This book "supplies a vocabulary of English words and idiomatic phrases 'arranged … according to the ideas which they express'. The thesaurus, continually expanded and updated, has always remained in print, but this reissued first edition shows the impressive breadth of Roget's own knowledge and interests."
Carstairs-McCarthy, Andrew. 1999. The Origins of Complex Language: An Inquiry into the Evolutionary Beginnings of Sentences, Syllables, and Truth, Oxford University Press. Paperback ISBN: 0-19-823821-5, xi+260pp, $19.95.
Reviewed by John Hammink, F-Secure Corporation, Helsinki
Linguistic origin and evolution has long been an issue of debate among a range of academic disciplines, including linguistics, anthropology, archeology, and biology. However, to my knowledge, few works have attempted to explain, in the broadest scope possible, certain aspects of language, which other evolutionary accounts either render ambiguous or simply leave unexplained. This book is unique in that it is able to merge many apparently distinct or specialized views, and more unique still in that it calls directly for more comment on the issue. Even though the book "makes no assumptions about the readers prior knowledge" a good general knowledge of linguistics, particularly phonology and syntax, are helpful. However the introductory chapter asks the reader to take "an outside view" and "visualize alternative directions in which language might plausibly have travelled" were it not for certain underlying peculiarities of human language.
Overview One of these peculiarities-namely, the universality of the syntactic distinction between sentences and noun phrases (NPs)- is a cornerstone of the inquiry, and the author's perspective continually supplies a reason for it consistent with the other arguments raised. This is eventually reconciled with a syllabic theory of syntactic development- that is, modern universal syntax is derived from a syllabic model associated with the descent of the larynx in proto-humans.
Towards the end of Chapter 1, where the real book seems to start, the author lays out the plan of the book and encourages the reader to proceed through the chapter in the order he/she chooses--some arguments may be better clarified by reading later chapters first and some chapters (namely chapter 3) provide auxiliary but not essential material to the course of inquiry.
In Chapter 2 the central argument begins as we look at three universal peculiarities of human language:
-duality of patterning: Linguistic expressions are analysable on two levels: 1. As composed of meaningless elements (sounds belonging to a finite inventory); 2. As composed of meaningful ones (words and phrases);
-the distinction between sentences and NPs.
So what is so peculiar about these "peculiarities"? First, vocabulary size. The point here may otherwise be brain size, but the author seems to allude to the concept of organization when he starts to talk about duality of patterning. Duality of patterning, of course, provides a mechanism whereby vocabulary can be extended, because elements are recombinable on both levels. Is duality of patterning a direct consequence of vocabulary expansion? Maybe. Or maybe not. While some new words are combined from purely phonological resources (nerd, gizmo, quark) most new vocabulary is derived from existing vocabulary items, either through extending the meaning of existing words, compounding, blending, and through meaning change and affixation. These mechanisms do not need duality of patterning to operate.
We are taken down a slightly different course when learning about the distinction between sentences and NPs. The author shows us 3 different scenarios for what could have been the evolution of syntax. First, no syntax: Languages might have large vocabularies and exhibit duality of patterning, but sentences are otherwise strung together in a "rough-and-ready" manner, with no grammatical constraints. To understand "Asyntactic", one must rely solely on common sense or pragmatic clues. The second syntax scenario makes use of the Spatiotemporal Co- ordinates and the Type/Token Distinction. A sentence in "Spatiotemporal" may be constructed as follows: Location in space + location in time + Object or Event. For example "In Riga + Yesterday + Man bite dog". The third syntactic scenario, called Monocategoric, lacks category distinctions between expressions. There are only simple expressions (Snake, you, John, Mary, story) and operators (YESTERDAY, DISAPPEAR, SEEM, SEE, TELL). Thus, the monocategoric [you snake SEE]YESTERDAY could be glossed as 1. 'You saw a snake yesterday' 2. 'your seeing a snake yesterday' 3. 'the snake you saw yesderday' 4. 'you who saw a snake yesterday'. Note that none of the above scenarios, asyntactic, spatiotemporal, or monocategoric require or make a distinction between sentences and NPs. However, the author demonstrates that each system works at least well enough so one cannot rule out the scenarios as "unworkable" (Carstairs-McCarthy, p. 27). Why then, did human languages universally evolve with this distinction?
