Evans and Levinson 2009 - The myth of language universals

BEHAVIORAL AND BRAIN SCIENCES (2009) 32, 429 -492
doi:10.1017/S0140525X0999094X
The myth of language universals:
Language diversity and its
importance for cognitive science
Nicholas Evans
Department of Linguistics, Research School of Asian and Pacific Studies,
Australian National University, ACT 0200, Australia
nicholas.evans@anu.edu.au
http://rspas.anu.edu.au/people/personal/evann_ling.php
Stephen C. Levinson
Max Planck Institute for Psycholinguistics, Wundtlaan 1, NL-6525 XD
Nijmegen, The Netherlands; and Radboud University, Department of
Linguistics, Nijmegen, The Netherlands
stephen.levinson@mpi.nl
http://www.mpi.nl/Members/StephenLevinson
Abstract: Talk of linguistic universals has given cognitive scientists the impression that languages are all built to a common pattern. In
fact, there are vanishingly few universals of language in the direct sense that all languages exhibit them. Instead, diversity can be found
at almost every level of linguistic organization. This fundamentally changes the object of enquiry from a cognitive science perspective.
This target article summarizes decades of cross-linguistic work by typologists and descriptive linguists, showing just how few and
unprofound the universal characteristics of language are, once we honestly confront the diversity offered to us by the world's 6,000
to 8,000 languages. After surveying the various uses of "universal," we illustrate the ways languages vary radically in sound,
meaning, and syntactic organization, and then we examine in more detail the core grammatical machinery of recursion,
constituency, and grammatical relations. Although there are significant recurrent patterns in organization, these are better explained
as stable engineering solutions satisfying multiple design constraints, reflecting both cultural-historical factors and the constraints of
human cognition.
Linguistic diversity then becomes the crucial datum for cognitive science: we are the only species with a communication system that
is fundamentally variable at all levels. Recognizing the true extent of structural diversity in human language opens up exciting new
research directions for cognitive scientists, offering thousands of different natural experiments given by different languages, with
new opportunities for dialogue with biological paradigms concerned with change and diversity, and confronting us with the
extraordinary plasticity of the highest human skills.
Keywords: Chomsky; coevolution; constituency; culture; dependency; evolutionary theory; Greenberg; linguistic diversity; linguistic
typology; recursion; universal grammar
1. Introduction
universals. Structural differences should instead be
accepted for what they are, and integrated into a new
approach to language and cognition that places diversity
According to Chomsky, a visiting Martian scientist would surely
at centre stage.
conclude that aside from their mutually unintelligible vocabularies,
Earthlings speak a single language.
The misconception that the differences between
-- Steven Pinker (1994, p. 232)
languages are merely superficial, and that they can be
resolved by postulating a more abstract formal level at
Languages are much more diverse in structure than cogni-
which individual language differences disappear, is
tive scientists generally appreciate. A widespread assump-
serious: it now pervades a great deal of work done in
tion among cognitive scientists, growing out of the
psycholinguistics, in theories of language evolution,
generative tradition in linguistics, is that all languages
language acquisition, neurocognition, parsing and speech
are English-like but with different sound systems and
recognition, and just about every branch of the cognitive
vocabularies. The true picture is very different: languages
sciences. Even scholars like Christiansen and Chater
differ so fundamentally from one another at every level of
(2008), concerned to demonstrate the evolutionary impossi-
description (sound, grammar, lexicon, meaning) that it is
bility of pre-evolved constraints, employ the term Universal
very hard to find any single structural property they
Grammar as if it were an empirically verified construct. A
share. The claims of Universal Grammar, we argue here,
great deal of theoretical work within the cognitive sciences
are either empirically false, unfalsifiable, or misleading
thus risks being vitiated, at least if it purports to be investi-
in that they refer to tendencies rather than strict
gating a fixed human language processing capacity, rather
# Cambridge University Press, 2009
0140-525X/09 $40.00
429

