Sign Language

Abstract

Sign language is a type of language expressed through the hands, face, and body rather than through sound, used chiefly within Deaf communities. This article sets out the evidence that natural signed languages are full languages in every structural sense, organised at every level a spoken language is, and that they are acquired, processed, and represented in the brain in the same ways. It describes the sub-lexical structure of signs, the manual babbling of signing infants, the way age of first exposure bounds ultimate attainment, the spontaneous birth of new signed languages, the role of visual iconicity, and the left-lateralised brain network that supports signing. Three interactive demonstrations let a reader build a minimal pair from the parameters of a sign, trace how age of acquisition shapes grammatical attainment, and place signs along the continuum from iconic to arbitrary.

Keywords: sign language, phonology, critical period, language emergence, iconicity

Sign language is the visual-gestural system through which Deaf communities encode meaning in the configuration and movement of the hands, the expression of the face, and the posture of the body, and exchange it with one another. In the Medical Subject Headings vocabulary the descriptor names a system of hand gestures used for communication by those with hearing loss, and it is filed as a narrower kind of language, alongside the study of language and the language arts. That placement is exactly right: a natural signed language such as American Sign Language (ASL) or British Sign Language (BSL) is not a manual code for a surrounding spoken language, nor a system of pantomime, but a distinct, fully grammatical language with its own vocabulary and its own syntax, unrelated to the spoken language of the hearing majority around it. ASL, for instance, is historically related to French Sign Language and is mutually unintelligible with BSL, even though the United States and Britain share a spoken tongue. The central finding of half a century of research is that the visual modality changes the outward form of language without changing its essential design.

The Structure of a Signed Language

The discovery that founded the field was that signs are not holistic pictures but are built, like spoken words, from a small inventory of meaningless contrasting units. William Stokoe, working at Gallaudet University, showed in 1960 that every sign of ASL can be analysed into a handful of simultaneous parameters — the handshape the hand assumes, the location on or near the body where the sign is made, and the movement the hand traces — with a fourth, palm orientation, added later (Stokoe, 2005). These parameters are the signed counterpart of phonemes: individually meaningless, they combine to distinguish one sign from another. Two signs that differ in exactly one parameter form a minimal pair, just as pat and bat do in English. The ASL signs glossed APPLE and ONION, for example, share a handshape and a movement but differ in location; CANDY, APPLE, and JEALOUS form a set contrasting in handshape at the same location. This is duality of patterning — a small set of meaningless elements combining into a large set of meaningful ones — realised in space rather than in time, and its presence in sign was the first decisive proof that signed languages have a phonology at all.

Above this sub-lexical level, signed languages show the full architecture of grammar. Edward Klima and Ursula Bellugi's landmark study demonstrated morphology, in which a single sign is systematically modulated for grammatical aspect and agreement, and a syntax organised in the signing space in front of the signer, where locations stand in for referents and verbs move between them to mark who did what to whom (Klima & Bellugi, 1979). Wendy Sandler and Diane Lillo-Martin's comparative synthesis places these findings in the frame of linguistic theory, showing that signed languages obey the same abstract universals — constraints on syllable structure, on phrase structure, on the mapping between meaning and form — that constrain spoken languages, while exploiting the visual modality in ways speech cannot (Sandler & Lillo-Martin, 2006). Signers also draw on fingerspelling, a manual alphabet used to borrow words from the surrounding written language, and on classifier constructions, in which a handshape represents a class of object and its motion depicts an event in space. Figure 1 shows how the three core parameters combine to specify a single sign, and Table 1 sets out each parameter with an illustrative minimal-pair contrast. The first demonstration lets a reader vary each parameter and see when the result is the same sign, a minimal pair, or an impossible form.

