Notes

INTRODUCTION: ‘THIS MARVELLOUS INVENTION’

‘this marvellous invention’: Arnauld and Lancelot, Grammaire générate et raisonnée (1660), 27: ‘Cette invention merveilleuse de composer de vingtcinq ou trente sons cette infinie variété de mots qui n’ayant rien de semblable en eux-mêmes à ce qui se passe dans notre esprit, ne laissent pas d’en découvrir aux autres tout le secret, et de faire entendre à ceux qui n’y peuvent pénétrer, tout ce que nous concevons, et tous les divers mouvements de notre âme.’ (All translations, unless otherwise stated, are by Guy Deutscher.)

‘every body perseveres’: Newton’s First Law of Motion (in the English translation of Andrew Motte, 1729), from his Principia Mathematica (1687): ‘Corpus omne perseverare in statu suo quiescendi vel movendi uniformiter in directum nisi quatenus a viribus impressis cogitur statum ilium mutare.’

The Sumerians: Suggestions for further reading: S. N. Kramer, History Begins at Sumer: Thirty-nine Firsts in Man’s Recorded History (1989), and The Sumerians: Their History, Culture and Character (1967). See also various articles about the language, culture and history of the Sumerians in J. M. Sasson, Civilizations of the Ancient Near East (2000).

Sumerian munintuma’a: from an inscription of Enannatum, ruler of Lagash in the twenty-fifth century BC, ed. Steible (1982), 185. I am grateful to Bram Jagersma for this example.

Uniformitarianism: On the history of this idea in linguistics, see Christy (1983).

No systems of communication today are evolving their first words: Some readers may wonder why I have not mentioned pidgin and Creole languages in this context. The reason is that pidgins are by no means systems of communication that are in the process of developing their first words. While they are indeed simplified languages, crucially, they incorporate all the fundamental ‘design features’ of language, and they are spoken by people who already have command of a fully complex human language (their mother tongue) and who are thus aware of the range of expression that a normal language affords. It is thus far from obvious that the linguistic behaviour of such speakers, even under conditions of extreme simplification, would reflect exactly a stage in the evolution of language when people had yet to discover the unbounded possibilities of expression that a complex language provides. In general, I will not rely on evidence from pidgins and Creoles in this book, since our understanding of their nature and especially of their genesis is still highly uncertain. In particular, an increasing body of evidence seems to suggest that much more of pidgin and Creole structure may be accounted for by the influence of substrate languages than has traditionally been assumed (see, e.g. Lefebvre (2004)), and this undermines attempts to present pidgins and Creoles as accurate reflections of earlier stages in the evolution of language.

Arguments for early emergence of language in Homo erectus: Bickerton and Calvin (2000), 104, Bickerton (1990), 138. For evidence for the control of fire from at least 700,000 years ago, see Goren-Inbar et al. (2004), 725.

Chimpanzee tool use as a culturally transmitted activity: Boesch (1993), McGrew (1993).

‘Hand axes’ produced by imitation: Davidson and Noble (1993), who also stress that these tools require much less planning than might at first appear.

Discrepancy between cognitive potential and realization: Renfrew (1996).

Earlier hominids could not produce the vowel i: Lieberman (1984), Lieberman (1991).

‘Explosion’ in arts and technology: Klein and Edgar (2002).

Symbolic artefacts as evidence of language: Davidson and Noble (1993).

Perforated shell-beads in South Africa: Henshilwood et al. (2004). These finds support the view held by archaeologists such as Davidson that the ‘microliths’ (little stone tools shaped into distinct geometric forms) from the Kassies River Mouth in South Africa, which date from around the same time, should also be seen as evidence of the use of symbols. See Davidson and Noble (1993).

‘When it comes to linguistic form, Plato…’: Sapir (1921), 219.

Brisset’s ‘coac coac’: Brisset (2001), 717, ‘Un jour que nous observions ces jolies petites bêtes, en répétant nous-même ce cri: coac, l’une d’elles nous répondit, les yeux interrogateurs et brillants, par deux ou trois fois: Coac. II nous était clair qu’elle disait: quoi que tu dis?’

Scenarios for the first emergence of language: Suggestions for further reading: An excellent introduction is J. Aitchison’s The Seeds of Speech (1996). Selections of articles representing different theories on the emergence of language can be found in M. Christiansen and S. Kirby, Language Evolution (2003); A. Wray, The Transition to Language (2002); T. Givón and B. F. Malle, The Evolution of Language out of Pre-language (2002). A collection of fascinating articles on the relation between tools and language, and the use of tools by hominids and primates is K. Gibson and T. Ingold, Tools, Language, and Cognition in Human Evolution (1993). Web-links to other resources and online articles can be found on www.unfoldingoflanguage.com.

Kanzi: Suggestions for further reading: For an engaging biography of Kanzi by his trainer, Sue Savage-Rumbaugh, and an account of his learning process and linguistic abilities, see Savage-Rumbaugh et al, Apes, Language, and the Human Mind (1998). See also Savage-Rumbaugh and Rumbaugh (1993), and the webpage of the Language Research Center of the Georgia State University www.gsu.edu/∼wwwlrc/biographies/kanzi.html, where it is also reported that Panbanisha, Kanzi’s younger sister, has developed language comprehension and production skills even more advanced than those of Kanzi. For a critical evaluation, see e.g. Steven Pinker’s The Language Instinct (1994).

The innateness controversy: Suggestions for further reading: For proponents of the ‘innatist’ view, see S. Pinker’s The Language Instinct (1994) and The Blank Slate (2002), or R. Jackendoff’s Foundations of Language (2002). For advocates of the opposite view, see G. Sampson’s Educating Eve: The ‘Language Instinct’ Debate (1997), T. Deacon’s The Symbolic Species (1997), or M. Tomasello’s Constructing a Language (2003). A recent issue of the journal The Linguistic Review (no. 19, from 2002) is devoted entirely to debating (in characteristically polemical tones) the ‘poverty of stimulus’ argument. Recently, Chomsky himself has offered a very different take on the question of what needs to be pre-wired in the brain. A co-authored article in the journal Science (Hauser, Chomsky and Fitch (2002)) argues that the only specific pre-wired linguistic mechanism is ‘recursion’ (the ability to re-apply an operation to a construction, in theory indefinitely, to create increasingly complex hierarchical structures, as in ‘this is the farmer that … that … that lay in the house that Jack built’). The article also suggests that recursion did not evolve specifically for language but for other cognitive abilities such as navigation. For a critical discussion see Pinker and Jackendoff (forthcoming).

Cultural evolution: Cavalli-Sforza and Feldman (1981).