---

Evans & Levinson: The myth of language universals
than just the particular form this takes in some well-known
psychologists learned from the linguistic wars of the 1970s
languages like English and Japanese.
(Newmeyer 1986) to steer clear from too close an associ-
How did this widespread misconception of language uni-
ation with any specific linguistic theory, the underlying
formity come about? In part, this can be attributed simply
idea that all languages share the same structure at some
to ethnocentrism - most cognitive scientists, linguists
abstract level has remained pervasive, tying in nicely to
included, speak only the familiar European languages, all
the modularity arguments of recent decades (Fodor 1983).
close cousins in structure. But in part it can be attributed
It will take a historian of science to unravel the causes
to misleading advertizing copy issued by linguists them-
of this ongoing presumption of underlying language uni-
selves. Unfortunate sociological splits in the field have left
formity. But a major reason is simply that there is a lack
generative and typological linguists with completely differ-
of communication between theorists in the cognitive
ent views of what is proven science, without shared rules
sciences and those linguists most in the know about lin-
of argumentation that would allow them to resolve the
guistic diversity. This is partly because of the reluctance
issue - and in dialogue with cognitive scientists it has
by most descriptive and typological linguists to look up
been the generativists who have been taken as representing
from their fascinating particularistic worlds and engage
the dominant view. As a result, Chomsky's notion of Univer-
with the larger theoretical issues in the cognitive
sal Grammar (UG) has been mistaken, not for what it is -
sciences. Outsiders have instead taken the articulate
namely, the programmatic label for whatever it turns out
envoys from the universalizing generativist camp to
to be that all children bring to learning a language - but
represent the consensus view within linguistics. But
for a set of substantial research findings about what all
there are other reasons as well: the relevant literature
languages have in common. For the substantial findings
is forbiddingly opaque to outsiders, bristling with
about universals across languages one must turn to the
arcane phonetic symbols and esoteric terminologies.
field of linguistic typology, which has laid bare a bewildering
Our first goal (sect. 2) in this article, then, is to survey
range of diverse languages, where the generalizations are
some of the linguistic diversity that has been largely
really quite hard to extract. Chomsky's views, filtered
ignored in the cognitive sciences, which shows how differ-
through various commentators, have been hugely influen-
ently languages can be structured at every level: phonetic,
tial in the cognitive sciences, because they combine philoso-
phonological, morphological, syntactic, and semantic. We
phically sophisticated ideas and mathematical approaches
critically evaluate (sect. 3) the kind of descriptive general-
to structure with claims about the innate endowment for
izations (again, misleadingly called "universals") that have
language that are immediately relevant to learning theorists,
emerged from careful cross-linguistic comparisons, and
cognitive psychologists, and brain scientists. Even though
we survey the treacherously different senses of "universal"
that have allowed the term to survive a massive accumu-
lation of counterevidence.
NICHOLAS EVANS is Professor of Linguistics at the
We then turn to three syntactic features that have
Australian National University. His more than 120
recently figured large in debates about the origin of
linguistic publications include grammars of Kayardild
language: grammatical relations (sect. 4), constituency
and Bininj Gun-wok; dictionaries of Kayardild and
(sect. 5), and recursion (sect. 6). How universal are these
Dalabon; edited books on polysynthesis, linguistic pre-
features? We conclude that there are plenty of languages
history, and grammar-writing; and the recent Dying
that do not exhibit them in their syntax. What does it
Words: Endangered Languages and What They Have
mean for an alleged universal to not apply in a given
To Tell Us (Wiley Blackwell, 2009). He has carried out
case? We will consider the idea of "parameters" and the
intensive fieldwork on a number of languages of Australia
idea of UG as a "toolkit" (Jackendoff 2002).
and Papua New Guinea. Current research projects focus
We then turn (sect. 7) to the question of how all this
on the encoding of psychosocial cognition in grammar,
song language traditions of Arnhem Land, and languages
diversity is to be accounted for. We suggest, first, that lin-
of South Coast New Guinea. Evans is a fellow of the
guistic diversity patterns just like biological diversity and
Australian Academy of the Humanities.
should be understood in the same sorts of ways, with func-
tional pressures and systems constraints engineering con-
STEPHEN C. LEVINSON is co-director of the Max Planck
stant small changes. Finally (sect. 8), we advance seven
Institute for Psycholinguistics, and Professor of Com-
theses about the nature of language as a recently evolved
parative Linguistics at Radboud University, Nijmegen,
bio-cultural hybrid. We suggest that refocusing on a
The Netherlands. He is the author of more than 150
unique property of our communication system, namely
publications on language and cognition, including the
its diversity, is essential to understanding its role in
books Pragmatics (Cambridge University Press [CUP],
human cognition.
1983), Politeness (CUP, 1987), Presumptive Meanings
(MIT, 2000), Space in Language and Cognition (CUP,
2003). In addition, he has co-edited the following collec-
2. Language diversity
tions: Language Acquisition and Conceptual Develop-
ment (CUP, 2001) with M. Bowerman; Grammars of
A review of leading publications suggests that cognitive
Space (CUP, 2006) with D. Wilkins; Evolution and
scientists are not aware of the real range of linguistic diver-
Culture (MIT, 2006) with P. Jaisson; and Roots of Soci-
sity. In Box 1, for example, is a list of features, taken from a
ality (Berg, 2006) with N. Enfield. Levinson has done
BBS publication on the evolution of language, that all
extensive fieldwork on languages in India, Australia,
languages are supposed to have - "uncontroversial facts
Mexico, and Papua New Guinea, and coordinated
about substantive universals" (Pinker & Bloom 1990; a
research on the typology of languages in New Guinea
similar list is found in Pinker 1994). But none of these
and Australia. He is a fellow of the British Academy
and the Academia Europaea.
"uncontroversial facts" are true of all languages, as noted
in the box.
430
BEHAVIORAL AND BRAIN SCIENCES (2009) 32:5

---

Evans & Levinson: The myth of language universals
Box 1. "Every language has X, doesn't it?": Proposed substantive universals (from Pinker & Bloom
1990) supposedly common to all languages
1. "Major lexical categories (noun, verb, adjective, preposition)" (! sect. 2.2.4)
2. "Major phrasal categories (noun phrase, verb phrase, etc.)" (! sect. 5)
3. "Phrase structure rules (e.g., "X-bar theory" or "immediate dominance rules")" (! sect. 5)
4. "Rules of linear order" to distinguish, for example, subject from object, or "case affixes" which "can take over
these functions" (! sect. 5)
5. "Verb affixes" signaling "aspect" and "tense" (including pluperfects) (! sect. 2.2.3)
6. "Auxiliaries"
7. "Anaphoric elements" including pronouns and reflexives
8. "Wh-movement"
There are clear counterexamples to each of these claims. Problems with the first three are discussed in section
2.2.4 and section 5; here are counterexamples to the others:
(4) Some languages (e.g., Riau Indonesian) exhibit neither fixed word-order nor case-marking (Gil 2001).
(5) Many languages (e.g., Chinese, Malay) do not mark tense (Comrie 1985, pp. 50-55; Norman 1988, p. 163),
and many (e.g., spoken German) lack aspect (Comrie 1976, p. 8).
(6) Many languages lack auxiliaries (e.g., Kayardild, Bininj Gun-wok).
(7) Many languages (e.g. Mwotlap; Francois 2005, p. 119) lack dedicated reflexive or reciprocal constructions
altogether, so that "they hit them dead" can mean "they killed them," "they killed themselves," or "they killed
each other" (Levinson 2000, p. 334 ff.). Some Southeast Asian languages lack clear personal pronouns, using
titles (of the kind "honorable sir") instead, and many languages lack third-person pronouns (Cysouw 2001).
Sign languages like ASL (American Sign Language) also lack pronouns, using pointing instead.
(8) Not all languages (e.g., Chinese, Japanese, Lakhota) move their wh-forms, saying, in effect, "You came to see
who?" instead of "Who did you come to see _" (Van Valin & LaPolla 1997, pp. 424 -25).
Some further universalizing claims with counterevidence:
(9) Verbs for "give" always have three arguments (Gleitman 1990); Saliba is a counterexample (Margetts 2007).
(10) No recursion of case (Pinker & Bloom 1990). Kayardild has up to four layers (Evans 1995a; 1995c).
(11) No languages have nominal tense (Pinker & Bloom 1990) - Nordlinger and Sadler (2004) give numerous
counterexamples, such as Guarani "my house-FUTURE-FUTURE" "it will be my future house."
(12) All languages have numerals (Greenberg 1978b - Konstanz #527). See Everett (2005; Gordon 2004) for
counterexample.
(13) All languages have syntactic constituents, specifically NPs, whose semantic function is to express general-
ized quantifiers over the domain of discourse (Barwise & Cooper 1981 - Konstanz #1203); see Partee (1995)
and sect. 5.
See also collection of "rara" at: http://typo.uni-konstanz.de/rara/intro/index.php
The crucial fact for understanding the place of language
Enfield & Levinson 2006; Laland et al. 2000; Levinson
in human cognition is its diversity. For example, languages
& Jaisson 2006).
may have less than a dozen distinctive sounds, or they may
Why should the cognitive sciences care about language
have 12 dozen, and sign languages do not use sounds at all.
diversity, apart from their stake in evolutionary questions?
Languages may or may not have derivational morphology
First, a proper appreciation of the diversity completely
(to make words from other words, e.g., run . runner),
alters the psycholinguistic picture: What kind of language
or inflectional morphology for an obligatory set of syntac-
processing machine can handle all this variation? Not the
tically consequential choices (e.g., plural the girls are vs.
conventional one, built to handle the parsing of European
singular the girl is). They may or may not have constituent
sound systems and the limited morphological and syntactic
structure (building blocks of words that form phrases),
structures of familiar languages. Imagine a language where
may or may not have fixed orders of elements, and their
instead of saying, "This woman caught that huge butter-
semantic systems may carve the world at quite different
fly," one says, something like: "Thatobject thissubject
joints. We detail all these dimensions of variation later,
hugeobject caught womansubject butterflyobject"; such
but the point here is this: We are the only known species
languages exist (sect. 4). The parsing system for English
whose communication system varies fundamentally in
cannot be remotely like the one for such a language:
both form and content. Speculations about the evolution
What then is constant about the neural implementation
of language that do not take this properly into account
of language processing across speakers of two such differ-
thus overlook the criterial feature distinctive of the
ent languages? Second, how do children learn languages
species. The diversity of language points to the general
of such different structure, indeed languages that vary in
importance of cultural and technological adaptation in
every possible dimension? Can there really be a fixed
our species: language is a bio-cultural hybrid, a product
"language acquisition device"? These are the classic ques-
of intensive gene:culture coevolution over perhaps the
tions about how language capacities are implemented
last 200,000 to 400,000 years (Boyd & Richerson 1985;
in the mind and in the brain, and the ballgame is
BEHAVIORAL AND BRAIN SCIENCES (2009) 32:5
431

