A review of William O’Grady’s (2022, version 3.2) Natural Syntax, an Emergentist Primer
http://ling.hawaii.edu/wp-content/uploads/Natural-Syntax-–-An-Emergentist-Primer-3rd-ed.-2022.pdf
O’Grady’s Natural Syntax traverses many of the topics that have been elaborated by Generative Syntax, from anaphora to islands, intended to demonstrate that Natural Syntax is a better theory (more about this later on). Evaluating Natural Syntax by itself, I find most processes perfectly plausible but will add some observations about ergative languages that Natural Syntax seems to struggle with, putting paid to the notion that any linguistic theory can account for any and all linguistic phenomena.
In the first instance, what I like about O’Grady’s Natural Syntax, is his basic assumption, i.e.
Semantic representations do not come ready-made of course; they must be built. My key proposal in this regard is that all mappings between form and meaning start with a maximally simple template, called a semantic base. As depicted below, the semantic base consists of a position for a predicate (PRED) and a position for a single argument – henceforth the base argument (represented by the symbol β).
The semantic base
PRED
<β>
The semantic base is the sine qua non of syntax – the minimal and least costly semantic representation needed for forming and interpreting a sentence of any type.
I have always assumed (studying and teaching languages) that the VERB is the centre of all sentence production/comprehension. It is the VERB that selects its nominal arguments. I have always wondered why a sentence is defined as a subject-predicate configuration (mirrored in generative syntax by NP VP) whereby the predicate (VP) is only the VERB when it is intransitive but is VERB + OBJECT (V NP) when transitive. Not sure why O’Grady chose the notation PRED when he clearly means the VERB (transitive or intransitive) which then selects its nominal arguments accordingly. Or else, since O’Grady terms PRED as the semantic base, the meaning (sic) of PRED is somewhat different from the narrow syntactic terminology. It doesn’t really matter, as he says these representations ‘must be built’ (why he is so opposed to ‘building’ tree structures will be discussed later). As we learn how to map the nominal arguments, and depending on word order, see the many permutations, we can only agree with the procedures in ‘building’ a sentence.
As mentioned above of particular interest for me is his treatment of ergative languages that place the patient (PAT) before the agent (AG), O’Grady showing how this is supposed to make sense in even tricky word order schemes. Having studied an ergative language myself, namely Niuean (a Polynesian language) I would have liked a wider and better treatment of this phenomenon. Let’s remind us of the basic scheme:
V(intransitive) N1(absolutive)
V(transitive) N1(ergative) N2(absolutive)
In traditional descriptions (e.g., Seiter, 1980 p.28) both N1 are declared ‘subjects’ which begs the question why N2 has the same case marking as N1. The other structure that needs explaining is that the N2 (absolutive) is often elided (or optional), rendering the structure as:
V(transitive) N1(ergative)
which gives rise to yet another question, i.e., is the VERB in this instance transitive or intransitive?
As is elaborated in some treatments of the ergative constructions (Chung, 1978, Seiter, 1980), this is related to the passive-to-ergative drift hypothesis for Polynesian languages, as exemplified by NZ Māori. Since the passive voice is not accounted for in Natural Syntax, let us remind us what the active-passive transformation is in English:
(a) The cat ate the mouse.
(b) The cat ate.
(c) The mouse ate.
(d) The cat slept.
(e) The mouse slept.
(d) The mouse was eaten by the cat.
(e) The mouse was eaten.
(f) *The mouse/cat was slept.
where ‘the mouse’ is the accusative object in the active voice, and the nominative subject in the passive voice. It seems to make sense that some languages drifted to a system whereby the canonical sentence with a transitive verb was in the passive voice, with the subject (patient) being marked by a special case, the ergative, while all agents were marked by the absolutive case. Such a system put into question the concept of transitivity (or valency), as the agent-less passive construction (e) only has one nominal argument, as have so-called intransitive constructions (d, e). Even in English (b, c) are questionable, i.e. while the verb ‘to eat’ is traditionally a transitive it can also be used as an intransitive. In terms of valency one can say that some verbs have one or two arguments. How such a system operates when the passive construction becomes the unmarked sentence (and the anti-passive the marked one) is a question that should occupy O’Grady in more detail, so his Natural Syntax has greater explanatory power, other than simply assigning argument slots that need to be filled (or elided). For example, as below, Niuean distinguishes (c - absolutive) and (e - ergative) by different case markings while the verb remains in the same form (as opposed to English where the difference is achieved via different verb forms). Note however that (c- absolutive) is a marked anti-passive construction not commonly used (to a degree similar to the English (e) sentence being marked/unusual).
