The functional nature of pronominal change: Innovative plural pronouns in English and Dutch Authors Gunther De Vogelaer Research Foundation Flanders (FWO) / Ghent University Department of Dutch linguistics Blandijnberg 2 B-9000 Gent (Belgium) Evie Coussé Ghent University Department of Dutch linguistics Blandijnberg 2 B-9000 Gent (Belgium) Contact details of corresponding author: Gunther De Vogelaer gunther.devogelaer@ugent.be tel. +32/9/264.40.75 fax. +32/9/264.41.70 Abstract At least since Gilliéron’s seminal work on lexical change, it is well-known that language change may be motivated by factors relating to the meaning and/or function of the elements involved. In more recent times, however, such functional accounts of language change have generally met with criticism. More specifically, there is disagreement as to whether grammar really is subject to processes of homonymy avoidance. This paper discusses examples of pronominal change from English and Dutch, in which new plural forms are formed which are, historically, compounds of a traditional pronoun and an element marking plurality (e.g., you guys in English, jij lieden or jullie in Dutch). Such innovative pronouns typically emerge in contact varieties, as reinforced pronouns or calques, and they appear to spread much more easily wherever they fill a disturbing gap in the pronominal paradigm, viz. in situations of so-called ‘horizontal homophony’. Thus, homonymy avoidance has played a crucial role in the processes by which innovative plural pronouns have diffused. This shows that pronominal change qualifies as a clear exception to the alleged rarity of functionally motivated change in grammar. Keywords language change, functional explanation, pronoun, Dutch, homonymy avoidance

The functional nature of pronominal change Innovative plural pronouns in English and Dutch 1. The functional nature of language change: controversy At least since Gilliéron’s seminal work on lexical change, it is well-known that language change may be motivated by factors relating to the meaning and/or function of the elements involved. With a famous example, Gilliéron pointed out that the fear for homonymy may stir processes of lexical replacement: in Gascon dialects the words gallus ‘rooster’ and cattus ‘cat’ both evolved to the form gat, giving rise to a rather inconvenient ambiguity in a rural society like the 19th century Gascony. Subsequently, the homonymy has been dispensed with by the replacement of gallus ‘rooster’ with words such as faisan ‘phaesant’ and vicaire ‘vicar’, which were originally only metaphorically used to refer to roosters (Gilliéron & Roques 1910:278-288; see also Bynon 1977:187-189, Geeraerts 2010:62-63). Such an instance essentially shows that language change may be ‘functional’ in nature: in some cases, changes appear to respond to language users’ need for efficient tools for communication.1

In more recent times, however, such functional accounts of language change have generally met with criticism. More specifically, there is disagreement as to whether grammar is subject to processes of homonymy avoidance. Lass (1997:355n), for instance, labels functionally motivated change at best a “trivial” phenomenon. In fact Lass (1997:359-361 and passim) strongly objects that a process like homonymy avoidance exists at all, and to the extent that it does, it is certainly restricted to lexical items from the same semantic field. Labov (1994:547-568), too, only attributes homonymy a minor role in language change. Somewhat less restrictively, Hopper & Traugott (2003:103) acknowledge homonymy avoidance in grammar, but they claim that it is a very infrequent phenomenon. Two of the examples they provide, come from the domain of pronouns. More specifically, Middle English appears to have borrowed the feminine pronoun she from northern English and plural they from Scandinavian to dispense with the merger of Old English masculine he, feminine heo and plural hie (see also Samuels 1972:114-116 on she and 1972:71-72 on they).

Since the adoptions of she and they are among the most far-reaching examples of change in the history of the English pronominal paradigm, this raises the question to what extent pronouns are exceptions to the general qualification of grammatical change as being rarely driven by homonymy. In order to shed light on this issue, this paper discusses a number of other examples of pronominal change, viz. the rise of new plural pronouns in English and Dutch. Especially in the second person plural, both English and Dutch have introduced new forms, typically compounds of older ambiguous pronouns and an element marking plurality (e.g., the adverb all, a plural -s or a noun such as guys in English, or the noun lieden in Dutch). For instance, while Standard English uses the second person pronoun you to refer to the second person plural, in regional varieties a plethora of innovative forms are found, including y’all , youse, you guys, etc. In Dutch, in the second but also in the first and third person plural, compounds have emerged of traditional, simplex pronouns and the noun lieden ‘men’ (e.g. 1pl. wij lieden, 2pl. gij lieden, 3pl. zij lieden). One of the descendants of these compounds, the form jullie (< jij /jou ‘you’ + lieden ‘men’), has become the Standard Dutch 2pl. pronoun. Both in English and in Dutch the simplex forms, originally 2pl. pronouns also used as honorifics, became 2sg. forms. For English, Wright (1997) and Hickey (2003) have highlighted the therapeutic effect of the introduction of such innovative 2pl. forms: indeed these forms disambiguate between the second person singular and plural. Lass (1997:360), 1 Throughout this paper, the term ‘functional’ is used in a narrow sense, as involving the solution of potential meaning ambiguities. Of course broader definitions of the term are found as well, in which ‘functional factors’ refer to all factors that involve some sort of ‘user-friendliness’ (cf. Haspelmath 1999) or even to system-internal factors in general.

however, includes 2pl. youse (and related forms) among the examples for which there was no functional need to avoid a collapse of categorical distinctions or to subsequently restore the old singular/plural distinction. He argues that speakers of English managed perfectly without such a distinction for centuries before the new complex pronominal forms emerged. Croft (2000:69) too claims that gaps in pronominal paradigms, including the absence of a separate 2pl. pronoun in English, do not motivate language change. Rather, innovative 2pl. pronouns are believed to be instances of grammaticalization of highly frequent colloquial forms of address.

For Dutch, similar contradictory opinions have been voiced. On the one hand, Van Hout (2003) hypothesizes that the compound pronouns with lieden must be older in the second person, and since they serve a disambiguating function in that person, this indicates that disambiguation has played a major role in their emergence. Van Loey (1958) on the other hand, emphasizes that the non-ambiguous compounds with lieden ‘men’ not only occur in the second, but also in the first and the third person. Especially in the first person, these compounds do not have a disambiguating effect, suggesting that their emergence is not due to homonymy avoidance.

The goal of this paper is to discuss the extent to which the emergence and/or use of plural pronouns such as English y’all , youse,... and Dutch jullie are indeed to be attributed to homonymy avoidance, thereby casting light on the functional nature of language change. This will be attempted in section 2, where contemporary data on the second person plural will be used to formulate a hypothesis on the way functional considerations affect language change. This hypothesis will then be tested in section 3 with historical data from the Dutch third person plural. Section 4 is an excursion addressing the issue of S-curves and the question whether the relevant changes are mainly implemented by speakers or by hearers. Section 5 concludes this paper. 2. Ambiguity and disambiguation: the second person plural case Section 2 of this paper focuses on 2pl. pronouns in English and Dutch, using contemporary or at least recent data from non-standard varieties of both languages. On the basis of these data, a hypothesis will be formulated on the way innovative 2pl. pronouns have replaced older forms ambiguous between a singular and a plural reading. This hypothesis will then be tested in section 3 with historical data from the Dutch third person plural, drawn from a corpus historical legal texts spanning the Middle Dutch and the Early Modern Dutch period (1250-1800). 2.1. ‘Horizontal homophony’ Both in English and in Dutch the present-day second person singular pronoun is historically a plural form that was also used as a honorific. These plural forms, such as English you and Dutch jij and gij,2 subsequently extended their use dramatically, turning the older singular pronouns thou and du (and the corresponding oblique and possessive forms) redundant, and neutralising the distinction between singular and plural (see Howe 1996:170-175 and 220-223 for discussion). The absence of a singular-plural-distinction is termed ‘horizontal homophony’ by Cysouw (2003:299-300). According to Cysouw, horizontal homophony is not typologically rare: in his sample of 117 pronominal paradigms there are 44 paradigms (37,6%) with homophony in at least one grammatical person. Despite the relatively high frequency with which horizontal homophony is observed, there appear to be constraints on the types of homophony. Homophony mainly occurs in the third person, and only to a lesser extent in the second and the first person. More specifically, an implicational relationship is observed 2 The pronoun jij ‘you’ is the present-day Standard Dutch 2sg. pronoun, gij ‘you’ is a historical form which is still widely used in the southern part of the Dutch speaking area.

between horizontal homophony in the third, second and first person plural: horizontal homophony in the first person is only found in paradigms in which the second person singular is homophonous to the plural, and horizontal homophony in the second person is only found in paradigms in which the third person singular is homophonous to the plural (cf. Cysouw’s 2003:160-162 ‘Horizontal Homophony Hierarchy’). The English and Dutch pattern of horizontal homophony is not in line with this hierarchy: indeed in English and Dutch the horizontal homophony involves the second person, whereas the third person does display a number distinction. This pattern is cross-linguistically rare.