Acquiring a Signed Language

If signed and spoken languages share a design, children should acquire them in the same way, and they do. Deaf infants exposed to sign from birth pass through the same milestones on the same timetable as hearing infants learning speech: first signs at around the same age as first words, a vocabulary spurt, and the same kinds of grammatical over-regularisation. The most striking parallel is in babbling. Laura Ann Petitto and Paula Marentette found that deaf infants acquiring sign produce manual babbling — rhythmic, reduced hand activity built from the phonetic units of sign, distinct from ordinary gesture — on the same developmental schedule that hearing infants babble vocally (Petitto & Marentette, 1991). Babbling is therefore not tied to the vocal apparatus at all; it is an amodal expression of an emerging linguistic capacity, and its manual form is powerful evidence that the language faculty is not specifically a speech faculty.

The parallel extends to the critical period. Elissa Newport's studies of ASL exploited a natural experiment: because most deaf children are born to hearing parents, they encounter their first accessible language at widely varying ages, from birth to late childhood. Comparing native signers (exposed from birth), early learners, and late learners who were matched on decades of subsequent daily use, Newport found that ultimate grammatical attainment declined steadily with the age of acquisition: native signers outperformed early learners, who outperformed late learners, on exactly the morphological structures that must be inferred rather than memorised (Newport, 1990). Rachel Mayberry extended the result to show that this is specifically a first-language effect: a late first exposure to any language, signed or spoken, leaves a lasting deficit that a late second language does not, because the earlier language provides scaffolding the first learner never had (Mayberry & Eichen, 1991). Sign language thus provides the cleanest available test of the sensitive-period hypothesis, disentangling age of first exposure from the confounds that make it hard to study in the hearing population. The second demonstration traces the resulting attainment curve against the age of first exposure.

The Emergence of New Sign Languages

The deepest evidence that the capacity for language is innate comes from cases where a signed language has come into being from almost nothing. A single deaf child born to hearing parents and never exposed to a signed language will spontaneously invent a homesign system — structured gestures with consistent, language-like properties — to communicate with the family. When a community of such children is brought together, homesign can bloom into a full language within a generation. The clearest documented case is Nicaraguan Sign Language, which arose after 1977 when deaf children were first gathered into schools in Managua. Ann Senghas and Marie Coppola showed that the language the children created grew in grammatical complexity across successive cohorts: the youngest children, learning from the first signers, systematically restructured what they received, breaking holistic gestures into discrete, recombinable elements and inventing a spatial grammar to mark who did what to whom — a grammar no adult had taught them (Senghas & Coppola, 2001). Language emergence of this kind demonstrates that children do not merely absorb a language; when the input is impoverished, they add the structure themselves.

These findings connect signed language to the wider question of how gesture and language relate. Susan Goldin-Meadow and Diane Brentari's synthesis distinguishes the categorical, combinatorial structure of a signed language from the gradient, holistic gesture that accompanies both speech and sign, arguing that the two coexist in signing just as speech and co-speech gesture coexist in hearing communication (Goldin-Meadow & Brentari, 2017). The birth of a signed language is, on this view, the imposition of categorical linguistic structure onto a gestural substrate that every human already possesses.

Iconicity in Sign

Because the hands can depict what they mean, signed languages are far more iconic than spoken ones: the ASL sign for TREE resembles a trunk and branches, and DRINK mimes raising a cup. This was long treated as a reason to doubt that sign was a real language, on the assumption that genuine language must be arbitrary. That assumption was mistaken. Pamela Perniss, Robin Thompson, and Gabriella Vigliocco reframed iconicity as a general property of all language, present in spoken vocabulary too (in onomatopoeia and sound symbolism) but far more available in the visual modality, where it aids learning and grounds abstract meaning in bodily experience rather than undermining grammatical structure (Perniss et al., 2010). Iconicity and arbitrariness are not opposites but ends of a continuum, and a signed language uses both: highly iconic signs sit alongside wholly arbitrary ones, and even transparent signs are subject to the same phonological constraints as opaque ones, so that a sign's iconic origin does not exempt it from the grammar. The third demonstration lets a reader place a range of signs along the iconic-to-arbitrary continuum and see how iconicity varies within a single vocabulary.