(d) V (eat) N1(mouse - ergative) N2(cat - absolutive) ‘The mouse was eaten by the cat/The cat ate the mouse’ (note that for translation purposes, one might select the English active construction as to convey the unmarked Niuean equivalent)
(e) V (eat) N1(mouse - ergative) ‘The mouse was eaten’ (unmarked)
(c) V (eat) N2(mouse - absolutive) ‘The mouse ate.’ (marked)
Also, of interest to O’Grady’s theory may be that the Niuean ergative (passive) system is drifting back to the English-type accusative (active) system because the Niuean education system under New Zealand control is largely English-medium based. A ramification of current efforts to reverse such trends, i.e. the renaissance of indigenous languages like Niuean and NZ Māori via more enlightened language policies, is that such indigenous languages are taught as second languages (English being the first) with the unintended consequence to further cause grammatical shifts, as mentioned above.
Given above observations, we can now address O’Grady’s apparent obsession with constantly comparing his sentence mapping procedures with that of Chomskian generative and/or minimalist processes, consigning the latter to the historical dustheap. While I understand the Popperian dictum that much of natural sciences is occupied by falsifying theories, I do not understand how this can be applied to linguistic theories that at best can have a psychological reality (i.e. linguistics being a branch of psychology). Sure, Chomsky alluded to what a linguistic theory should accomplish, e.g., have explanatory power and be elegant, defending his theories against the onslaught of competing theories. Chomsky always prefaced his theories with ‘assuming that X is true, then y will follow’, e.g., in its present incarnation, assuming that MERGE is a basic computation, then most sentence structures can be accounted for by following various procedures. In the same way O’Grady assumes that PRED is the basic building block for processing most sentences. To make categorical statements about the value of one linguistic theory over another seems to contradict O’Grady’s introductory notes, namely that the ‘quest to explain language may lie beyond the reach of the only creatures who are able to use it’.
I wouldn’t call this ‘a deep irony’ though, rather a paradox, in that any explanation of any phenomena requires a meta-language (actually, better called sub-language) so that the snake doesn’t bite its own tail. The natural sciences have evolved ever more sophisticated meta/sub-languages, not to speak of the various attempts of explaining mathematics (cf. Russell’s Principia Mathematica). That we now have a plethora of computer languages is testament to the ironic idea that subsets of natural language can be the basis of programming AI to learn natural languages – i.e., one can detect here an irony in that the acquisition of language by humans appears to be child’s play. If we could only figure out how the brain gives rise to language even at such an early age! Assuming that the human brain is or harbours anything like a computational system, some neuro-linguists have embarked on a wild goose chase to attempt the impossible. The reason being that language cannot be explained by language – a system cannot explain itself. I have likened this paradox to the biblical (nonsensical) story of the prohibition of eating fruit from the tree of knowledge, lest one is a god. In other words – in the beginning was the word – only a god can explain language. Not that this confers a certain status on linguists like O’Grady or Chomsky who at least agree on the notion that we should nevertheless try our best to explain what cannot be explained. Being a lesser linguist myself, I would – nevertheless – claim that linguistics is the crème de la crème of all sciences. Einstein and Co. might laugh at the suggestion, given that they are busy explaining the universe down to the last particle at CERN. That they too communicate their findings in terms of language – or a subset of language incomprehensible to common man – would escape their view of the world, likely to disagree with Nietzsche’s contention that science too is a form of storytelling. On that count we wouldn’t have Einstein begin his paper Zur Elektrodynamik bewegter Körper with an exposition on the philosophical assumptions underlying his technical claims. Writers on highly technical subjects tend to launch into the details of their proposals, leaving the layman behind in quick time. Chomsky’s seminal Aspects of the Theory of Syntax took no prisoners on that account either.
