Prescribed reading

The new millennium has brought with it a new academic interest in the later phases of the standardisation process - as witness the Leiden University project on the codification of English and a series of conferences elsewhere on the theme of prescriptivism.² Prescriptivism itself, until recently the factor in language-change that academic linguists preferred to deride or ignore, has become a major site of investigation. Traditional textbook assumptions – for instance, that all eighteenth-century codifiers were (a) prescriptive and (b) effective – have given way to serious enquiries into the history of the prescriptive movement in linguistic thought and the processes of prescriptive change in linguistic practice. What are the mechanisms of prescriptive language-change? how do they differ from, or interact with, mechanisms posited as system-internal or socially-driven (such as “drift” and “functional ecology” on the one hand or “accommodation” and “prestige” on the other)? These are questions now being addressed and they have raised, in turn, important questions of methodology: how do we identify, monitor or measure the impact of prescription on practice?

Some of the pioneer work in this area is being carried out by members of the Leiden research group, with a range of projects that measure the prescriptions of eighteenth-century grammarians against (for example) their private linguistic practice (e.g. Tieken-Boon van Ostade 2003, 2006), or the popularity of their grammars (Navest forthc.), or the timing of a change in general contemporary usage (Auer 2006). One of my aims in this paper is to add a new test to this repertoire. I want to propose that we may be able to gauge how far prescriptive rules have influenced a speech-community by examining the interpretive habits of members of that community. Readings – and in particular misreadings – can, I believe, tell us much about the presuppositions and expectations of the reader, including the degree to which they have internalised specific rules of grammar.

The case-study I have selected to illustrate this method has its own independent interest as a chapter in the history of pronouns and in the Great Gender Shift of English (often summarised as the language’s evolution “from grammatical to natural gender”).³ The eighteenth century was a pivotal moment in the final phase of this process, being the period when the notion of natural gender was explicitly formulated and when the pronoun started to be commonly defined as “a word standing instead of a Noun, as its Substitute or Representative” (Lowth 1762:31).⁴

Gender can be defined, in grammatical terms, as a system for signalling the declensional class to which a noun belongs and the constructional relationship between a given noun and (for instance) its determiner or modifying adjective, or it can be defined semantically as the conceptual classification system that a language’s noun classes themselves putatively reflect.⁵ Old English (OE) is commonly thought of as a language with grammatical gender (Hogg 1992:124-125), whereas Present-day English (PDE), with its lack of formal variation in nouns or adjectives, lends itself more readily to the semantic approach. Crystal’s popular grammar (1988) strikes the following balance:

Crystal here crystallises two important aspects of the PDE gender system: its specific semantic bases and the key role of the pronoun as its exponent. Pronouns are important in PDE in being the category of the language where gender is still grammaticised, in the sense that they permit or require a speaker to make formal choices, not only between third person pronouns (he/she/it) but also between relative pronouns (who/which) and interrogative pronouns (who?/what?). These choices are rooted in the conceptual categories of animacy, person-hood and sex, and the patterns of co-occurrence which Crystal has in mind - and which are taught as the norms of English usage to foreign learners - are illustrated below:⁷

The “we” of Crystal’s last sentence in (1) is cast in the role of reader/hearer rather than speaker. What (2) shows is that this is an active not a passive role. The highly standardised pronoun selection rules of PDE allow the semantic properties of gender to be read off the pronouns and projected on to the nouns they are in construction with (or the real-world entities that they, in turn, refer to).⁸ The reader can predict, that is, with fair confidence of success, that in (2a) X will have the properties of “inanimate” or at least “non-human”, while in (2b) X will be classifiable as both “human” and “male”. In a Cloze test, a competent hearer/speaker of English would be expected to replace X in (2a) with a noun like table and to replace X in (2b) with a noun like brother or uncle.

It is this high degree of predictability in the semantic content of pronouns that licenses us as readers/hearers to interpret deviations from the co-occurrence norms of (2) as expressive violations, in which a speaker has chosen to reclassify a referent, as, for example in (3):

Finding utterances like these in the context of a novel (let’s say) a modern reader would be likely to interpret the use of it in (3a) as a de-personalising or de-animating gesture, expressive of the wife’s hostility towards her husband; whereas the use of she in (3b) might well be read as endowing the car with animacy, person-hood and sex, and hence as an indicator of affection, contrapuntal to the deprecatory tone of the utterance as a whole (which has a notably different emotional interpretation if transferred to the context of “husband referring to wife”). In (4) – to turn to some actual, and publicly prominent, examples – a similar flouting of expectation produced shock effects:

In (4a), taken from the Camillagate transcript, Camilla Parker Bowles was widely interpreted as expressing her alienation from and contempt for her then husband when she referred to him as it in her telephone conversation with Prince Charles. In (4b), one of the most effective slogans of equal rights activists in the 1970s, the traditional image of God as a white male was challenged not only by assertion (“God ... is black”) but also, and perhaps even more powerfully, by the choice of she, rather than the expected he, in the anaphoric pronoun slot, which retrospectively reclassified “God” as a female being.

