what model of biological plausibility

8/12/2019 What Model of Biological Plausibility

1/37

WHICH MODEL OF BIOLOGICAL PLAUSIBILITY FOR LANGUAGE?

THE CASE OF WHAT DARWIN GOT WRONG

Abstract

The goal of this paper is to discuss some of the conceptual consequences of the arguments put

forward in What Darwin got Wrong, for a broader theory of the biolinguistic approach. The book

offers arguments against New Synthesis approaches to Evolutionary Theory, which are particularly

germane to biolinguistic matters. One main contention is that only approaches to evolutionary facts

that capture the organisminternal !laws of form" observed across living organisms can be

theoretically and empirically adequate. #pproaches that only focus on organisme$ternal principles

cannot reach this goal. %owever, the book does not investigate whether this contention applies to

linguistic matters as well. This issue is addressed in the paper, and it is argued that organism

internal properties, which can be captured via the formal notion of !&onservativity", must be

found in language as well. Therefore, it is argued that only those linguistic theories that capture

these properties, be they about syntactic, semantic or acquisition matters alike, can be considered

as biolinguistically plausible.

1 I!tr"#$ct%"!& T'( Pr"b)(* "+ B%")",%ca) P)a$s%b%)%t- +"r )a!,$a,(

'erry (odor and )assimo *iattelli*almarini+s oint effort, What Darwin got wrong,presents a

criticism of neodarwinist approaches to Evolutionary Theory. This criticism is based on the

!New Synthesis" model, classic and contemporary alike -)ayr, /012 3awkins, /40, /502

among others6. (odor and *iattelli*almarini7s -henceforth8 !the authors"6 two main

arguments against the New Synthesis -henceforth8 NS6 and its applications, as proposed in the

book, can be informally summed up as follows. (irst, NS offers an e$planatory model that

works like Skinner7s theory of operant conditioning -Skinner, /946. Since operant

conditioning is inadequate as an e$planatory tool, the NS model is equally inadequate to


2/37

account the data it purports to account. Second, the NS model is logically flawed8 it aims to

e$plain data by resorting to principles that are claimed not to be part of the model -artificial

selection6. So, no putative !recalibration" of the model is possible, in order to amend these

theoretical flaws. # new evolutionary framework is called for, the authors of the book claim.

One problem is that (odor : *iattelli*almarini -;


3/37

aspects of modern evolutionary theory including, among others, a misinterpretation of Evo

3evo as mutually e$clusive with NS models. %ence, its invitation to drop old models of

Evolutionary Theory for new ones is unustified ->lock : @itcher ;


4/37

relatively precise arguments on how a biologically plausible theory of language could be

defined, it is not clear how one could use these arguments to e$press such a theory. Two

questions seem particularly stringent. # first open question is which strands of the

biolinguistic approach, broadly conceived, are consistent with the authors+ main argument. #

second, consequent open question is which parts of the minimalist program, which we will

suggest as being the main strand of >iolinguistics to be consistent with the authors+

arguments, can offer an optimal level of !conceptual fitness" with these arguments.

Our goal in this paper is to answer these two questions. %ence, we want to verify whether

the authors7 arguments can also give us an important insight on which !parts" of the >io

linguistic #pproach and, more specifically, of the minimalist program are consistent with the

author7s arguments. Therefore, in answering the first question we aim to show that the

authors+ arguments are a more appropriate account of biolinguistic matters than the NS ones.

#t the same time, we also aim to show that their arguments are more appropriate than the

evolutionary psychology ones, although necessarily antithetic to them. An answering the

second question, we aim to show that only certain approaches within the minimalist program

are compatible with the authors+ arguments.

De aim to answer to these two questions via a twostep plan. (irst, we wish to discuss

which strands of >iolinguistic #pproach are consistent with the arguments presented in

(odor : *iattelli*almarini -;


5/37

offer us powerful theoretical tools for answering our two questions. The remainder of the

paper is geared towards reaching these goals, accordingly.