O’Grady’s title Natural Syntax, an Emergentist Primer sounds technical in the first part and non-technical in the second. The word ‘syntax’ does not ordinarily evoke common understanding, being mainly associated with linguistics at the graduate level, to distinguish the concept from the more lowbrow term of ‘grammar’ (cf. the use of Grammar School in the British education system). With regards to the ‘Primer’ idea, O’Grady informs us that
As its subtitle suggests, Natural Syntax is intended for an audience with little or no background in the study of emergence or its possible relevance to the understanding of language.
He then seems to contradict himself by elevating his subject as follows:
Linguistics – and especially syntax – has been a hotbed of controversy for many decades, for reasons that await scrutiny and assessment by scholars of the history and philosophy of science.
Now we are at the level of ‘scholars’, warning an unsuspecting audience that what we are dealing with awaits ‘scrutiny and assessment by scholars of the history and philosophy of science’. Based on this observation one might expect O’Grady to submit his syntax proposals and then await scrutiny and assessment. Alas, there is quite a way to go before we get to that. Given that linguistics is not an exact science like physics, chemistry or mathematics – Chomsky proposed it being part of psychology, cognition included – we can of course expect some loose talk, especially when directed at competing theories. Chomsky himself and authors allied with him are not above such remarks, dutifully quoted by O’Grady, such as:
… there is good reason to think …
… There is no longer a conceptual barrier to the hope that …
… any linguistic theory is going to have to meet two conditions …
--- We can all agree that …
… it is impossible to draw any conclusion …
… No rational person can believe that …
… there is no coherent alternative …
Not that O’Grady and his fellow travellers are immune from it either:
… The phenomena of language are best explained by …
… the structure of human language must inevitably be shaped around …
… Clearly no one denies that …
… there are compelling reasons to believe that …
… Everyone pretty well agrees that …
… All researchers agree that …
Neither O’Grady’s nor Chomsky’s linguistic theories are hardly at the level where we could legitimately say that ‘Clearly no one denies that … 1 + 1 =2. In today’s anti-scientific, conspiracy-driven, fake world there would be plenty influencers who even deny that. As such, the more extreme versions of academic revenge-competitiveness have resulted in the infamous ‘linguistics wars’ promoted by the likes of R A Harris and C Knight (I engaged in minor skirmishes with both). In this context it is also interesting that O’Grady seems to suggest that various linguistic talents were wasted due to a supposed adherence to a particular school of thought – mostly generative syntax a là Chomsky, until he supposedly reversed his theory to the Minimalist approach (to be discussed in more detail below). Academia, especially in the human sciences, is replete with academic departments hiring only adherents of the school of thought (ideology), the chair of the department represents. So it is not only anti-Chomsky academics who missed out on being hired but also pro-Chomsky ones. I might include myself here for the latter, getting my linguistics degrees at the University of Auckland, at a time when the anthropological linguistics department was dominated by descriptivists who derided Chomsky for his theories (from a satirical poem composed by A Pawley who had attended Chomsky’s lectures in Bloomington, Indiana in 1971):
My kernels appear in most of the journals
My trees can be seen in their pages
No transformations but by my operations
will be permitted for ages and ages
No wonder when a teaching position came up, and I applied, I was turned down in favour of a religious SIL-type descriptivist (my chance at last to affect some revenge). Equally, somewhat earlier on when I attended LMU in Germany to study psychology in 1970, hardcore behaviourists dominated the proceedings and beat out all my enthusiasm I had garnered from Reich and the likes. Little did I know then that a certain Noam Chomsky had already debunked Skinner’s behaviourism in 1959, so I joined the APO under an anarchist flag instead, breaking off my studies at LMU and leaving Germany to escape the draft, and eventually settling in New Zealand.