Cysouw (2003) does not provide an explanation for the Horizontal Homophony Hierarchy. One possibility is that the first and the second person are more central in human thinking, i.e. they are less ‘marked’ than the third person (cf. the ‘person hierarchy’, as discussed for instance in Croft 1990:136-139). Hence it seems unnatural that the third person would distinguish categories that are not kept apart in the first and the second person. Another explanation can be that it seems rather problematic in a conversation if one cannot distinguish between reference to first person alone or to first person and another referent (addressee or third person), or between one or more addressees, whereas not being able to distinguish between one or more third persons poses less problems, certainly since unambiguous reference to third persons is often achieved by means of non-pronominal items (nouns, proper names).

Whatever be the explanation, the rarity of the horizontal homophony that is found in English and Dutch indicates that, cross-linguistically speaking, paradigms are avoided in which the number distinctions are neutralised in the second person but not elsewhere. 2.2. Disambiguation in non-standard English: y’all , you guys etc. Apart from a rare number of dialects, the vast majority of the varieties of English have lost the old 2sg. form thou (and corresponding forms such as thee and thine), and have replaced it with the former 2pl. pronoun you. Subsequently, in many varieties an innovative 2pl. pronoun was introduced. According to Kortmann & Szmrecsanyi (2004:1154), the emergence of 2pl. pronouns ranks among the most widespread features in worldwide varieties of English: in an overview of 46 varieties, 34 are mentioned in which a special 2pl. form is used. The 2pl. forms in many of these varieties are discussed in detail by Wright (1997) and Hickey (2003). The relevant forms include the following: ye, youse/yez, y’all and you’uns/yins. In addition, a number of forms are used that still are clearly recognizable as compounds, viz. you all, you guys, you lot and you ... together.

There is variation as to the etymology of these forms. For instance, ye is an originally reduced form of you which has assumed 2pl. meaning, while its strong counterpart you is used for 2sg. (Hickey 2003:349). The form originated in Irish English varieties, where it is still used, and from where it was diffused, for instance to Newfoundland English. In other varieties, again most notably in Irish English, youse and yez are found, both formed by attaching the plural suffix -s to, respectively, you and ye. The oldest attestations of youse are from 19th century Irish English; according to Hickey (2003:347, 350-351) the form was probably created in Ireland by non-native speakers of English. Today, youse is found in many varieties of English, for instance in Irish-influenced dialects in Scotland and England, and also in South Africa, Zimbabwe, Australia and New Zealand, which are believed to be influenced by Irish English too (see, again, Wright 1997 and Hickey 2003 for discussion). Two forms commonly associated with the United States are y’all , a southern form, and you’uns (or yins), which is used, for instance, in Pennsylvania. The most obvious etymology for y’all , which is thought to be of African American origin, seems to be that it is a reduced form of you all, another 2pl. expression attested extensively in the southern USA (cf. Lipski 1993:45; see also Hickey 2003:360-361 for arguments). According to Montgomery (1992), however, y’all does not

derive from you and all, but from ye and all, a combination frequently found in Irish sources. A similar story may hold for you’uns (< ‘you + ones’). You’uns is thought to be of American origin (Wright 1997:176), but the tendency to attach ’un ‘one’ to pronouns and adjectives was also found in Scots (Hickey 2003:357). Whatever be the precise source, it is believed that both the rise of y’all and you’uns was stimulated by the presence of a separate 2pl. pronoun unu (or una) in Gullah, Caribbean Creoles or other varieties of English that were spoken by the black slaves in the south (Wright 1997:175). Unu/una is part of the African substrate in these varieties (cf. the use of una in, for instance, Nigerian Pidgin; see Faraclas 2004:850). Other forms found in Pidgins and Creoles are aayu (< ‘all of you’) in the Caribbean (Acetto 2004:446), yupela (or the dual yutupela, < ‘you (+ two) + fellow’) in Tok Pisin (cf. Hickey 2003:363), the related forms iufala, iutufala and iutrifala in Solomon Islands Pidgin (Jourdan 2004:707), and youfla and you mob in Aboriginal English (Burridge 2004:1117).

Strikingly, most innovative forms appear to have emerged in contact situations. This is not to say that there are no interesting forms that have emerged in native varieties. Apart from you all, the linguistic literature mentions you lot, you ... together and you guys. Unfortunately, serious linguistic research on the etymology and the history of these forms seems to be lacking, but they appear to be quite young. You lot and you ... together are British forms (cf. Trudgill 2004:147). You guys is used in North America. It may derive from a vocative NP. As the parts of these compound forms are still clearly recognizable, it is probably not accurate to consider them pronouns in their own right, although at least you guys has certainly grammaticalized to some extent. One argument for this is the fact that you guys has become a gender-neutral form for many North American speakers of English. It is well-known that the use of you guys is spreading rapidly over the United States (Maynor 1996, Tillery et al. 2000).

It is tempting to see the innovative 2pl. forms as a result of speakers’ attempts to re-introduce the number distinction in the second person, thereby, in Croft’s (2000) terms, filling a gap in the pronominal paradigm. This is indeed implied in the accounts by Wright (1997) and Hickey (2003). Lass (1997:359-360) hypothesizes that there are three possible scenarios with which functional pressures can be implemented in language change: precognition (‘prophylaxis’), repair (‘therapy’), and selective variation control. In prophylaxis, speakers anticipate on intelligibility problems by blocking any changes that might lead to homonymy. In contrast, therapy restores distinctions that have gone lost. Finally, selective variation control implies that speakers tend to select user-friendly variants more often. For instance, speakers may be more reluctant to implement changes in environments in which they cause homonymy, leading to a higher probability that a conservative variant will be preserved.

Specifically with respect to plural pronouns, the facts discussed above clearly argue against overly simplistic functional accounts, such as accounts stipulating that, in Lass’ (1997) terms, a sort of ‘therapy’ is observed here. Indeed there are at least three arguments against an analysis of these forms as the mere product of language users’ attempts to fill a gap in the English pronominal paradigm. The first argument against such an analysis is that the majority of 2pl. forms appears to have originated in contact situations. Some of the very widespread forms, e.g. youse, y’all and you’uns are even claimed to have emerged as the result of, in Thomson & Longacre’s (1985:205) terms, “a tendency on the part of bilinguals to create patterns in one of their languages which are structurally parallel to those found in the other.” Hence, as Lass (1997:360) points out, native speakers of English seem to manage very well without a special 2pl. pronoun. The second argument is related to this: a considerable time gap is observed between the loss of the old 2sg. form and the rise of the new 2pl. form. Indeed both Wright (1997:181-182) and Hickey (2003:357-359) mention periods in which no attempts are found to disambiguate between the second person singular or plural, neither with innovative pronouns, nor with syntagmatic sequences whatsoever, again indicating that horizontal homophony in the second person is far from a powerful trigger for language

change. Finally, a number of strategies to compose a special 2pl. form are also used to form other elements. For instance, apart from y’all/you all, forms are found such as we-all/we-all’s, who-all or what-all (Howe 1996:174), and even unuaal (= 2pl. unu + all; Wright 1997:175). As a counterpart for you’uns, Wright (1997:176) mentions the existence of we-uns. Less grammaticalised forms such as you guys or you lot co-occur with them guys or them lot.