---

Evans & Levinson: The myth of language universals
fundamentally changed when the full range of language
in human languages are based on a maximal 500 languages
diversity is appreciated.
sample (in practice, usually much smaller - Greenberg's
The cognitive sciences have been partially immunized
famous universals of language were based on 30), and
against the proper consideration of language diversity by
almost every new language description still guarantees
two tenets of Chomskyan origin. The first is that the differ-
substantial surprises.
ences are somehow superficial, and that expert linguistic
Ethnologue, the most dependable worldwide source
eyes can spot the underlying common constructional
(http://www.ethnologue.com/), reckons that 82% of the
bedrock. This, at first a working hypothesis, became a
world's 6,912 languages are spoken by populations under
dogma, and it is wrong, in the straightforward sense that
100,000, 39% by populations under 10,000. These small
the experts either cannot formulate it clearly, or do not
speaker numbers indicate that much of this diversity is
agree that it is true. The second was an interesting intellec-
endangered. Ethnologue lists 8% as nearly extinct, and a
tual program that proceeded on the hypothesis that
language dies every two weeks. This loss of diversity, as
linguistic variation is "parametric"; that is, that there are
with biological species, drastically narrows our scientific
a restricted number of binary switches, which in different
understanding of what makes a possible human language.
states project out the full set of possible combinations,
Equally important as the brute numbers are the facts of
explaining observed linguistic diversity (Chomsky 1981;
relatedness. The number of language families is crucial to
see also Baker 2001). This hypothesis is now known to
the search for universals, because typologists want to test
be false as well: its predictions about language acquisition,
hypotheses against a sample of independent languages.
language change, and the implicational relations between
The more closely two languages are related, the less inde-
linguistic variables simply fail (Newmeyer 2004; 2005).
pendent they are as samplings of the design space. The
The conclusion is that the variation has to be taken at
question of how many distinct phylogenetic groupings
face value - there are fundamental differences in how
are found across the world's languages is highly controver-
languages work, with long historico-cultural roots that
sial, although Nichols' (1992) estimate of 300 "stocks" is
explain the many divergences.
reasonable, and each stock itself can have levels of diver-
Once linguistic diversity is accepted for what it is, it can
gence that make deep-time relationship hard to detect
be seen to offer a fundamental opportunity for cognitive
(English and Bengali within Indo-European; Hausa and
science. It provides a natural laboratory of variation in a
Hebrew within Afroa-Asiatic). In addition, there are
fundamental skill - 7,000 natural experiments in evolving
more than 100 isolates, languages with no proven affilia-
communicative systems, and as many populations of
tion whatsoever. A major problem for the field is that we
experts with exotic expertise. We can ask questions like:
currently have no way of demonstrating higher-level phy-
How much longer does it take a child to master 144
logenetic groupings that would give us a more principled
distinctive sounds versus 11? How do listeners actually
way of selecting a maximally independent sample for a
parse a free word order language? How do speakers plan
set smaller than these 300 to 400 groups. This may
the encoding of visual stimuli if the semantic resources
become more tractable with the application of modern
of the language make quite different distinctions? How
cladistic techniques (Dunn et al. 2005; Gray & Atkinson
do listeners break up the giant inflected words of a poly-
2003; McMahon & McMahon 2006), but such methods
synthetic language? In Bininj Gun-wok (Evans 2003a),
have yet to be fully adopted by the linguistic community.
for instance, the single word abanyawoihwarrgahmarne-
Suppose then that we think of current linguistic diver-
ganjginjeng can represent what, in English, would consti-
sity as represented by 7,000 languages falling into 300 or
tute an entire sentence: "I cooked the wrong meat for
400 groups. Five hundred years ago, before the expansion
them again." These resources offered by diversity have
of Western colonization, there were probably twice as
scarcely been exploited in systematic ways by the scientific
many. Because most surviving languages are spoken by
community: We have a comparative psychology across
small ethnic groups, language death continues apace. If
species, but not a proper comparative psychology inside
we project back through time, there have probably been
our own species in the central questions that drive cogni-
at least half a million human languages (Pagel 2000), so
tive science.
what we now have is a non-random sample of less than
2% of the full range of human linguistic diversity. It
would be nice to at least be in the position to exploit
2.1. The current representation of languages
that sample, but in fact, as mentioned, we have good infor-
in the world
mation for only 10% of that. The fact is that at this stage of
Somewhere between 5,000 and 8,000 distinct languages
linguistic inquiry, almost every new language that comes
are spoken today. How come we cannot be more
under the microscope reveals unanticipated new features.
precise? In part because there are definitional problems:
When does a dialect difference become a language differ-
ence (the "languages" Czech and Slovak are far closer in
2.2. Some dimensions of diversity
structure and mutual intelligibility than so-called dialects
In this section we illustrate some of the surprising dimen-
of Chinese like Mandarin and Cantonese)? But mostly it
sions of diversity in the world's languages. We show how
is because academic linguists, especially those concerned
languages may or may not be in the articulatory-auditory
with primary language description, form a tiny commu-
channel, and if they are how their inventories of contras-
nity, far outnumbered by the languages they should be
tive sounds vary dramatically, how they may or may not
studying, each of which takes the best part of a lifetime
have morphologies (processes of word derivation or inflec-
to master. Less than 10% of these languages have decent
tion), how varied they can be in syntactic structure or their
descriptions (full grammars and dictionaries). Conse-
inventory of word classes, and how varied are the semantic
quently, nearly all generalizations about what is possible
distinctions which they encode. We can do no more here
432
BEHAVIORAL AND BRAIN SCIENCES (2009) 32:5