What the examples of (2) – (4) demonstrate is the apparently paradoxical blend of fixity and flexibility in the PDE pronoun system. Pronouns have fixed semantic values in terms of gender but this leads neither to invariant usage nor to judgments of non-grammaticality or speaker-incompetence when the norms of usage are not observed. Instead, norm-violations are readily interpreted as expressive and there is a high degree of consensus among native speakers about what is being expressed in each case, whether attitudinal meaning, as in (3) and (4a), or ideational meaning, as in (4b). Taken together, the examples of (3) and (4) illustrate what we might call “post-prescriptivist variability”, in which an expressive flexibility of usage depends on fixed form-meaning values within the system, coupled with a mutual consciousness of the system shared by speaker and hearer. In fact, the rules of the pronominal system illustrated in this section are largely those inherited from the eighteenth-century codifiers of English (though, as we shall see later, the two systems have some interesting points of divergence in their conceptual basis). If we look at the situation that the eighteenth century itself inherited, taking a snapshot of the language at around 1600, we find pronoun variability of a quite different kind.

It can be argued that in 1600 the process of standardising English was already well under way. English was established both as the national language and as the focus of nationalist pride and there were role models for a standard variety available in late-medieval Chancery English, in Tudor Court English, or (on the testimony of Puttenham) in the usage of what was later to be called “the golden triangle” of Oxford-Cambridge-London.⁹ But the forms of the Early Modern English (EModE) prestige variety were not (or not typically) disseminated and enforced by prescription. Hence, in many areas of the language, variability was the norm.¹⁰ Pronoun usage was one such area. The long-term drift of English away from the grammatical gender of OE was proceeding at different rates in different dialects and some of the variability we find in EModE texts may be attributable to the legacy of Middle English (ME) regional systems (especially in colloquial usage), while other examples (especially in educated usage) seem to point to the influence of the Latin gender system on the emergent prestige variety.¹¹ In this situation – which we might call “pre-prescriptive variability” – while variants may have social connotations, there is unlikely to be a fixed and predictable semantic value attached to the choice of one variant form rather than another. Something close to free variation (at least, at the local level) is what we find in the case of all the pronoun types involved in PDE agreement patterns.¹² The examples given below are largely taken from the usage of Shakespeare but they do, I believe, illustrate the general state of affairs in the embryonic Standard English of circa 1600.

In OE and Early Middle English (EME), what? was the interrogative used predicatively in questions about the identity of things (what is it?) or of persons (what is he?/ what are you?). By later ME a competing form with who? was available, but Early Modern usage remained variable so that in 1600 the question “what is X?” could still have either animate or inanimate reference. For an animate-male-person, for instance, the forms what is he?/who is he? could be used apparently interchangeably:

Conversely, the question “what is that?” could still have either animate or inanimate reference, a variability exploited by Shakespeare to create a comic misunderstanding in Twelfth Night,when Sir Andrew asks “What’s that?” referring to a word (accost) but Sir Toby Belch interprets and answers his question as referring to a person (My neice’s chambermaid).

One reason why this exchange can seem merely silly to modern audiences is that the ambiguity on which Shakespeare is playing has been lost. By 1700, what’s this/that? was restricted to non-personal referents and although the form what is he/she? survived, it became a question about social role or occupation (in exchanges such as Q: what is he? A: a lawyer) rather than name or identity (as in EModE Q: what is he? A: Falstaff).¹³

In EME, the indeclinable form that was a universal complementiser, gradually becoming supplemented in its relativiser function by the WH- forms (which developed from OE interrogative pronouns).¹⁴ Which began to rival that in frequency by the fifteenth century, especially in formal prose where it was common from the fourteenth century. Like that, it was originally a gender-neutral form, occurring with both animate and inanimate antecedents, and it was still the only English relativiser explicitly recognised in Ben Jonson’s grammar (1640 [1947]:513) and in the first edition of Wallis’s grammar (1653/1674 [1972]),¹⁵ though their contemporaries, Bullokar (1586:346) and Butler (1634:41), register the rise of relative who. First attested in very specific and formulaic contexts, who gradually became established as a generally available option in the sixteenth century, thus providing the formal possibility for the expression of the personal/non-personal contrast we have today. But in 1600, usage was still unregulated and it was not uncommon to find collocations of the type the man which and the thing who, as illustrated below, where in (7a) who/which both refer to inanimates (caskets), but in (7b) to animates/persons (princes).

There was, however, a general drift in usage across the two centuries 1500-1700. A corpus-based study by Dekeyser suggests, for instance, that the rate of occurrence of which with human antecedent dropped from over 30% in the early sixteenth century to 10% in the early seventeenth (Dekeyser 1984:71).¹⁶

In the sixteenth century we find the remains of earlier practice in which, though it is rarely collocated with nouns denoting humans, he/she are not uncommonly used with inanimate antecedents.¹⁷ By the end of the seventeenth century this variability had become so infrequent as to escape the sampling methods of the Helsinki Corpus.¹⁸ Nevalainen and Raumolin-Brunberg (1994:183-184) exemplify the variability and describe the change as follows:

In the case of personal pronouns, then, as with interrogatives and relatives, what we find in the EModE data is a variability of usage coupled with a drift towards functional specialisation. The strength of that drift is confirmed by the case of the possessive form of the personal pronoun.

