Peter opened Paul the door

English-speakers who have not had the good fortune to be exposed early to Greek or Latin, or even to their own language as it existed before the Norman Conquest, tend to find the notion of grammatical case baffling despite the survival in English of a genitive case (renamed possessive) and the distinction between subject and object pronouns in the first and third persons. Evidently, the alleged Irish saying that when it comes to politics the English are born three whiskeys down applies no less to grammar. Others, made of sterner stuff, having learned to distinguish between in urbem and in urbe, in die Stadt and in der Stadt, will find that this is only the beginning. Some languages, such as Finnish, not only have cases of destination, position and departure, but distinguish interior from exterior: ‘into’, ‘in’, ‘out of’ demand a different set of cases from ‘to’, ‘on/at’, ‘off/from’ (at least, that is the theory; actual usage is far more complex). There are also languages in which nouns may take more than one case-ending, as when in Old Georgian a noun in the genitive further adds the same case-ending as the noun it qualifies; but even English can contribute the double genitives hers and theirs, and shares with Danish the phrasal genitive the king of Spain’s daughter.

The nature and the manifestations of case are the subject of the new Oxford Handbook; anyone who wishes to understand the phenomenon of case from any point of view will find something of interest in its 57 chapters (not counting the introduction) by 62 authors or co-authors. The breadth of treatment will be apparent even from the titles of the seven parts into which the book is divided: ‘Theoretical Approaches to Case’; ‘Morphology of Case’; ‘Syntax of Case’; ‘Case in (Psycho)linguistic Disciplines’; ‘Areal and Diachronic Issues’; ‘Individual Cases: Cross-Linguistic Overviews’; ‘Sketches of Case Systems’. To make sense of the whole, readers would have to be familiar with a plethora of technical terms, not all of which are explained, and abbreviations, not all of which are expanded; what f-structures and c-structures are the profane must find out for themselves. It is not evident why some languages are the focus of whole chapters and others merely furnish examples, even if, like Finnish and Hindi/Urdu, they are cited by several contributors. But such things are inevitable in multi-author volumes not written to a formula.

Even a reviewer competent in the 500 or so languages mentioned, and in all the branches of linguistics considered, could hardly produce a readable account of the book by working through it methodically from beginning to end. But in the first chapter Barry Blake sets out the basic facts and problems with admirable clarity, noting the origin of the term ‘case’ in the Greek notion of a falling away (ptōsis; in Latin, casus; in German, Fall) from the ‘upright’ or nominative form, and the Sanskrit grammarians’ still useful distinction between morphological case and semantic function. They recognised six semantic functions: the patient, the implement, the agent, the recipient, the source and the place. In the active voice of transitive verbs, the agent is in the nominative and the patient in the accusative (‘I hit him’); in the passive, the patient is in the nominative and the agent in the instrumental (‘he was hit by me’); we may add that in the participial past the nominative expresses the agent of an intransitive verb but the patient of a transitive, the agent appearing in the instrumental, an alignment known as ergative. Thus gatā gopī means ‘the milkmaid went,’ but hatā gopī sainikena means ‘the milkmaid was killed by the soldier’ or, with a different emphasis, sainikena gopī hattā ‘the soldier killed the milkmaid.’

The most specific details concerning an individual language in the first section, on theoretical approaches, appear in Anna Wierzbicka’s analysis of the use, and non-use, in Polish of the dative for the person affected favourably or unfavourably by the action, what Latin grammarians call dativus commodi vel incommodi. Polish uses the dative both when Peter opens a tin of sardines for Paul and when he opens the door for him, Paul being in either instance enabled to perform an action. English uses the dative only when the action Paul is enabled to perform will be carried out on the direct object, so that Peter opened Paul a tin of sardines is (she says) legitimate (I should find it easier with him, but no matter), since Paul eats the sardines (though not the tin), but Peter opened Paul the door is not, since he merely passes through the doorway. On the other hand, Polish would not use the dative if Peter shut the door on Paul; whereas in Latin Massilienses portas Caesari clauserant, ‘the Marseillais had closed the gates against Caesar,’ is perfectly idiomatic. Polish admits other datives of disadvantage, but not that; and although several differences are noted with Italian and French (but not between Polish and other Slavonic languages), no unitary meaning of the Polish dative is supplied to account for these inclusions and exclusions, probably because none can be found.