Lass (1997:361), however, takes an even stronger position against functional accounts: indeed he deems not only ‘therapy’ to be an implausible scenario, but the other two scenarios as well. Croft (2000:69-70) too denies that functional considerations have played a role in the rise of new 2pl. forms in English, and he provides an alternative explanation: in his view, the innovative 2pl. forms are examples of grammaticalisation, which is said to be typical for all ‘grammatical’ expressions, especially highly frequent ones. This explanation suffers from drawbacks, however. First of all, there is no evidence for stating that non-grammaticalised expressions of 2pl. (you all, you ones,...) are particularly frequent, or more frequent than similar expressions in other functions (we all, we ones,...). In addition, denying the role of horizontal homophony leaves unexplained why neither of the non-2pl. compounds have diffused as successfully as the 2pl. compounds did (see, e.g. Maynor 1996 and Tillery et al. 2000 for recent data concerning the increasing use of 2pl. pronouns in the United States). Hence, despite the fact that horizontal homophony does not seem to have triggered the emergence of the innovative 2pl. pronouns, it seems hard to imagine that horizontal homophony has played no role whatsoever in the developments observed in the English pronominal paradigm. 2.3. Disambiguation in Dutch: jullie and dialectal variants In Standard Dutch, as in English, the originally second-person plural pronoun (Middle Dutch ghy, nowadays jij or gij ‘you’) has replaced the former second-person singular pronoun (in Dutch, du). Most dialects behave like Standard Dutch in this respect, although some eastern and northern dialects have indeed preserved du as the second person singular pronoun. Many dialects have developed a new second person plural pronoun, such as jullie, the Standard Dutch 2pl. pronoun. The precise etymology of jullie is a matter of debate (see Goossens 1994:160-161; Vermaas 2002:43; Barbiers et al. 2005:26 and De Vogelaer 2008:184-187, 206-209 for discussion): the first element may derive either from the 2sg. subject pronoun jij or the object pronoun jou; the second element from lie (< liede(n) ‘men’) or lui (< lude(n), which is in turn a phonological variant of liede(n)). Standard Dutch jullie is a strong pronoun: it can carry stress and can be used in non-clitic positions. Unlike other strong pronouns, jullie does not have a weak counterpart.

In the dialects, many morphological variants of 2pl. jullie are found, such as gieder, gulder, julder, etc. (‘you (plural)’). The Syntactic Atlas of Dutch Dialects (Barbiers et al. 2005: map 46b) distinguishes no less than 14 different forms in the Dutch language area, not even taking into account phonological differences. Unlike in English, however, these forms appear to derive from a single source: like jullie, they are all original compounds of the 2sg. pronoun combined with the noun lieden (‘men’) or a morphological variant. Most of the variation in the present-day dialects originated later on, as a result of different processes. First, the original compounds have undergone reduction, leading to divergence (e.g., ghi lieden > gullie, gulle). In addition, all dialects have severely reduced the number of case forms of the lieden-compounds, but there are differences as to the forms that have been preserved, which has been an important source of divergence as well (e.g. forms such as gulder derive from the genitive ghi lieder, and differ significantly from former nominatives or accusatives). Importantly, no weak forms have emerged of the compounds. Rather, the original simplex

variants remain in use. Thus, the strong 2pl. pronoun gullie has ge as its weak counterpart.3 Diachronic data indicate that the compounds on lieden have spread from the southwest (i.e. the Belgian provinces of East and West Flanders). Historical sources provide instances of compounds with lieden from this region as early as the 13th century (Van Loey 1958, Goossens 2000), an era in which the southwest was a dominant economic region, and a region where intensive contact with French took place. Strikingly, unlike in English the Dutch compounds start appearing in the sources in a time when the ‘old’ 2sg. pronoun du was still used, albeit rarely in some regions (see Aalberse 2009:207 for figures for the 13th century).

For Dutch too, there are a few arguments against the assumption that the mere emergence of these innovative pronouns is triggered by a tendency in language users to avoid homophony between the second person singular and plural, i.e. against an account suggesting that some sort of ‘therapy’ is observed. Van Loey (1958) suggests that the compounds were calqued on the French reinforced pronouns nous autres ‘we’ and vous autres ‘you (plural)’, which were certainly used by the French-speaking elite in 13th century Flanders.4 Significantly, the oldest instances of compounds with lieden include all grammatical persons, the first, the second and the third. The third person resembles the second person ones in that the traditional simplex pronouns were ambiguous between a singular and a plural reading in some grammatical cases in Middle Dutch, and even in present-day Standard Dutch, the subject pronoun zij may refer both to the third person singular feminine and the third person plural. In the first person plural, however, the compounds with lieden clearly did not have a disambiguating function: in the first person there has never been the slightest chance of ambiguity between the singular pronoun ik ‘I’ and the plural pronoun wij ‘we’.

Thus a straightforwardly functional explanation in terms of therapy does not account for the data. But an alternative explanation along the lines of Croft (2000) is not very convincing either. In Croft’s opinion, the liability of 2pl. compound pronouns to grammaticalise is due to their usage frequency. The fact that in Dutch the 2pl. innovative pronoun jullie has become standardised allows to test Croft’s (2000) hypothesis with corpora. In these Standard Dutch corpora, however, the second person plural pronoun does not rank among the most frequent items, since most of the references with pronouns are taken care of by weak forms, which do not show a tendency towards disambiguation.5 In the Spoken Dutch Corpus, a corpus of roughly 10 million words, jullie is found 5229 times, a figure collapsing subject, oblique and possessive usages of the form (with this score, it ranks 190th on the frequency list). Thus the second person plural appears the most infrequent plural person in spoken language, far less frequent than the first person plural with over 25000 strong

3 The pronoun gullie is a non-standard form used in the south of the Dutch language area. Both strong gullie and weak ge combine with an inflectional -t, and in some dialects both pronouns can be combined in a clitic doubling construction (e.g. ge zijt gullie ziek ‘you are you (plural) ill’). Thus weak ge can safely be considered a second person plural pronoun, which is identical to the second person singular weak form. In Standard Dutch, the situation is more complex. The weak equivalent to Standard Dutch jullie is je. Both pronouns combine with different inflectional endings, however: jullie combines with an inflectional -n whereas je still triggers the traditional -t ending on the verb. Thus it appears more accurate to consider je a ‘general’ second person pronoun in stead of a second person plural form. 4 Something similar seems to have happened in the Walser German dialects that are spoken in Northern Italy. Dal Negro (2004:161) attests instances of pronouns such as wir andre ‘we’ and ir andre ‘you (plural)’, which are also considered calques of French nous autres and vous autres. The Walser German situation does not provide a complete parallel with the Flemish one, however, since the second part of the Walser German compounds relates much more obviously to the French pronouns with autres than in the Dutch case, and, like in French, a third person plural compound seems to be lacking in Walser German but not in Dutch. 5 To provide some figures: 2sg./pl. je ranks 16th (108400 tokens), 3sg.fem./pl. ze 24th (68568 tokens), and 1pl. we 39th (47949 tokens). Of these forms, je is used in subject and oblique function and as a possessive, ze as a subject and in oblique function, and we only as a subject.

pronouns and the third person plural with some 12000 strong pronouns.6 Given the simultaneous emergence of lieden compounds in the first, second and third person plural, Croft’s (2000) frequency driven grammaticalisation scenario implies that, in Dutch, the innovative pronouns should form most easily in the first person plural and least easily in the second person plural. In reality, the opposite situation is observed.

Some crucial evidence for a proper understanding of the relevant changes comes from dialect geography. De Vogelaer (2006) discusses the geographical distribution of the Dutch compound pronouns: in the contemporary dialects they are found well outside the borders of the Belgian provinces of East and West Flanders. Strikingly, the area in which compounds are found, differs for each grammatical person. This is, in a way, unexpected: given that the compounds on lieden originate in one single region and simultaneously for the different grammatical persons (Van Loey 1958), it might be expected that they have experienced similar if not identical histories. But this is only partly the case: after having emerged in the southwest, the Dutch compound pronouns have diffused over the rest of the language area, but there are significant differences as to the success with which this process of diffusion has taken place. The structure of the pronominal paradigm seems to have played an important role: the compounds are indeed much more widespread in the grammatical persons in which they disambiguate between the singular and the plural. They are quite generally found in the Dutch language area in the second person, where they disambiguate between singular and plural readings of gij and jij . The compound pronouns have also diffused rather successfully in the third person, where forms such as zij ‘they/she’, haar ‘their/his/her’ and hem ‘their/him/her’ displayed horizontal homophony.7 The fact that 3pl. compounds are less widespread than their 2pl. counterparts, is in line with the Horizontal Homophony Hierarchy, which captures the fact that ambiguity is tolerated less well in the second person than in the third.8 The distribution of the compound pronouns is minimal in the first person, where the traditional, non-compound pronoun wij ‘we’ is never homophonous to the singular pronoun ik ‘I’. Thus the geographical distribution of the Dutch compounds clearly relates to their user-friendliness, more specifically to the extent to which the relevant pronouns have a disambiguating function. This can be interpreted as evidence in favour of a scenario of ‘selective variation control’, in which speakers favour non-ambiguous forms over ambiguous ones, which over time gives rise to different degrees of diffusion for different forms.