---

Evans & Levinson: The myth of language universals
than lightly sample the range of diversity, drawing atten-
linguistic diversity (Maddieson 1984). Jakobson's distinc-
tion to a few representative cases.
tive features - binary values on a limited set of (largely)
acoustic parameters - were meant to capture the full set
2.2.1. Sound inventories. We start by noting that some
of possible speech sounds. They were the inspiration
natural human languages do not have sound systems at
for the Chomskyan model of substantive universals, a con-
all. These are the sign languages of the deaf. Just like
strained set of alternates from which any particular
spoken languages, many of these have developed indepen-
language will select just a few. But as we get better infor-
dently around the world, wherever a sufficient intercom-
mation from more languages, sounds that we had thought
municating population of deaf people has arisen, usually
were impossible to produce or impractical to distinguish
as a result of a heritable condition. (Ethnologue, an
keep turning up. Take the case of double-articulations,
online inventory of languages, lists 121 documented sign
where a consonantal closure is made in more than one
languages, but there are certainly many more.) These
place. On the basis of evidence then available, Maddieson
groups can constitute both significant proportions of
(1983) concluded that contrastive labial-alveolar conso-
local populations and substantial populations in absolute
nants (making a sound like "b" at the same time as a
terms: in India there are around 1.5 million signers.
sound like "d") were not a possible segment in natural
They present interesting, well-circumscribed models
language on auditory grounds. But it was then discovered
of gene-culture coevolution (Aoki & Feldman 1994;
that the Papuan language Yeli Dnye makes a direct con-
Durham 1991): Without the strain of hereditary deafness,
trast between a coarticulated "tp," and a "tp" where the t
the cultural adaptation would not exist, whereas the


is further back towards the palate (Ladefoged & Maddie-
cultural adaptation allows signers to lead normal lives,
son 1996, pp. 344 -45; Maddieson & Levinson, in
productive and reproductive, thus maintaining the
preparation).
genetic basis for the adaptation.
As more such rarities accrue, experts on sound systems
The whole evolutionary background to sign languages
are abandoning the Jakobsonian idea of a fixed set of par-
remains
fascinating
but
obscure - were
humans
ameters from which languages draw their phonological
endowed, as Hauser (1997, p. 245) suggests, with a capa-
inventories, in favor of a model where languages can
bility unique in the animal world to switch their entire
recruit their own sound systems from fine phonetic
communication system between just two modalities, or
details that vary in almost unlimited ways (see also
(as the existence of touch languages of the blind-deaf
Mielke 2007; Pierrehumbert et al. 2000):
suggest) is the language capacity modality-neutral?
Do phoneticians generally agree with phonologists that we will
There have been two hundred years of speculation that
eventually arrive at a fixed inventory of possible human speech
sign languages may be the evolutionary precursors to
sounds? The answer is no. (Port & Leary 2005, p. 927)
human speech, a view recently revived by the discovery
And,
of mirror-neurons (Arbib 2005). An alternative view is
Languages can differ systematically in arbitrarily fine phonetic
that language evolved from a modality-hybrid communi-
detail. This means we do not want to think about universal
cation system in which hand and mouth both participated,
phonetic categories, but rather about universal phonetic
as they do today in both spoken and signed languages
resources, which are organized and harnessed by the cognitive
(cf. Sandler 2009). Whichever evolutionary scenario you
system . . . . The vowel space - a continuous physical space
favor, the critical point here is that sign languages are an
rendered useful by the connection it establishes between
existence proof of the modality-plastic nature of our
articulation and perception - is also a physical resource. Cul-
language capacity. At a stroke, therefore, they invalidate
tures differ in the way they divide up and use this physical
such generalizations as "all natural languages have oral
resource. (Pierrehumbert 2000, p. 12)
vowels," although at some deeper level there may well
be analogies to be drawn: signs have a basic temporal
organization of "move and hold" which parallels the rhyth-
2.2.2. Syllables and the "CV" universal. The default
mic alternation of vowels and consonants.
expectation of languages is that they organize their
Returning to spoken languages, the vocal tract itself
sounds into an alternating string of more versus less sonor-
is the clearest evidence for the biological basis for
ant segments, creating a basic rhythmic alternation of
language - the lowering of the larynx and the right-angle
sonorous vowels (V) and less sonorous consonants (C).
in the windpipe have been optimized for speaking at the
But beyond this, a further constraint was long believed
expense of running and with some concomitant danger
to be universal: that there was a universal preference for
of choking (Lenneberg 1967). Similar specializations
CV syllables (like law /l :/ or gnaw /n :/) over VC sylla-
exist in the auditory system, with acuity tuned just to the
bles (like awl / :l/ or awn / :n/). The many ways in which
speech range, and, more controversially, specialized
languages organize their syllable structures allows the
neural pathways for speech analysis. These adaptations
setting up of implicational (if/then) statements which
of the peripheral input/output systems for spoken
effectively find order in the exuberant variation: No
language have, for some unaccountable reason, been mini-
language will allow VC if it does not also allow CV, or
mized in much of the discussion of language origins, in
allow V if it does not also allow CV:
favor of an emphasis on syntax (see, for example, Hauser
et al. 2002).
CV . V . VC
The vocal tract and the auditory system put strong
This long-proclaimed conditional universal (Jakobson &
constraints on what an articulatorily possible and percep-
Halle 1956; cf. Clements & Keyser 1983; Jakobson 1962)
tually distinguishable speech sound is. Nevertheless, the
has as corollary the maximal onset principle (Blevins
extreme range of phonemic (distinctive sound) inven-
1995, p. 230): a /. . ..VCV. . ../ string will universally be
tories, from 11 to 144, is already a telling fact about
syllabified as /. . .V-CV. . ./. An obvious advantage such a
BEHAVIORAL AND BRAIN SCIENCES (2009) 32:5
433