CHAPTER 1: A CASTLE IN THE AIR

‘Basque is really a strange language’: Scaligerana, or the collection of Scaliger’s aphorisms: Scaliger (1695), 48–9 (s.v. Basque). On Scaliger’s linguistic skills see the Funeral Oration by Daniel Heinius (Heinius 1927 [1609]).

Aardvark: From Dutch aard-vark,’ earth-pig’, this is the name of the ‘ant-bear’, an ant-eating mammal found in South Africa, famous mostly for its prominent position at the head of English dictionaries.

Turkish word ‘you are one of…’: Some of the glosses are only approximations. In particular, dik is a non-future participle marker that nominalizes the clause ‘can’t cause to…’ Ler is a morpheme which pluralizes the whole clause nominalized by dik. And imiz is a possessive pronoun ‘our’ (rather than ‘we’). In connection with the participle dik, however, imiz marks the subject in the participle clause. So more literally, the Turkish word should be translated ‘you are one of those of our not being able to turn into someone from town.’ On the structure of such participle constructions in Turkish see Haig (1998), 50–70.

Sensitivity to hierarchical structure: Chomsky (1971), 29, who uses a hungry dog as an example, rather than a seal. But the structure of the example is the same.

Neo-Aramaic example (Alqosh dialect): E. Coghill (p.c).

Gender systems: Corbett (1991), 8–12. Among the languages of the world which have gender distinctions, there seem to be only a few which have a consistent or nearly consistent logical classification. Tamil is an example of an almost entirely consistent system, and in some sense, English is another example, although a fairly marginal one, since gender in English is only manifested by the pronouns he, she, it, and is not formally marked on nouns themselves.

Jemez number ending -sh: Mithun (1999), 81, 443.

CHAPTER 2: PERPETUAL MOTION

‘Eppur si muove!’: Galileo is reputed to have made this defiant statement in 1632, after he was forced by the Inquisition to recant his claim that the earth moves around the sun.

Polychronicon: Rawson Lumby (1876), 297.

Translations of Wycliffe and Ælfric: Both translated not from the original Hebrew, but from the Latin Vulgate, where Noah is called Noe.

Seventeenth-century pronunciation of all as {owl}: Lass (1999), 94ff.

‘Men of mature age’: Eastlake (1902), 992. The h-less pronunciation of words such as ‘ospital should not be taken to mean that the speakers Eastlake describes dropped all their h’s, since the words that Eastlake mentions are all of French origin. The process by which h reappeared in pronunciation, because of the influence of spelling, can be seen more recently with the adjective ‘historic’, which was originally pronounced without an h, hence the still current phrase ‘an historic moment’.

‘I take it you already know’: This poem appeared in the Manchester Guardian on 21 June 1954, in the Miscellany section on page 3, with the title ‘Brush up your English: Hints on pronunciation for visiting foreigners’. It is signed T. S. W. Taylor and Taylor (1983), 99 attribute it to T. S. Watt (and change the word ‘lough’ in the original on line 4 to ‘laugh’).

German biblical translations: AD 1000: translation of Notker, from Tax (1981), 333 and Tschirch (1955), 122; Modern German: Elberfelder translation. French biblical translations: AD 400: Vulgate; AD 1200, from Michel (1860), 131; Modern French: Darby translation.

Languages multiply happily of their own accord: The account given here for divergence due to geographical separation is of course highly simplified. For discussions of the different patterns of change and divergence resulting from geographical dispersal, see Dixon (1997), Nettle (1999).

‘Give us this day’: Indo-European data from Lockwood (1972).

Mbabaram: Suggestions for further reading: for a fascinating account of the search for Mbabaram and other aboriginal languages, see R. M. W. Dixon, Searching for Aboriginal Languages: Memoirs of a Field Worker (1989). On the change from gudaga, which is still the word for ‘dog’ in the neighbouring language Yidin, to dog, see ibid., pp. 126ff., Dixon (1972), 347–52, Dixon (1991).

Role of contact in language change: Labov (2001).

Change emerging from the accumulated behaviour of individuals: Keller (1994). For a general discussion of the reasons for language change, see also Aitchison (2001).

‘This is my loved son that liketh me’: Oxford English Dictionary, s.v. like, v.1 1.a. (The pylgremage of the sowle; from the French of G. de Guilleville c. 1400) with spelling normalized from ‘This is my loued sone that lyketh me’.

‘Should we not be monstrously ingratefull’: Oxford English Dictionary, s.v. resent, v. II.9 (Isaac Barrow, Sermons).

‘I was sure that this instance of his friendship’: Oxford English Dictionary, s.v. resent, v. II.9 (Bishop William Warburton, Letters from a late eminent prelate to one of his friends).

‘the process of linguistic change has never been directly observed’: Bloomfield (1933), 347.

The role of variation in sound change: I do not wish to imply that the realization that change proceeds through synchronic variation has settled all questions regarding sound change. One area that is still controversial is to what extent sound change proceeds across all words at the same time, or progresses word by word (‘lexical diffusion’). For a review of the state of the art, see Labov (1994), Part D.

Charles I quotes: Oxford English Dictionary, s.v. resent v. II.9, and resent v. I.2.

CHAPTER 3: THE FORCES OF DESTRUCTION

‘Beloved men, know that this is the truth’: From the opening of Wulfstan’s Sermo Lupi ad Anglos.

‘Tongues, like governments’: Samuel Johnson: Dictionary of the English Language, 1755. Reprinted in Bolton (1966), 154.

‘a mistake was a mistake’: Clive James, ‘The unmysterious suicide’, Times Literary Supplement, 21 June 2002.

‘most people who bother with the matter’: George Orwell, ‘Politics and the English Language’, reprinted in Bolton and Crystal (1969), 217.

Schleicher quotes: Schleicher (1850), 11, 231.

‘The greatest improprieties…’: Thomas Sheridan’s General Dictionary of the English language (1780), quoted from MacMahon (1998), 383.

‘I do here, in the Name of all the Learned’: Jonathan Swift’s Proposal for Correcting, Improving and Ascertaining the English Tongue, reprinted in Bolton (1966), 108.

‘six hundred years ago…’: Grimm (1819), x.

Koster quotes: Koster (2001), 29, 31. French ‘alteration’ is not a neutral term, so the more accurate meaning of ‘altération qui se mue en altérité’ is ‘debasement which turns into otherness’.

Discussion between Hugo and Cousin: Victor Hugo, Choses vues 1830–1846, Séance du 23 Novembre 1843 (Hugo (1972), 273).

‘gradually lost the proper and instinctive sense’: Paris (1862), 3–4.

‘practically everyone … in those days’: Cicero, Brutus 258 (ed. Kytzler (1977), 194): ‘sed omnes turn fere … recte loquebantur. sed hanc certe rem deteriorem vetustas fecit.’