. W'at Dar/%! G"t Wr"!,& R()(0a!c( +"r A B%")%!,$%st%c Ar"ac'

The goal of this section is twofold. The first goal is to offer a brief summary of the authors+

arguments, and their import for the biolinguistic approach -section ;..6. The second goal is to

discuss the import of the argument for three dimensions of linguistic analysis8 synta$,

semantics and language acquisition. De start with synta$ -section ;.;.6, proceed with

semantics -section ;.1.6, and conclude with acquisition -section ;.B.6.

.1 W'at Dar/%! G"t Wr"!,& A br%(+ S$**ar-2 a!# A! A!a)-s%s "+ T'( Ar,$*(!ts

The goal of this section is to offer a brief summary of the arguments offered in the book.

The first argument offered in the book can be summed up as follows. The NS model is

not a valid theory of evolution, because it is inconsistent with recent and not so recent

discoveries in biology. The analysis of the Evo3evo research program offered by the authors

suggests that evolutionary processes are the result of rules of !wellformedness" on the

combination of genes, the organisminternal laws of form that govern the growth and

development of organisms. Af the authors+ analysis of these findings is correct, then we should

take into account the importance of genome and its structure, when proposing plausible

evolutionary models. #ccording to this view, then, evolution is first and foremost an

endogenous or organisminternalprocess, on which organismexternalprocesses -e.g. NS6

can intervene on the resultsof this internal process. Evolution is also a process in which

emergent properties play a key role. >oth the combinationof genetic material into more


6/37

comple$ units, and the possible permutationsof this material into different structures may

result into phenotypical variation. So, we should e$pect that such operations should have a

linguistic counterpart that applies onto linguistic structures -e.g. sentences6, as well.

The second argument can be summed up as follows. NS is usually defined as a process

that can select life forms for their fitness with respect to the environment, as if it were akin to

artificial selection. %owever, if NS can evaluate and select life forms with respect to their

fitness in an environment, then it acts much like a rational, agentbased process, which is

precisely what NS should notbe like. NS appears to be inconsistent, since it encompasses two

opposing principles that work at the same time8 random selection and nonrandom, !select

for" or agentive selection. #s the authors discuss at length, if the environment acts as an

agentivelike principle of selection, then by definition it cannot also be a random, !mindless"

process at the same time. So, NSbased theories of evolution are inconsistent, and cannot

e$plain the data they purport to e$plain.

The authors+ view of Evo3evo is far from uncontroversial, and tends to create a tension

between the internal, developmental properties of an organism and the e$ternal influences that

can play a role on these processes -(odor : *iattelli*almarini, ;


7/37


8/37

theoretical kernel of a biologically plausible theory of language. Therefore, the second

argument strongly suggests that a biologically plausible theory of language must principally

be an internalist theory of language first and foremost, and not ust an externalist theory

-#soulin ;arwise : &ooper, /52 %eim : @rat?er, //58 ch.;2 a.o.68

-6 D(A,B!D(A,A"B

An words, if a settheoretic relation between setsA andB holds, then it must hold between

the first set and the more restrictive intersection of both sets -i.e. A"B6. # biological

application can be formulated as follows. Af a biological propertyA-having wings6 is related

to a propertyB -being a pig6, then it must hold for the combination of these two properties, if

this combination does not define an empty property set. Since wings are incompatible with


9/37

being a pig, the set of pigs with wings -i.e.A"B6 is empty. De are not aware if the authors had

in mind this definition, too, but we think that possible conceptual divergences are negligible.