Perhaps the most striking pronoun development in the EModE period - in that it involves a formal innovation rather than a shift in the balance of usage between existing variants - is the emergence of its as a distinctive “neuter” possessive. (For fuller discussion, see Nevalainen and Raumolin-Brunberg 1994.) Striking, too, is the speed and completeness with which the change was implemented. In 1586 Bullokar reports (and uses) only his (e.g. “when such participle is resolved by his verb”, 1586:342) and the first instance of its attested by the Oxford English Dictionary in any context is 1598. But by 1672 the change had already embedded itself so securely in usage and consciousness that in his Defense of the Epilogue, Dryden, apparently ignorant of the grammar of his parents’ generation, rebukes Ben Jonson for “ill syntax” in using his with the antecedent heaven (Watson 1962:Vol. 1, 175). Between these two dates, there is considerable variability both between writers (some opting for the receding, some for the rising form) and within idiolects. In the example below, we see Butler (1634:3) using the rising form its in apparently free variation with hir:

The eighteenth century continued to implement the changes outlined in §2. But whereas the seventeenth-century’s grammaticisation of gender in the pronoun system can be understood as a process of drift (to use Sapir’s term) or of unconscious consensus on the part of the speech-community, eighteenth-century developments were affected by the rise of deliberate policies of standardisation and the concomitant emergence of a pedagogical tradition for teaching English (rather than Latin) grammar. This change had major consequences.

In general terms, whereas the generation of 1600 acquired the English pronominal system by intuition or imitation, for the generation of 1800 this natural acquisition process was reinforced, or modified, by precept. Among educated speakers at least, variable practices were replaced by categorical rules and increasingly - especially from the mid-century onwards - many of the variants that had been available in the seventeenth century became stigmatised as either incorrect or “vulgar”. Furthermore, as the rules of “correct” pronominal choice were expounded by successive grammarians, the semantic oppositions on which they were based became more explicitly, and more narrowly, defined.

As I mentioned in my introduction, a crucial moment (symbolically if not causally) is when Lye in 1671 writes “a Pronoun is a word put for a Noun and supplying its stead” (Michael 1970:320). In one sense, this is not new. As Michael points out, “standing for a noun is the literal meaning of pronoun in English, Latin and Greek, and was the basis of Dionysius Thrax’s definition in the second century B.C.”. But the eighteenth-century grammarians made two innovations: firstly by making this the pronoun’s central if not exclusive function, secondly by re-construing the notion of “standing for”. When they say that a pronoun “stands for” a noun, they are not making the distributional point that a pronoun is the syntactic equivalent of a noun, able to occupy the same slot in a construction (say, subject or object position in an SVO sequence). Rather, their definitions “increasingly stress the substitutive function of the pronouns” (Michael 1970:321) - just as their definitions of the noun increasingly stress its substitutive function for a real world entity. The pronoun is thus bound to the noun in such a way that for any given noun there will be a consistent pronominal representative (e.g. John = he/who, the girl = she/who, a tree = it/which) and this representative will encode the properties of the real-world entity that the noun in turn represents.

This link between gender in the language system and sex in the world is most influentially proclaimed by Harris in the period’s linguistic philosophy, followed by Lowth in its pedagogical grammar. For both it is a mark of the superiority of English over other languages that the correspondence between nature and language is so close:

In the must of Lowth’s final sentence we see how this linguistic theory translates into prescriptive grammar. It’s possible to detect the same bias in Dr Johnson’s apparently descriptive observations on changes that have taken place in the usage of both relative and personal pronouns. In the grammar prefaced to the Dictionary (1755:44), he notes:

Like many other grammarians of the period, Johnson formulates contemporary agreement norms between noun and pronoun in terms of a binary opposition in the phenomenal world between “persons” and “things”. Though Lowth in (10b) might appear to be proposing a rather different dissection of the animacy continuum in classing humans with “animals” as opposed to “things inanimate” (Lowth 1762:29), when discussing the pronouns themselves he makes no mention of non-personal animates. Johnson does not discuss the place of animals directly either, but his own pronominal practice, as exemplified below, suggests a persistence, in their case, of something like seventeenth-century free variation.

In the second half of the century, however, consistency of practice is more insisted on by grammarians - a noun should always be represented by the same pronoun - and in selecting the pronoun, the criterial property of “person”-hood is, increasingly, not the attribute of animacy but the attribute of rationality. In both respects, Johnson’s practice is too lax to suit Cobbett, for instance. In commenting on (12), Cobbett says:

For further examples of late eighteenth-century grammarians objecting to similar collocations in early eighteenth-century writers, see Sundby et al. (1991:117). Other grammarians are even more restrictive, excluding children from the category of person-hood, again on the criterion of rationality. An early and influential example is Priestley (1768:98-99):

Translated into the pedagogical practice of the period, the grammarians’ judgments were represented as “rules”, which were learnt by heart and driven home by repeated exercises in the correction of “false grammar”. First introduced by Fisher ([1745] 1750), these exercises became the most common classroom activity after parsing, appearing, on Michael’s calculation, in “about eighty texts” before 1800 (Michael 1987:325-330). A typical textbook is Buchanan’s The British Grammar of 1762, which inculcates the rules of pronoun-noun agreement by requiring pupils to correct examples such as the following:

(For extensive exemplification of eighteenth century pronominal agreement rules, see Sundby et al. 1991:107; 109-110; 116-120.) Buchanan’s example in (15c) is particularly interesting because it represents a mistake that no pupil was likely to make. Amidst all the variability of the preceding EModE period, it had never been normal usage to collocate it with nouns denoting humans and, as we have seen in §2, the form its was a recent innovation seemingly inspired by a growing spontaneous preference for distinguishing inanimate from animate possessors. But by the later eighteenth century, spontaneity had hardened into dogma; the aim of schoolteachers was to inculcate constant vigilance against lapses from a system which would consistently and categorically separate out the rational human from the rest of the animacy continuum. Hence there were ongoing grumblings about collocating whose with nouns denoting animals or inanimates, even though the paradigm provided no direct alternative (Sundby et al.1991:117, 118); and objections were raised against the gender-neutral relativiser that, precisely because it did not explicitly signal person/non-person distinctions (Sundby et al.1991:173).