Another piece of dialect geographical evidence strengthens this view. Significantly, the distribution of 2pl. jullie and dialectal variants also relates to the geographical distribution of the older 2sg. pronoun du. In present-day dialects, 2sg. du is restricted to the northern and

6 The first person plural pronouns include 14953 times subject wij ‘we’, 8979 times oblique/possessive ons ‘us/our.neuter’ and 3755 times possessive onze ‘our.common’. The third person plural pronouns are zij ‘she/they’ (7764 tokens of which 3691 are third person plural), hen ‘them’ (1444 tokens) and hun ‘them/their’ (7183 tokens). Hence, even if one counts individual pronouns rather than grammatical persons, alternatives to wij (e.g. wijlie) and zij (e.g., zijlie) are expected to grammaticalise more easily than jullie. 7 In the case of zij ‘they/she’, the homophony is still found in present-day Standard Dutch. The pronouns haar ‘their/his/her’ and hem ‘their/her/him’ have lost their plural meaning in Standard Dutch and have also specialized for gender, haar ‘her’ being a feminine form and hem ‘him’ a masculine one. 8 This statement even holds if the fact is taken into account that Standard Dutch has a compound in 2pl. but not in 3pl. This Standard Dutch 2pl. compound, jullie, is mainly found in the central, Hollandic dialect area on the one hand, and in the north eastern provinces of Drenthe and Groningen on the other hand. In the former area jullie is part of the authentic dialects; it co-occurs with non-standard compounds such as 3pl. zullie. Hence in these dialects the use of jullie cannot be explained as a standardisation effect. In the latter area, the north eastern dialects, standardisation very likely does play a role. But if one disregards the instances of jullie which are due to standardisation, the distributional differences remain. Indeed large areas in the east of the Netherlands have a 2pl. compound at their disposal different from jullie, without having a compound form in 3pl. This confirms that the distributional differences between 2pl. compounds and 3pl. compounds relate to the different place of the second and the third person on Cysouw’s (2003) Horizontal Homophony Hierarchy.

south eastern periphery of the Dutch language area. Precisely in these areas, the 2pl. compounds are hardly used (see Goossens 1994:158, De Vogelaer & Coussé 2008:32). Hence the distribution of the 2pl. compounds over the Dutch language area is strongly determined by the presence or absence of horizontal homophony: almost all dialects in which the erstwhile plural pronoun jij or gij has replaced 2sg. du have adopted a 2pl. compound, whereas almost none of the dialects with du have done so. Here too, the diffusion of an innovative pronoun over the Dutch language seems to be motivated by the structure of the pronominal paradigm. 2.4. The history of unambiguous plural pronouns: a hypothesis To sum up, English and Dutch resemble each other in having developed new second person pronouns which unambiguously have plural number. At first sight, these developments appear quite different. For instance, the use of the English innovative 2pl. pronouns is more or less confined to non-standard varieties, whereas one of the new 2pl. forms in Dutch, viz. jullie ‘you (plural)’, has infiltrated in the standard language. In addition, the English 2pl. pronouns have emerged relatively recently when compared to their Dutch counterparts. Thus, in English there has been a long period of time in which the pronoun you was used to refer both to the second person singular and plural, which in many varieties continues to this very day. In Dutch, 2pl. compounds such as ghy lieden ‘you (plural)’ are attested from eras in which du was still (sometimes) used as the 2sg. form, and jij /gij was a 2pl. pronoun that could also be used as a honorific.

On a more abstract level, however, there appear to be similarities in the motivations behind the relevant changes, and in the way the changes have been implemented. Indeed in both cases, it is hard to see a direct causal relationship between horizontal homophony and the mere emergence of the innovative 2pl. forms, as implied in the type of accounts that Lass (1997) labels ‘therapy’: (complete) horizontal homophony does not seem to be a necessary condition for the emergence of new 2pl. pronouns (cf. Dutch), nor does it seem to be a sufficient one (cf. English). In the regions in which the innovative pronouns originate, the processes responsible for the formation of these pronouns appear to be operating on non-ambiguous pronouns as well. Indeed the use of formatives such as English all and uns ‘ones’ and Dutch lieden ‘men’ extends to non-ambiguous elements, such as the 1pl. pronoun (e.g. we-all and we-uns in non-standard English, wullie (< wy lieden) in non-standard Dutch). Significantly, however, disambiguating forms such as 2pl. y’all , you lot in English, 2pl. jullie, gullie in Dutch and 3pl. zij lieden in Dutch have more successfully diffused.

These similarities call for similar accounts, and more specifically for a scenario in which not the emergence but the diffusion of these forms is influenced by functional considerations. It seems indeed hard to explain the fact that in both languages the 2pl. compounds have spread over much larger cohorts of speakers than the other pronouns with all, ones, etc. in English or lieden in Dutch without making reference to the fact that the former are very useful from a functional point of view. Hence, if not the emergence, at least the diffusion of the innovative 2pl. forms seems driven by functional considerations on the language users’ part. Thus, it is very well possible that, again using a term from Lass (1997), a case of ‘selective variation control’ is witnessed here, in which growing numbers of speakers start using a non-ambiguous form in ever increasing circumstances. Section 3 tests whether this is indeed a plausible scenario. 3. Diffusion in Dutch: diachronic data 3.1. Data gathering Basically, the scenario laid out in section 2 for the rise of unambiguous pronouns implies that language users, over the course of time, prefer unambiguous pronouns over ambiguous ones, leading to a gradual shift to the unambiguous forms. This claim was predominantly based on

contemporary or at least recent data. In order to further underpin this scenario, the evidence given in section 2 needs to be supplemented with diachronic data, which should allow to locate the actual locus of language change more precisely. This is the goal of section 3, where diachronic data are discussed concerning the development of non-ambiguous pronouns in Dutch. Unlike in English, the use of non-ambiguous forms, both in the second and the third person plural, traces back to the Middle Ages. Since texts were written in regional varieties of Dutch in that era, it should be possible to document the rise of non-ambiguous forms. The investigation takes a quantitative perspective: it compares the ratio of ambiguous vs. non-ambiguous plural pronouns. Importantly, the investigation only takes strong pronouns into consideration: while the history of Dutch has witnessed the introduction of a series of unambiguous strong pronouns, no such development has taken place in the weak pronouns. The important geographical variation with respect to the phenomenon under scrutiny imposes rather strict requirements on the data that can be used in the investigation: these data need to come from texts which can be located precisely. Ideally, the relevant texts can also be dated quite accurately. We have extracted all strong pronouns from a corpus of historical legal texts that meets these requirements, viz. the Compilation Corpus Historical Dutch (Coussé, to appear). The subcorpus of legal texts spans from the middle of the thirteenth century until the end of the eighteenth century, covering the whole Middle Dutch period and the beginning of the Early Modern Dutch period. The texts were selected evenly from the dialect regions of Flanders, Brabant and Holland, which were influential cultural and political centres in the medieval period. The corpus search yielded 6819 strong pronouns in total.

While contemporary data especially revealed interesting developments in second person pronouns, in the historical data only third person plural pronouns occur frequently enough in the corpus to warrant a quantitative investigation (n=3752). Thus this section focuses on the Dutch third person plural. Since extensive homonymy is observed in the third person in Dutch, the rise of third person plural pronouns does present a valuable case to document pronominal change. Section 3.2, 3.3 and 3.4 subsequently address subject pronouns, object pronouns and possessives. In section 3.5, the question will be raised to what extent the changes discussed in these sections are representative for the other grammatical persons. 3.2. Subject pronouns In our corpus, 1629 instances are found of third person plural pronouns in subject function. The most frequent form is zij ‘they’ (and orthographic variants). In addition, compounds are used of zij ‘they’ and the noun lieden ‘men’. The latter display some formal variation: Flemish dialects typically use zijlieden whereas Hollandic dialects have zijluiden. Table 1 shows the proportion of simplex zij ‘they’ vs. compounds with lieden ‘men’. In general, very few compounds are found: the proportion of compounds never exceeds 15% in Flanders and Holland, and Brabant seems to use no compounds at all. Table 1. Subject pronouns in Flanders, Brabant and Holland, 13th-18th century

FLANDERS BRABANT HOLLAND

zij compound with lieden

zij compound with lieden

zij compound with lieden

13th C. 257 22 76

100% 100% 100%

14th C. 85 1 60 86

98,8% 1,2% 100% 100%

15th C. 125 52 228

100% 100% 100%

16th C. 106 7 7 217 25

93,8% 6,2% 100% 89,7% 10,3%

17th C. 44 3 21 98 13

93,6% 6,4% 100% 88,3% 11,7%

18th C. 42 9 43 2

100% 100% 95,6% 4,4% These data do not lead one to believe that compounds with lieden ‘men’ are indeed diffused successfully in the third person. There is a slight increase in the use of compounds in the 16th and 17th century, but this increase is far from impressive, and in the 18th century the compounds more or less disappear from the data.