Chapter 3 discusses the philosophical distinction between Truth and Reference. Section 3.1 asks "Are Truth and Reference Distinguishable Nongrammatically?" Throughout the chapter, several philosophers are called into play: Frege (section 3.2); Wittgenstein (3.3); and more recently Strawson (3.5). Of particular interest were Plato's views (3.5) about Falsity and Non-existence and (3.6)Syllables and Sentences. The latter carries over to the author's view, e.g. that "the sentence/NP distinction arises from a carry-over into syntax of a pattern of organization that evolved in the first instance for the neural control of phonetic articulation." (Carstairs- McCarthy p. 63). Apparently Plato noticed a resemblance between sentence and syllable structure. The way in which Plato illustrates the difference between syllables and certain phonemes seems to reflect the difference between saying and naming. (Ryle, 1960). In Thaetus, Plato shows that while the So- in Socrates is divisible into s- and o-, there are also sounds like b- that essentially can't be pronounced without an accompanying vowel.
Chapter 4 is appropriately titled Attempts to Solve the Problem, and it contains some previous explanations for the 3 pecularities of human language: vocabulary size, duality of patterning and the sentence/NP distinction. The author illustrates that these explanations do not account for the peculiarities in a unified fashion. One explanation for vocabulary size relates to the concept of "synonomy avoidance" whereby it is rarely the case that two (or more separate words) are exchangable perfectly in all contexts. Duality of patterning may turn out to be on by-product of contrasting meanings for a growing number of distinct calls. Several separate concepts are weighed and either critiqued and or dismissed here, including call blending, sound symbolism, phonetic assimilation, semantic drift and phonological self- organization. The longest section in the chapter had to do with the Sentence/NP distinction, because far more scholastic effort has been devoted to syntactic structure than either of the previous two concepts. The discussion begins with Universal Grammar (UG), which was postulated to explain why all human languages have characteristics in common. Unfortunately, it seems that most Chomskyan linguists seem to avoid discussions of the evolution of UG, except for Frederik Newmayer (1991, 1998) Stephen Pinker(1994, 1995) and Paul Bloom (Pinker and Bloom, 1990). The author sides with his colleagues as to whether the evolution of UG is worth discussion, but disagrees with them on a number of fronts as to how and why it happened as it did.
In Chapter 5, the author presents his solution. In the first section, Scope of the Solution, we are presented the three peculiarities of human language as the result and reminded that no other account of language evolution account for these peculiarities. The following sections begin by exploring the reasons for exact synonomy in adult vocabulary, particularly as they relate to language acquisition. 5.2 proposes a language where objects are named disjunctively. This brings up the question: Why is it intuitive to name things separate in appearance with separate names? Is this part of our biological endowment? The next subsection deals with principles guiding vocabulary acquisition. These principles are applied in a study of Vocabulary Acquisition Principles in Inflectional Morphology. Again, in this section, the author proposes another scenario language where disjunctive categories are applied to a wordclass. The author is able to show clearly how this is counterintuitive. Then we look at a study of several real world languages, which seem to apply these very disjunctive classes to their inflectional morphology of nouns, however, on closer examination, it is more systematic than first appears-some inflections appear to be what the author calls 'class defaults' whereby others appear to be 'class identifiers'. It is the range of these identifiers that varies.
There are evolutionary implications to this, as we first learn about the descent of the larynx in walking hominids. The larynx in other mammals is positioned high in the back of the mouth which allows them to breathe while eating, something humans cannot do. However something we can do (besides walk upright) is form vowels. Vowels are formed in the supralaryngal cavity, and where and how the tongue splits this is what partially determines the frequencies of the first and second formants of vowels. For example, for the vowel [a] the tongue is low in the mouth, so the oral air volume is large while the pharyngal volume is small. (Carstairs-McCarthy p. 127). This makes the first formant (the pharyngal one) high relative two other vowels, and the second (the oral one), low. There is even a little "exercise" provided to allow one to listen to the first formant. Tilt your head back so that the front of the neck is taut. Place the lips and tongue silently in position for each of the vowels [a], [e], [i], [o], [u] while simultaneously tapping the neck with a finger or pencil for each. (Ladefoged, 1993). Essentially we are shown, without the new L-shaped configuration of the supralaryngal cavity, there would be no clearly defined first and second formants. Without these, there would be no distinctions in vowel timbre.