Af these arguments are on the right track, then their import for linguistic theories can, and

perhaps mustbe evaluated with respect to their ability to be consistent with the authors7

arguments, #uabiological principles applied to the subdomain of language. The reason is as

follows. =et us assume that linguistic properties are a part of biological properties. Then, we

represent the set of linguistic properties as $, and that of biological properties asB%. De take

that one specific type of relation, thesub-set relation !", can represent the parthood relation

between these sets -i.e. we have $B%6. # logical consequence of this assumption is all the

properties that apply to the set B%, including conservativity, monotonically apply to each of its

subsets, hence they identify possible restrictionsof this set -=andman, //8 ch. ;6. An other

words, A suggest that conservativity in language is a consequence of language being part of

our cognitive, and hence biological faculties, which in turn have the conservativity property1.

# less formal and more intuitive approach to this assumption is as follows. De e$pect that a

biologically plausible theory of language must also be an internalist theory, which also

implements a languagespecific notion of conservativity. The reason is simple8 if language is

an e$pression of our internal biology, and our biology has the conservative property, then

language must have the conservative property. An this regard, the authors+ arguments seem to

surprisingly converge with theories apparently distant from their position, such as embodied

cognition -=akoff : 'ohnson, ///2 Evans ;


10/37

discuss this aspect more in detail, as it would lead us too far afield. Dhat matters for us, then,

is that if this assumption about biological conservativity and its e$tension is notcorrect, then

we would e$pect that the NS model, instead, can give an accurate account of linguistic

phenomena. Af linguistic properties do not depend on -i.e. are not part of6 our biological

properties, they should share any features as well. The ne$t sections discuss this hypothesis,

and its validity, in detail.

.. B%")",%ca))- P)a$s%b)( t'("r%(s "+ )a!,$a,(& S-!ta3

The goal of this section is to discuss which syntactic theories meet the requirements for

biological plausibility that the authors present, and that A discussed in the previous section.

Since A take a broad perspective to the >iological #pproach to language, A will discuss a broad

set of theories, including e$ternalist ones.

A start from those theories that could be labeled as having an -organism6e$ternal

perspective to sentence structure. )ost, if not all of these theories contend that there are no

languagespecific principles by which distinct constituents are combined together by internal

autoorgani?ational properties. Some e$amples include construction or usagebased

grammars -oldberg //9, ;


11/37

that these theories also appear to be logically inconsistent. These accounts focus on e$ternal

factors such as !pragmatic principles", very broadly construed and defined. So, they seem to

lack mechanisms that account for why one can distinguish between wellformed and ill

formed sentences, one case being syntactic islands -olderg, ;


12/37

De now turn our attention to those theories that do take an internalist perspective to

syntactic matters. De start from nonminimalist theories, which to an e$tent do not place a

great emphasis on biological or evolutionary plausibility -if any6, at least in our

understanding. E$amples of such theories that we briefly discuss are %*S -%ead *hrase

rammar8 Sag, Dasow : >ender ;


13/37

consequence of this approach is that these theories face nontrivial problems in accounting

simple forms of inversionorfronting, such as locative inversion in -B6-968

-B6 >ehind the car, the boys were playing cricket

-96 Anto the room a unicorn came

The e$amples in -B6-96 involve a *repositional *hrase that is moved -or copied6 into

sentenceinitial position, as a topic or as a subect respectively -den 3ikken, ;


14/37

define themselves as a substantial departure from its predecessor framework, the'rinciples

and 'arameters heory, and those approaches that consider it a natural evolution of *rinciples

and *arameters Theory -e.g. >alari : =oren?o, ;urraco : =onga ;


15/37

basic combinatoric property. #lthough most variants of the cartographic approach assume that

the related operation &ovecan permute syntactic elements, the range of this operation is

subect to rather specific constraints. Dhen movement occurs, syntactic units move to

!higher" positions in the clausal spine that attract the features of the moved constituents.