The Signing Brain

If signed languages are true languages, they should recruit the brain's language network rather than a general visuospatial one, and the neuroimaging evidence confirms that they do. Ruth Campbell, Mairéad MacSweeney, and Dafydd Waters' review established that signing engages the classic left-hemisphere perisylvian regions — the frontal and temporal areas long associated with spoken language — despite sign being a visual, spatial, and motoric activity that might have been expected to lateralise to the right (Campbell et al., 2008). MacSweeney and colleagues drew the theoretical moral: the neural organisation of language is driven by linguistic structure, not by the sensory and motor channel through which language happens to be transmitted, so the signing brain and the speaking brain overlap where it matters most (MacSweeney et al., 2008). Karen Emmorey's programme has mapped both the deep overlap and the genuine, modality-specific differences — the greater involvement of parietal cortex in the spatial grammar of sign, for instance — that a complete account must also explain (Emmorey, 2021).

The quantitative case has since been consolidated. Patrick Trettenbrein and colleagues' meta-analysis pooled the imaging literature and confirmed that sign language comprehension reliably activates a left-lateralised fronto-temporal network largely shared with spoken language, while also recruiting bilateral regions tied to the visual-spatial demands of the modality (Trettenbrein et al., 2021). The picture that results is the neural counterpart of the structural one: language is language in the brain, whatever the modality through which it enters.

Worked Example

Consider how the age of first exposure to a signed language bounds ultimate grammatical attainment, following the logic of Newport's ASL studies. Represent attainment as a score U on a 0-100 scale, where 100 is native-signer performance, and model it as a schematic function of the age of first exposure a in years. Native signers, exposed from birth, are at ceiling; attainment then declines at a roughly constant rate as first exposure is delayed, until it reaches a floor. A simple piecewise model captures the shape: U(a) = 100 for a ≤ 4, and U(a) = 100 − 2.5 × (a − 4) thereafter, not falling below a floor of 40.

Take a native signer first exposed in infancy (a = 0). Since 0 ≤ 4, U = 100: full native attainment is expected. Take an early learner first exposed at age 6: U = 100 − 2.5 × (6 − 4) = 100 − 2.5 × 2 = 100 − 5 = 95. Take a late learner first exposed at age 12: U = 100 − 2.5 × (12 − 4) = 100 − 2.5 × 8 = 100 − 20 = 80. Now apply a native-like threshold of 90: a learner counts as native-like only if U ≥ 90. The infant learner (100) and the age-6 learner (95) clear it; the age-12 learner (80) does not. Solving 100 − 2.5 × (a − 4) = 90 gives 2.5 × (a − 4) = 10, so a − 4 = 4 and a = 8 — the model's crossover age, beyond which native-like attainment is no longer predicted. The numbers are illustrative rather than measured effect sizes, but they reproduce the central regularity Newport documented: attainment is flat and high for exposure in early childhood and then falls steadily with the age of first exposure. The second demonstration lets a reader move the age of exposure and watch U cross the threshold at exactly this point.

Discussion

The study of signed languages settled a question that had stood since antiquity: whether language is inextricably bound to speech, or whether the vocal channel is one contingent vehicle for a deeper, amodal capacity. The answer is now clear. Signed languages have phonology, morphology, and syntax (Stokoe, 2005; Klima & Bellugi, 1979; Sandler & Lillo-Martin, 2006); they are babbled (Petitto & Marentette, 1991) and acquired (Newport, 1990; Mayberry & Eichen, 1991) on the same schedule as spoken languages; they can arise spontaneously in a community of children (Senghas & Coppola, 2001); and they are processed by the same left-lateralised brain network (Campbell et al., 2008; MacSweeney et al., 2008). Language is therefore an abstract computational system that can be realised in sound or in sight with equal completeness.