In §5, I turn to the consequences of these developments in linguistic thought for the generation of 1800 as an interpretive community. How did pupils subjected to the prescriptive pedagogic practices of the later eighteenth century respond when they became readers of texts produced within the variable grammar of 1600?

4. Prescriptive grammar and interpretation: some eighteenth century readings

We have already seen one example of an eighteenth-century reader in action. In (11) above, Johnson, while recording the confusions of “ancient” pronoun usage, comments: “at least it was common to say the man which though I remember no example of the thing who”.¹⁹ This should strike us as odd. As an avid reader of texts from the Age of Shakespeare, Johnson would certainly have come across sequences of the type the thing who, as in example (7a) from the Merchant of Venice. Indeed, the figures given for Shakespeare by Hope (1994:36) and for Spenser by Sugden (1936:56-57) suggest that as minority variants, the man which and the thing who had a comparable rate of incidence in writers of that generation (around 10%). How is it then that Johnson remembered the one but not the other? One explanation might be that he was reading the two collocations differently. Whereas constructions of the type the man which struck him as a straightforward grammatical anomaly or archaism, those of the type the thing who were less visible because they were interpretable. The key to their interpretability can be found, I think, in Harris’s Hermes (1751).

Harris defends the pronoun-based gender system of English against the grammatical gender of other (and at that time more prestigious) languages on two counts. First, English provides a more accurate – natural - representation of the world (he points out the absurdity of “mind”, which has no inherent sex, being represented by a “masculine” noun in one language and a “feminine” noun in another); secondly, the English system allows for expressive reclassification. This is the significance of his parenthetical comment in (10a) above. In “figures of speech”, the properties of animacy, sex and reason can be attributed to non-animate beings by the simple flick of a pronoun. This possibility, he suggests, provides English with two distinct styles of writing, correlating with two distinct genres:

Harris here makes the by now familiar eighteenth-century equation between animacy, personhood and reason (made explicit when he glosses “personified” things as “a kind of intelligent Beings”), and the absence or presence of reason is for him sufficiently signalled by pronoun selection (Virtue ... its v. Virtue ... her). He turns his theory into a reading practice when he decides to exemplify the ornamental personification of poetic style from Milton:

But is this an accurate interpretation of Milton’s intentions? We may contrast Harris’s reading on the one hand with Dryden’s 1672 interpretation of Jonson’s heaven ... his (which Dryden notes not as a personification but as a straightforward grammatical error) and with a marginal comment on (16) and (17) written in my copy of Hermes by its mid-nineteenth-century owner:

This nineteenth-century commentator is aware of the history of the language and is indeed correct in supposing that its is barely represented in Milton’s repertoire, so that there is no norm-violation in the sequence thunder … his and hence no inevitable presumption of personification.²⁰ Dryden’s reading of Jonson represents an early application of the prescriptive mind-set to variable usage: he sees an anomaly in what was, by 1672, a receding use of his, but he does not semanticise it (although the context in Jonson – “though heaven should speak with all his wrath at once” - might provide legitimate grounds for doing so). Harris, however, places and interprets Milton’s his within the context of his theory of natural-gender-with-expressive-violation. It should be added, though, that he is not concerned with the kind of expressive violation illustrated in (3) and (4) above. In Harris’s account, norm-violations are restricted in acceptability and codified in meaning: they signal the ornamental personifications of poetic style.

We have some indications that Harris’s contemporary, Johnson, may have shared his stylistic beliefs about pronouns: as when he describes whose as “rather the poetical than regular genitive of which”(Johnson 1755:44). It seems plausible, then, to suppose that his semantic interpretation ran along the same lines as Harris’s, too, so that the reason why he “remember[ed] no instance” of the type the thing who is that whenever he encountered such a collocation he read it as an instance not of anomalous or archaic usage (as in the case of a man which) but of personification. This hypothesis about Johnson’s reading habits gains support from his Dictionary entries for who and which. He gives them strictly complementary definitions, which being “the pronoun relative, relating to things” and who “a pronoun relative, applied to persons”. In keeping with his comments in the prefatory grammar (in (11) above), he notes, under which, that “it formerly was used for who, and related likewise to persons”; but he has no equivalent acknowledgement that who was formerly applied to non-persons. Yet among his illustrative quotations for who, we find:

Did Johnson simply not notice that who was here applied to heart? Or is he reading Shakespeare’s sequence heart ... who in the same way as Harris reads Milton’s sequence thunder ... his, that is, as a poetic personification?²¹

Lowth provides us with an interesting commentary on the examples and influence of both Johnson and Harris. He cites the passage in (16) with great admiration (thereby transplanting Harris’s views from language philosophy into classroom pedagogy). But, unlike Harris (and most of his contemporaries), he is aware of the history of the English language and so knows that a collocation such as the thunder ... his is not necessarily a marker of personification but may, instead, represent obsolete usage. He experiences difficulty, however, in deciding which is the case in earlier texts. A long footnote demonstrates his confusions:

In the first of Lowth’s three examples, a modern reader might well detect a personification in Bacon’s text, since learning is consistently figured as a human being, passing through the life-phases of infancy, youth, maturity and old age. Surprisingly, Lowth does not appear to read it this way, declaring instead that “his is evidently used as the Possessive Case of it”.Can this be a genre-constrained reading? Bacon is writing prose and is regarded as a philosopher; is he therefore taken as an exponent of what Harris called the “logical style” and so not suspected of indulging in the ornamental fancies of personification?²²

Lowth’s response is markedly different when it comes to the Bible and Shakespeare, both revered texts, but both classed (by Lowth at least) as poetry.²³ They appear together in his footnote because both make heart the antecedent of “incorrect” and inconsistent pronouns (heart ... it ... her in the Bible and heart ... who ... it in Shakespeare). In the culturally more sensitive case of the Bible, Lowth abstains from comment, leaving the example as an interpretive conundrum; but in expounding the problem he detects in Shakespeare, he makes it clear that he is reading the sequence heart ... who as a personification rather than an example of archaic grammar. What causes him to jib at this particular example is that he is sensitive to the inconsistency of collocating heart with who and it in the same sequence (unlike Johnson, whom Lowth’s analysis silently rebukes).

Cobbett’s similar rejection of Johnson’s own variable usage (in (13) above) represents the response of the generation of 1800, who, having internalised the system bequeathed by Lowth and his contemporaries, are intolerant of inconsistency and oblivious (at least in the genre-context of a philosophical prose essay) to the possibility of personification. An even more telling example of unconscious interpretive processes at work in the same generation is a misquotation of Donne by Coleridge. Prefixing lines from Donne’s verse letter “To Sir Henry Goodyere” to an essay in The Friend, he renders Donne’s line (20a) as (20b):

Is Coleridge misremembering here? or making a conscious revision? In either case, the change suggests that he found Donne’s original collocation of body ... her difficult to interpret either as a literal sequence or a personification and therefore resorted to the plural bodies ... their as a way out of the impasse.²⁴

Such collocations have presented a continuing difficulty for the post-1800 interpretive community, inheriting the semantico-grammatical norms of prescriptivist grammar. In the later nineteenth century, for example, we find Abbott, in A Shakespearian Grammar, declaring firmly that in Shakespeare’s usage “Who personifies irrational antecedents” (1870:179). He pursues this interpretive strategy doggedly through increasingly challenging cases until finally forced to concede that “sometimes who is used where there is no notion of personality” (he is apparently defeated by the internal inconsistency in “the world who of itself is peised well”). The local alternations illustrated in (7) pose particular problems for him and in the case of (7a), he finally resorts to a non-semantic rationale, asking: “does euphony prefer which in the accented, who in the unaccented syllables?” (Abbott 1870:181). What he can’t do, apparently, is to concede the possibility of a free variation of forms in such cases.

Finally, in case we think that a century of declared descriptivism in grammar has freed our own generation from the pitfalls of prescribed readings, let’s look at an example of a modern misreading of an eighteenth-century misreading of Shakespeare. The case in point is the passage we have already seen quoted by Johnson (18) and Lowth (19). Lowth’s discussion is cited by Sundby et al (1991:118) as an instance of normative responses to “gender conflict between coreferential personal and relative pronouns”. However, when they quote the passage (1991:108), the form of their entry is follows:

Their truncation of the Shakespeare quotation makes clear that these modern authors’ reading of the passage differs from Lowth’s in taking ghost rather than heart as the antecedent to the pronouns who/it, despite the fact that in the original passage, heart immediately precedes who, while ghost is separated from it by much intervening material. Can we account for this? The explanation I propose is that modern readers, having internalised the norms of PDE usage (exemplified in (2)), are primed to read pronouns semantically rather than syntactically and so to look for a noun/referent that might appropriately cause “a gender conflict” between who and it. A ghost, as liminal entity hovering between the worlds of living and dead, human and non-human, precisely fits the bill, so that ghost ... who ... it represents what in editorial theory is called the “easy reading” or “banalisation”. The sequence heart ... who ... it, by contrast, though the accurate reading syntactically, is the more difficult combination semantically because for the modern reader a heart is a much less good representative of a “bigeneric” entity. In other words, Sundby et al. inherit Lowth’s reading of who as an animising pronoun, but by the end of the twentieth century, the notion of a personified body part (already difficult for Coleridge to accept) has become so deviant as to render Lowth’s reconstruction of Shakespeare’s construction uninterpretable, - or rather misinterpretable.

In the long history of the Great Gender Shift in English, the eighteenth century marks a distinctive new phase, the moment when, in Labovian terms, “change from below” gave way to “change from above” and the process of change became affected by various kinds of metalinguistic awareness. This paper has charted some of the consequences of that development. In the most influential language philosophy of the time, the notion of natural gender emerged as part of the wider belief that language should (and could) reflect the order of things in the world. In the pedagogic grammars of the time, this belief took the form of a systematisation of usage in third person and relative pronouns that linked specific pronouns with specified semantic values so as to reflect the period’s key opposition between “person” and “non-person” (an issue discussed more fully in Adamson 2001b). In the reading practices of subsequent generations – including our own - we see evidence of the pedagogues’ success, in the personifying force that attaches to occurrences of he/she/who.

