Introduction

We have seen that causativization is one of the important factors underlying verb alternations in English, and one which I have argued is built in to the interpretation of verbal decompositional structure in a fundamental way. In this chapter, I look more closely at the morphology associated with causativization in one language, Hindi/Urdu, which productively constructs transitive verbs from simpler, usually intransitive bases. The argument here will be that in accounting for the regular morphology and its syntactic/semantic consequences, we can get some justification for the abstract system of primitives being argued for in this book. At the same time, the comparison between English and Hindi/Urdu will allow us to formulate some specific hypotheses about the nature of parametric variation in constructing verbal meaning. The larger picture will be, firstly, that there is explicit evidence for decomposition from morphological and analytical constructions, and secondly, once again, that languages vary only in the ‘size’ of their lexical items, not in the fundamental building blocks of eventive meaning.

In chapter 4, I discussed the debate in the literature concerning the direction of the causative–inchoative alternation. Recall that Levin and Rappaport Hovav (1995), Chierchia (2004) and Reinhart (2002) all agree in deriving the inchoative alternant from a lexically causative base. I argued in chapter 4 that the structure of the conceptual argument dissolves once one moves to a nonlexical, structure building framework. While the morphological evidence from some languages where the inchoative version of a verb transparently contains the verb itself plus a piece of ‘reflexivizing’ morphology (e.g. si in Italian, se in French, sja in Russian) seems to support a detransitivization story, typological work shows that this is not generally the case across languages (Haspelmath 1993).

Given the outlines of the system presented so far, we can use the primitives at our disposal to discuss the different natural classes of verb that emerge from this kind of syntactic organization. I wish to emphasize that this chapter is not intended primarily as a detailed investigation of any one set of phenomena. The purpose is to explore the ways in which this system can be used to analyse different possible verb types – the emphasis will be on the flexibilities and constraints on the system itself. In each case I indicate what verb types are possible and how they might be instantiated in English. I describe what the most natural mode of analysis would be for many common verb classes in English that have received treatment in the literature, pointing out where the system forces one to make choices between various analytical options. In all cases, the particular structures proposed here are intended to be starting points for more detailed research. In chapters 5 and 6, I take up the issues of path construction and causative formation respectively in more detail, and attempt to make some more substantive proposals.

In general, by taking seriously the event-structure participanthood of arguments, I aim to show that a somewhat different classification of verb types emerges. In addition, the system I am arguing for will allow flexibility in a verb's syntactic behaviour, within a system of constraints. Importantly, some of the previous principles of mapping between lexical information and syntax assumed in the literature will be abandoned.

Words and texts cited in this book generally follow established conventions of transliteration or citation and are not given in IPA transcription. The following notes are intended to guide the reader to the pronunciation of forms cited in this book. Since in many cases the languages are no longer spoken, there is often uncertainty about the precise realisation of certain sounds, and the pronunciations given here can only at best be approximate. It should be noted that we have not attempted to give comprehensive accounts of the phonologies of the languages concerned, but merely to aid readers to understand how a particular sign is used. In general we have avoided giving details of signs which are not used in this book. Where no information is given on the pronunciation of a sign, the reader can assume that it has a value approximately equivalent to its IPA equivalent. In all cases we have tried to follow the standard orthography used in the scholarly literature, except in the case of Greek, for which we have not used the Greek alphabet, but a transliteration which should make it accessible to all and enable readers who know Greek to recognise the original words.

The PIE lexicon

Comparative reconstruction begins with the comparison of lexical items across different languages, but most works on IE operate largely on the basis of roots and affixes rather than lexemes. The standard etymological dictionaries of IE (see Pokorny (1959), Rix et al. (1998) and Watkins (2000)) present the IE lexicon as a collection of roots, each of which has limited semantic scope and from which a number of derived stems can be formed. For example, the correspondence between Latin agō ‘I drive, lead’ and Greek ágō ‘I lead’ and Sanskrit ájāmi ‘I drive’ is listed under a root *h2eg′- (or *ag′-) rather than as a separate lexical entry, and a unified meaning is given for the root as a whole. Listed in the same dictionary entry one may find nominal derivations such as *h2eg′-mn, which is the hypothetical ancestor of Latin agmen ‘procession, military column’ and Sanskrit ájman- ‘course, procession’, or the adjectival form *h2(e)g′-to-, continued as the Latin participle āctus ‘driven’ and found in the Celtic compound ambactos ‘servant’. Furthermore, the word-equation of Latin ager ‘field’, Greek agrós ‘field’, Gothic akrs ‘field’ and Sanskrit ájras ‘plain’ is given by Pokorny and Watkins as a derivative of the same root, on the supposition that the original root was connected with animal herding, and *h2eg′-ro- designated the space into which animals were driven.

Introduction

We suspended discussion of the nature of agrammatic comprehension impairment in chapter 12, in order to consider the role of working memory in sentence processing and to review the methodology of on-line language assessment. With a clearer picture of the alternative (but not necessarily irreconcilable) concepts of working memory resources utilized in volitional and automated language processing, together with an appreciation of the current state-of-the-art in behavioural and neural techniques for monitoring moment-by-moment fluctuations in processing load, we are better equipped to critically evaluate competing theories of receptive agrammatism. But to avoid needless confusion that often attends discussion of this topic, let us be clear what we mean by ‘receptive agrammatism’, how it relates to the clinical classification of aphasia (Broca's, Wernicke's, anomic, conduction and transcortical aphasia), and why this particular language syndrome has preoccupied neurolinguistic research more than any other over the last quarter century or so.