This conclusion reshaped cognitive science well beyond the study of deafness. It removed speech-specific mechanisms from the core of the language faculty, forcing accounts of universal grammar and of critical periods to be stated in modality-neutral terms. It showed that iconicity, once thought to disqualify sign from languagehood, is a general and even useful property of language (Perniss et al., 2010; Goldin-Meadow & Brentari, 2017). And it turned signed languages into an indispensable comparative tool: because they vary the transmission channel while holding the linguistic computation constant, they let researchers ask which properties of language follow from its abstract structure and which from the accident of how it is sensed and produced.

Current Directions

Contemporary work is quantitative and comparative. Large lexical databases now put sign research on the same empirical footing as psycholinguistics of speech: Naomi Caselli, Karen Emmorey, and colleagues built ASL-LEX, a database of the frequency, phonological structure, and iconicity of thousands of ASL signs, enabling the well-controlled experiments on lexical access and recognition that spoken-language research has long relied on (Caselli et al., 2017). On the neural side, the pooling of two decades of imaging into meta-analysis (Trettenbrein et al., 2021) has moved the field from demonstrating that sign engages the language network to specifying precisely where sign and speech converge and where the visual-spatial modality leaves its own signature. Emmorey's recent synthesis frames the open questions that follow: how the brain integrates the linguistic and the depictive within a single signed utterance, and how bimodal bilinguals — people fluent in both a signed and a spoken language — manage two languages in two modalities at once (Emmorey, 2021). Across these fronts the agenda has shifted from proving that sign is language to using sign to learn what language is.

Common Misconceptions

Sign language is universal.

There is no single sign language. Deaf communities have developed many distinct, mutually unintelligible signed languages, and ASL and BSL are unrelated despite a shared spoken tongue (Sandler & Lillo-Martin, 2006).

Sign language is a manual code for a spoken language.

A natural signed language has its own vocabulary and grammar, unrelated to the surrounding spoken language; it is not spoken words rendered on the hands (Klima & Bellugi, 1979).

Signs are pantomime, not real words.

Signs are built from a small set of contrasting, meaningless parameters that form minimal pairs, exactly the duality of patterning that characterises spoken words (Stokoe, 2005).

Iconicity means sign cannot be a true language.

Iconicity is a general property of language, present in speech too, and iconic signs obey the same grammar as arbitrary ones; visual resemblance does not exempt a sign from structure (Perniss et al., 2010).

Glossary

Age of acquisition.

The age at which a person is first exposed to a language; in signed languages it varies widely and strongly predicts ultimate grammatical attainment.

Arbitrariness.

The absence of resemblance between a linguistic signal and its meaning; one end of the continuum whose other end is iconicity.

Classifier.

A handshape that represents a class of object, whose movement in the signing space depicts the motion or arrangement of that object in an event.

Critical period.

An early developmental window within which a first language is acquired most fully; often reframed as a gradually closing sensitive period.

Deaf community.

A cultural and linguistic community whose members use a signed language as a primary means of communication and share a common identity.

Duality of patterning.

The organisation of language on two levels, meaningless units combining into meaningful ones; in sign, contrasting parameters combine into signs.

Fingerspelling.

A manual alphabet in which each letter has a distinct handshape, used within a signed language to borrow words from the surrounding written language.

Handshape.

The configuration of the fingers and hand in forming a sign; one of the core sub-lexical parameters that distinguish one sign from another.

Homesign.

A structured gesture system invented by a deaf child with no access to a conventional signed language, showing language-like regularities of its own.

Iconicity.

A resemblance between a linguistic form and its meaning; more available in the visual modality of sign than in speech, but present in both.

Language emergence.

The spontaneous arising of a new, fully grammatical language, documented most clearly in the birth of Nicaraguan Sign Language among deaf children.

Location.

The place on or near the body at which a sign is made; one of the core sub-lexical parameters of a sign.

Manual babbling.

The rhythmic, reduced hand activity produced by deaf infants acquiring sign, built from the phonetic units of sign and paralleling vocal babbling.

Minimal pair.

Two signs (or words) that differ in exactly one contrasting element, demonstrating that the element carries a distinction of meaning.