The next section 5.4, Our Ancestors' Dilemna, we recall that our hominid ancestors had a vocal call system which obeyed synonomy-avoidance principles (principles strong enough, we are reminded, in modern language, to create intralinguistic "meanings" such as collocational restrictions and inflection-class distinctions) (Carstairs-McCarthy p. 129). What therefore happens to our hominids' vocal call system when the new supralaryngal cavity vowel potential is combined with strong synonomy avoidance?
Because our brains are finite, memory capacity is limited. A modest increase in muscular precision means that we have exponentially more potential sounds we can combine into new calls. In particular, labial, alveolar, and velar sounds like [p] [t] and [k] respectively are robust enough that the sound characteristic doesn't change much if the place of articulation varies by as much as a centimeter one way or the other. An educated adult nowadays knows between 50,000 and 250,000 words (Aitchinson 1987:7). Differences in sounds between two words or two affixes tends to cause a meaning to come about to differentiate them. Given this, it seems likely that synonomy avoidance principles are part of our linguistic inheritance. Thus the dilemna: on one hand, with our new supralaryngal cavity and enhanced motor control, our ancestors had a lot of new sounds to try out, but with the expectation that all the distinct calls should differentiate in meaning. On the other hand, with those finite brains, it's impossible for all those possible distinct calls to be memorized with distinct meanings.
The dilemna is resolved in section 5.5. Vocabulary grows, while duality of patterning is introduced. The author elaborates: "Cultural development uninhibited by any vocabulary size limits is just what we will expect if...potential vocabulary size is driven not by environmental or social factors but rather by inherited synonomy-avoidance principles (Carstairs-McCarthy p. 132).
However, this in itself creates yet another dilemna: If language now evolves such that longer calls are rendered usable by being strings of shorter calls (with meaning), how will the combined meaning of these strings be determined? Syntax. But syntax isn't arbitrary. The author thus argues that these early syntactic features (and ultimately the sentence/NP distinction) are a natural outcome of speech innovations due to vocal-tract changes. But what specifically?
In 5.6.1. we are introduced to syllables themselves as the phonetic motivation. Syllables embody a rapid alteration in sound sonority, whereby (usually) a sound with higher sonority (a vowel) can be surrounded by a sound with lesser sonority (for example a plosive or a stop). Syllables can be broken up into margins (onsets and codas) with a nucleus (sonority peak) in the middle. In 5.6.2 we are shown the internal organization of the syllable and given several prevaling views in phonetics about syllable weight. However, the brass tacks come at 5.6.3, where we finally see the syllable as a model for sentence structure.
Linguists have found similarities for some time between phonology and syntax, and the author, in this section proposes the influence of phonology on syntax. At first, he holds some other views to light before he substantiates his own views. In so doing, he provides, in detail, many structural similarities between syllables and sentences. He disproves further the proposed languages Spatiotemporal and Monocategoric by attempting, and failing, to map them to compliance with syllabic models. He finally reaches some conclusions with this: ''...if human language is descended from something like a primate call system, there is no stage in its evolution at which the sentence/NP distinction would have had to emerge in order for it to express everything that human language can...Monocategoric, just like Spatiotemporal, is not the sort of language we would expect to evolve in an environment where syntax was modelled on syllable structure, so as to favor the three asymmetries." (Carstairs-McCarthy p. 150). Farther on, he draws a 6- point checklist on how syntax would comply with the syllabic model and reflect the three peculiarities. Most of the rest of the chapter is more expansion on these proofs and their implications.