Receptive agrammatism refers to a pattern of comprehension impairment that is revealed by psycholinguistic investigations of the kind described in detail in chapter 12. Subjects manifest an inability to use syntactic cues for sentence comprehension in tests of thematic role assignment, where pragmatic and lexical cues to meaning are rigorously controlled by selection of sentence materials and other aspects of the testing situation. A pattern of comprehension impairment that can be identified as receptive agrammatism has the following attributes: (1) not better than chance performance for agent identification on reversible passive constructions, (2) poor performance on object relative clauses and other structures involving departures from canonical word order, and (3) selective ‘blindness’ to the presence of semantically opaque function words or grammatical affixes.

Introduction

A satirical piece in the New Yorker, “Talking chimp gives his first press conference,” parodies trained primates, the media, the scientific community and a number of other institutions. “Hello? Can everyone hear me? Anyone? Check, check. Check, one two. Is this thing on? Not the microphone – I mean my Electronic Larynx Implant device … The development of the ELI was a long and arduous process, and there were more than a few times – usually after being shot with a tranquillizer dart and then waking up hours later with excruciatingly painful bleeding stitch holes in my neck and chest regions – when I wasn't sure if it was worth it. But I guess it was, because here we are today, in this beautiful conference room at the Sheraton” (Simms 2005, 44). The humor of this monologue derives not so much from the reportorial scientific discourse or the monkey-business of the speaker (whose preoccupations are kibbles, orange wedges and bodily functions), but rather from the irony of our closest relative's inability to produce fluent language. Trained primates have learned vocabulary such as red toy or orange, and might be able to communicate (with signs) “hello” or “check, check,” but they could never articulate the complex syntax, temporal displacement and sophisticated vocabulary of the first full sentence.

Introduction

Indo-European (IE) is the best-studied language family in the world. For much of the past 200 years more scholars have worked on the comparative philology of IE than on all the other areas of linguistics put together. We know more about the history and relationships of the IE languages than about any other group of languages. For some branches of IE – Greek, Sanskrit and Indic, Latin and Romance, Germanic, Celtic – we are fortunate to have records extending over two or more millennia, and excellent scholarly resources such as grammars, dictionaries and text editions that surpass those available for nearly all non-IE languages. The reconstruction of Proto-Indo-European (PIE) and the historical developments of the IE languages have consequently provided the framework for much research on other language families and on historical linguistics in general. Some of the leading figures in modern linguistics, including Saussure, Bloomfield, Trubetzkoy and Jakobson, were Indo-Europeanists by training, as were many of those who taught in newly founded university departments of linguistics in the second half of the twentieth century. Despite this pedigree, IE studies are now marginalised within most university linguistics courses and departments. In most US and European institutions, Indo-Europeanists with university posts do not teach in linguistics departments but in classics, oriental studies, celtic studies or the like. Historical linguistics courses may include a section on PIE, or Saussure's work on laryngeals as an example of internal reconstruction, but few students will engage in any current work on IE in any depth.

Reconstruction and the comparative method

Current research into the Indo-European language family largely involves linguistic reconstruction. Reconstructing aspects of the parent language is both an end in itself and an aid to understanding the links between the languages in the family and explaining their historical development. In Indo-European studies, reconstruction has enabled linguists to interpret texts in languages which have left only scanty linguistic remains and which would be otherwise largely obscure (as in the case of Lusitanian discussed in section 1.2). It is possible to reconstruct any aspect of the parent language, but the crowning achievement of comparative linguistics is phonological reconstruction. There is a broad consensus among scholars that the phonemic inventory of PIE can be reconstructed fairly accurately, although there is still debate about the phonetic realisation of the phonemes. Most Indo-Europeanists would place greater confidence in the reconstructed phonemic system than in many of the reconstructions of individual lexemes or morphological or syntactic phenomena.

How does this confidence in reconstructed phonemes come about? As an example, let us consider the comparison of English, Dutch and German, which are all members of the Germanic branch of Indo-European. Any speaker of one of these languages will see similarities in the vocabulary and grammar of the other two. An English speaker learning Dutch and German, for example, cannot fail to notice that the words for ‘bread’ and ‘water’ in the two languages (brood and water in Dutch, Brot and Wasser in German) are extremely close.

Introduction

In the chapter “Baby born talking – describes heaven,” Pinker (1995, 269–270) quotes Adam, the toddler whose corpus of evolving speech has furnished data to generations of scholars (the first number after Adam's name refers to the year and the second to the month of his age).

Adam, like all children, was not born talking, but by two years three months he was combining words in a toddler-like fashion; a year later his language was nearly that of the adults and children he had been listening to for the first three years of his life.

Introduction

In the two preceding chapters, we have explored in a preliminary way two different paths to understanding the human ‘language faculty’ (Chomsky, 1965; Jackendoff, 1997) or our capacity for spoken language communication. The linguistic approach seeks to isolate and describe the elements of a system of spoken communication by studying varieties of linguistic expressions in the world's languages and human language in general. The neuropathological approach examines types of language breakdown in response to brain damage of various kinds. It is hoped that the search for parallels or correspondences in these two very different domains will yield empirical constraints on a theory of language that could not otherwise be discovered if these two strands of inquiry were conducted in isolation from one another. For example, a fundamental distinction that grammarians draw between lexis and rule in the architecture of the language faculty may turn out to have a correspondence – or not – in the classification of language pathologies, reflecting the organization of language capacities in the human brain. We have already provided you with some classical findings from these two domains, which provides at least a foundation for speculation and further inquiry.

However, it is time to draw some critical methodological distinctions in the interests of making our search for correspondences and a cross-disciplinary theory of language more precise. The distinctions that we draw here will anticipate issues discussed more fully in subsequent chapters.