---

Evans & Levinson: The myth of language universals
universal principle would give the child is that it can go
is mirrored by differences in grammatical organization
right in and parse strings into syllables from first exposure.
right through to the deepest levels of how meaning is
But in 1999, Breen and Pensalfini published a clear
organized.
demonstration that Arrernte organizes its syllables
around a VC(C) structure and does not permit consonantal
2.2.4. Syntax and word-classes. Purported syntactic uni-
onsets (Breen & Pensalfini 1999). With the addition of this
versals lie at the heart of most claims regarding UG, and
one language to our sample, the CV syllable gets down-
we hold off discussing these in detail until sections 4
graded from absolute universal to a strong tendency, and
through 6. As a warm-up, though, we look at one funda-
the status of the CV assumption in any model of UG
mental issue: word-classes, otherwise known as parts of
must be revised. If CV syllables really were inviolable
speech. These are fundamental to grammar, because the
rules of UG, Arrernte would then be unlearnable, yet chil-
application of grammatical rules is made general by formu-
dren learn Arrernte without difficulty. At best, then, the
lating them over word-classes. If we say that in English
child may start with the initial hypothesis of CVs, and
adjectives precede but cannot follow the nouns they
learn to modify it when faced with Arrernte or other
modify (the rich man but not Athe man rich), we get a gen-
such languages. But in that case we are talking about
eralization that holds over an indefinitely large set of
initial heuristics, not about constraints on possible
phrases, because both adjectives and nouns are "open
human languages. The example also shows, as is familiar
classes" that in principle are always extendable by new
from the history of mathematical induction (as with the
members. But to stop it generating Athe nerd zappy we
Gauss-Riemann hypothesis regarding prime number den-
need to know that nerd is a noun, not an adjective, and
sities), that an initially plausible pattern turns out not to be
that zappy is an adjective, not a noun. To do this we
universal after all, once the range of induction is suffi-
need to find a clearly delimited set of distinct behaviors,
ciently extended.
in their morphology and their syntax, that allows us to dis-
tinguish noun and adjective classes, and to determine
2.2.3. Morphology. Morphological differences are among
which words belong to which class.
the most obvious divergences between languages, and lin-
Now it has often been assumed that, across all
guistic science has been aware of them since the Spanish
languages, the major classes - those that are essentially
encountered Aztec and other polysynthetic languages in
unlimited in their membership - will always be the same
sixteenth-century Mexico, while half a world away the Por-
"big four": nouns, verbs, adjectives, and adverbs. But we
tuguese were engaging with isolating languages in Vietnam
now know that this is untenable when we consider the
and China. Isolating languages, of course, lack all the
cross-linguistic evidence. Many languages lack an open
inflectional affixes of person, number, tense, and aspect,
adverb class (Hengeveld 1992), making do with other
as well as systematic word derivation processes. They
forms of modification. There are also languages like Lao
even lack the rather rudimentary morphology of English
with no adjective class, encoding property concepts as a
words like boy-s or kiss-ed, using just the root and
sub-sub-type of verbs (Enfield 2004).
getting plural and past-tense meanings either from
If a language jettisons adjectives and adverbs, the last
context or from other independent words. Polysynthetic
stockade of word-class difference is that between nouns
languages go overboard in the other direction, packing
and verbs. Could a language abolish this and just have a
whole English sentences into a single word, as in Cayuga
single word-class of predicates (like predicate calculus)?
Eskakheh na'tayethwahs "I will plant potatoes for them
Here controversy still rages among linguists as the bar
again" (Evans & Sasse 2002). Clearly, children learning
for evidence of single-class languages keeps getting
such languages face massive challenges in picking out
raised, with some purported cases (e.g., Mundari) falling
what the "words" are that they must learn. They must
by the wayside (Evans & Osada 2005). For many languages
of the Philippines and the Pacific Northwest Coast, the
also learn a huge set of rules for morphological compo-
argument has run back and forth for nearly a century,
sition, since the number of forms that can be built from
with the relevant evidence becoming ever more subtle,
a small set of lexical stems may run into the millions (Han-
but still no definitive consensus has been reached.
kamer 1989).
A feeling for what a language without a noun-verb dis-
But if these very long words function as sentences,
tinction is like comes from Straits Salish. Here, on the
perhaps there's no essential difference: perhaps, for
analysis by Jelinek (1995), all major-class lexical items
example, the Cayuga morpheme -h na- for "potatoes" in
simply function as predicates, of the type "run,"
the word above is just a word-internal direct object as
"be_big," or "be_a_man." They then slot into various
Baker (1993; 1996) has claimed. However, the parallels
clausal roles, such as argument ("the one such that he
turn out to be at best approximate. For example, the pro-
runs"), predicate ("run[s]"), and modifier ("the running
nominal affixes and incorporated nouns do not need to be
[one]"), according to the syntactic slots they are placed
referential. The prefix ban- in Bininj Gun-wok ka-ban-
in. The single open syntactic class of predicate includes
dung [she-them-scolds] is only superficially like its
words for events, entities, and qualities. When used
English free-pronoun counterpart, since kabandung can
directly as predicates, all appear in clause-initial position,
followed by subject and/or object clitics. When used as
mean both "she scolds them" and "she scolds people in
arguments, all lexical stems are effectively converted into
general" (Evans 2002). It seems more likely, then, that
relative clauses through the use of a determiner, which
much of the obvious typological difference between poly-
must be employed whether the predicate-word refers to an
synthetic languages and more moderately synthetic
event ("the [ones who] sing"), an entity ("the [one which
languages like English or Russian needs to be taken at
is a] fish"), or even a proper name ("the [one which] is
face value: the vast difference in morphological complexity
Eloise"). The square-bracketed material shows what we
434
BEHAVIORAL AND BRAIN SCIENCES (2009) 32:5