‘every age claims’: Weigel (1974), 7.

‘With the Third Declension…’: Sayers (1963), 180.

Date of Latin change from s to r: Touratier (1975).

Grimm’s law: The formulation of the sound shifts appeared in the second edition of Grimm’s Deutsche Grammatik (1822), three years after the first edition (1919).

Table of ‘home grown’ and ‘borrowed later’ cognates: The reconstructions are based on Kluge (1995) and Watkins (2000). Only the roots are given here, in order to facilitate the comparison with the modern languages. The symbols , stand for long r and m sounds which behave like vowels, somewhat like in the English word bottom. The k of roots like kerd and kwon was originally palatalized (that is, pronounced {ky}, as in the English word ‘cube’), so the Indo-European roots should properly be written erd and won. But in the so-called ‘centum’ branches of Indo-European, which include all the ones mentioned in the table, the palatal velar merged with the plain velar k (and likewise ĝ merged with the plain g, and ĝh with the plain gh), so I will not specifically mark the palatalization in any of the etymologies here or elsewhere in the book.

A system where ‘everything holds together’ (tout se tient): This is often attributed to Saussure, but apparently was articulated first by Antoine Meillet (1903), 407.

‘The Germans’ mighty progress and urge for freedom’: Grimm (1848), 417: ‘… mit dem gewaltigen das mittelalter eröfnenden vorschritt und freiheitsdrang der Deutschen zusammenhängt, von welchen Europas umgestaltung ausgehn sollte.’

Japanese para-kiri: Martin (1987), 10–12, 399.

‘By leaving out a Vowel to save a Syllable’: Jonathan Swift’s A Proposal for Correcting, Improving and Ascertaining the English Tongue (1712). Reprinted in Bolton (1966), 113–15.

Danish song: ‘Fo’ ajle di små blomster’ (Opvåvni), by Mads Hansen (1870).

‘Clinton Sends Vowels’: This piece originally appeared in the American satirical magazine The Onion (December 1995), but was then widely circulated unattributed on the internet.

Czech phrase ‘stick finger down throat’: Strictly speaking, it is inaccurate to say that the Czech phrase does not contain vowels, since the sound r in words like strč, prst, is long and builds the core of a syllable as vowels do.

The Complaints of Khakheperre-seneb: Lichtheim (1973), 146, Parkinson (1996), Parkinson (1997).

French intensifiers: In modern French, see Koster (2001), 32. In Old French, see Gamillscheg (1957), 753.

Middle English emphatic combination ne … nawt: Fischer (1992), 280, Jack (1978), 29–39. Notice, incidentally, that the use of ‘double negation’, which would be frowned upon by purists today, was entirely normal then.

‘At all’ as an intensifier: Originally, ‘at all’ was used to add emphasis also to ‘normal’ affirmative sentences, as in ‘They were careless at all, they thought all things were cock-sure’ meaning ‘they were very careless’, or ‘careless in every way’. Oxford English Dictionary, s.v. all A.9.b (H. Latimer, Sermons and remains c.1555).

‘one may truthfully say that…’: Quoted in Olender (1997), 51–9.

‘What a dreadful pity…’: Voltaire’s Letter to Catherine the Great, 26 May 1767, ed. Besterman (1974): ‘Je ne suis pas comme une dame de la cour de Versailles, qui disait: c’est bien dommage que l’aventure de la tour de Babel ait produit la confusion des langues; sans cela tout le monde aurait toujours parlé français.’

Saussure’s theory: His Mémoire sur le système primitif des voyelles dans les langues indo-européennes is dated 1879, but appeared in 1878. Saussure’s term for what I have called ‘rogue sounds’ was coefficients sonantiques (‘sonant coefficients’).

The root pā(s) ‘protect’: This root has a short form pā and an extended form pās. The example used by Saussure (1879, 129–30) is the short form pā (as in Sanskrit pā-tár ‘protector’), but for ease of comparison I use the extended form pās here.

Critical acclaim of Saussure’s theory: Mayrhofer (1988).

Contribution of Hermann Möller: Szemerényi (1973), 7. Möller also made significant improvements to Saussure’s system, increasing the number of the coefficients sonantiques. He didn’t use the term ‘laryngeals’ initially (in 1879 he called them ‘glottal’, a year later ‘guttural’, and it was only in 1911 that the term ‘laryngeal’ appeared) but he made the connection with the Semitic sounds from the beginning.

The Hittites and the discovery of Hittite: Suggestions for further reading: C.W. Ceram, Secret of the Hittites: The Discovery of an Ancient Empire (1973), O. R. Gurney, The Hittites (1981), J. G. Macqueen, The Hittites and Their Contemporaries in Asia Minor (1986). For an introduction to the cuneiform script, see C. B. F. Walker, Cuneiform (1987).

The Hittite phrase NINDA-an ettsa-tteni…: Hrozný’s report on his decipherment appeared in December 1915, in the Mitteilungen der Deutschen Orient-Gesellschaft. It is often claimed that ‘now you will eat bread and drink water’ was the very first Hittite phrase that Hrozný managed to decipher, but in his own report, he only mentions it as one of the first. The actual transcription of the Hittite sentence is: nu NINDA-an e-iz-za-at-te-ni wa-a-tar-ma e-ku-ut-te-ni. Since the Babylonians themselves borrowed the cuneiform script from the Sumerians, the word-signs that they used, and which were then adopted by the Hittites, were based on Sumerian words. NINDA is Sumerian for bread. The Old High German cognate which I spell ettsan for transparency was actually spelled ezzan. Hrozný also thought that eku looked rather like Latin aqua, which strengthened his conviction that it meant ‘drink’. But this was just a coincidence. Linguists today think that eku came from the Indo-European root *egwh-’drink’, which in Latin (after some sound changes) ended up in the word ēbrius ‘drunk’, from which we have English inebriated.

An ‘almost unbelievable accident’: KuryŁowicz (1927), 101: ‘Cette question serait insoluble dans l’état actuel de la grammaire comparée si, par suite d’un accident presque invraisemblable, le hittite ne semblait pas avoir conservé le caractère consonantique de 2.’

Schleicher quotes: ‘at first sight we observe precisely the opposite’, ‘precisely the fact that we find language already fully constructed’, ‘Only when a nation has perfected its language’, all from Schleicher (1850), 11–12; ‘Languages are natural organisms’: Schleicher (1863), 6; ‘history, that enemy of language’: Schleicher (1850), 134.

‘languages are historical creations, not vegetables’: Bonfante (1946), 295. His criticism of Schleicher is aimed mainly at the details of the family-tree model. Similar criticism was expressed much earlier, most forcefully by the founders of the Neo-Grammarian movement, H. Osthoff and K. Brugmann (1878), iii-xx. For a history of nineteenth-century linguistics, see Morpurgo Davis (1998).