It is the question of reading practice that I have primarily been concerned with in the later sections of this paper. My general claim is that misreading is an important, and productive, function of reading within the history of the language. Readers at any given historical point are overwhelmingly habituated to the form of the language they themselves speak, and when reading texts from earlier and different forms of it, they naturally seek out interpretations compatible with their own. In many instances, this will not cause difficulties; when it does, as many examples given in this paper demonstrate, the discrepancy highlights not only the inner structure of the inherited text, but also that of the inheritor reading. Misreadings therefore are to be cherished and collected as the closest thing we have to Cloze tests for earlier generations. They open a window on the internalised grammars of successive interpretive communities and so provide crucial aids to our understanding of the historical changes through which languages pass.

Bullokar, William (1586). Bref Grammar of English. In: M. Plessow, (ed.), Geschichte der Fabeldichtung in England bis zu John Gay (Palaestra LII). Berlin: Mayer and Müller, 1906: 333-385.

Cobbett, William (1823). A Grammar of the English Language, ed. R. Burchfield, 1984, Oxford: Oxford University Press.

Fisher, Ann (1745). A New Grammar. Facsimile of 1750 edition, published Newcastle upon Tyne, ed. R. Alston (English Linguistics 1500-1800) Menston: Scolar Press, 1968.

Harris, James (1751). Hermes or A Philosophical Inquiry concerning Universal Grammar. 3^rd edn. 1771. London.

Jonson, Ben (1640). The English Grammar. In: C. H. Herford and P. & E. Simpson (eds.), Ben Jonson, Vol. 8. Oxford: Clarendon Press, 1947.

Lowth, Robert (1762). A Short Introduction to English Grammar. Facsimile edition, ed. R. Alston (English Linguistics 1500-1800) Menston: Scolar Press, 1967.

Priestley, Joseph (1768). The Rudiments of English Grammar [2^nd ed.; 1^st ed. 1761]. London.

Wallis, John (1653/1674). Grammar of the English Language, ed. and transl. by. J.A. Kemp, 1972, London: Longman.

Adamson, Sylvia. 2001a. “Historicising Shakespeare’s Language”. Paper presented at a plenary panel of the Shakespeare Association of America, Miami.

Adamson, Sylvia. 2001b. “Animacy, Animism and Natural Language”. Plenary paper given at the First International Conference on Late Modern English, Edinburgh.

Adamson, Sylvia. 2004. “Who is Sylvia? What is she?: Negotiating Identity in Early Modern English”. Paper presented at the English Department research seminar, University of Helsinki.

Auer, Anita. 2006. “Precept and Practice: the Influence of Prescriptivism on the English Subjunctive”. In: C. Dalton-Puffer, D. Kastovsky, N. Ritt & H. Schendl (eds.), Syntax, Style and Grammatical Norms: English from 1500-2000. Linguistic Insights. Frankfurt; Bern, etc.: Peter Lang. 33-53.

Curzan, Anne. 2003. Gender Shifts in the History of English. Cambridge: Cambridge University Press.

Dekeyser, Xavier. 1984. “Relativisers in Early Modern English: A Dynamic Quantitative Study”. In: J. Fisiak (ed.), Historical Syntax. Berlin: Mouton. 61-87.

Fischer, Olga. 1992. “Syntax”. In: N. Blake (ed.), The Cambridge History of the English Language, Vol. 2, 1066-1476. Cambridge: Cambridge University Press. 207-408.

Hogg, R. M. 1992. “Phonology and Morphology”. In: R. M. Hogg (ed.), The Cambridge History of the English Language, Vol. 1, The Beginnings to 1066. Cambridge: Cambridge University Press. 67-167.

Hope, Jonathan. 1994. The Authorship of Shakespeare’s Plays. Cambridge: Cambridge University Press.

Huddleston, Rodney and Geoffrey K. Pullum. 2002. The Cambridge Grammar of the English Language. Cambridge: Cambridge University Press.

Ihalainen, Ossi. 1985. “He took the boittle and put ’n in his pocket’: The Object Pronoun it in Present-Day Somerset”. In: W. Viereck (ed.), Focus on: England and Wales. Amsterdam: Benjamins.

Ihalainen, Ossi. 1991. “On Grammatical Diffusion in Somerset Folk Speech”. In: P. Trudgill and J. K. Chambers (eds.), Dialects of English: Studies in Grammatical Variation. London: Longman.

Jespersen, O. 1909-1949. A Modern English Grammar on Historical Principles. Copenhagen: Munksgaard; London: George Allen and Unwin.

Lass, Roger (ed.). 1999. The Cambridge History of the English Language, Vol. 3, 1476-1776. Cambridge: Cambridge University Press.

Mausch, Hanna. 1986. “A Note on Late Middle English Gender”. Studia Anglica Posnaniensia 18. 89-101.

Michael, Ian. 1970. English Grammatical Categories and the Tradition to 1800. Cambridge: Cambridge University Press.

Michael, Ian. 1987. The Teaching of English: from the Sixteenth Century to 1870. Cambridge: Cambridge University Press.

Mustanoja, Tauno F. 1960. A Middle English Syntax. Helsinki: Société Néophilologique.