---

Evans & Levinson: The myth of language universals
need to add to the English translation to convert the
Once again, then, the great variability in how languages
reading in the way the Straits Salish structure lays out.
organize their word-classes dilutes the plausibility of the
There are thus languages without adverbs, languages
innatist UG position. Just which word classes are supposed
without adjectives, and perhaps even languages without
to be there in the learning child's mind? We would need to
a basic noun-verb distinction. In the other direction, we
postulate a start-up state with an ever-longer list of initial
now know that there are other types of major word-class -
categories (adding ideophones, positionals, coverbs, classi-
e.g., ideophones, positionals, and coverbs - that are unfa-
fiers, etc.), many of which will never be needed. And,
miliar to Indo-European languages.
because syntactic rules work by combining these word-
Ideophones typically encode cross-modal perceptual
class categories - "projecting" word-class syntax onto the
properties - they holophrastically depict the sight, sound,
larger syntactic assemblages that they head - each word-
smell, or feeling of situations in which the event and its par-
class we add to the purported universal inventory would
ticipants are all rolled together into an undissected gestalt.
then need its own accompanying set of syntactic
They are usually only loosely integrated syntactically,
constraints.
being added into narratives as independent units to spice
up the color. Examples from Mundari (Osada 1992) are
2.2.5. Semantics. There is a persistent strand of thought,
ribuy-tibuy, "sound, sight, or motion of a fat person's but-
articulated most forcefully by Fodor (1975), that languages
tocks rubbing together as they walk," and rawa-dawa,
directly encode the categories we think in, and moreover
"the sensation of suddenly realizing you can do something
that these constitute an innate, universal "language of
reprehensible, and no-one is there to witness it." Often
thought" or "mentalese." As Pinker (1994, p. 82) put it,
ideophones have special phonological characteristics, such
"Knowing a language, then, is knowing how to translate
as vowel changes to mark changes in size or intensity,
mentalese into strings of words and vice versa. People
special reduplication patterns, and unusual phonemes or
without a language would still have mentalese, and
tonal patterns. (Note that English words like willy-nilly or
babies and many nonhuman animals presumably have
heeby-jeebies may seem analogous, but they differ from
simpler dialects." Learning a language, then, is simply a
ideophones in all being assimilated to other pre-existing
matter of finding out what the local clothing is for universal
word classes, here adverb and noun.)
concepts we already have (Li & Gleitman 2002).
Positionals describe the position and form of persons and
The problem with this view is that languages differ enor-
objects (Ameka & Levinson 2007). These are widespread
mously in the concepts that they provide ready-coded in
in Mayan languages (Bohnemeyer & Brown 2007;
grammar and lexicon. Languages may lack words or
Brown 1994; England 2001; 2004). Examples from
constructions corresponding to the logical connectives
Tzeltal include latz'al, "of flat items, arranged in vertical
"if" (Guugu Yimithirr) or "or" (Tzeltal), or "blue" or
stack"; chepel, "be located in bulging bag," and so on. Posi-
"green" or "hand" or "leg" (Yeli Dnye). There are
tionals typically have special morphological and syntactic
languages without tense, without aspect, without
properties.
numerals, or without third-person pronouns (or even
Coverbs are a further open class outside the "big four."
without pronouns at all, in the case of most sign
Such languages as Kalam (PNG; Pawley 1993) or the Aus-
languages). Some languages have thousands of verbs;
tralian language Jaminjung (Schultze-Berndt 2000) have
others only have thirty (Schultze-Berndt 2000). Lack of
only around 20 to 30 inflecting verbs, but they form
vocabulary may sometimes merely make expression more
detailed event-descriptors by combining inflecting verbs
cumbersome, but sometimes it effectively limits expressi-
with an open class of coverbs. Unlike positionals or ideo-
bility, as in the case of languages without numerals
phones, coverbs are syntactically integrated with inflecting
(Gordon 2004).
verbs, with which they cross-combine in ways that largely
In the other direction, many languages make semantic
need to be learned individually. In Jaminjung, for
distinctions we certainly would never think of making.
example, the coverb dibird, "wound around" can
So Kiowa, instead of a plural marker on nouns, has a
combine with yu, "be," to mean "be wound around," and
marker that means roughly "of unexpected number": on
with angu, "get/handle," to mean "tangle up." (English
an animate noun like "man" it means "two or more," on
"light verbs," as in take a train or do lunch, give a feel
a word like "leg," it means "one or more than two," and
for the phenomenon, but of course train and lunch are
on "stone," it means "just two" (Mithun 1999, p. 81). In
just regular nouns.)
many languages, all statements must be coded (e.g., in
Classifiers are yet another word class unforeseen by the
verbal affixes) for the sources of evidence; for example,
categories of traditional grammar - whether "numeral
in Central Pomo, whether I saw it, perceived it in
classifiers" in East Asian and Mesoamerican languages
another modality (tactile, auditory), was told about it,
that classify counted objects according to shape, or the
inferred it, or know that it is an established fact (Mithun
hand-shape classifiers in sign languages that represent
1999, p. 181). Kwakwala insists on referents being coded
the involved entity through a schematized representation
as visible or not (Anderson & Keenan 1985). Athabaskan
of its shape. And further unfamiliar word classes are con-
languages are renowned for their classificatory verbs,
tinuously being unearthed that respect only the internal
forcing a speaker to decide between a dozen categories
structural logic of previously undescribed languages.
of objects (e.g., liquids, rope-like objects, containers, flexi-
Even when typologists talk of "ideophones," "classifiers,"
ble sheets) before picking one of a set of alternate verbs of
and so forth, these are not identical in nature across the
location, giving, handling, and so on (Mithun 1999, p. 106
languages that exhibit them - rather we are dealing with
ff.). Australian languages force their speakers to pay atten-
family-resemblance phenomena: no two languages have
tion to intricate kinship relations between participants
any word classes that are exactly alike in morphosyntactic
in the discourse - in many to use a pronoun you must
properties or range of meanings (Haspelmath 2007).
first work out whether the referents are in even- or
BEHAVIORAL AND BRAIN SCIENCES (2009) 32:5
435

