Language is characterized by: Language is characterized by: Language. Why is language a major area of research?

Similar documents
Evolution of Symbolisation in Chimpanzees and Neural Nets

NAME: East Carolina University PSYC Developmental Psychology Dr. Eppler & Dr. Ironsmith

While you are waiting... socrative.com, room number SIMLANG2016

Language Development: The Components of Language. How Children Develop. Chapter 6

The dilemma of Saussurean communication

Age Effects on Syntactic Control in. Second Language Learning

English Language and Applied Linguistics. Module Descriptions 2017/18

Language Acquisition Fall 2010/Winter Lexical Categories. Afra Alishahi, Heiner Drenhaus

Seminar - Organic Computing

Module 12. Machine Learning. Version 2 CSE IIT, Kharagpur

CS 598 Natural Language Processing

A Neural Network GUI Tested on Text-To-Phoneme Mapping

Unit 14 Dangerous animals

Food Chain Cut And Paste Activities

On Human Computer Interaction, HCI. Dr. Saif al Zahir Electrical and Computer Engineering Department UBC

Correspondence between the DRDP (2015) and the California Preschool Learning Foundations. Foundations (PLF) in Language and Literacy

Psychology and Language

Ling/Span/Fren/Ger/Educ 466: SECOND LANGUAGE ACQUISITION. Spring 2011 (Tuesdays 4-6:30; Psychology 251)

1 Signed languages and linguistics

Speech Recognition at ICSI: Broadcast News and beyond

Listening and Speaking Skills of English Language of Adolescents of Government and Private Schools

Abstractions and the Brain

INTRODUCTION TO PSYCHOLOGY

Dyslexia/dyslexic, 3, 9, 24, 97, 187, 189, 206, 217, , , 367, , , 397,

Knowledge-Based - Systems

L1 and L2 acquisition. Holger Diessel

Python Machine Learning

An Empirical and Computational Test of Linguistic Relativity

Artificial Neural Networks

Linguistics. Undergraduate. Departmental Honors. Graduate. Faculty. Linguistics 1

Proposal of Pattern Recognition as a necessary and sufficient principle to Cognitive Science

A Case-Based Approach To Imitation Learning in Robotic Agents

Degeneracy results in canalisation of language structure: A computational model of word learning

MADERA SCIENCE FAIR 2013 Grades 4 th 6 th Project due date: Tuesday, April 9, 8:15 am Parent Night: Tuesday, April 16, 6:00 8:00 pm

VB-MAPP Guided Notes

Evolutive Neural Net Fuzzy Filtering: Basic Description

Designing a Rubric to Assess the Modelling Phase of Student Design Projects in Upper Year Engineering Courses

Language Acquisition by Identical vs. Fraternal SLI Twins * Karin Stromswold & Jay I. Rifkin

Planting Seeds, Part 1: Can You Design a Fair Test?

Tracy Dudek & Jenifer Russell Trinity Services, Inc. *Copyright 2008, Mark L. Sundberg

Artificial Neural Networks written examination

How the Guppy Got its Spots:

Learning Structural Correspondences Across Different Linguistic Domains with Synchronous Neural Language Models

Language Acquisition Chart

DEVELOPING AN INTERACTIVE METHOD TO MAP THE STUDENT PERSPECTIVES ON EVOLUTION

Knowledge Transfer in Deep Convolutional Neural Nets

babysign 7 Answers to 7 frequently asked questions about how babysign can help you.

Objectives. Chapter 2: The Representation of Knowledge. Expert Systems: Principles and Programming, Fourth Edition

SARDNET: A Self-Organizing Feature Map for Sequences

Learning Methods in Multilingual Speech Recognition


AP PSYCHOLOGY VACATION WORK PACKET UNIT 7A: MEMORY

The Complete Brain Exercise Book: Train Your Brain - Improve Memory, Language, Motor Skills And More By Fraser Smith

Cooperative evolutive concept learning: an empirical study

A Minimalist Approach to Code-Switching. In the field of linguistics, the topic of bilingualism is a broad one. There are many

Going back to our roots: disciplinary approaches to pedagogy and pedagogic research

An Embodied Model for Sensorimotor Grounding and Grounding Transfer: Experiments With Epigenetic Robots

Neuro-Symbolic Approaches for Knowledge Representation in Expert Systems

Reading Horizons. A Look At Linguistic Readers. Nicholas P. Criscuolo APRIL Volume 10, Issue Article 5

OUTLINE OF ACTIVITIES

Calibration of Confidence Measures in Speech Recognition

Recommended Guidelines for the Diagnosis of Children with Learning Disabilities

The Strong Minimalist Thesis and Bounded Optimality

SNAP, CRACKLE AND POP! INFUSING MULTI-SENSORY ACTIVITIES INTO THE EARLY CHILDHOOD CLASSROOM SUE SCHNARS, M.ED. AND ELISHA GROSSENBACHER JUNE 27,2014

Holy Family Catholic Primary School SPELLING POLICY

Parsing of part-of-speech tagged Assamese Texts

Accelerated Learning Course Outline

Chapter 9 Banked gap-filling

Rajesh P. N. Rao, Aaron P. Shon and Andrew N. Meltzoff

What is Thinking (Cognition)?