The s-r irregularity in Old English ‘choose’: As it happens, there are also cognate forms in closely related languages, such as Dutch kiezen ‘choose’ and German auserkoren ‘chosen’, which could have suggested even without records from older stages that there was something untoward with s and r in the Germanic languages. Still, without the historical records, we would never have any cause to reconstruct all the idiosyncrasies of ‘choose’ in Old English.

CHAPTER 4: A REEF OF DEAD METAPHORS

The definition of metaphor: Linguists have debated whether, and to what extent, two types of metaphoric processes should be distinguished in language: ‘metaphor’ and ‘metonymy’ (see e.g. Kövecses (2002), Ch. 11). Metaphor is meant to involve a transfer of a concept to a distant domain, whereas metonymy is the transfer to an adjacent domain (as, for example, in referring to the monarchy as ‘the crown’, or to someone with a drink problem as ‘on the bottle’). Since the metaphors in this chapter generally have a ‘basis in experience’, some may prefer to call them metonymies. But for our purposes, the niceties of the distinction are of no particular concern, since the crucial aspect of the process is the movement from concrete to abstract domains, not whether the abstract domain is ‘adjacent’ or ‘distant’.

‘Metaphors, I said!’: Skármeta (1988), 16 (translation: K. Silver).

Conceptual metaphors: Lakoff and Johnson (1980), Metaphors We Live By.

Molière quote: Le Bourgeois Gentilhomme (‘The Middle Class Gentleman’), Act II, scene IV. Translation: Project Gutenberg.

‘if the house be discovered by tempest’: Oxford English Dictionary, s.v. discovery. 1. (Edward Coke, The institutes of the lawes of England, 1628.)

‘decide’ from ‘cut’: Basque example from Larry Trask (p.c.), Indonesian example from A. Gianto (p.c), Endo example from J. Zwarts (p.c.), Mandarin Chinese jué (more commonly jué dìng) (‘cut/breach’, ‘decide’) from G. Sampson (p.c.). Compare also French trancher (‘cut/slice’, but also ‘settle a matter once and for all’).

Transfer from concrete to abstract: Metaphors in the other direction are much rarer, but travelling is one area where such metaphors are found. For example, we can say that ‘Cambridge is at an hour’s distance from London’. Here we measure the distance (space) by the time it takes to traverse it.

Metaphor as an analogical process: Gentner et al. (2001).

Most languages don’t have a ‘have’ verb: Heine (1997a), 75 reports that from a sample of no languages, about 14 per cent expressed possession with a transitive verb.

Examples of possession: Russian, So, Mupun, Quechua, Breton, Nama, from Heine (1997a), 92–5. Turkish, Irish, Tamil, Dullay, from Heine (1997b), 48–59. Waata from Heine et al. (1991), 37. In Turkish, this is not the most common construction for expressing possession, and is used only in specific circumstances. Akkadian example: from an Old Babylonian letter: Frankena (1974), no. 57.

‘dictionary of faded metaphors’: ‘Wörterbuch erblasseter Metaphern’, literally ‘turned-pale metaphors’. From Vorschule der Aesthetik, §50 (Paul (1967), 184).

‘Outside … is used in a sense not known to the language’: Elwyn (1859), 82, quoted from the Oxford English Dictionary, s.v. outside D. II. 4. b.

Ewe megbé: Heine et al. (1991), 65. Megbé itself seems to be a compound, as according to Westermann (1954), me itself can mean ‘back’.

Hebrew ‘face’: The sound p has weakened to/in some environments, but for transparency, I use p throughout.

Body-part examples: Anttila (1972), 149 (Hungarian); Heine and Kuteva (2002) (all the rest except Hebrew). While parts of the body are the most common source of spatial terms, other objects can also serve in this function. One example is the French preposition chez, meaning ‘at’ or ‘by’, whose origin is the Latin word casa ‘house’ (Gamillscheg (1928), 309). The Scandinavian preposition hos (‘by’, ‘at’) has the same origin, the word hus ‘house’. The Akkadian bīt (‘house’) also developed into a preposition meaning ‘at’ in later stages of the language.

CHAPTER 5: THE FORCES OF CREATION

‘Let us put our trust in the eternal spirit’: The ending paragraph of Bakunin’s first publication, ‘Die Reaktion in Deutschland’ (Bakunin (1842), 1002).

The history of ‘going to’: Scheffer (1975), 270, Danchev and Kytö (1994) (but note their error on p. 63: the alleged early example from 1567 is actually from 1657: the Parliamentary Diary of Thomas Burton, entry of 12 January 1657: ‘Especially when you are going to lay a tax upon the people, it is fit you should be unanimous’ (Rutt (1828), 339). See also Traugott and Dasher (2002), Hopper and Traugott (2003), 1–3, 93.

‘as they were goynge to bringe hym there’: Rolls of Parliament (Rotuli Parliamentorum) 1278–1503, ed. Strachey (1767), Vol. 5, 16.

‘was goyng to be broughte into helle’: From The Revelation to the monk of Evesham (Aber (1869), 43). This was a translation of a Latin work composed much earlier, around 1196, called Visio Monachi de Eynsham. The Latin source has ‘infelix … agitur in gehennam’ (‘the unhappy one is driven into hell’).

Fletcher quote: Women Pleased, act III, scene ii.

‘To be short, You see that My Magazine…’: Speech of Charles I to the Gentry of Yorkshire on 12 May 1642, Fulman and Perrinchief (1662), 401.

‘“going to” is the signe of the Participle of the future’: Poole (1646), 26.

‘Going to’ more restricted in use before nineteenth century: Traugott and Dasher (2002), 84. For example, ‘raising’ phrases with inanimate subjects, such as ‘it is going to rain’ only seem to appear in the latter half of the nineteenth century.

‘Now Willie lad, I’m ganna gie You’: Alexander Douglas (of Strathmiglo), Poems, Chiefly in the Scottish Dialect (1806), Cupar-Fife, p. 70, quoted from Wright (1900), 659, s.v. go 4.

Basque, Tamil, Zulu examples of ‘going to’: Heine and Kuteva (2002), 161–3.

Report of the Inquisition: Schiaparelli (1929), 74 (no. 19, AD 715, giugno 20, Siena. Breve de Inqiositione): ‘Uuarnefrit gastaldius mihi dicebat: Ecce missus uenit inquirere causa ista, et tu, si interrogatus fueris, quomodo dicere habes? Ego respondi ei: Caue ut non interroget, nam si interrogatus fuero, ueritatem dicere habeo.’.