Wierzbicka commends the use of ‘natural semantic metalanguage’ to explicate contexts using only simple words supposedly possessing equivalents in all languages. The outcome is an insufferably long-winded ultra-basic English poorer in vocabulary and more rudimentary in syntax than the average child’s. It is allegedly preferable to the use of terms not part of ordinary English speech; but Wierzbicka’s examples of these, experiencer and affected, are considerably less demanding than eight-line rigmaroles about ‘someone’ and ‘this other someone’ that provoke the Spartan response to a lengthy speech: ‘Forgotten the beginning, don’t understand the end.’

In Part II, on the morphology of case, Andrew Spencer considers (among other things) Australian languages in which cases not only correspond to verb-categories such as tense or the syntactical status of the clause, but are themselves given the morphology of verbs. At this point the reader realises that the marvels of language match those of nature in wonder and exceed them in mystery, for though we may tell how such things evolved we cannot tell why. James Blevins demonstrates the inadequacy of theoretical approaches to the relations between case-forms within a paradigm; strangely, although he discusses the notion of stem, he does not notice its origin in Sanskrit grammar, or its acceptance in Indo-European studies. He would rather derive a large number of Avar cases from the ergative than from an oblique stem coinciding with it; but no one would derive the Latin genitives plural rosarum equorum dierum from the ablatives singular rosa equo die.

Matthew Baerman examines case syncretism, the use of the same case-form for different cases. In Sanskrit, as he notes, the ablative is identical in the singular with the genitive except in a-stems (the counterpart of the Greek and Latin second declension); in the dual it is identical with the instrumental and dative, in the plural (as in Latin) with the dative. (In fact, the general separation of ablative and genitive singular familiar from Latin is a secondary development confined to Italic and Iranian.) Indo-European languages are found to be atypical in admitting identity between cases that express neither the agent nor the patient without extending it to all such ‘non-core’ or even all oblique cases. Baerman’s insistence on distinguishing such syncretisms from those produced by phonetic or morphological change would be better taken if we knew enough about the history of the languages concerned. In Pushto, the oblique singular often resembles the nominative plural; Baerman considers and dismisses a theoretical explanation, but does not ask whether, as in Old French, this is the result of linguistic development.

Edith Moravcsik tests hypotheses for the use of case-markers in a language, finding them usually but not always true; Oliver Iggesen looks at case-asymmetry, in which case-distinction is richer, or poorer, or at least different in certain pronouns when compared with other nominals. In citing English pronouns he overlooks thou/thee and ye/you (I have heard the former pair used in natural if semi-dialectal speech), but otherwise all goes well until he looks for motivations: his theory of why at least first and second-person pronouns ‘rightfully’ tend to have richer case-inflection than other nominals, even third-person pronouns, cannot explain why in French only the third person distinguishes accusative from dative and the first and second persons plural do not distinguish case at all, nor why (to cite his own data) in Chittagong Bengali nouns have a separate ergative and pronouns have not, nor why in Pitjantjatjara (spoken in the Central Australian desert) nouns distinguish an ergative from an absolute case, pronouns a nominative from an accusative.