BIOS 104 Biology for Non-Science Majors Spring 2016 CRN Course Syllabus

Let's Learn English Lesson Plan

A Pipelined Approach for Iterative Software Process Model

Linking Task: Identifying authors and book titles in verbose queries

A Bootstrapping Model of Frequency and Context Effects in Word Learning

Understanding the Relationship between Comprehension and Production

Learning Methods for Fuzzy Systems

AGENDA LEARNING THEORIES LEARNING THEORIES. Advanced Learning Theories 2/22/2016

Accelerated Learning Online. Course Outline

STUDIES WITH FABRICATED SWITCHBOARD DATA: EXPLORING SOURCES OF MODEL-DATA MISMATCH

Spring 2015 IET4451 Systems Simulation Course Syllabus for Traditional, Hybrid, and Online Classes

PHYSICS 40S - COURSE OUTLINE AND REQUIREMENTS Welcome to Physics 40S for !! Mr. Bryan Doiron

Introduction to Simulation

GOLD Objectives for Development & Learning: Birth Through Third Grade

OPTIMIZATINON OF TRAINING SETS FOR HEBBIAN-LEARNING- BASED CLASSIFIERS

CSC200: Lecture 4. Allan Borodin

Ch VI- SENTENCE PATTERNS.

Examinee Information. Assessment Information

The Acquisition of English Grammatical Morphemes: A Case of Iranian EFL Learners

A Reinforcement Learning Variant for Control Scheduling

Semi-Supervised GMM and DNN Acoustic Model Training with Multi-system Combination and Confidence Re-calibration

KLI: Infer KCs from repeated assessment events. Do you know what you know? Ken Koedinger HCI & Psychology CMU Director of LearnLab

Rendezvous with Comet Halley Next Generation of Science Standards

have to be modeled) or isolated words. Output of the system is a grapheme-tophoneme conversion system which takes as its input the spelling of words,

The Acquisition of Person and Number Morphology Within the Verbal Domain in Early Greek

CS Machine Learning

ARCHITECTURES FOR COSY ROBOTS

2.B.4 Balancing Crane. The Engineering Design Process in the classroom. Summary

9.85 Cognition in Infancy and Early Childhood. Lecture 7: Number

Grade 8: Module 4: Unit 1: Lesson 11 Evaluating an Argument: The Joy of Hunting

Transcription:

Language Language is: a a rule based system of symbolic codes used for communication. Language is characterized by: Semantics Rules used to communicate meaning. Grammar (syntax) A limited set of rules describing how we can combine the symbols in certain orders. Arbitrariness No inherent relationship between symbol & referent. 1 2 Language is characterized by: Generative A limited number of symbols can be combined in (infinitely) novel ways. Applies to syntax and semantics Dynamic (Changes occur over time) New words are added to the lexicon (but not new sounds; Xhosa) Rules of grammar change Displacement Ability to refer to objects not physically present. Vervet monkey (Cheney( & Seyfarth, 1990). Why is language a major area of research? Language requires many other capacities Perception, categorization, memory, motor skill It can be explored from MANY different perspectives. It may be the one ability unique to humans. It may be the heart of thought. 3 4 1

The Relationship of Language & Thought Sapir-Whorf (1956) Hypothesis Linguistic relativity (strong version) Distinctions encoded in one language are not found in any other language. Translation programs--sometimes fine distinctions are missed The Relationship of Language & Thought Sapir-Whorf (1956) Hypothesis Linguistic determinism (weak version) The structure and complexity of a language determines how we perceive and think about the world. Language may influence, but not determine, our perception of the world. ambiguous figures 5 6 Now that we have an idea of what language is and isn t, where does it come Did language evolve? (How do children learn language?) Did language evolve?. The emergence of a "language" in an evolving population of neural networks. Connection Science, 10, 83-97. [Section on Evolution from textbook] 7 8 2