Fredegar’s account of Daras: Wolfram (1982), 70 (book II, par. 62). The historical reality behind this story is a bit more complex. Fredegar does not mention Kavadh by name (and refers to him only as the Persian king), and he also seems to have conflated two Byzantine emperors into one story. Dara(s) was founded (or at least fortified) in AD 507, by the Byzantine emperor Anastasius, a predecessor of Justinian, after he had made a temporary peace with Kavadh. Twenty-three years later, the same Kavadh was again engaged in war with the Byzantines, this time with emperor Justinian. Justinian’s legendary general Belisar routed Kavadh in a famous battle near Daras, and two years and a few military reverses later, Justinian again negotiated a peace (with Kavadh’s son), ending the war. In Fredegar’s story, all three events seem to have been conflated. Today the ruins of Daras lie in the small village of Oğuz near the town of Mardin in eastern Turkey, not far from the Syrian border. (Encyclopaedia Iranica, Vol. VI, 1993, s.v. dārā.)

Contraction of dare habes to daras: the modern French for ‘you will give’ is not daras but donneras, because at some later stage, the verb dare dropped out of use and was replaced by donare.

Table of Late Latin to Modern French: Based on Valesio (1968), 159.

Table with ‘mwa jem’: based on Lambrecht (1981), 15, Schwegler (1990), 112, Klausenburger (2000), 25, 83. The original pronouns now form a phonological unit with the verb, they cannot be stressed, they never appear in isolation.

Origin of case ending -ibus: Szemerényi (1996), 165.

Latin ‘wolf’: The Latin lupo actually started out in Indo-European as *wlkwo, and has the same origin as English wolf.

‘I knew I had to say thank you’: Mann (1986), 185.

Aujour d’aujourd’ hui: the phrase has in fact quite a long pedigree, and is even attested in literary genres, for instance in Alphonse de Lamartine’s Méditations poétiques (1820): ‘[Dieu] le sait, il suffit: l’univers est à lui / Et nous n’avons à nous que le jour d’aujourd’hui!’

Cycles of auxiliaries fusing with the verb: In the era separating Indo-European from modern French, three such cycles took place. The Latin future endings (as in canta-bo ‘I will sing’) must have originated from the fusion of an auxiliary ‘be’ with the verb (or as an analogical formation on the fusion of ‘be’ in another construction). By the time of Late Latin, these future endings had been weakened, but then a cycle started again when another auxiliary, ‘have’, merged with the verb to give a new set of endings in early French (chanter-ax ‘I will sing’). In modern colloquial French, a new cycle is underway, with constructions such as je vais chanter ‘I’m go(ing to) sing’.

‘That which one calls construction comes’: Paul (1880), 351: ‘Das, was man Aufbau nennt, kommt ja … nur durch einen Verfall zu Stande, und das, was man Verfall nennt, ist nur die weitere Fortsetzung dieses Prozesses.’

‘Merging of two words into one is excessively rare’: Bloomfield (1933), 415.

CHAPTER 6: CRAVING FOR ORDER

Role of analogy in cognition and intelligence: Holyoak and Thagard (1995), Gentner et al. (2001).

Grotty and grotesque: ‘grotesque’ itself came from a noun ‘grotto’. The Oxford English Dictionary, s.v. grotesque speculates that ‘the etymological sense of grottesca would be “painting appropriate to grottos”. The special sense is commonly explained by the statement that grotte, “grottoes”, was the popular name in Rome for the chambers of ancient buildings which had been revealed by excavations, and which contained those mural paintings that were the typical examples of “grotesque”’.

One ‘Weetabick’ and ‘a chee’: Burridge (2004), 10.

The Semitic languages: Suggestions for further reading: On the language, culture, and history of the Akkadians (Babylonians and Assyrians), see A. L. Oppenheim, Ancient Mesopotamia: Portrait of a Dead Civilization (1964); J. Oates, Babylon (1986), and for the legacy of this culture S. Dalley, The Legacy of Mesopotamia (1998). For a collection of articles on the language, culture, writing, and history of the Ancient Near East, see J. M. Sasson, Civilizations of the Ancient Near East (2000). On the history of Hebrew, see A. Sáenz-Badillos, A History of the Hebrew Language (1993). On the culture and history of the Phoenicians, see S. Moscati, The World of the Phoenicians (1999). On Aramaic and the Aramaeans, see E. Lipinski, The Aramaeans: Their Ancient History, Culture, Religion (2000). For an introduction to Arabic and its history, see K. Versteegh, The Arabic Language (1997). A linguistic survey of the Semitic languages can be found in R. Hetzron, The Semitic Languages (1997).

‘This is what Babi says to Shartum’: The text of this Old Akkadian letter was published by Veenhof (1975) (see also Kienast and Volk (1995), 153). The tablet is currently kept in the Vrije Universiteit, Amsterdam. The photograph was taken by Guy Deutscher.

Semitic is family with longest history: the only serious contestant to the length of attested history is ancient Egyptian, whose written records start around 3000 BC, and whose later descendant, Coptic, was spoken until around AD 1500.

Template variants and In fact, there is also a third alternative, but for simplicity this will be ignored here.

Arabic verbs aktum, aftil: The form given here is the short form, called the ‘jussive’. The normal form has an ending -u. Note, moreover, that due to a secondary development in the history of the West-Semitic languages (including Arabic, see tree on p. 179), the Arabic forms in question have come to denote the non-past rather than past tense. But the transformation in the tense-aspect system of the West-Semitic languages is a later (and very complex) development which need not concern us here.

The emergence of the simple past-aktum: Forms like aktum may not originally have had a past-tense meaning, but could have been non-differentiated for tense. The past nuance may have developed later, when a new form emerged with a specific present/future meaning, thus displacing the original to the territory of proper past.

Quirk vowel as an indication for the original stem vowel: In some Semitic languages (such as Hebrew), the root vowels of the original verb have undergone thorough transformation, and have been regularized and standardized, so as to make them quite unreliable for determining the identity of the vowel in the ancient stem.

a-ā (‘I will laugh’): In the earlier stages of Akkadian, the future appears as a-iā, which later contracts to a-ā. See Appendix B: Laryngeals Again?