Part III, on the syntax of case, begins with Beatrice Primus’s study of precise relationships within such categories as ‘agent’ and ‘patient’, which may be differently grouped in different languages: in Guaraní, ‘I remember’ is literally ‘it remembers me’ (one recalls Latin pudet me ‘I am ashamed,’ literary German mich hungert, ‘I am hungry’). Ad Neeleman and Fred Weerman examine the relationship between morphological distinctiveness and freedom of word order and construction, in particular contrasting Dutch, which lacks a dative-accusative distinction, and German, which still has one. The student of classical Greek will raise an eyebrow at the assertion that in languages with determiners (for instance a definite article) adjectives cannot be separated from their nouns, and the rule stated on the facing page that debars English adjectives like proud from taking a nominal complement without a preposition (one does not say ‘he is proud his name’) does not apply to worthy and unworthy, stilted as their use without of may now seem. Nor is it true that we cannot say He put the meat into the oven hot (i.e. the meat was already hot when he put it in), though in the Icelandic sentence Neeleman and Weerman cite as legitimate, the masculine eldheitan refers to the oven not the meat; were we meant to imagine a parallel to Tolkien’s ‘O! Water Hot is a noble thing’?

Anna Siewierska and Dik Bakker consider the alternative strategies to case-marking, namely agreement-marking and word order. In Gumawama, a language of Papua New Guinea, the verb takes both a prefix and a suffix agreeing in number respectively with agent and patient. When both are of the same number, they are distinguished by word order: agent patient verb (making it an APV language, along with Latin and Turkish). This is an exception to the tendency they find for APV languages to have case-marking, for AVP languages to lack it, and VAP languages to come between the two, being somewhat likelier than either to mark agreement instead. No examples of VAP languages are given; of those nearest to home, the Celtic languages do not mark agreement, but even Irish, which retains four cases, has lost its former nominative-accusative distinction. So has the Romance family, after post-classical Latin had adopted AVP order, though French and Occitan retained a discrete nominative till the late Middle Ages.

The remaining chapters in Part III would require more explanation than can be reasonably given here, and the psycholinguistic articles in Part IV, on the acquisition of case, case in language production, case in language comprehension and case in aphasia, lie outside my competence. Part V begins with Leonid Kulikov on the evolution of case systems, after which Bernd Heine considers the generation of case-markers by grammaticalisation; a neat example is the degeneration of a Finnish noun meaning ‘at the time’ into a postposition meaning ‘with’, and in Karelian into a suffix. But if prepositions meaning ‘from’ (such as of) and ‘at’ (e.g. Faeroese hjá) count as possessive case-markers (add Norwegian til, ‘to’), we may note how the colloquial Dutch zijn or haar/d’r between possessor and possession (like early modern English his/her/their), as in mijn broer zijn vriend and mijn zuster haar vriend, gave rise in Afrikaans to the universal possessive marker se: my broer se vriend, my suster se vriend, functionally matching ‘my brother’s friend’, ‘my sister’s friend’. The process is also attested in reverse, as when in the Tibeto-Burman language Newari the instrumental suffix na becomes (it seems) the postpositive conjunction ‘when’. Kulikov returns with Jóhanna Barðdal to discuss the decline of case, which can easily be illustrated in Indo-European and in non-classical Arabic, and of less frequent case usages, which are demonstrated from Germanic. Other chapters treat of geographical distribution and language contact.

The use of particular cases is the theme of Part VI, of terminology, and of case polysemy, the different meanings that the same case can express. The authors, Andrej Malchukov and Heiko Narrog, reproduce a ‘semantic map’ of English to and the Russian dative, but without a key it is less informative than it ought to be. They include meanings expressed only with the aid of prepositions, such as directional k in Russian (strangely misprinted v, which in the sense ‘into’ takes the accusative) and nach in German; but what of zu, when not used with a place-name? Malchukov concludes the section with a survey of rare and exotic cases, such as Australian aversive cases (‘for fear of X’) and Himalayan altitudinal cases that distinguish motion up, down and on the level; it is no mere freakshow, for we are warned that by world standards European languages are unusual in making the experiencer the subject of the verb. Indeed within Europe the tendency runs in that direction: whereas German still says es gefällt mir and Russian mne nravitsya, older English ‘it likes me’ has become ‘I like it.’