Chapter 6, entitled apes, anthropology, and the brain, has subsections pertaining to exactly these three things. These, of course, provide some complementary evidence to the author's claims, as section 6.1 indicates. 6.2, Archeology and Biological Anthropology provides complimentary evidence from these fields. Some linguistic developments can be mapped similarly to tool development, at least in relative terms of complexity. As it turns out, the impetus for vocal-tract changes may be traceable to our hominid ancestors attempts to walk upright, as the larynx can be placed by measuring the base of the skull. Some of the dates related to the dropping of the larynx can provide correlating evidence to various "Growth spurts" in early human culture. 6.3, The Brain looks at aspects of the brain, which allow the language evolutionist to study from a more abstract point of view. Some issues discussed here include the effects of Broca's Aphasia and some points brought up from the study of the grammatical structure of sign language. 6.4 Apes and language, discusses how chimpanzees avoid synonymy, the nature of Ape's sign combinations and how a bonobo understands spoken English.
The final chapter, 7, is titled "Just How Unique are We?" and is intended to summarize the whole of the author's argument and discuss its implications. The outcome is discussed, some parallels drawn between linguistic and cognitive evolution, and there is even a section that explains, among other things, how the nature of knowledge may certainly affect our interpretation of the evidence (one example being the way chimpanzees react to commands). The last section, "Why Humans?" we are shown two questions which must be independently resolved if one is to dig deeper into the evidence surrounding language evolution: "How did language originate?" and "Why was it in humans that language evolved?"
In a sense, this was a difficult book to summarize, in that there were so many points to be considered that it was often difficult to follow the main thread of the discussion without going too far in depth. There were too many "sub-threads" supporting the main argument, which I felt were expanded a bit too far. This is not in detriment to the work itself. In fact, I believe that if one of the goals of this book is to spark debate, I am sure there will be reactions from most if not all of the disciplines represented.
Despite the fact that the book seemed to lack empirical evidence in some places, I could find little to argue with. In fact, the author himself points out that collecting data for this sort of "experiment" may prove to be a bit different than doing controlled laboratory research, and I'd have to say that I agree.
If I were editing this book for non-specialist readers, I would probably order the chapters more or less in the opposite order, starting with the conclusion, working next to the supporting evidence in other fields, and then getting on with the three pecularities (actually chapter 2) and the proof itself (chapter 5), and finishing by showing the conclusions presented in the context of their implications. I would probably omit Chapter 3 altogether, or at least move it to the end of the book as a supplementary chapter, because it doesn't provide directly supporting evidence to the already very dense text in the supporting chapters. In fact reading this book in the order it is written is something of a liability if one is trying to follow the author's impressive argument.
On the whole I found the book both accessible and informative, if a bit dense. It should be of interest of anyone attempting to follow the language evolution debate. However to be of more general (or even popular) interest, it could at least provide more background for non- linguists if not be written in a more general vain. This book itself (when combined with some of its bibliographical references) could be the basis for a useful graduate-level course in language evolution, or its method of study, if this is not already the case.
Aitchinson, Jean. 1987. Words in the Mind: An Introduction to the Mental Lexicon. Oxford: Blackwell.
Hurford, James R, Studdert-Kennedy, Michael, and Knight Chris. 1998. (eds.). Approaches to the Evolution of Language: Social and Cognitive Bases, Cambridge: Cambridge University Press.
Ladefoged, Peter. 1993. A course in Phonetics (3rd International Edn.). Fort Worth: Harcourt Brace.
Newmeyer, Frederick J. 1991. Functional explanation in linguistics and the origins of language [with peer commentary and response]. Language and Communication, 11: 3-114.
- -1998. On the supposed 'counterfunctionality' of Universal Grammar: some evolutionary implications. In Hurford et al.(1998), 305-19.
Pinker, Steven. 1994. The Language Instinct. New York: William Morrow.
- -1995. Facts about human language relevant to its evolution. In Changeux and Chavallion (1995), 262-83.
- -and Bloom, Paul. 1990. Natural language and natural selection [with peer commentary and response]. Behavioral and Brain Sciences 13: 707-84.
Ryle, Gilbert. 1960. Letters and syllables in Plato. Philosophical Review, 69: 431-51.
John Hammink holds a B.Sc. in Linguistics from Eastern Michigan University. He has been working for the last five years in the areas of data-security software localization and validation. His work with validation (particularly test automation) involves both training the corpora of the end application as well as developing a trainable "test" corpus for the automation machine. His interests include Corpus Linguistics, Linguistic Evolution and the. Finno-Ugric Language group.