Antiquity of the a-mutation: On the reconstruction of a-mutation to Proto-Cushitic, see Zaborski (1975), 163–5. Diakonoff (1988), 86 postulates the same for Proto-Berber. There is also evidence from within the Semitic languages themselves to suggest that the pattern is very old. One reason is that when records begin in the third millennium BC, the a-mutation pattern seems already to be receding. In Akkadian itself, and even more so in the later attested Semitic languages, speakers used different means to bring more and more of the hollow verbs into the fold of the regular three-consonant system (for example, by repeating the second consonant of the root twice). A further indication for the antiquity of the a-mutation comes from some of the most common verbs in Akkadian, such as ‘go’ and ‘give’. These verbs do not misbehave in exactly the same way as the hollow verbs, but they do show the a-mutation in the future tense: a-ddin (‘I gave’) a-ddan (‘I will give’), a-llik (‘I went’) a-llak (‘I will go’). Since the simplest and most common words often cling on to ancient patterns that have been abandoned by the rest of the language, the a-mutation in verbs like ‘give’ and ‘go’ is a strong reason to suspect that we are dealing with a very old relic. There are also indications that the a-mutation in Semitic originally applied to three-consonant verbs as well, see Knudsen (1984).

Origin of the vowel mutation (ablaut) in Germanic verbs: Szemerényi (1996), 87, 121.

Grounds for suspecting that there were more two-consonant verbs in prehistory: First, the decline in the ranks of the hollow verbs during the historical period itself suggests that the further back in time one goes, the more numerous they were. This suspicion is strengthened when one looks at other branches of the Afro-Asiatic family, which even today have more verbs with two consonants than with three. It is thus possible that they reflect the more ancient state of affairs, in which two-consonant verbs were the majority. For a critical assessment, see Zaborski (1991).

Swelling cycles: In a few cases, the old and the new verbs still coexist in the attested period. One example is the verb kūn (‘to be firm, upright’), and ša-kun (‘to make firm’, ‘install’) – it seems clear that the latter is a swollen version of the former. In other cases, such as with *pil (or perhaps pal or pul), the original verb is no longer attested, but we can suspect it was once there, because it has swollen in two different direction, both with the prefix ša-, to give the root š-p-1 (‘become lowly’), and with the prefix na-to give the root n-p-1 (‘fall’).

The psychological reality of consonantal roots: Prunet et al. (2000).

CHAPTER 7: THE UNFOLDING OF LANGUAGE

Conceptual distinction between ‘things’ and ‘actions’: On possible neurological precursors and correspondences to the thing-action distinction, see Hurford (2003), and Givón (2002b), 17.

M. Jourdain’s skills: See also Hawkins (1994), 440–41.

Choice between ‘man spear throw’ and ‘man throw spear’: Givón (1979), 275ff. has suggested, in fact, that the order ‘man spear throw’ was the dominant order in earlier periods. His reasoning is based partly on the observation that in historical time, changes in word order from ‘man spear throw’ to ‘man throw spear’ are very common, but changes in the opposite direction, from ‘man throw spear’ to ‘man spear throw’ are exceedingly rare.

The reliance of ancient languages on time iconicity (Caesar’s principle): Deutscher (2000), 175ff.

Murshili’s aphasia text: R. Lebrun (1985), 103–37, Farber et al (1987), 289. The Hittite conjunction nu is cognate with the English ‘now’, but its use corresponds more closely to English ‘and’. In Hittite, sentences always have to start with some linking particle (except for the first sentence in the ‘paragraph’). Nu is the most common of these sentence initial particles, and in this sense, ‘and’ is the closest translation. The place name Kunnu is preceded by the logogram DU6 which means ‘ruin mound’, so some scholars have normalized it as Tell-Kunnu.

‘I am Tarzan of the Apes. I want you…’: Burroughs (1914), 195. In fact, the proverbial ‘me Tarzan, you Jane’ line does not even appear in this precise form in the 1930s Tarzan film. The origin of the line is apparently much less glamorous, and seems to have been a quip made in a Hollywood car-park by Johnny Weissmuller (Tarzan) to Maureen O’Sullivan (Jane). Seeing her struggling to lift a heavy suitcase, Weissmuller effortlessly swooped it up and tossed it into her car, muttering, ‘Me Tarzan, you Jane.’

‘Monsieur est étranger?…’: ‘You are a foreigner, Sir?’… ‘Foreigner? Certainly not. I’m English!’

Pointing words (demonstratives): On children’s use of demonstratives, see Clark (1976), Diessel (1999), 111, Tanz (1980). For the primitive status of demonstratives, which are not known to develop from any construction that does not already include a pointing (deictic) element, see Diessel (1999). There have been some suggestions in the literature (Heine and Kuteva (2000), 159) for a development from the verb ‘go’ to a demonstrative, but these seem quite speculative.

Equivalence of ‘this’ and ‘here’, ‘that’ and ‘there’: Diessel (1999), 14, Heine and Kuteva (2002), 172. In English, both ‘there’ and ‘that’ derive from the same Indo-European root *to-. In Korean, ieki ‘here’ comes from i-eki ‘this place’, and keki ‘there’, from ke-eki ‘that-place’ (Diessel (1999), 21), but derivation in the opposite direction is also found, where ‘this’ comes from something like ‘thing here’, and ‘that’ from ‘thing there’. Very common is a development in which an element from one pair is reinforced by elements from the other pair. This can be seen, for instance, in French ceci ‘this’ and cela ‘that’, which come from a combination of ce ‘that’ with ici ‘here’ and là ‘there’. Similarly, Swedish den här (literally ‘that here’) is used for ‘this’, and den där (literally ‘that there’) is used for ‘that’.

Origin of pronouns: For Vietnamese, see Nguyen (1992), 181 (có is an affirmative particle which I left untranslated). Another example is the Akkadian ending -am, which originally meant ‘towards here’, but has developed into a pronominal ending ‘to me’, and in fact became the usual way of expressing first person indirect object, see Kouwenberg (2002). But pointing words are not the only source from which pronouns can develop. First person pronouns are known to derive from various designations for ‘man’, ‘person, ‘people’, e.g. Latin homo ‘man’ > colloquial French on ‘we’.

Latin ille: Harris (1978), 100–101.

Akan ‘give’: Lord et al. (2002).

Chinese ‘give’: Peyraube (1988), 207, Peyraube (1996), 178–82.

Collapse of two actions into one: Thai examples from Blake (2001), 161–2, see also Matisoff (1991), 439 for similar use of the Thai verb paj ‘go’. Efik example from Heine and Kuteva (2002), 289, Chinese from Peyraube (1996), 191. On the perception of the two original actions as one event, see Givón (1991).

‘The day is short and it is passed pryme’: Chaucer’s ‘The Friar’s Tale’. In modern times, a similar process can be observed with the verb ‘following’, see Olofsson (1990). In a phrase such as ‘the people following us’, ‘following’ is still a verb denoting movement, and one can replace it by other verbs, such as ‘chasing’/’trailing’/’leading’, or by ‘who follow’: ‘the people who follow us’. But today, ‘following’ is used more widely, in phrases such as ‘she dumped him following their row’, where ‘following’ is no longer in a verb slot (one cannot say ‘she dumped him who follow their row’) and must be understood as a preposition. The more established preposition ‘during’ started out just like ‘following’, as a participle of the verb ‘to dure’ (which has since gone out of use).