In any case, O’Grady seems to have escaped the linguistics wars unscathed, happily ensconced at UH, developing his brand of Emergentism. As noted above, in the human sciences it seems desirable to first state one’s adherence to a certain school of thought, if not outright ideology, be it Marxist or McCarthyite, before getting down to the technicalities (Chomsky famously divorced his political activism from his linguistics). As such O’Grady is at pains to first establish his philosophical credentials for Emergentism, citing British philosophers Lewes and Mills who advocated a distinction between ‘resultants’ and ‘emergents’, by way of saying that ‘resultants’ clearly show their components while ‘emergents’ don’t. O’Grady uses Mill’s example of water not showing its components of hydrogen and oxygen, hence water being an ‘emergent’. Personally, I find this rather obscure. In Chemistry I cannot think of one example that would demonstrate a ‘resultant’ rather than an ‘emergent’, other than two elements that do not react to form a compound. In Physics there are no resultant elements (as in the periodic table) either. Sure, there are resultants like ‘mass’, similar to the result of 1kg + 1kg = 2kg (arithmetic is also an example of a complex system, the components (numbers) of which are inherent to that system only. The rainbow is cited as an example that is a ‘resultant’ showing its component colours, and yet the ‘organisation’ of these colours gives rise (emerge) to the concept of the rainbow. As such it would be fairly obvious that any painting is a resultant, as one can see the different colours that make up the painting. The literature on these matters seems quite uncertain as to the definitions of ‘resultant’ versus ‘emergent’ (cf. https://eldervass.com/wp-content/uploads/2020/12/Elder-Vass-2005-Emergence-Realist-Account-of-Cause-JCR-PPV.pdf.
Nevertheless, O’Grady goes on to explain ‘emergence’ as a particular phenomenon in nature so that ‘things are often not what they appear to be’. He calls this ‘understanding the mysteries of nature’. The two statements either add up to an oxymoron or to a truism: a ‘mystery’ is at best a phenomenon not yet explained by science, or more commonly a phenomenon which has a scientific explanation but is denied by various mystics (nowadays called conspiracy theorists). Obviously, old wives’ tales or certain aspects of folk science are myths that are debunked by science every day – albeit to no great effect as currently evidenced by climate crisis deniers (possibly in the same league as former holocaust deniers). Not that academics are immune from inventing myths - see Chomsky’s critique of Skinner.
Maybe O’Grady wants to point out that the ‘mystery’ of language is just an old wives’ tale of Chomskian proportions, and that his brand of linguistic science will debunk any such mystery and come up with the scientific goods. In that case we are back to square one, or so it seems. If we were to accept that ‘resultants’ and ‘emergents’ are neutral scientific terms for certain natural phenomena – and in which in my book refer to quite different categories – then we might be quite open to O’Grady’s suggestion that language is an ‘emergent’, were it not for his next claim that ‘emergents’ solve mysteries while ‘resultants’ are part of ‘essentialism’ which in turn are items ‘unique to a complex phenomenon itself’. In other words, a complex system that is defined by components that cannot occur by themselves. Here one might straight away jump to a linguistic (synchronic) phenomenon where certain roots of words do not occur by themselves or whereby many prefixes/suffixes/infixes do not occur by themselves. Is that a case of essentialism? Or is it a case of emergentism in that one can explain many of these items in terms of diachronic developments? Not surprisingly there is also the suggestion that complex system harbour both resultant and emergent processes – an idea not commensurate with O’Grady’s either-or theory.
In any case, this is the trail that leads again to Chomsky (of the generative syntax and UG era) who is accused of being an essentialist, one who claims that the components of language are unique to language and cannot be derived from phenomena outside language. I presume this also makes Chomsky a ‘resultant’ orientated scientist who claims that language ‘resulted’ suddenly and inexplicitly from a genetic mutation in the brain some 150,000 years ago, while O’Grady is an ‘emergentist’ who claims that language evolved/emerged over time from cognitive processes that predate language and/or work in parallel with language. I am not even sure if this amounts to some sort of fundamental difference. In common and scientific language use one can just as well claim that the ‘result’ of combining certain cognitive processes is language – just as much the ‘result’ of a genetic mutation is language. Maybe descriptivists have a point here: we don’t care where language comes from, we just want to describe what is in front of our eyes and ears. The child that acquires their language has no inkling (nor does it need it) where this language comes from, nor that it evolved/emerged/resulted from a single cell billions of years ago, nor that they resulted/emerged synchronically from a couple of cells (sperm and egg).
