Origins, structure and evolution of langauge
What is language?
• Communication
– Signals emitted by an organism whose function is to influence other
organisms (of the same species)
• Symbolic
– There is little or no relationship between nature and meaning of the
signal
• Generative
– Different combinations of the same signals
Do other primates have language?
• Vervet monkeys
– Warning vocalisations specific to eagle, snake, leopard
• Rhesus monkeys
– Can distinguish vocalisations such as [p] from [b] (like infants)
• Chimpanzees
– Wild animals highly vocal
– Nim Chimpsky signing patterns (food item first, name at end)
– Kanzi (bonobo) word order preferences using pictograms
Human-specific language innovations
• Physical changes
– Lower larynx gives greater range of vocalisations
– Nasal cavity can be sealed off giving greater range of vowel sounds
– Muscular tongue
• Elaboration of brain areas (Left Cortex)
– Broca’s area (production)
– Wernicke’s area (reception)
Language fossils
• Artefacts
• Genes
– Language must be at least as old as the ancestor of modern humans
– Genetic similarities between populations reflect linguistic similarities
• Possible universal cognates
– AQ’WA = water
– TIK = finger, one
– MAMA, PAPA
Linguistic diversity
• 5,000+ extant languages; many dialects
• 100s of language groups (Indo-European is just one!)
• Huge grammatical and phonetic diversity
–
–
–
–
Immutable/inflective/agglutinating
Fixed/free word order
Subject/topic prominent
SVO/SOV/VSO
He is eating for her
English
Näïkìmlyìïà
Kivunjo (Bantu)
Language groups in Europe
• Sir William Jones, 1786
– Recognised relationships between different languages pointed to
common origin
Semitic
Arabic
Sanskrit
Avestan
Clas. Greek
Latin
Gothic
Turkish
Turkic
Old Irish
Indo-European
Hebrew
How do languages evolve?
• Changes in word use
– Nouns become verbs (mail), adjectives change meaning (cool),
• Origin/loss of cognates
– Hound original term, ‘dog’ appears later
• Changes in pronunciation
– The great vowel shift (15th C.). E.g. mood previously pronounced
‘mode’, house pronounced with vowel sound of ‘loose’
• Changes in grammar
– Loss of gender in English, disuse of ‘thou’, ‘whom’
Why do languages evolve?
• Cultural exchange
– Denim, Tennis, Dandelion, Assassin
• Novel situations
– Surfing, cookie, blog
• Phonetic laziness!
– Water pronounced as ‘wadr’
• Identity and slang
– Coded meaning (e.g. rhyming slang)
Universal Grammars
• All individuals have an inbuilt potential to generate grammatical
structures
– These consist of nouns, verbs, subject-object-indirect object relations,
clause structures, etc.
• To a large extent this universal grammar (UG) is hard-wired
– Wild-children (e.g. Genie) develop protolanguage in absence of stimulus
– Genetic basis of specific language impairment
– Children’s generalising ability (e.g. wugged)
• Differences between languages in phonemes, words and
grammatical parameters
Language trees
S
VP
NP
det
NP
N
The student
V
failed
det
N
the
test
Stochastic grammars
• We can describe sentences by sets of generative rules
Rule
Meaning
S → NP VP
Sentences consist of a noun phrases and a verb phrase
NP → N [PP]
A noun phrase consists of a noun and possible a
preposition phrase
PP → P NP
A Preposition phrase consists of a preposition and a
noun phrase
VP → V NP
A verb phrase consists of a verb and a noun phrase
S → if S then S
A sentence can consist of two sentences joined by an
if .. then .. construction
Language parameters
• Major differences in grammatical structure can be due to small
differences in grammatical parameters
Baker (2003)
• Major difference is presence/absence of polysynthesis
– If present word order flexible but word structure complex and rigid
– E.g. in Mohawk participant must be named in the verb that names the
event
Rukwe’
wa-sh-ako-hsir-u
ne owira’a
Man
past-he-her-blanket-gave
the baby
Baker (2003)
Why do grammars evolve?
• Drift in small populations
– Individuals make errors
– In small populations errors may be passed between generations
• Encryption (Baker, 2003)
– Encoding messages so that only a select few can interpret the answer
may be important
– Letter substitution = Phoneme differences, Transposition = parameter
differences
– Navajo code talkers of WWII
• Grammatical structures are an example of stochastic grammars
– Generate algorithms for constructing strings
– E.g. stochastic context-free grammars (SCFGs) consist of terminal
states generated by internal states independent of context
E.g
Rule
Meaning
S → aWb
Start produces terminals (a,b) and internal state (W)
W → aYb
Internal states generate terminals (a,b) and convert to
other states (Y)
W→n
STOP
Note that there might be many different types of internal state, each of which
generates certain sets of terminals and other states. E.g. NP generates
terminals (nouns, determiners, adjectives) and other states (PP, NP,…)
Stochastic Grammars in Bioinformatics
S → LS | L
F → dFd | LS
L → s | dFd
s = unpaired bases
d = paired bases
S = non-terminal producing loops
F = non-terminal producing stems
L = non-terminal producing unpaired
bases or new stems
Knudsen and Hein (1999). Bioinformatics 15:446
How do we understand language?
• For stochastic grammars there is a dynamic programming solution
that allows us to consider all possible parsings of a given string
S
VP
VP
NP
V*
PP
NP
NP
N Adv
V det
N
P det
N
I once shot an elephant in my pyjamas
Groucho Marx
Is this how humans learn and process?
• Th hmn cpcty fr dcdng mprfct sgnls s rmrkbl
• We unexpected grammar can even sense make of
• Humans … predictive … to … in … gaps
• iasudiahsuandcanquicklyspotpatternssdfskkdflgdfkgmfk
References
•
•
•
•
•
•
•
Pinker,S. (1994) The Language Instinct. Penguin
Burchfield, R. (1985) The English Language. OUP
Aitchison, J. (1996) The Seeds of Speech. CUP
Ruhlen, M. (1994) The Origin of Language. Wiley
Crystal, D. (1987). The Cambridge Encyclopedia of Language. CUP
Purves, D. et al (2000). Neuroscience. Sinauer
Baker, M. (2003). Linguistic differences and language design. Trends
Cog. Sci. 7:349
• Nowak et al. (2001). Evolution of universal grammar. Science 291:114
• Durbin et al. (1999). Biological Sequence Analysis. CUP
Descargar

Slide 1