Nevalainen, Terttu and Helena Raumolin-Brunberg. 1993. “Early Modern British English”. In: M. Rissanen, M. Kytö and M. Palander-Collin (eds.), Early English in the Computer Age. Berlin & New York: Mouton de Gruyter. 53-73.

Nevalainen, Terttu and Helena Raumolin-Brunberg. 1994. “Its strength and the beauty of it: the Standardisation of the Third Person Neuter Possessive in Early Modern English”. In: Dieter Stein and Ingrid Tieken-Boon van Ostade (eds.), Towards a Standard English, 1600-1800.Berlin & New York: Mouton. 171-216.

Poussa, Patricia. 1992. “Pragmatics of this and that”. In: M. Rissanen, O. Ihalainen, T. Nevalainen and I. Taavitsainen (eds.), History of Englishes: New Methods and Interpretations in Historical Linguistics. Berlin and New York: Mouton de Gruyter. 401-417.

Quirk, R., S. Greenbaum, G. Leech and J. Svartvik. 1972. A Grammar of Contemporary English. Harlow: Longman.

Robinson, F.C. 1968. “Coleridge, Personification and the Personal Genitive”. Die Neueren Sprachen. 564-566.

Rydén, Mats. 1983. “The Emergence of who as Relativiser”, Studia Linguistica 37 (2). 126-134.

Shelley, Percy Bysshe [1821]. “A Defence of Poetry”. In: Mrs Shelley (ed.), Essays, Letters from Abroad, Translations and Fragments.London: Moxon, 1840. Vol. 1. 1-57.

Smith, A. J. (ed.). 1975. John Donne: The Critical Heritage. London: Routledge and Kegan Paul.

Smith, Jeremy. 1996. An Historical Study of English: Function, Form and Change. London: Routledge.

Sugden, Herbert W. 1936. The Grammar of Spenser’s “Faerie Queene”. Special supplement to Language, published by the Linguistic Society of America.

Sundby, Bertil, Anne K. Bjørge and Kari E. Haugland (1991). A Dictionary of English Normative Grammar 1700–1800, Studies in the History of the Language Sciences 63. Amsterdam: John Benjamins.

Tieken-Boon van Ostade, Ingrid. 2003. “Lowth’s Language”. In: Marina Dossena, Maurizio Gotti and Charles Jones (eds.), Insights into Late Modern English, Frankfurt am Main: Peter Lang. 241–264.

Tieken-Boon van Ostade, Ingrid. 2006. “Eighteenth-Century Prescriptivism and the Norm of Correctness”. In: Ans van Kemenade and Bettelou Los. The Handbook of the History of English. Oxford: Blackwell. 539–557.

Trench, Richard Chevenix. 1855. English Past and Present. London: John W. Parker and Son.

Wales, Katie. 1996. Personal Pronouns in Present-day English. Cambridge: Cambridge University Press.

Watson, George (ed.). 1962. Of Dramatic Poesy and other Critical Essays by John Dryden, (2 vols.). London: J. M. Dent & Sons.

1. I am grateful to the Leverhulme Trust for funding the research project of which this study forms part (Poetry and Parts of Speech: the Study of Grammar and the Practice of Literature 1570-1970). Various sections of my argument were first aired in the research papers listed in the bibliography and, in their present form, at the research seminar of the VICI project at the University of Leiden (referenced below) I have to thank the hosts who invited me to speak on these occasions and those who listened and asked questions. For comments on draft versions of this paper, I am grateful to Roger Lass, Terttu Nevalainen, John Woolford and to my Sheffield colleagues in the Lang/Lit Work-in-Progress seminar: Joe Bray, Jane Hodson, Richard Steadman-Jones and Katie Wales.

2. See the VICI project “The Codifiers and the English Language: Tracing the Norms of Standard English”, run by Ingrid Tieken-Boon van Ostade. Notable conferences on prescriptivism have been: “Histories of Prescriptivism”, Sheffield, 2003; “Language History from Below – Linguistic Variation in the Germanic Languages from 1700–2000”, Bristol, 2005; and “Perspectives on prescriptivism,” Ragusa, 2006.

3. For monograph studies of personal pronouns and of “the great gender shift” in English, see, respectively, Wales (1996) and Curzan (2003). The term “the Great Gender Shift” was, as far as I know, coined by Poussa (1992) on the analogy of the more familiar term, “Great Vowel Shift”. The coinage accurately reflects, I believe, the equivalent importance of the two events in the shaping of the modern language.

4. The first formulation in the English tradition of the notion that the pronoun “represents” the noun is attributed by Michael (1970:320) to Thomas Lye (1671).

5. The standard introduction to the complex questions of linguistic gender is Corbett (1991). Corbett argues that “there is no purely morphological system” of gender (Corbett 1991:34) and that all noun classification systems are ultimately semantic in origin. At first glance, however, some of the distinctions made by grammatical gender systems can seem not only non-semantic but even counter-semantic, as with Old English, which appears to class stones as masculine, women as neuter.

6. For a fuller version of a similar view, see Quirk et al. (1972:187-192), who distinguish ten co-occurrence patterns corresponding to ten gender classes; they posit gender as a “covert” category in English nouns, made overt by the pronouns with which they co-occur. For a more complex and nuanced account of PDE, see Huddleston and Pullum (2002:484-499).

7. Our overwhelming familiarity with the PDE system should not blind us to the fact that other languages base their gender systems in quite different semantic distinctions, as do some non-standard varieties of English, notably the well-attested count-mass distinction which underlies pronominal usage in South Western dialects (see Ihalainen 1985, 1991).