---

Evans & Levinson: The myth of language universals
odd-numbered generations with respect to one another, or
to construct a meaning recognizable to us as "banana" or
related by direct links through the male line. On top of
"candle."
this, many have special kin terms that triangulate the
In the light of examples like these, the view that "linguis-
relation between speaker, hearer, and referent, with
tic categories and structures are more or less straightfor-
meanings like "the one who is my mother and your daugh-
ward mappings from a pre-existing conceptual space
ter, you being my maternal grandmother" (Evans 2003b).
programmed into our biological nature" (Li & Gleitman
Spatial concepts are an interesting domain to compare
2002, p. 266) looks quite implausible. Instead, languages
languages in, because spatial cognition is fundamental to
reflect cultural preoccupations and ecological interests
any animal - and if Fodor is right anywhere, it should
that are a direct and important part of the adaptive char-
be here. But, in fact, we find fundamental differences in
acter of language and culture.
the semantic parameters languages use to code space.
For example, there are numerous languages without
notions of "left of," "right of," "back of," "front of " -
words meaning "right hand" or "left hand" are normally
3. Linguistic universals
present, but don't generalize to spatial description. How
The prior sections have illustrated the surprising range of
then does one express, for example, that the book you
cross-linguistic variability at every level of language, from
are looking for is on the table left of the window? In
sound to meaning. The more we discover about languages,
most of these languages by saying that it lies on the table
the more diversity we find. Clearly, this ups the ante in the
north of the window - that is, by using geographic
search for universals.
rather than egocentric coordinates. Research shows that
There have been two main approaches to linguistic uni-
speakers remember the location in terms of the coordinate
versals. The first, already mentioned, is the Chomskyan
system used in their language, not in terms of some fixed,
approach, where UG denotes structural principles which
innate mentalese (see Levinson 2003; Majid et al. 2004).
are complex and implicit enough to be unlearnable from
Linguists often distinguish between closed-class or func-
finite exposure. Chomsky thus famously once held that
tion words (like the, of, in, which play a grammatical role)
language universals could be extracted from the study of
and open-class items or general vocabulary which can be
a single language:
easily augmented by new coinages or borrowing. Some
researchers claim that closed-class items reveal a recurrent
I have not hesitated to propose a general principle of linguistic
set of semantic distinctions, whereas the open-class items
structure on the basis of observation of a single language. The
inference is legitimate, on the assumption that humans are not
may be more culture-specific (Talmy 2000). Others claim
specifically adapted to learn one rather than another human
effectively just the reverse, that relational vocabulary (as in
language. . . . Assuming that the genetically determined
prepositions) is much more abstract, and thus prone to cul-
language faculty is a common human possession, we may con-
tural patterning, whereas the open-class items (like nouns)
clude that a principle of language is universal if we are led to
are grounded in concrete reality, and thus less cross-linguis-
postulate it as a "precondition" for the acquisition of a single
tically variable (Gentner & Boroditsky 2001). In fact, neither
language. (Chomsky 1980, p. 48)1
of these views seems correct, for both ends of the spectrum
Chomsky (1965, pp. 27-30) influentially distinguished
are cross-linguistically variable. Consider, for example,
between substantive and formal universals. Substantive
the difference between nouns and spatial prepositions.
universals are drawn from a fixed class of items (e.g., dis-
Landau and Jackendoff (1993) claimed that this difference
tinctive phonological features, or word classes like noun,
corresponds to the nature of the so-called what versus
verb, adjective, and adverb). No particular language is
where systems in neurocognition: nouns are "whaty" in
required to exhibit any specific member of a class. Conse-
that their meanings code detailed features of objects,
quently, the claim that property X is a substantive univer-
while prepositions are "wherey" in that they encode abstract,
sal cannot be falsified by finding a language without it,
geometric properties of spatial relations. These researchers
because the property is not required in all of them. Con-
thus felt able to confidently predict that there would be
versely, suppose we find a new language with property
no preposition or spatial relator encoding featural properties
Y, hitherto unexpected: we can simply add it to the inven-
of objects, for example, none meaning "through a cigar-
tory of substantive universals. Jackendoff (2002, p. 263)
shaped object" (Landau & Jackendoff 1993, p. 226). But
nevertheless holds "the view of Universal Grammar
the Californian language Karuk has precisely such a spatial
as a "toolkit" . . . : beyond the absolute universal bare
verbal prefix, meaning "in through a tubular space"
minimum of concatenated words . . . languages can pick
(Mithun 1999, p. 142)! More systematic examination of
and choose which tools they use, and how extensively."
the inventories of spatial pre- and post-positions shows
But without limits on the toolkit, UG is unfalsifiable.
that there is no simple universal inventory, and the mean-
Formal universals specify abstract constraints on the
ings can be very specific; for example, "in a liquid," "astrad-
grammar of languages (e.g., that they have specific rule
dle," "fixed by spiking" (Levinson & Meira 2003) - or
types or cannot have rules that perform specific oper-
distinguish "to (a location below)" versus "to (a location
ations). To give a sense of the kind of abstract constraints
above)" versus "to (a location on a level with the speaker)."
in UG, consider the proposed constraint called Subjacency
Nor do nouns always have the concrete sort of reference
(see Newmeyer 2004, p. 537 ff.). This is an abstract prin-
we expect - for example, in many languages nouns tend to
ciple meant to explain the difference between the gram-
have a mass or "stuff"-like reference (meaning, e.g., any
maticality of the sentence (6) and (7), below, versus the
stuff composed of banana genotype, or anything made of
ungrammaticality (marked by an asterisk) of sentence (8):
wax), and do not inherently refer to bounded entities. In
(6) Where did John say that we had to get off the bus?
such languages, it takes a noun and a classifier (Lucy
(7) Did John say whether we had to get off the bus?
1992), or a noun and a classificatory verb (Brown 1994),
(8) AWhere did John say whether we had to get off the bus?
436
BEHAVIORAL AND BRAIN SCIENCES (2009) 32:5

