what (or possibly which), since relative that is highly marginal in this area. Edwards (1993a: 214-238) notes that the relative pronoun in Southern British English “can be omitted in dialect speech not only in object position as in StE but also in subject position”, as in There’s a train goes through without stopping and It ain’t the best ones finish first.

[41] Tagliamonte’s (2002: 147-165) study of relatives, drawing data from six different localities from the north and south of Britain, confirms existential constructions as a favourable context for zero relatives. Each variety (north, south, standard and peripheral) has more zero relatives in existential constructions, with cleft-constructions coming in second place. Tagliamonte’s (2008: 107-113) discussion of relatives in the Roots Archive (see Tagliamonte, Smith & Lawrence 2005b) also includes evidence of zero relatives in existential sentences, as exemplified by There was two women Ø lived in it (PVG/1) from Northern Ireland and The is no many folk Ø liked going to the pit to work (CMK/c) from Southwest Scotland. (For the Roots Archive, see also note [33].) Miller (1993: 99-139) also notes that, in Scottish English: “[t]hat (as subject relative pronoun) can be omitted in existential constructions: we had this French girl came to say, my friend’s got a bother used to be in the school, there’s only one of us been on a chopper before (informally recorded).” Similarly, Harris (1993: 148-149) states that in Irish English “omission extends to subject relatives, as in I’ve a friend lives over there”.

[42] For more information on the Somerset corpus data, collected by Ihalainen in the 1970s and 1980s and now part of the Helsinki Corpus of British English Dialects (HD), see http://www.helsinki.fi/varieng/CoRD/corpora/Dialects/somerset.html. Extraposed what-clauses are also discussed by Peitsara (2006: 267) in her study on traditional Suffolk dialect, which is based on the Suffolk sub-corpus data of the HD. Peitsara illustrates the usage with the example They was up to a farm yere then, what they call Wood Farm (Kingsbury BT).

[43] The and-coordination exemplified in 7.52 (a-b) resembles the alternative to non-restrictive relatives discussed in Miller & Weinert (1998: 349). Miller and Weinert’s alternative is the insertion of one clause in the middle of another. This inserted clause is introduced by and “but is not smoothly coordinated with the matrix clause”. There is typically a pause before and after the inserted element and “the separate pitch pattern over the latter”. The dashes represent the pauses in their example The boy I was talking to last night – and he actually workeds in the yard – was saying it’s going to be closed down. The same example is given by Miller in his Grammar of Scottish English (1993: 99-139) with the explanation “instead of non-restrictive relative clauses, speakers of Scots use co-ordinate clauses”. The and-coordination alternative is also noted as “a ‘quasi-relative’ clause formed with and” by Harris in his Grammar of Irish English (1993: 139-187).

[44] This structure is also infrequent in Suffolk. Kekäläinen (1985a: 354) found only one instance So he was – he was one they called ’im Russian Bob or Russian Finn (Snape “J”W), and explained this example as follows: “The personal pronoun, the initial aspirate of which is suppressed, is the object of the relative clause”. Kekäläinen’s example does not exactly correspond to that by Ihalainen. In her later study on relative clauses in Suffolk, Peitsara (Kekäläinen) (2002a: 168) notes this structure, “the existence of personal pronouns as relativizers”, but, as Peitsara explains, “because of their relative infrequency and the difficulty of distinguishing them”, she does not include them in her study.

[45] Biber et al. (1999: 14.4.5.3) use the term intrusive what and illustrate the usage with the examples It’s harder than what you think it is this (BrE) and Orange flowers that’s about as much as what we’d ever get in here! (BrE).

[46] Ihalainen (1990b: 86-99) studied comparative clauses like these in three samples: (a) Somerset (some 23,000 words collected by Ihalainen in the 1970s and 1980s) (b) Cambridgeshire (some 27,000 words from the present writer’s data) and (c) Yorkshire (some 14,700 words collected for the SED and transcribed by Melchers (1972)). Ihalainen discovered that in Somerset and Cambridgeshire comparative clauses were often introduced by the structures than what and what rather than simply by than, whereas this redundant structure was rare in Yorkshire.