Verbal semantics aside, what really seems to irk O’Grady is that Chomskian bio-linguistics and Universal Grammar explain language as a self-contained system. However, before we continue, we should qualify language here as syntax, the rules of constructing a sentence. This so-called Chomskian essentialism presumes that categories like verbs, nouns and combinatory processes are quite unique to language. So, what is wrong with that? Isn’t O’Grady using the very same concepts in his sentence processing?
Again, O’Grady and others submit to the Popperian obsession (for the human sciences) to have to falsify a competing theory, in order to verify their own. O’Grady is even more encouraged to do so as it appears that Chomsky himself saw the light and abandoned UG in favour of the Minimalist Program, a program that could almost be called Emergenist – if only! O’Grady quotes Chomsky:
There is no longer a conceptual barrier to the hope that the UG might be reduced to a much simpler form, and that the basic properties of the computational systems of language might havea principled explanation instead of being stipulated in terms of a highly restrictive language-specific format for grammars.
(Chomsky 2005:8)
The problem is that silly Chomsky and Co. haven’t quite given up on UG, what O’Grady calls ‘Rebooting Universal Grammar’. Since Chomsky’s biolinguistics under the Minimalist Program posits a Language Faculty that has UG as a basis, we are back to square one:
The term Universal Grammar (UG) is a label for [the] striking difference in cognitive capacity between “us and them [humans and animals].”
(Chomsky, Gallego & Ott 2019:230)
It has taken O’Grady some 16 pages to deal with Chomsky, and only now we get down to what is O’Grady’s theory, introduced as The Strict Emergenist Protocol. The first axiom is ‘direct mapping’ between sound and meaning, as opposed, alas, to Chomsky’s ‘mediated mapping’ as explained by a Jackendoff quote:
… the correlation of sound and meaning is mediated by syntactic structure ...
(Jackendoff 2007:3)
Now, I don’t know if binary tree structures are still alive in Chomsky’s Minimalist Program but I since the MERGE operations are also binary in nature, I wouldn’t be surprised if they still were, as a useful metaphor. Be this as it may be, O’Grady’s solution is called, as mentioned above, ‘direct mapping’, represented like this:
FORM | | MEANING |
Harry left | ⇐⇒ | LEAVE <h> |
Since we are now wondering how ‘direct mapping’ might work – other than saying so – we see O’Grady backpedalling a bit:
To avoid possible confusion, two clarifications are in order. First, the rejection of syntactic structure applies specifically to ‘tree structures.’ It does not deny that speech involves words of particular types (nouns, verbs, etc.) that are inflected and linearized in particular ways. Second, I am not proposing that syntax can bedispensed with, only that it should be reconceptualized as a set of operations that map strings of words directly onto semantic representations and vice versa in ways to be explored in the chapters that follow.
So, we still have N (noun) and perhaps NP (noun phrase) and V and VP, and T (tense) just like in the rejected tree structures – and we still have syntax! This seems to contradict O’Grady’s note:
[Language] maps a string of words directly onto a semantic representation without the mediation of grammatical principles or syntactic structure. (O’Grady 2015:102)
In conclusion then, to get back to the beginning, I find O’Grady’s theory quite appealing due to his assumption of PRED being the basic template. Other than that, as he uses the concepts of syntax just like anybody else, be it Chomsky or Panini, I don’t see the need to assert his notion that his theory is any better than any other. Linguistics, like many other human sciences, benefit from Mao’s dictum to ‘let a thousand schools of thought blossom’ simply because human nature is, and always will be, as un-speakable and contradictory as the traits that make us uniquely human, namely language (langue) and language use (parole). While the likes of O’Grady and Chomsky are very good at de-mystifying language (langue) as systems of syntax, we have no one to explain why humans use language to shoot themselves in the foot as much as to elevate themselves beyond the gravity of earth (Chomsky as a political activist, tries his best but in his long and distinguished career he has not been able to make the world a better place, only warning us of pending, man-made, language-mediated catastrophes).
References
Chung, S. L. (1978). Case marking and grammatical relations in Polynesian. Austin: University of Texas Press.
Seiter, W. (1980). Studies in Niuean Syntax. Garland Publishing, New York & London.