The question should be addressed to what extent these data reflect actual language usage in the course of history. While it is commonly assumed that older texts closely resemble the dialect spoken in a certain region, there are some indications that in this case, they do not: first of all, Van Loey (1958) and Goossens (2000) provide 13th century instances of compounds with lieden ‘men’ whereas these forms do not show up in our corpus. This is especially surprising for Flanders, since the compounds are believed to have originated in this region. Thus, it is likely that our corpus data underestimate the use of the compounds in the older periods in the table. For the more recent centuries, it is insightful to compare the data from the written sources with contemporary dialect data, such as the ones gathered for the Syntactic Atlas of Dutch Dialects, vol. 1 (Barbiers et al. 2005, map 47b) and the Morphological Atlas of Dutch Dialects, vol. 2 (Goeman et al. 2008, map 41). The atlas maps show that compounds with lieden are nowadays used in large parts of the Dutch language area. Flanders, being the region where the pronouns have originated, shows the most productive usage, followed by Brabant and Holland, where compounds are still found in the majority of the sampling points.9 This casts serious doubt over the results in table 1: first, it seems unlikely that the compounds with lieden, after having emerged in the Middle Ages, have disappeared from language usage to become pervasive in contemporary dialects. And second, the fact that Holland is the region where most of the compounds are found in the corpus does not correspond to the contemporary picture either, in which Holland uses less compounds than Flanders and Brabant. Yet in these regions, substantially less or even no instances are found of pronouns such as zijlieden or zijluiden. At least from the sixteenth century onwards, this may be due to the development of a written standard, from which pronouns such as zij lieden apparently were excluded, although they were certainly used in spoken Dutch (cf. the dialect data). Thus the dialect data do lead to believe that the compounds with lieden ‘men’ have diffused over the Dutch language area, but this process of diffusion is not visible in the written sources of Dutch. While it is not uncommon for written language to lack phenomena productively used in colloquial speech, the absence of compound pronouns is rather surprising in this case since compounds do indeed appear in oblique function and in possessives (cf. below). We have as yet no satisfactory explanation for this discrepancy. 3.3. Oblique pronouns

9 In most sampling points in Holland, the compound is a form not attested in the historical corpora, typically hullie or hunnie. These forms are compounds of oblique hun ‘them’ which have been extended to subject function (De Rooij 1990). During the 20th century another non-ambiguous pronoun was introduced in subject function, viz. the simplex hun ‘them’. This form has diffused rapidly in recent decades (see Van Hout 1989:235-237 for discussion), which might be due to the fact that it ‘solved’ the homophony between 3pl. zij ‘they’ and 3sg. feminine zij ‘she’. Thus, in large parts of the Dutch language area it is common to use former 3pl. oblique pronouns as subjects, e.g. in sentences such as hullie komen ‘they (lit. ‘them’) come’ or hun hebben gelijk ‘they (lit. ‘them’) are right’.

In our corpus, 1026 strong oblique pronouns were found. About half of these forms come from the Flemish texts in the corpus. In Flanders, four types of forms are used throughout the centuries under investigation, viz. simplex hem and haar, compounds of these pronouns with lieden ‘men’ (e.g. hemlieden, haarlieder), plural hen/hun, and compounds of hen/hun with lieden ‘men’.10 Unlike for the subject pronouns, compounds with lieden ‘men’ are already clearly present in the 13th century texts in the corpus. They gain in popularity until the 17th C., thereby pushing the simplex forms hem and haar towards the margins, at least in their usage as plural pronouns. From the 17th century onwards, a shift is witnessed to hen/hun, the forms used in present-day Standard Dutch. Interestingly, the 17th century also shows forms such as henlieden and hunlieden which appear to be contaminations of hen/hun and the compounds with lieden. It is likely that the compounds with lieden ‘men’ had undergone significant phonological reduction by the 17th century, leading to forms resembling the present-day compounds with lieden ‘men’ (e.g. hullie, hulder, hunder) rather than unreduced hem lieden and haar lieden (see, for instance, Kloeke 1941:169). Hence the use of henlieden and hunlieden may be due to writers wrongfully linking reduced forms to compounds with hen and hun rather than with hem and haar, two forms which had assumed a purely singular meaning by the 17th century. Table 2. Oblique pronouns in Flanders, 13th-18th century

Flanders traditional hem/haar

compound with lieden

hen/hun hen/hun +

lieden 13th C. 110 16

87,3% 12,7% 14th C. 40 25

61,5% 38,5% 15th C. 21 68

23,6% 76,4% 16th C. 8 127 2

5,8% 92,7% 1,5% 17th C. 1 19 10 25

1,8% 34,5% 18,2% 45,5% 18th C. 6 41

12,8% 87,2% In Brabant, the data are more sparse. Nevertheless, it seems warranted to conclude that compounds with lieden ‘men’ are hardly used as a means to disambiguate between singular and plural interpretations of hem and haar, at least not in texts.11 In stead, Brabantic texts make ample use of the pronouns hen and hun, which appear to be Brabantic innovations. It is unclear how these forms have originated. While they allegedly derive from hem ‘him/them’ (Van Bree 1987:256, Howe 1996:206), the fact that these forms are not used in the third person singular, unlike hem, reveals that they have emerged as plural pronouns. At least from the 14th century onwards, hen and hun gradually replace plural usages of hem and haar, a process reaching completion in the 16th century. Table 3. Oblique pronouns in Brabant, 13th-18th century 10 It is at present still unclear from which cases these forms originate: according to Van Bree (1987:256), hen developed phonologically out of dative hem. With respect to haar, Howe (1996:206) assumes that analogy with the accusative singular has played a role. Neither of these accounts explains why the given forms are only introduced in one specific grammatical case. Already in Middle Dutch, however, both hen and haar were used in environments in which dative and genitive were normally not used. 11 Dialect atlases show that compounds with lieden ‘men’ do occur in present-day dialects, making it likely that they were also used in historical varieties of the language, especially in more recent centuries.

Brabant traditional hem/haar

compound with lieden

hen/hun hen/hun +

lieden 13th C. 3

100,0% 14th C. 9 8

52,9% 47,1% 15th C. 5 17

22,7% 77,3% 16th C. 13

100,0% 17th C. 1 28

3,4% 96,6% 18th C. 13

100,0% The third area under investigation, Holland, displays similarities with both Flanders and Brabant. While showing a clear preference for simplex hem and haar in the 13th century, the 14th-16th century witness a tendency to take over the compounds with lieden ‘men’ that had become the dominant device in reference to third person plural subjects in Flanders. From the 15th century onwards, the Brabantic forms hen and hun are infiltrating. These forms become the dominant pronouns in the 17th century. As in Flanders, small proportions are found of compounds of hen/hun with lieden ‘men’. Table 4. Oblique pronouns in Holland, 13th-18th century

Holland traditional hem/haar

compound with lieden

hen/hun hen/hun +

lieden 13th C. 34

100,0% 14th C. 44 1

97,8% 2,2% 15th C. 57 13 8 2

71,3% 16,3% 10,0% 2,5% 16th C. 62 38 17 5

50,8% 31,1% 13,9% 4,1% 17th C. 18 5 46 7

23,7% 6,6% 60,5% 9,2% 18th C. 17 36

32,1% 67,9% Figure 1 provides an overview of the proportion of non-ambiguous pronouns used between the 13th and the 18th century, in Flanders, Brabant and Holland. In all three of the regions, there is a continuous move towards more frequent usage of non-ambiguous forms. Flanders and Brabant are ahead in this development. The very introduction of non-ambiguous pronouns in Flanders appears to have taken place before the 13th century, and hence this part of the process cannot be documented with written sources. In Brabant, it is unclear when non-ambiguous forms emerged, since the oldest texts only yield three instances of 3pl. oblique pronouns. Holland lags behind: on average, it takes Holland two to three centuries longer to reach the same proportions of non-ambiguous pronominal reference as Flemish and Brabantic varieties of Dutch. Figure 1. Oblique pronouns in three dialect areas, 13th-18th century