Modality.

The sensory-motor channel through which a language is transmitted, visual-gestural for sign and auditory-vocal for speech.

Movement.

The path or internal motion of the hand in forming a sign; one of the core sub-lexical parameters of a sign.

Nicaraguan Sign Language.

A signed language that emerged after 1977 among deaf children in Managua, growing in grammatical complexity across successive cohorts of learners.

Phonology.

The level of language governing the inventory of contrasting sub-lexical units and how they pattern; in sign, the parameters of handshape, location, and movement.

Key Researchers

Ursula Bellugi (1931-2022). Director of the Laboratory for Cognitive Neuroscience at the Salk Institute; her work with Edward Klima demonstrated the grammatical structure of ASL and pioneered the study of its neural basis. Google Scholar - Wikipedia - Wikidata

Karen Emmorey (contemporary). Distinguished professor at San Diego State University and director of the Laboratory for Language and Cognitive Neuroscience; a leading authority on the psycholinguistics and neurobiology of sign language. ORCID - Google Scholar - Wikipedia

Rachel I. Mayberry (contemporary). Professor of linguistics at the University of California, San Diego, and director of the MultiModal Language Development Laboratory; her work established the lasting effect of age of first-language acquisition. Faculty page - Wikipedia

Elissa L. Newport (contemporary). Director of the Center for Brain Plasticity and Recovery at Georgetown University; her ASL studies of native, early, and late learners provided decisive evidence for a sensitive period in language acquisition. ORCID - Google Scholar - Wikipedia

Ann Senghas (contemporary). Professor of psychology at Barnard College, Columbia University; her longitudinal documentation of Nicaraguan Sign Language showed a new language acquiring grammar across cohorts of children. ORCID - Wikipedia - Wikidata

William C. Stokoe (1919-2000). Linguist at Gallaudet University whose 1960 analysis of American Sign Language into contrasting parameters founded the linguistic study of signed languages and established that signs have a phonology. Wikipedia - Wikidata

Frequently Asked Questions

Is sign language a real language?
Yes. Natural signed languages have phonology, morphology, and syntax, are acquired on the same schedule as spoken languages, and are processed by the same brain network, so they are full languages in every structural sense (Stokoe, 2005; Sandler & Lillo-Martin, 2006).

Is there one universal sign language?
No. Deaf communities have developed many distinct signed languages that are mutually unintelligible; American Sign Language and British Sign Language, for example, are unrelated despite the shared spoken language (Sandler & Lillo-Martin, 2006).

Is sign language just gestures or pantomime?
No. Signs are built from a small set of contrasting, meaningless parameters, handshape, location, and movement, that combine to form minimal pairs, which is the same duality of patterning found in spoken words (Stokoe, 2005).

Do deaf babies babble?
Deaf infants exposed to sign produce manual babbling, rhythmic hand activity built from the units of sign, on the same developmental timetable that hearing infants babble vocally, showing that babbling is not specific to speech (Petitto & Marentette, 1991).

Does the age of learning sign language matter?
Yes. Ultimate grammatical attainment declines steadily with the age of first exposure, so native signers outperform early learners, who outperform late learners, evidence for a sensitive period in first-language acquisition (Newport, 1990; Mayberry & Eichen, 1991).

Can a new sign language be created?
Yes. Nicaraguan Sign Language emerged spontaneously after 1977 when deaf children were gathered together, and it grew in grammatical complexity across successive cohorts of child learners (Senghas & Coppola, 2001).

Does iconicity make sign less of a language?
No. Iconicity is a general property of language, present in speech as well, and iconic signs obey the same grammar as arbitrary ones, so visual resemblance does not disqualify a sign from linguistic structure (Perniss et al., 2010).

Does the brain process sign like speech?
Largely yes. Sign language recruits the same left-lateralised fronto-temporal network as spoken language, with additional recruitment of regions serving the visual-spatial demands of the modality (MacSweeney et al., 2008; Trettenbrein et al., 2021).

References