8. The whole discussion of gender and pronouns is bedevilled by the problematics of the sense-reference distinction. I shall avoid it, wherever possible, in this paper by using the traditional shorthand of terms such as “animate nouns” and “anaphoric pronouns”, without committing myself to any ontological or theoretical presuppositions that those terms might carry.

9. For discussion of these models, see Smith (1996:68). Puttenham’s widely quoted advice to Elizabethan writers is that they should model their language on the “usual speech” found within 60 miles of London (Smith 1996:93).

10. Rissanen (1999:187) takes the “richness of variant forms and constructions” as a keynote of his account of EModE syntax.

11. For a distinction between colloquial and classically-influenced “sexualisation”, see Jespersen (1909-1949:VII, 5.8.3). Jespersen makes the interesting speculation (not, as far as I know, subsequently investigated) that “it is characteristic of Shakespeare and Ben Jonson that when they differ in the ‘sex’ which they attribute to the same thing, we see the free play of the imagination in the case of the former, and strict adherence to classical gender in the latter”. Some of the complexities of gender agreement patterns in Late Modern English (LME) are addressed by Mausch (1986) and Curzan (2003:124-131); but there is no in-depth study of the situation in EModE, either for regional dialects or for the proto-standards.

12. I am not, of course, suggesting that “free variation” is an appropriate label for the larger picture of pronoun usage. In fact, the pattern of “drift” I record here points to the opposite conclusion. My argument is simply that the local choices made by EModE speakers do not necessarily carry the implications that are predictably attached to the choices of modern speakers exemplified in (3) and (4).

13. A fuller discussion of the who?/what? option can be found in Adamson (2004). The most complete published account of the interrogatives remains Karlberg (1954). Poussa (1992) describes the related development of restrictions in the animate reference of this/that.

14. For a succinct summary of the development of relativisers, see Fischer (1992:295-304); for the emergence of who, see Rydén (1983).

15. The relativising use of that is noted by Jonson (1640:537) and, in later editions, by Wallis (1653/1674 [1972]:328-329), but they both interpret it as a stand-in for which (thus showing the degree to which the later form had displaced the earlier in grammatical consciousness). It is sometimes noted that in his fourth edition (1674) Wallis adds to his discussion of which the note “et (de personis) who et whom” (1653/1674 [1972]:326-327); it is less often noted that this is in the context of describing the interrogative use of “relativum which”, so may be a recapitulation of his earlier recognition (1653/1674 [1972]:322-323) of a distinction in the interrogative pronoun between who/whom? (“de personis”) and what? (“de rebus”).

16. If one were looking for a rationale behind the distribution of variants in (7) a possible candidate is hierarchy. It’s worth noting that who is allocated to the king and to gold (the king of metals), which to the subordinate in each hierarchy (Gaunt/silver). I don’t know of any study that pursues this possibility but it might well be worth doing.

17. On the position in OE, Mitchell (1985:37) quotes and endorses Heltveit’s view that “the use of hit in reference to persons was in conflict with the speaker’s recognition of personality and sex to a greater degree than was the use of he and heo in reference to asexuals”. For EME, Mustanoja (1960:45) notes that “the definite article when it accompanies wif, maiden, woman or child, usually agrees with the grammatical gender of the noun, but the personal pronouns referring to these nouns seldom do so”.

18. The electronic Helsinki Corpus of English Texts together with an online manual is available from the Oxford Text Archive. The Early Modern sections of the Corpus represent the years 1500-1700 in approximately half a million words drawn from a range of genres. See Nevalainen and Raumolin-Brunberg (1993).

19. It is possible that Johnson is referring to what he “remembers” hearing spoken around him in his youth in Lichfield; but the context of his remarks is clearly a discussion of the pronoun usage of “ancient writers”.

20. Ingram and Swaim’s Concordance to Milton’s English Poetry (1972) lists only three instances of its in Milton’s poetry. Trench (1855:114-115) comments on the eighteenth century’s ignorance of the late emergence of its, pointing out that a knowledge of this fact would have been sufficient to expose Chatterton’s medieval forgeries.

21. The quotation is from 2Henry VI, 3.2.161-165. The speaker, Warwick, is drawing on contemporary physiological theory in which the heart, as the principal seat of life, draws the blood to its aid as it resists the onset of death. In an “untimely” death, the heart has no opportunity to do this, hence the corpse’s face will be black (suffused with blood) instead of pale. Warwick uses this theory to argue that Gloucester must have been murdered rather than dying a natural death.

22. In A Defence of Poetry, Shelley (1840 [1821]: Vol. 1, 11) tried to reverse these received oppositions. In proclaiming that “Lord Bacon was a poet”, he testifies to – and protests against - the then current opinion both of Bacon and of poetic language.

23. Lowth was, of course, the author of De Sacra Poesie Habraeorum (1753), which approached Biblical language from an aesthetic rather than doctrinal viewpoint.

24. Coleridge’s resort to the gender-neutral plural prefigures the tactic used by many late twentieth-century writers, who try to avoid the “sexist” overtones of the writer ... he by using writers ... they. Evidence of Coleridge’s sensitivity to the gender implications of pronouns in other cases is supplied by Robinson, who argues that “for Coleridge neuter whose always implied personification” (1968:565).