Different (types of) pronouns are responsible for the increase in the usage of unambiguous forms: compounds with lieden ‘men’ in Flanders, hen/hun in Brabant, and both types of pronouns in Holland. From the 17th century onwards, hen/hun begins to replace the compounds with lieden ‘men’ as the 3pl. oblique pronoun. Given that hen/hun only exists marginally in Flemish dialects, its infiltration in the Flemish part of the corpus reflects a tendency towards standardisation in the written language, which inevitably involves a shift towards Hollandic variants, since already in that era Holland was the centre of the Dutch language area. But by that time the different non-ambiguous pronouns were already well established in language usage. Thus, it appears as if the increase of non-ambiguous forms can be interpreted as an independent development in the regions under investigation, which provides a strong indication for a functional motivation behind pronominal change. 3.4. Possessives In the corpus, 1097 instances occur of strong, third person plural possessive pronouns, most of them from the areas Flanders and Holland. In Flanders, possessives by and large behave identical as oblique pronouns: compounds with lieden ‘men’ exist alongside the simplex pronoun haar from the 13th century onwards, but it takes several centuries, more precisely until the 16th century, before these compounds become the dominant device to refer to the third person plural. From the 17th century, the compounds are replaced by hun, probably as a result of the standardisation of the written language. Again, a few compounds are found of hun and lieden ‘men’. Table 5. Possessive pronouns in Flanders, 13th-18th century

Flanders traditional haar

compound with lieden

hun hun + lieden

13th C. 2 6 25,0% 75,0%

14th C. 11 7 61,1% 38,9%

15th C. 56 71 44,1% 55,9%

16th C. 56 155 26,5% 73,5%

17th C. 1 55 9 6 1,4% 77,5% 12,7% 8,5%

18th C. 5 34 12,8% 87,2%

For Brabant the data are rather sparse, but nevertheless reveal a clear tendency to dispense with the traditional, ambiguous pronoun haar. Minor proportions are observed of compounds with lieden ‘men’. The most important alternative for haar is hun, which becomes the dominant form in the 17th century. As in Flanders, a few compounds are used of hun and lieden. Table 6. Possessive pronouns in Brabant, 13th-18th century

Brabant traditional haar

compound with lieden

hun hun + lieden

13th C. 3 1 75,0% 25,0%

14th C. 13 4 76,5% 23,5%

15th C. 28 1 96,6% 3,4%

16th C. 7 2 1 2 58,3% 16,7% 8,3% 16,7%

17th C. 1 20 4,8% 95,2%

18th C. 2 13 13,3% 86,7%

Holland shows a similar picture, but it lags behind a few centuries in comparison to Flanders and Brabant: haar is replaced, initially with compounds with lieden ‘men’, but in the long run the pronoun hun prevails. The shift from haar to hun does not reach completion before the 19th century. Table 7. Possessive pronouns in Holland, 13th-18th century

Holland traditional haar

compound with lieden hun

hun + lieden

13th C. 7 100,0%

14th C.

15th C. 190 100,0%

16th C. 172 19 1 89,6% 9,9% 0,5%

17th C. 53 6 12 74,6% 8,5% 16,9%

18th C. 29 1 35 44,6% 1,5% 53,8%

Figure 2 summarizes the developments in the three areas by plotting the proportion of unambiguous pronouns vis-à-vis the traditional form haar. No data are included for the 13th century since in neither of the regions more than 10 possessives are attested in the corpus. Figure 2. Possessive pronouns in three dialect areas, 13th-18th century

Flanders is clearly leading. In contrast to the oblique pronouns, Brabantic possessives maintain their ambiguity a bit longer. Apparently the third person plural pronouns hen and hun first gained firm ground as oblique forms before the usage of hun was extended towards the possessive function. Holland is the slowest region to adopt non-ambiguous forms. Also, Holland takes over Flemish and Brabantic forms rather than developing its own non-ambiguous forms. As for the oblique forms, the shift towards non-ambiguous forms involves several pronouns. Thus figure 1 supports the conclusion from section 3.3, viz. there is a functional motivation behind the pronominal change observed in the history of Dutch. 3.5. Overview In sum, highly similar developments are observed in oblique pronouns and in possessives. Given that the oldest stages of the shift towards non-ambiguous pronouns are not documented, no insight can be gained from historical records in the motives behind the mere emergence of the relevant forms. It seems unlikely, however, that disambiguation has played a role there (see section 2 for argumentation). In the course of history the unambiguous strong forms gradually replace the traditional forms. Figure 3 shows how this process has taken effect; oblique and possessive pronouns are pooled. Flanders and Brabant have more or less simultaneously shifted towards non-ambiguous forms. At least in the first stages, different pronouns are involved in this process, so the developments are essentially independent of each other. From the 17th century onwards, Brabantic forms such as hen and hun become dominant. Holland has taken over the non-ambiguous forms originating in Flanders and Brabant, and follows the same pathway as did the other two regions, with a delay of two to three centuries. Figure 3. Proportion of non-ambiguous oblique and possessive pronouns, 13th-18th century

Figure 3 shows that the replacement process starts in a slow fashion (cf. especially the curve for Holland), subsequently speeds up, and slows down when it reaches completion. Thus, when plotted on a time axis, the replacement process patterns like many other instances of linguistic change in showing an S-shaped curve (see Bailey 1973, Kroch 1989, Denison 2003 for a discussion of S-curves). This entails that the shift towards non-ambiguous pronouns is a gradual one, not only in space (cf. section 2), but also in time, in which subsequent generations of language users, independently in different regions, overuse unambiguous pronouns in comparison to previous generations.

The current data situation for Dutch does not allow to draw similar graphs for second and first person pronouns: these pronouns are too rare in the Compilation Corpus Historical Dutch, and other existing corpora do not locate and/or date the included texts with enough precision. However, detailed descriptions of 13th century Dutch and of the distribution of pronouns in contemporary dialects provide enough information to be able to check whether similar developments as for the third person plural are conceivable there. Both in the second and in the first person plural, the relevant regions only use compounds with lieden ‘men’ as alternatives to the original simplex pronouns. In the first person plural, compounds such as wij lieden ‘we’ and ons lieden ‘us/our’ were already used in the 13th century, in Flanders (Van Loey 1958). While ons lieden ‘us/our’ never diffused from the region and eventually was lost in Flanders as well (Goossens 2000), wij lieden ‘we’ did spread to Brabantic. In Holland, however, wij lieden is not used. Thus, generally speaking, the first person plural compounds have been spread much less succesfully over the Dutch language area. The second person shows a completely different picture (cf. supra): here too, the forms are believed to originate in Flanders, in the 13th century or even earlier, but the compounds deriving from 2pl. jij lieden and gij lieden (e.g., jelui, julder, gulder, gieder, etc.) are covering the larger part of the Dutch language area, including areas where no other compounds with lieden ‘men’ are used (e.g. the east and the north of the Netherlands). In addition, the 2pl. pronoun jullie is the only compound to have assumed standard status. Thus the compounds with lieden ‘men’ reveal how the historical process of pronominal replacement has taken place. The fact that they originate simultaneously for all grammatical persons suggests that ambiguity did not play a role in their emergence. Yet as explained in section 2.4, the different success with which they have spread can hardly be related to other factors than the presence or absence of an ‘ambiguity-solving’ effect. 4. Excursion: mechanisms of change and S-curves

Apart from the mere existence of functionally motivated change in pronominal paradigms, some remarks are in order as to the way any alleged processes of homonymy avoidance may have taken effect. The fact that the change takes several centuries to reach completion shows that we are certainly not dealing with, in Lass’ (1997) terms, a case of therapy. In fact of the three scenarios for functionally driven language change that Lass (1997) provides, only ‘selective variation control’ can logically account for the observed changes. In such a scenario, preferences on the part of speakers determine which variants will be successfull in the long run. There are, however, other conceivable scenarios than the three discussed by Lass. A particularly interesting one is provided by Labov (1994:569-599), who emphasizes the role of the hearer, more specifically during the process of language learning. According to Labov, language learning involves probability matching, i.e. learners align the frequencies with which they use certain variants to the frequencies observed in their language input. Since homonymous variants in the input tend to be misunderstood more easily than non-homonymous ones, the proportion of homonymous forms that is ‘pooled’ in the set of data from which language learners draw their frequencies, will be lower than the proportion in the language input, leading to a small decrease of the homonymous variant’s usage frequency in the language of the learner. Labov’s (1994) scenario is more mechanistic than the ones by Lass, in that it does not assume functionally motivated change to be the result of intentional speaker actions. Thus it essentially denies that language users avoid homonymy. Rather, the fact that homonymy is sometimes dispensed with, is the naturalistic outcome of the fact that homonyms tend to give rise to misunderstandings and unambiguous pronouns do not.