Part VII begins with Malchukov and Spencer on the typology of case systems; they broach – and leave unresolved – the relation between morphology and syntax in establishing cases: whether, for instance, Pitjantjatjara and similar languages should be said to exhibit ergative syntax in nouns and nominative syntax in pronouns, or to possess three cases, agentive, nominative and accusative, of which the second and third are alike in nouns and the first and second in pronouns. In Russian, the accusative singular of most masculine nouns (and the accusative plural of all nouns) resembles the nominative if they denote inanimate beings, but the genitive if they denote animates; qualifying adjectives behave the same way, even for those nouns that have a specific accusative. Historically, the nominative-like form is derived from a once-marked accusative and the genitive-like form is a special function of the genitive, but within the language as it is the authors are clearly right to say that both are accusative by syntax.

Here we may compare the Finnish ‘total object’ (meaning one affected in its entirety by the action), for which the normal accusative in singular nouns and adjectives (now resembling the genitive, but historically distinct from it) is replaced in certain syntactical contexts by what some authorities call a form resembling the nominative, others the nominative itself. There are theoretical difficulties either way; further to complicate them, the accusative plural of Finnish nouns and adjectives is always the same as the nominative, but personal pronouns both singular and plural have a specific accusative form. Opposite views of the problem are taken in two contributions, but with arguments in neither.

Michael Daniel and Dmitry Ganenkov review the wealth of cases in the languages of Daghestan; Peter Arkadiev considers languages with just two cases, after which other contributors focus attention on particular languages or language groups, including Ik (Christa König), spoken in north-eastern Uganda, with seven cases that may be taken by almost any part of speech – including adverbs, conjunctions, adpositions and verbs – but that has defective encoding of subject, agent and object. Thus in the Ik for ‘I see the children’ both ‘I’ and ‘children’ are nominative, but in the command ‘see (you) the children!’ both ‘you’ and ‘children’ are in the case called oblique; in ‘when I see the children I go’ both ‘I’ (expressed only once) and ‘children’ are accusative.

Akio Ogawa, in writing about Japanese, is happy to accept as case-markers the numerous postpositions that relate invariable nouns to their clauses. Spencer, by contrast, is unwilling to allow Japanese any cases except as possibly required ‘for theory-internal reasons’, which sounds uncommonly like ‘when the theorist needs to twist the facts’; but most other contributors who discuss the language allow it to have cases. Once again, the problem of morphology and syntax arises; Baerman, one of those who speak of Japanese nominatives and accusatives, even asserts that in the English sentence I gave the picture to Anne’s brother and the book to Susan’s the understood brother is marked for the dative case, presumably by the preposition. But if the speaker changed the emphasis from the indirect objects to the direct by saying I gave Anne’s brother the picture and Susan’s the book, would brother be marked as dative (and not merely as indirect object) by the word order? If so, we are reverting to the days when the sentence The White King and Queen gave Humpty Dumpty a cravat was said to exhibit successively the nominative, dative and accusative cases because they would have been used in Latin (or indeed Old English).

On the other hand, when Lewis Carroll makes Humpty Dumpty say They gave it me, we still distinguish the nominative they and the objective me. Hence the discomfort caused H.W. Fowler by 1 Corinthians 2.9 in the Authorised Version, ‘Eye hath not seen, nor ear heard, neither have entered into the heart of man, the things which God hath prepared for them that love him’), for ‘the things’ would have to be represented by two different pronoun-forms: ‘eye hath not seen them’/‘neither have they entered’. But between English accusative and dative the morphological distinction is entirely lost; the functions can still be told apart, but if we speak of separate cases we shall have to say that in some sense the classical languages preserved the Indo-European instrumental because, as Quintilian noted more than 1900 years ago, the naming of the implement does not accord with the proper sense of the Greek dative and the Latin ablative.

Nevertheless, fuzzy edges remain. The well-known exchange, ‘My mother made me a homosexual.’ ‘If I gave her the wool, would she make me one too?’, turns on the use of the single case-form me for both direct and indirect object; it would lose its point in German, which distinguishes mich from mir (though translated into Berlinisch it would mock the local use of mir for both). But would ‘My mother made me a homosexual and a nice warm sweater’ be grammatical?