E$amples -;6-96 would thus involve the operation&ove, which would permute the target wh-

elements and *repositional *hrases to a !noncanonical" position. Therefore, *rinciples and

*arameters approaches fall in what we could perhaps call a !bidimensional space" of

biological comple$ity. Dhile they assume that operations of permutation are possible, they

also assume that these operations range over a fi$ed sequence of linguistic units. This aspect

places these proposals within a gray theoretical space, since they include certain conceptual

!remnants" of the *rinciples and *arameters approach, most notably the !sequencing" of

building units. %owever, versions of cartography and nanosynta$ seem to overcome this

problem, if only in via rather finegrained theories of movement -Svenonius, ;


16/37

apply to sets of le$ical items -e.g. into, the, room, etc.6, of which the output is a sentence such

as -96. These approaches do not postulate any other organi?ing principles than the rules of

sentence formation applied to le$ical items, taken as !building blocks" of sentences, and

derive the wellformedness of these sentences as a result of these combinatoric principles. An

doing so, such approaches indirectly meet the goals of biological adequacy we discussed so

far, as they place emphasis on -sentence6 structurebuilding as a simple, !blind" process that

generates regular -syntactic6 forms.

#s matters stand, then, formal theories both within and outside the minimalist program

offer an internalist approach to syntactic matters. %ence, they can be seen as biologically

plausible theories that are consistent with the authors7 arguments. iven this consistency, they

do not incur in the empirical problems that constructionLusagebased grammars incur.

%owever, these internalist theories form a conceptual cline, in which only a certain subset of

)inimalist proposals seem to fully meet criteria of biological plausibility that the authors+

arguments outline. Dhile %*S and =( seem to only include a combinatoric principle

-)erge6 in their theoretical apparatus, the cartographic approach only a permuting principle

-)ove6, other approaches seem to include both -e.g. categorial grammars6. De leave aside a

more finegrained analysis of these differences, and turn to semantic matters.

.4 B%")",%ca))- P)a$s%b)( t'("r%(s "+ )a!,$a,(& S(*a!t%cs

The goal of this section is to discuss biologically plausible theories of semantics that are

consistent with the syntactic considerations we have offered in the previous section. #gain,

given our broad perspective on the >iological #pproach, we will cover several such theories.

#ny considerations about the psychological and biological plausibility of semantic

theories touch a rather thorny issue. Cecent analyses in the literature observe that standard


17/37

!philosophical" approaches take an e$ternalist position to meaning -#soulin ;


18/37

-46 =uigi has bought the Jespa from )ario

-56 )ario went to the pub

-/6 )ario is at the pub

E$amples -06-46 display an antonymous relation that holds between the Jerbs buyand

sell. These sentences can be true in the same e$ternal conte$t, as the same individuals -)ario

and =uigi6 may be involved in the same event of commercial e$change. %owever, these

sentences describe the same e$ternal event under two different !internal" points of view.

Coughly, )ario7s point of view is e$pressed in -06, while =uigi7s point of view is e$pressed in

-46. Note, furthermore, that the prepositions to and from also contribute to capture this

distinction in points of view, in a basically compositional way. So, e$ternalist approaches

would face a problematic challenge in modeling the notion of !perspective" that underpins

antonym relations when defined over verbs and prepositions. *ossibly, they would also face

this challenge with other parts of speech as well.

# similar challenge would also emerge from e$amples -56-/6, amply discussed in the

literature on events at least since *arsons -//


19/37


20/37

(or these reasons, they both fail to meet the requirements for biological plausibility that the

authors7 argument outline. #s e$tensions of OT, they will never be able to offer correct

e$planations of semantic data, but rather false or, at most, empty ones. *hilosophical

perspectives on semantics are so close to standard OTLevolutionary accounts, that they may as

well as be considered the authors7 ideal targets of criticism. An other words, the systematic

reliance of e$ternal factors and a NSlike account that these theories employ provides the

chief obstacle to a correct analysis of data, for these theories.

Dhile the import for these e$ternalist approaches to semantics should be clear, a

different picture emerges once we move within the domain of linguistic approaches to

semantics. An this domain, modeltheoretic semantics have been employed by scholars with

quite different perspectives. &lassical and modern modeltheoretic perspectives place a strong

emphasis on semantic derivations and processes. %owever, these perspectives tacitly consider

modeltheoretic semantics as attempting to model facts in the world, rather than mental

representations thereof. Some modern variants known as !dynamic semantics" e$plicitly

consider the process of interpretation as a way to modify what information we share, in the

e$tralinguistic -e$ternal6 conte$t -roenendik : Stockhof //2 Stalnaker ///2 a.o.6.