In Labov’s account, each generation computes a new probability to use non-ambiguous forms that differs from the probability in the previous generation. The account is precise enough to warrant a computational simulation. If it is assumed that ambiguous tokens are understood as singular forms with a proportion p, the formula to compute the proportion F2 with which innovative, unambiguous pronouns are used vis-à-vis the proportion F1 found in the previous generation, can be computed with the following formula:

100.F1 F2 = 100-(p.(100-F1))

with F1= usage frequency unambiguous variant in previous generation (in %), F2= usage frequency unambiguous variant in new generation (in %) and P= proportion of ambiguous forms misunderstood (0 < p < 1) This formula indeed yields an S-shaped curve when plotted on the time axis, and hence it seems at least theoretically plausible to capture the shift towards unambiguous pronouns in Dutch. However, S-curves remain a somewhat enigmatic concept. Denison (2003) discusses a number of unsolved questions with respect to S-curves, including alternative mathematical functions underlying them, and the nature of the actual mechanism behind S-curves. It is beyond the scope of this paper to contribute to an answer to these questions. It will merely be tried to reveal the implications of Labov’s (1994) argumentation for the present data, thereby hoping that some light can be cast over the issue whether functionally driven change is to be located in the actions of speakers or in misunderstandings on the hearer’s part. It should be noted in advance that, whatever be the outcome of the simulation, it remains difficult to rule out one of these scenarios: first, the two scenarios are not mutually exclusive. And second, since the relevant change has been completed centuries ago, we have no direct access to speakers or hearers involved in the change. To make things even more complex, at certain moments in history, variation in pronoun systems is bound to assume social significance, since pronouns typically (also) function as markers of social identity (cf. Mühlhäusler & Harré 1990). Hence the parameter ‘homonymy’ may have conspired with or been inhibited by factors such as overt and covert prestige.

Figure 4 plots five implementations of the scenario along the lines of Labov (1994) in which p is taken to be .10, .20, .25, .30 and .40, respectively. It is assumed that a complete shift boils down to a transition from 5% to 95% usage of the innovative forms.12 The graph also includes the data from figure 3. Within the time span investigated in the corpus, Flanders and Brabant evolve from some 15% unambiguous pronouns in the 13th century to 95% in the 17th century, i.e. in 400 years time; the graph includes pooled data. Holland evolves from 5% around 1400 (2,2% in the 14th century; 8,5 in the 15th) to 61% in the 18th century (i.e. +56% in some 350 years time). Figure 4. From ambiguous to non-ambiguous pronouns in a ‘Labovian’ scenario

The data for Holland on the one hand, and Flanders/Brabant on the other, roughly fall within the scenarios with p=.20 or p=.25. This means that, in order for Labov’s (1994) scenario to work, and hence for hearer misunderstandings to be the only factor involved in pronominal change, some 20 to 25% of the ambiguous forms need to be misunderstood. Although it would require a perception experiment to test whether this is a plausible figure, it seems as if this proportion is unrealistically high.13 Indeed it is likely that, in cases in which they run a

12 A note should be added about the ramifications of equaling a completed change with a shift from 5% to 95% usage of innovative forms. More specifically, especially in the early and final stages of a linguistic change, the shifts in usage frequencies between different generations become very small. For instance, when p is set at .25 and F1 at 5%, the first generation of learners ends up with F2=6,6%, which boils down to a shift of 1,6%. If F1 is lowered to 2%, F2 is 2,6%, involving a mere 0,6% shift. It remains to be seen whether language learners are indeed capable of perceiving and storing such detailed frequencies. 13 Figure 4 only contains data for oblique and possessive pronouns. Even though the corpus data do not reveal this, contemporary dialect data indicate that the relevant changes have occurred in subject pronouns as well. In subject pronouns, the fact should be taken into account that, in Dutch, verbal endings distinguish between singular and plural readings of third person subject pronouns (e.g. zij werk-t ‘she works’ vs. zij werk-en ‘they work’). This makes misunderstandings rather unlikely. Thus, a scenario along the lines of Labov (1994) does not

20-25% chance to be misunderstood, speakers will consciously shift towards an unambiguous variant. Thus the simulation carried out in figure 4 leads to believe that functionally driven language change is not simply explained by the fact that hearers sometimes misunderstand homonymous pronouns. 5. Conclusions The main goal of this paper was to show that both the English and the Dutch pronominal system provide several instances of functionally driven change. Apart from 3pl. they and 3sg.feminine she, both addressed in Samuels (1972), the data discussed in section 2 and 3 make it likely that homonymy, more precisely ‘horizontal homophony’, has played a role in the rise of innovative plural pronouns such as 2pl. y’all , yous, you guys,... in English, and 2pl. gij lieden (and modern variants including Standard Dutch jullie) and 3pl. zij lieden (and variants) in Dutch. The current data situation does not allow a complete reconstruction of the relevant diachronic developments, especially because little is known about the factors that have caused the innovative plural pronouns to emerge. Given the fact that most innovative plural pronouns are by no means generally found in the relevant language areas, horizontal homophony appears a rather weak motive for the coinage of new pronouns, at least in situations in which no language contact is observed. Their diffusion, however, clearly seems to be driven by their user-friendliness. The Dutch case leaves little doubt that homonymy avoidance plays a role in diffusion. Hence a similar account seems plausible for obviously very similar developments in two closely related languages. The scenario runs as follows: in both languages new plural pronouns emerge in contact varieties as reinforced pronouns or calques. These innovative pronouns diffuse easily wherever they fill a disturbing gap in the pronominal paradigm, viz. in places with horizontal homophony. The second person case also makes it clear that in both languages a tendency is observed to comply with the Horizontal Homophony Hierarchy. Cysouw (2003:315) speculates that, as speakers are consciously aware of the referential properties of independent words, unusual mixtures of referential properties are disfavoured. The present data show that Cysouw’s (2003) proposal is essentially correct, showing that the Horizontal Homophony Hierarchy is not just a statistical fact about the structure of pronominal paradigms, but that it has diachronic relevance as well. His hypothesis can be refined in two ways: first, unusual mixtures of referential properties are easily dispensed with rather than avoided. And second, while violations of the Horizontal Homophony Hierarchy appear to be severely restricted, the Dutch third person plural shows that the avoidance of homophony extends beyond the hierarchy. Indeed Cysouw’s (2003) hierarchy allows for the third person to be the only place where horizontal homophony is found, but still the homophony is dispensed with in Dutch. It is instructive that in Dutch, disambiguation is only observed in the series of strong pronouns: while 2pl. jij /gij changes in gij lieden/ jij lieden and variants, and 3pl. zij in zij lieden and variants, their weak counterparts je and ze do not undergo change. This is typologically common: Cysouw (2003:314) shows that homophony is more often found in reduced person markers, i.e. inflectional endings and clitics. Indeed languages almost universally not only have independent pronouns at their disposal, but also a series of such ‘reduced’ person markers (see Siewierska & Bakker 1996 and Siewierska 1999 for discussion). The independent pronouns are typically used in cases where the referent is less easily accessible in the discourse context, requiring extra referential efforts from the speaker, or in situations where the pronoun is placed in focus and/or carries stress, in which misunderstandings are particularly disturbing.

explain why an overwhelming majority of contemporary dialects uses subject pronouns such as zijlie (< zij lieden) (or even hun ‘them’, see note 9) without having adopted wijlie (< wij lieden).