‘With’ as a source of ‘and’: For Chinese gēn, see Peyraube (1996), 191; for Turkish and Swahili, see Heine and Kuteva (2002), 80–83; for African languages, see Lord (1993), Ch. 4.

‘Red’ and ‘Adam’: The connection between ‘red’ and ‘Adam’ was probably the word adama ‘(red) soil’. The Book of Genesis related Adam to adama by claiming that God ‘formed man of the dust of the ground, and breathed into his nostrils the breath of life’. This scenario for the origin of human beings may not seem very likely, but modern linguists believe that the etymology is in fact quite plausible. A similar etymological relation is found in Latin humanus (‘human’) and humus (‘soil’).

Combination of demonstrative and noun into a complex referring expression: Himmelmann (1997), Diessel (1999), 69. For looser status of ‘that’ in Old English, see Traugott (1992), 173. It is interesting that among the two-sign combinations that the bonobo Kanzi produces, by far the most common ones involve using a symbol for an object (say ‘peanut’) followed by a physical pointing gesture ‘that’, see Savage-Rumbaugh and Rumbaugh (1993), 101.

Origin of articles, quantifiers, plural markers: Haspelmath (1995), 363–82, Heine (1997a), 71ff. Note that unlike property-words, elements such as plural markers and articles are not purely subordinate ‘modifiers’, since they do interact with other elements on the main level of the sentence. For example, the verb changes according to whether the subject noun is singular (the duck swim-s) or plural (the ducks swim). So as opposed to a modifier (such as ‘old’ in ‘the old duck swims’), which has no direct relation with any other element in the sentence except the noun it modifies, the relation between the noun and the plural marker -s is a symbiosis in which each of the elements determines part of the phrase’s profile towards the rest of the sentence. (On how such a symbiosis develops historically, see Himmelmann (1997).) Nevertheless, for simplicity of presentation, I subsume here under the general term ‘appendage’ not only pure modifiers, but also all the other elements in the noun phrase except for the head noun, including determiners (such as the definite article) and quantifiers (such as ‘all’, ‘each’). Concerning the development of ‘all’ from ‘whole’, note that the terms for the property ‘whole’ itself often develop from even simpler physical properties such as ‘healthy’ or ‘undamaged’. The English word ‘whole’ has the same origin as ‘health’, and its close synonym ‘entire’ is borrowed from Latin in-teger, literally ‘un-touched’.

‘Of these three states’: from the homily Hali meidenhad (c. 1230), ed. Furnivall (1922), 23.

Asymmetry between nominalization and verbalization: Hopper and Thompson (1985), 176, Woodworm (1991), 62ff. It is true that English can turn a verb into a noun without any ending: ‘take a walk’, ‘have a go’, and so on, but many languages cannot.

Development of French -age. The ultimate origin of -age is the Latin ending -(a)ticus, which already in Latin had long been established as a thoroughly grammatical marker, whose function was to turn a noun or adjective X into ‘someone/something of or belonging to X’ (as in the Late Latin silva ‘forest’ and silvaticus ‘someone/something belonging to the forest’, which ended up in modern French as sauvage ‘savage’). On the history of -age in French, see Meyer-Lübke (1966), 61–3. It is not known with which noun(s) exactly the passage of -age to verbs started. I chose manage because it is a familiar word, but Meyer-Lübke uses a different example, the measure ‘aune’ (English ‘ell’, the measure of an outstretched arm), to which -age was added to give aunage (like English ‘mileage’). Since aune was also turned into a verb auner ‘to measure by the aune’, speakers could have associatied the abstract noun aunage directly with the verb auner.

Origins of nominalization markers: Various types of nominal markers are extended to verbs, for instance case markers (Blake (1999), for European and Australian examples), nominal classifiers (Aikhenvald (2000), 332), and various abstract-noun markers. The latter themselves often originate from collectives or designations of time and place, such as the Germanic -ing/ung (which must originally have been denominal, see Jespersen (1948), 205 and Kluge (1995), s.v. -ung: ‘Letztlich liegen ig. k-Erweiterungen zu n-Stämmen vor, so daß das Suffix ursprünglich denominal gewesen sein muß’), German ge- (as in Ge-spräch), or the Basque suffixes -te, -tze, and -keta (Trask 1995). Even zero-nominalization (as in English ‘a walk’, ‘a run’) could have originated ultimately from an equivalent process of analogical back formation: a noun ‘an X’ can be turned into a verb ‘to X’, but if the verb at a later stage is perceived to be the basic element in the pair, speakers can assume that the derivation went in the other direction, and so by analogy extend the pattern to other verbs.

‘she was on the riding’: Strictly speaking, the presentation of such developments should have waited until the next section, because they involve embryonic subordination, where one verb functions as a dependent of another. But as the next section will argue, the ability to present a verb as a dependent of another has in fact already been achieved in the previous section, with the ability to nominalize verbs.

‘Finish’ developing into a markers of completed action: Chinese example from Li and Thompson (1981), 186. For the verb ‘finish’ as the source of tense and aspect markers in other East Asian languages, see Matisoff(1991), 436. Markers of completed action often themselves turn into more general indicators of past tense, as in modern German and French, where what had originally been the markers of a completed action ‘I have eaten’ (Ich habe gegessen, j’ai mangé) has come to be used as a past tense more generally. On the sources of other tense, mood, and aspect markers across languages, see Bybee, Perkins and Pagliuca (1994).

Shift from possession to obligation: Further examples in Matisoff (1991), 427.

Fidel Castro on a state visit to Moscow: Based on Lukes and Galnoor (1987), cited in Smith and Tsimpli (1995), 77.

Demonstratives (pointing words) in apposition as the source of relative markers: Deutscher (2001).

Expansion of nominalized verbal forms to relative clauses: Matisoff (1972), DeLancey (1986), Genetti (1991), Harris and Campbell (1995), 291–3, 310–13, Noonan (1997), Blake (1999).

Origin of question words and negation words: Since interrogatives (question words) did not feature in the mammoth story, their origin was not discussed in the chapter. But in fact, it seems that like demonstratives (see note for p. 228), they also have a primitive status, since they are not known to develop from nouns and verbs, or from anything else that does not have an interrogative element to start with (Diessel (2003)). This means that in addition to the demonstratives ‘this’ and ‘that’, one interrogative element (such as ‘what’) should also be assumed as part of our ‘raw materials’. The whole range of question words could then develop from this interrogative element, on the lines of ‘what-place’ > ‘where’, ‘what-person’ > ‘who’, and so on. Negation words are known to develop from more specific (intrinsically negative) verbs such as ‘leave’, ‘abandon’, ‘lose’, ‘fail’ (see Heine and Kuteva (2002), 188, 192).