---

Evans & Levinson: The myth of language universals
The child somehow has to extrapolate that (6) and (7)
the field as a whole has not kept a generally accepted
are okay, but (8) is not, without ever being explicitly told
running score of which putative universals are left standing.
that (8) is ungrammatical. This induction is argued to be
In short, it has proven extremely hard to come up with
impossible, necessitating an underlying and innate prin-
even quite abstract generalizations that don't run afoul of
ciple that forbids the formation of wh-questions if a wh-
the cross-linguistic facts. This doesn't mean that such gen-
phrase intervenes between the "filler" (initial wh-word)
eralizations won't ultimately be found, nor that there are
and the "gap" (the underlying slot for the wh-word).
no genetic underpinnings for language - there certainly
This presumes a movement rule pulling a wh-phrase out
are.2 But, to date, strikingly little progress has been made.
of its underlying position and putting it at the front of
We turn now to the other approach to universals, stem-
the sentence as shown in (9):
ming from the work of Greenberg (1963a; 1963b), which
(9) AWhere did John say whether we had to get off the
directly attempts to test linguistic universals against the
bus ____?
diversity of the world's languages. Greenberg's methods crys-
However, it turns out that this constraint does not work
tallized the field of linguistic typology, and his empirical gen-
in Italian or Russian in the same way, and theorists have
eralizations are sometimes called Greenbergian universals.
had to assume that children can learn the specifics of the
First, importantly, Greenberg discounted features of
constraint after all, although we do not know how
language that are universal by definition - that is, we
(Newmeyer 2004; Van Valin & LaPolla 1997, p. 615 ff.).
would not call the object in question a language if it
This shows the danger of extrapolations from a single
lacked these properties (Greenberg et al. 1963, p. 73).
language to unlearnable constraints. Each constraint in
Thus, many of what Hockett (1963) called the "design
UG needs to be taken as no more than a working hypoth-
features" of language are excluded - for example, discre-
esis, hopefully sufficiently clearly articulated that it could
teness, arbitrariness, productivity, and the duality of
be falsified by cross-linguistic data.
patterning achieved by combining meaningless elements
But what counts as falsification of these often abstract
at one level (phonology) to construct meaningful elements
principles? Consider the so-called Binding Conditions,
(morphemes or words) at another.3 We can add other
proposed as elements of Universal Grammar in the
functional features that all languages need in order to
1980s (see Koster & May 1982). One element (condition
be adequately expressive instruments (e.g., the ability to
A) specifies that anaphors (reflexives and reciprocals)
indicate negative or prior states of affairs, to question,
must be bound in their governing category, whereas a
to distinguish new from old information, etc.).
second (condition B) states that (normal nonreflexive)
Second, Greenberg (1960, see also Comrie 1989: 17-23)
pronouns must be free in their governing category.
distinguished the different types of universal statement
These conditions were proposed to account for the
laid out in Table 1 (the terminology may differ slightly
English data in (10a-c) and comparable data in many
across sources):
other languages (the subscripts keep track of what each
Although all of these types are universals in the sense
term refers to). The abstract notion of "bound" is tied to
that they employ universal quantification over languages,
a particular type of constituent-based syntactic represen-
their relations to notions of "universal grammar" differ
tation where the subject "commands" the object (owing
profoundly. Type 1 statements are true of all languages,
to its position in a syntactic tree) rather than the other
though not tautological by being definitional of language-
way round, and reflexives are sensitive to this command.
hood. This is the category which cognitive scientists
Normal pronouns pick up their reference from elsewhere
often imagine is filled by rich empirical findings from a
and so cannot be used in a "bound" position.
hundred years of scientific linguistics - indeed Greenberg
(10a) Johnx saw himy. (disjoint reference)
(1986, p. 14) recollects how Osgood challenged him to
(10b) Johnx saw himselfx (conjoint reference)
produce such universals, saying that these would be of fun-
(10c) AHimselfx saw Johnx/himx.
damental interest to psychologists. This started Greenberg
This works well for English and hundreds, perhaps
thousands, of other languages, but it does not generalize
Table 1. Logical types of universal statement (following
to languages where you get examples as in (11a, b) (to rep-
Greenberg)
resent their structures in a pseudo-English style).
(11a) Hex saw himx,y
(11b) They
Absolute
Statistical
x,y saw thema,b/x,y/y,x.
Many languages (even Old English; see Levinson 2000)
(exceptionless)
(tendencies)
allow sentences like (11a) and (11b): the same pronouns
Unconditional
Type 1. "Unrestricted
Type 2. "Unrestricted
can either have disjoint reference (shown as "a,b"), con-
(unrestricted)
absolute
tendencies"
joint reference ("x,y") or commuted conjoint reference
universals"
Most languages
("y,x," corresponding to "each other" in English). Does
All languages have
have property X
this falsify the Binding Principles? Not necessarily,
property X
would be a typical response in the generativist position -
it may be that there are really two distinct pronouns (a
Conditional
Type 3.
Type 4. "Statistical
normal pronoun and a reflexive, say) which just happen
(restricted)
"Exceptionless
implicational
to have the same form, but can arguably be teased apart
implicational
universals"
in other ways (see, e.g., Chung [1989] on Chamorro).
universals"
If a language has
But it is all too easy for such an abstract analysis to presup-
If a language has
property X, it will
pose precisely what is being tested, dismissing seeming
property X, it also
tend to have
counterexamples and rendering the claims unfalsifiable.
has property Y
property Y
The lack of shared rules of argumentation means that
BEHAVIORAL AND BRAIN SCIENCES (2009) 32:5
437

---

Evans & Levinson: The myth of language universals
Box 2. The challenge of sign languages
Many proposed universals of language ignore the existence of sign languages - the languages of the deaf,
now recognized to be full-blown languages of independent origin (Klima & Bellugi 1979). Studies of, for
example, American Sign Language, British Sign Language, and Indo-Pakistani Sign Language (Zeshan 2002)
show that these are unrelated, complex systems of their own. They can even be said to have "phonologies" -
patterns of hand shape, facial expression, and so on, which, although individually meaningless, can be combined
to make morphemes or words (Padden & Perlmutter 1987).
The typology of sign languages is in its infancy (see, e.g., Perniss et al. 2008; Perniss & Zeshan 2008; Schwager
& Zeshan 2008; Zeshan 2006a; 2006b). The Ethnologue lists 121 sign languages, but there are certainly many
not yet listed. The major sign languages show some typological similarities, but the smaller ones, only now
coming under scrutiny, are typologically diverse (see, e.g., Meir et al., in press).
Sign languages offer a model "organism" for understanding the relation between biological and cultural
aspects of language (Aoki & Feldman 1994). They also offer unique opportunities to study the emergence of
new languages under different conditions: (a) where home-signers (Goldin-Meadow 2003) are congregated
and a sign language emerges among themselves, as in Nicaragua (Senghas et al. 2004); and (b) where a localized
hereditary deaf population lives among hearers who also sign, as in Bali (Marsaja 2008) or in a Bedouin group in
Israel (Sandler et al. 2005). These studies show that although word order constraints may show early, it may take
three generations or more to evolve syntactic embedding and morphology.
When due allowance is made for the manual-visual interface, sign languages seem to be handled