Taking all these instances of pronominal change together, it appears that homonymy avoidance has played a crucial role in the processes that have shaped the pronominal paradigms of present-day varieties of English and Dutch. Given the fact that typological research as conducted by Cysouw (2003) has detected universal tendencies in the structure of pronominal paradigms, it seems likely that diachronic research in other languages will yield similar instances. Therefore, pronominal change qualifies as a clear exception to the alleged rarity of functionally motivated change in grammar (Lass 1997:359-361 and passim; Hopper & Traugott 2003:103). References Aalberse, S. (2009). Inflectional economy and politeness. Utrecht: LOT. Acetto, M. (2004). Eastern Caribbean English-derived language varieties: morphology and

syntax. In B. Kortmann, E. Schneider, et al. (Eds.), A handbook of varieties of English. Volume 2: Morphology and syntax (pp. 439-453). Berlin/New York: Mouton de Gruyter.

Bailey, C.-J. (1973). Variation and linguistic theory. Washington, DC: Center for applied linguistics.

Barbiers, S., Bennis, H., De Vogelaer, G., Devos, M., van der Ham, M. (2005). Syntactic atlas of Dutch dialects. Volume 1 (pronouns, agreement and dependencies). Amsterdam: Amsterdam University Press.

Bynon, T. (1977). Historical linguistics. Cambridge: Cambridge University Press. Burridge, K. (2004). Synopsis: morphological and syntactic variation in the Pacific and

Australasia. In B. Kortmann, E. Schneider, et al. (Eds.), A handbook of varieties of English. Volume 2: Morphology and syntax (pp.1116-1131). Berlin/New York: Mouton de Gruyter.

Coussé, E. (to appear). Een digitaal compilatiecorpus historisch Nederlands. Lexikos. Croft, W. (1990). Typology and universals. Cambridge: Cambridge University Press. Croft, W. (2000). Explaining language change: An evolutionary approach. Harlow: Longman. Cysouw, M. (2003). The paradigmatic structure of person marking. Oxford: Oxford

University Press. Dal Negro, S. (2004). The decay of a language: the case of a German dialect in the Italian

Alps. Bern: Lang. Denison, D. (2003). Log(ist)ic and simplistic S-curves. In R. Hickey (Ed.), Motives for

language change (pp.54-70). Cambridge: Cambridge University Press. De Rooij, J. (1990). Over hun en hen, en hun. Vorm en functie van de niet-gereduceerde

voorwerpsvormen van het persoonlijk voornaamwoord in de derde persoon meervoud, in standaardtaal, ouder Nederlands en dialect. Taal en Tongval, 42, 107-147.

De Vogelaer, G. (2006). Actuation, diffusion, and universals: change in the pronominal system in Dutch dialects. Zeitschrift für Dialektologie und Linguistik, 73, 259-274.

De Vogelaer, G. (2008). De Nederlandse en Friese subjectsmarkeerders: geografie, typologie en diachronie. Gent: Koninklijke Academie voor Nederlandse Taal- en Letterkunde.

De Vogelaer, G., Coussé, E. (2008). De kracht van disambiguering: nieuwe meervoudspronomina van het Middelnederlands tot nu. Taal en Tongval, Theme issue 21 (Dialectgeografie en interne factoren), 13-35.

Faraclas, N. (2004). Nigerian Pidgin English: morphology and syntax. In B. Kortmann, E. Schneider, et al. (Eds.), A handbook of varieties of English. Volume 2: Morphology and syntax (pp.828-855). Berlin/New York: Mouton de Gruyter.

Geeraerts, D. (2010). Theories of Lexical Semantics. Oxford: Oxford University Press Gilliéron, J., Roques, M. (1910). Études de géographie linguistique 12-13. Revue de

philologie française, 24, 278-307.

Goeman, T., van Oostendorp, M., van Reenen, P., Koornwinder, O., van den Berg, B., van Reenen, A. (2008). Morphological atlas of the Dutch dialects. Volume 2. Amsterdam: Amsterdam University Press.

Goossens, J. (1994). Sprachatlas des Nördlichen Rheinlands und des Südöstlichen Niederlands. Zweite Lieferung. Marburg: Elwert Verlag.

Goossens, J. (2000). Onslieden en ons leden. In S. Gillis, J. Nuyts, J. Taeldeman (Eds.), Met taal om de tuin geleid: een bundel opstellen voor Georges De Schutter ter gelegenheid van zijn pre-emeritaat (pp.187-191). Antwerpen: UIA.

Hickey, R. (2003). Rectifying a standard deficiency: Second-person pronominal distinction in varieties of English. In I. Taavitsainen, A. Jucker (Eds.), Diachronic perspectives on address term systems (pp. 345-374). Amsterdam: John Benjamins.

Hopper, P., Traugott, E. (2003). Grammaticalization [second edition]. Cambridge: Cambridge University Press.

Howe, S. (1996). The personal pronouns in the Germanic languages: a study of personal pronoun morphology and change in the Germanic languages from the first records to the present day. Berlin: de Gruyter.

Jourdan, C. (2004). Solomon Islands English: morphology and syntax. In B. Kortmann, E. Schneider, et al. (Eds.), A handbook of varieties of English. Vol. 2: Morphology and syntax (pp.702-719). Berlin/New York: Mouton de Gruyter.

Kortmann, B., Szmrecsanyi, B. (2004). Global Synopsis: Morphological and syntactic variation in English. In B. Kortmann, E. Schneider, et al. (Eds.), A handbook of varieties of English. Volume 2: Morphology and syntax (pp.1142-1202). Berlin/New York: Mouton de Gruyter.

Kloeke, G. (1941). Over ‘jullie’ en enige andere pronomina. Nieuwe Taalgids, 35, 161-170. Kroch, A. (1989). Reflexes of grammar in patterns of language change. Language Variation

and Change, 1, 199-244. Labov, W. (1994). Principles of Linguistic Change. Volume 1: Internal factors.

Oxford/Cambridge: Blackwell. Labov, W. (2001). Principles of Linguistic Change. Volume 2: Social factors.

Oxford/Cambridge: Blackwell. Lass, R. (1997). Historical linguistics and language change. Cambridge: Cambridge

University Press. Lipski, J. (1993). ‘Y’all’ in American English: From black to white, from phrase to pronoun.

English World-Wide, 14, 23-56. Maynor, N. (1996). The pronoun y'all: questions and some tentative answers. Journal of

English Linguistics, 24, 288-294. Montgomery, M. (1992). The etymology of ‘y’all’. In J. Hall, N. Doane, D. Ringler (Eds.),

Words of honor: essays presented to Frederic G. Cassidy (pp.356-369). New York: Garland.

Mühlhäusler, P., Harré, R. (1990). Pronouns and people: The linguistic construction of social and personal identity. Oxford: Basil Blackwell.

Samuels, M. (1972). Linguistic evolution, with special reference to English. Cambridge: Cambridge University Press.

Siewierska, A. (1999). From anaphoric pronoun to grammatical agreement marker: why objects don’t make it. Folia Linguistica, 33, 225-251.

Siewierska, A., Bakker, D. (1996). The distribution of subject and object agreement and word order type. Studies in Language 20(1), 115-161.

Thompson, S., Longacre, R. (1985). Adverbial clauses. In T. Shopen (Ed.), Language typology and syntactic description. Volume II (pp.171-234). Cambridge: Cambridge University Press.

Tillery, J., Wikle, T., Bailey, G. (2000). The nationalization of a Southernism. Journal of English Linguistics, 28, 280-294.

Trudgill, P. (2004). The dialect of East Anglia: morphology and syntax. In B. Kortmann, E. Schneider, et al. (Eds.), A handbook of varieties of English. Volume 2: Morphology and syntax (pp.142-153). Berlin/New York: Mouton de Gruyter.

Van Bree, C. (1987). Historische Grammatica van het Nederlands. Dordrecht: Foris Publications.

Van Hout, R. (1989). De structuur van taalvariatie: een sociolinguïstisch onderzoek naar het stadsdialect van Nijmegen. Dordrecht: Foris.

Van Hout, R. (2003). Pronominale meervoudsvorming in de Nederlandse dialecten. Leuvense Bijdragen, 92, 257-270.

Van Loey, A. (1958). Over pronomina van het type ‘wijlieden’. In W. Pée, et al. (Eds.), Album Edgard Blancquaert (pp.319-322). Tongeren : Michiels, Tongeren.

Vermaas, H. (2002). Veranderingen in de Nederlandse aanspreekvormen van de dertiende t/m de twintigste eeuw. Utrecht: LOT.

Wright, S. (1997). ‘Ah’m going for to give youse a story today’: remarks on second person plural pronouns in Englishes. In J. Cheshire, D. Stein (Eds.), Taming the vernacular: from dialect to written standard language (pp.170-184). London/New York: Longman.