EPILOGUE

Sumerian munintuma’a: From an inscription of Enannatum, ruler of Lagash, c.2450 BC (Steible (1982), 185). On the structure of such forms in Sumerian, see Attinger (1993), 192ff., Edzard (2003), 87ff. The actual developments which created the structure of the Sumerian verb must have been extremely complex. An attempt at reconstructing a few of the details is offered in Coghill and Deutscher (2002). For a general discussion of the investment of zero-morphemes with specific meaning, see Bybee (1994).

Erriplen in Gurr-goni in the ‘vegetable gender’: Aikhenvald (2002), 408.

Drift towards simpler word structure as ‘debasement’: Towards the end of the nineteenth century, some linguists tried to swing the pendulum to the other extreme and claim that the drift was actually ‘progress’. Most prominent was the Danish scholar Otto Jespersen (1894).

Examples for new endings and prefixes emerging in Indo-European languages: Persian, for instance, has acquired a new dative (and later accusative) case ending from an erstwhile postposition, and in Czech, prepositions have merged with the noun to give case prefixes, although this is not reflected in the spelling.

The variety of subordinate clauses has increased in modern European languages: Kortmann (1997) for adverbial subordination, Deutscher (2000), Chs.10–11 for complementation.

Patterns of communication in small societies: Givón (2002a), Ch.9.

Correlation between simple societies and complex morphology: Dixon (1997), Kusters (2003), Perkins (1992), Trudgill (1992). On contact as a factor that encourages simplification, see Trudgill (1992), 195–212, who takes as one example Norwegian and Faroese (a very small and until recently very isolated community), which have both developed from a common ancestor fairly recently. But whereas Faroese has kept much more of the endings of the ancestor language, Norwegian has lost almost all of them.

Factors that may contribute to the drift towards simpler morphology: Another factor which has been advanced to explain the slower rate of new fusions in the historical period is the change in word order from OV to VO (see Appendix E: The Turkish Mirror) that many Indo-European languages have undergone during the historical period. There seems to be a marked asymmetry between the rate of suffixation and prefixation, with postpositions, for instance, tending to fuse more easily than prepositions (see Cutler et al. (1985), Hall (1988), Bybee et al. (1990)). As OV languages tend to have postpositions and post-verbal auxiliaries, whereas VO languages prepositions and pre-verbal auxiliaries, the change from OV to VO could have decreased the rate of new fusions.

Languages of the world are vanishing: For estimations of the rate of language loss, see Crystal (2002), 19, and Dixon (1997), 116.

APPENDIX A: FLIPPING CATEGORIES

Use of ‘gonna’ in American English: In African American English, gonna has been further shortened to gon, and the auxiliary ‘is’ is no longer present in constructions such as ‘you gon come’. But in questions and negative sentences, an auxiliary is inserted, as in ‘you ain’t gon(na) come’ (Green (2002), 36, 40).

‘All grammars leak’: Sapir (1921), 38.

APPENDIX B: LARYNGEALS AGAIN?

Effect of laryngeals in changing vowels to a in Semitic languages: Lipinski (2001), 45.10,14, Mosacti et al. (1964), 16.110.

The pattern aī – aīa extended to verbs without a laryngeal: For reasons of presentation, I use a hollow verb as an example. But this is not meant to imply that the a-mutation had its source specifically with hollow verbs. I could just as well have used a three-consonant verb with a final laryngeal as an illustration.

APPENDIX C: THE DEVIL IN THE DETAIL

Reflexive template: Lieberman (1986), 610ff. offers a survey of the t-sterns in Semitic and Afro-Asiatic languages.

Intensive template: for a comprehensive discussion of gemination (doubling of the middle consonant) in Semitic, see Kouwenberg (1997), and especially pp. 445ff. for a historical scenario. Trukese data from Goodenough (1963). For Indo-European, see Niepokuj (1997).

Causative template: For the view that ša comes from a verb ‘make’, see Hodge (1971), 41. Heine and Reh (1984), 276 reconstruct a verb *iss/*ass ‘make’ for Proto-Cushitic, and think it is the origin of the Somali causative suffix -is. For discussion see Tropper (1990), 8ff. Some scholars have argued that ša is pronominal in origin, but cross-linguistic evidence makes this unlikely.

Passive template: Testen (1998), Kouwenberg (2004). Zaborski (2001), 595 suggests as etymology the Bedja verb n/nV ‘to be’.

APPENDIX D: THE COOK’S COUNTERPOINT

‘I remember when Mrs Cibber’: Burney (1785) (‘Sketch of the Life of Handel’), 33.

‘once he had dressed…’: Beowulf, line 1472; ‘because he wanted to send to us him himself’: Christ I, line 129. We know that sylfne is emphasized, since it alliterates with sendan.

Norwegian hans kokk and sin kokk: It is much more usual in Norwegian to say ‘kokk-en hans’ (cook-the his) than ‘hans kokk’ (his cook), but for ease of comprehension, I chose the more formal (ultimately Danish influenced) version.

APPENDIX E: THE TURKISH MIRROR

Greenberg’s discovery of word-order correlations: The original publication detailing these discoveries appeared in Greenberg (1963). For a modern assessment and revision, see Dryer (1992), Croft (2003), Ch.3. For diachronic explanations for these correlations, see Aristar (1991), Givón (2001), Ch.5, and for a general discussion of the role of diachrony in linguistic universals see Givón (2002a), Ch.6.

Two possessive constructions in English as relics of a change from OV to VO: The actual picture is much more tangled than presented here. For a history of the two constructions and their use, see Rosenbach (2002), Ch.7, who shows, among others, that even in the Old English period there was already some flexibility in the placement of the head-noun and the genitive, and that in later stages of English, the status of the ‘5 possessive marker changes from a case marker to a marker with a phrasal scope. For a discussion of how and why languages change from OV to VO, see e.g. Harris and Campbell (1995), Ch.8.

Correlation between HEAD-APPENDAGE and prepositions: The picture presented here is highly simplified. For example, cumbersome constructions such as ‘of-of-Ruritania ruler son’ and ‘son ruler Ruritania-s-s’ only refer to a situation where: (a) the marker of possession appears on the appendage rather than on the head noun, (b) this marker is a preposition or postposition that has a whole phrase as its scope, rather than a case marker that applies only to just one word. When either of these conditions is not met, different constructions arise. But the examples here are only meant to demonstrate the type of processing difficulties that may arise with ‘inconsistent’ ordering of certain pairs. For the most thorough examination of processing issues relating to word order, see Hawkins (1994), Hawkins (2004).