&onsequently, the model of 3iscourse that emerges from these theories is an e$ternal one, in

that it models our use of meanings of utterances in conte$t. So, these theories also fall a

within the range of both arguments, as being inherently e$ternalist.

Not all semantic theories, however, take this perspective on models, so some

distinctions are necessary. )odeltheoretic semantics can be approached as the result of

evaluating syntactic structures and their interpretation against a model of 3iscourse. This

intermediate syntactic step can be skipped8 words and sentences are interpreted directly,

against a model. The first approach usually has the (regean property of compositionality,

while the second usually lacks it. Since we are assuming that a biologically grounded faculty


21/37

of language coincides with derivational processes, we also think that noncompositional

theories miss an important generali?ation. =et us assume that sentences are the end result of

syntactic processes merging together basic le$ical units. Then, the interpretation of these

sentences against a mental model of 3iscourse can only be the result of combining together

the single interpretations in a coherent structure. (urthermore, if semantic relations among

sentences can be defined, then these relations must reflect the semantic relations among the

words and phrases they include in their structures. The interpretation of sentences such as -06

-/6, and their relations of entailment and antonymy must emerge as the compositional result of

combining Atems such as toandfrom, or buy andsell together. E$tralinguistic factors should

play no relevant role, in an analysis of these processes.

Several theories, being noncompositional in nature, fail to meet this requirement.

E$amples of such theories are the )ental )odels theory -'ohnson=aird /51, //;62

&onceptual Semantics -'ackendoff /51, //


22/37

approaches, other than the cartographic ones. (or instance, *ietroski -;


23/37

of conservativity that was originally devised to capture the semantic properties of enerali?ed

Kuantifiers ->arwise : &ooper, /52 onwards6. Our e$aptation of this notion may be a

useful conceptual tool, but not necessarily a correct one. An this regard, it seems a bit of a

stretch to suggest a direct logical connection between the properties of the Kuantified Noun

*hrase every man, and the wingless nature of pigs. %owever, semantic conservativity is

defined as a property that restricts the properties of enerali?ed Kuantifiers in their

interactionwith other parts of speech. (urthermore, the enerali?ed Kuantifier approach can

be e$tended, with minor provisos, to adverbs and other parts of speech that can denote

relations among events ->ach, 'elinek, @rat?er : *artee, //9, a.o.6. Therefore, its conceptual

import for our discussion turns out to be quite vast, empirically speaking. The e$amples in

-


24/37

*ietroski, ;


25/37

internalist grammar to be one that builds sentences as mental obects that, in turn, denote

-structured6 mental representations, or !meanings", which stand in certain internal -le$ical6

relations -e.g. entailment6. *erhaps, in such a scenario these semantic relations could be

interpreted as further dimension of linguistic organi?ation, to an e$tent ancillary to the

dimensions of synta$. De leave aside a more thorough discussion of these comple$ matters,

and focus on more empiricallyoriented topics. (or this reason, we turn our attention to

acquisition matters.

.5 B%")",%ca))- P)a$s%b)( t'("r%(s "+ )a!,$a,(& )a!,$a,( Ac6$%s%t%"!

The goal of this section is to discuss theories of language acquisition that meet the authors7

arguments for biological plausibility. There is an ample consensus that acquisition problems,

#uapsychological problems, represent a clear case in which psycho and biological factors

must play a vital role in linguistic e$planations -ang, ;


26/37

process consists, to a large e$tent, in the child selecting which syntactic constructions are

present in a language. #ccording to these theories, then, a child acquires a language via the

directed, guided input of adults surrounding him. Such models often e$plicitly define their

hypotheses about language acquisition processes as being conceptually isomorphic to NS

-Tomasello, ;


27/37

of language acquisition cannot be a theory that hinges on e$ternal factors driving acquisition.