Given limited evidence (no fossils) it is difficult to prove language evolved, so simulations may provide useful evidence. What was done in the present simulation? 100 simple organisms (ANN) try to eat mushrooms Feedforward net w/14 input, 5 outputs, 5 hidden units Edible & poisonous mushrooms look similar. 2 one-word utterances are communicated among NN Location of mushroom can be determined before edibility, so the words avoid & approach would be useful 9 10 At the end of life (a certain number of time steps), the 20 individuals with the most energy (from eating mushrooms) produce 5 offspring each. Offspring have the same connection weights as parents (with 10% genetic mutations of weights). This is repeated for 1000 generations. 11 12 3

3 populations were examined: 1) No language language inputs always =.5, everyone must learn about mushrooms on their own. 2) Language is externally provided 3) Language evolves A creature sees a mushroom and labels it for another randomly selected creature nearby. 13 14 Results After 1000 generations the creatures were very good at discriminating mushrooms. 28 mushrooms and 1 toadstool on average No-language group had 150 energy units whereas those with language had more than 250 units. No difference between outside- and evolvedlanguage, but outside-language group reached plateau faster. 15 Ability to categorize objects in the environment based on perceptual properties, repeated social interactions, and language may have co-evolved. A consistent/distinct signal for mushroom/toadstool increases reproductive chances. Analysis of the output units in the no-language group shows some support for this idea; increases in the quality of the signal coincide with an increase in fitness. The signal can act as a substitute for perceptual information when the object cannot be perceived This is the power of language an abstract symbol. 16 4

Production and perception must have evolved in parallel. One w/o the other is useless All languages have both abilities How can language evolve if its informative function may be advantageous to the receiver but not the producer? Why not lie to get all the food for yourself? Vervet monkeys don t lie, they signal lion when they see it, snake when they see it, etc. 17 The results are consistent with Burling (1993) Human language evolved from the cognitive (sensory-motor) capacities of pre-lingual ancestors rather than primate level communication. 18 Where did language come Ulbaek (1998). The origin of language and cognition. In Hurford, Studdert- Kennedy & Knight (eds). Approaches to the evolution of language. Where did language come Language evolved from animal cognition NOT animal communication. 19 20 5

Where did language come Language grew out of cognitive systems already in existence and working. It formed a communicative bridge between animals that were already cognitive. Where did language come Consider the cognitive processes used for: Tool-using and making Cognitive maps Learning through imitation Social knowledge Deception Theory of mind Similar processes are used with language. 21 22 Where did language come Language was used by early species to communicate thoughts. Sharing thoughts can be disadvantageous. So why bother? Where did language come Greater cognitive intelligence outweighed that disadvantage. Social conditions such as reciprocal altruism (increased fitness by sharing and helping) lead to greater cognitive intelligence. Was the first language even spoken? 23 24 6

Is sign language a real language? YES! (LSA in 1970 s) Same developmental trajectory as spoken language Babble (9 m.o.) 50 word stage ( ( 2 y.o.) It has syntax, semantics, morphology, phonology. Uses same parts of the brain as spoken language Signed languages emerge spontaneously How do we learn/acquire language? Language acquisition involves several stages: Babbling 4-6 m. <all sounds> 9 m <in/out & less/more common> First words: 1 year old Overextension (milk, juice, glass, cup = bati ) Underextension (Only Spot is doggie ) 25 26 How do we learn/acquire language? Displacement & Over-regularization are also exhibited about this time (18 m.) A rule is applied to a word that is an exception to the rule. In applying the rule X + s = plural a child might say "mouses" instead of "mice." How do we learn/acquire language? Holophrastic Speech One word means many things Telegraphic Speech: 18-24 months old Children have a vocabulary explosion after about 50 words (expressive); 18 months. 27 28 7

What can Physics tell us about language? Watts (2004). The new science of networks. Annual Review of Sociology, 30, 243-270. Important ideas to understand from Watts (2004) This article highlights the importance of interdisciplinary cross- Each field reinvented the wheel A network is a useful tool to represent various systems Biological, Social, Technological, etc. systems 29 30 Important ideas to understand from Watts (2004) Note that this type of network is more general or abstract than the artificial neural networks we discussed earlier. These networks do NOT learn, do NOT have activation levels, etc. Nodes = an entity Links = a relationship between entities In general these networks describe structures, not processes. However, structure does have implications for processing 31 Important ideas to understand from Watts (2004) Many real-world networks are interesting mixtures of ordered and random networks. Two types of networks that have received much recent attention are small-world and scale-free networks. Small-world Although the network is very large, there are random shortcuts that allow one to traverse the network very quickly. Scale-free There are a few highly-interconnected nodes that contribute to the robustness (to damage) of the network. Growth and preferential attachment are two mechanisms that lead to this type of structure. 32 8