Of course, children require the e$ternal input of other speakers, so that their mental grammar

can -and must6 store this information, to !grow" over developmental time. %owever, this

process of !internali?ation" of these e$ternal inputs is likely to be guided by the specific

properties that also govern adultlike language production and comprehension. Theories of

language acquisition such as &rain : Thornton -///6 or ang -;


28/37

that preserves the internal organi?ation of grammar, while allowing grammar to build up a

!bigger" set of le$ical items. This process should also build up grammatical relations among

these items -e.g. whether they are syntactic or semantic6 that a child learns when he learns a

language. Af one takes seriously the authors7 argument for acquisition matters as well, then

would e$pect that the structural properties of grammar -#uaa biological structure6 remain the

same, although the !si?e" of the grammar grows over time. Such principles are deeply

embedded in all the theories based on minimalist assumptions that we reviewed so far,

although they seem to be the main guiding principle in &rain : Thornton -///6 and ang

-;


29/37

of this theory. Cather different accounts e$ist, and may be considered the maority view in the

field -Tooby, &osmides : >arrett ;


30/37

constraints as to which theoretical approach, within >iological #pproach theori?ing, appears

to have a sounder biological plausibility.

Overall, What Darwin .ot Wrong presents two arguments which seem to be very

important for the biolinguistic approach and one of its specific incarnations, the minimalist

program. These two arguments enrich the debate on what constitutes desiderata for a

biologically sound approach to language. An doing so, What Darwin .ot Wrong appears useful

in stressing out the importance of formulating biolinguistic theories that are at the same time

empirically adequate and theoretically sound, since such theories need to be entirely

naturalistic. An doing so, the book spells out a very precise logical and empirical space in

which linguistic, and more precisely minimalist theories must move, in order to be

biologically plausible. At is of course an open question on whether a specific form of

minimalism can be easily defined, within this conceptual space. %owever, if the authors are

correct in their assessment, a theory that is defined within this space will be a theory that

meets a much sought, and yet elusive bio and logical validity. %ence, it will be a theory

worth pursuing in detail.

R(+(r(!c(s

#ndrews, #very. -;arbara %. *artee, eds. -//96.
http://arts.anu.edu.au/linguistics/People/AveryAndrews/Papers/mtr.pdfhttp://arts.anu.edu.au/linguistics/People/AveryAndrews/Papers/mtr.pdf


31/37

+uantification in 1atural languages. 3ordrecht8 @luwer #cademic *ublishers.

>arwise, 'on : Cobin &ooper. -/56. enerali?ed quantifiers and natural language.

$inguistics and 'hilosophyB -;6, 9/;/.

>Inite?>urraco, #ntonio : Jictor ). =onga -;iolinguistic

#pproach and Evo3evo.Biolinguistics B -B6, 1alari, Sergio : uillermo =oren?o. -;oston, )#8

#llyn and >acon.

&arroll, Sean >. -;


32/37

)AT *ress.

&rain, Stephen, : *aul *ietroski. -;lackwell.

Evans, Jyvyan. -;


33/37

and irau$.

(ong, Jivienne. -//46. he order of things3 What directional locatives denote. *h3

dissertation, Stanford niversity.

eurts, >art. -///6.'resuppositions and 'ronouns. Elsevier, O$ford.

oldberg, #dele E. -//96. onstructions7 A onstruction .rammar Approach to

Argument *tructure7 niversity of &hicago *ress.

oldberg, #dele E. -;


34/37

%ornstein, Norbert. -;


35/37

)illikan Cuth . -;


36/37

ompass B-06, B41;/.

Sle?ak, *eter. -;arrett. -;


37/37

Anterface in =anguage 3esign.Biolinguistics4 -6, 1B.

ang, &harles. -;

what model of biological plausibility

Documents