Download - Lang process for human and comp
LANGUAGE PROCESSING: HUMANS and COMPUTERS
PSYCHOLINGUISTICS
linguistic performance
Syntactic processingcomprehension speech
production
when we
speak
when we
listen
The Human Mind at Work
We access the lexicon and grammar to assign a
structure and meaning to the sounds we hear
We access our lexicon to find words
We use the rules of grammar to construct
sentences & to produce the sounds that expresses
the message
“ I quite agree with you, said the Duchess; “ and the moral of that is- ‘Be what you seem to be’ – or, if you’d like it more simply – ‘Never imagine yourself not to be otherwise than what it might appear to others….to be otherwise.’”“I think I should understand that better,” Alice said very politely, “if I had written down: but I can’t quite follow it as you say it.”
* A perfectly grammatical sentence and yet it is hard to understand.
Comprehension
Speech signal
can be described in physical or acoustic term.
Speech sound
Physically, a sound is produced when there is a disturbance in the position of air molecules.
Acoustic phonetics – is concerned only with speech sounds, all of which can be heard by the normal human ear.
Fundamental frequency
Spectrograms/ voiceprints
pitch
intensity
TERMINOLOGIESHow fast the variation of the air
pressure occur
Fundamental frequency of the sound when perceived
Loudness of the sound
Patterns of sounds
- difficulty of the listener to carve up the continuous speech signal into meaningful units.
- difficulty to distinguish two physically distinct instances of a sound are the same.
Speech Perception andComprehension
A sniggle blick is procking a slar.
Segmentation problem
Lack of invariance problem
Bottom-up and Top-down Models
the process by which we obtain information about themeaning and syntactic properties of a word from our mental
lexicon
The gypsy read the young man’s palm for palm for only a dollar.
Lexical access and word recognition
Semantic priming
syntactic and semantic relations among words and phrases in a sentence
The warehouse fires….. could continue in two ways: 1….were set by an arsonists 2….employees over sixty After the child visited the doctor prescribed a course of injections.
parsing
Syntactic processing
garden path sentences
• AnotAnother example of processing difficulty is a rewording of a Mother Goose poem.
Original form:
Rephrased
This is the dog that worried the cat that killed rat ate the malt that lay in the house that Jack built.
Jack built the house that the malt that the rat that the cat that the dog worried killed ate lay in.
Shadowing task
Average shadower can do with a delay of (500 ms)
Speech sound Subjects are asked to repeat what they hear as rapidly as possible
Exceptionally good shadowers can follow what is being said only a syllable behind (300 ms)
Technique to study sentence comprehension
starts with a speaker who produces an acoustic signal that represents a thought , idea, or message to be conveyed to the listener.
The hiring of minority faculty. The firing of minority faculty.
big and fat pig and vat
Speech error
Speech production
blends• produce part of one word and
part of another; selecting two or more words instead of deciding between them
1. splinters/ blisters splisters 2. edited/annotated editated 3. a swinging/ hip chick a swip chick
Lexical selection
Spontaneous errors show that the rules of morphology and syntax may also be applied when we speak.
1. groupment instead of grouping2. We swimmed in the pool.3. An burly bird instead of an early bird
Application and Misapplication of
Rules
nonlinguistic factors can be involved insometimes interfere with – linguistic processing. 1. He made hairlines He made headlines.
2. It can deliver a large payroll. It can deliver a large payload.
Nonlinguistic Influences
Computer Processing of Human Language
COMPUTATIONAL LINGUISTICS
Computational phonetics and phonology
Computational morphology Computational syntax
Computational sign language
Computational semantics
Computational pragmatics
• 20• process of analyzing the speech signal into it’s component phones and phonemes, and
producing in effect, a phonetic transcription of speech.
process of creating electronic signals that simulate the phones and prosodic features of
speech and assemble them into words and phrases for output to an electronic speaker.
Computational phonetics and phonology
Speech recognition
Speech synthesis
• \
converts written text into basic units of the synthesizer. It translates the input text into phonetic representation.
Text-to-speech
Difficulties: 1. The problem of words that spelled alike but
pronounce differently. e.g. She has read the book. She will read the book.
2. Inconsistent spelling e.g. I take it you already know Of tough and bough and cough and dough
Computational Morphology
The processing of word structures by computers.Programmed to look for roots and affixes. All words are placed into computer’s dictionary.e.g. walking - walk+ing fondness – fond+nesse.g. fax – faxes, fax’s, faxing, faxed, refax, faxable
Computational Syntax
Parser - is a computer program that attempts to replicate a grammar to assign in a phrase structure of words
Rules: S NP VP, NP Det Ne.g. The child found the kittens. top-down parser – child found the kittens bottom-up parser- The child found,the, kittens
the representation of the meaning of words and morphemes in the computer, as well as the meanings derived from their combinations.
• The computer tries to find concepts in its semantic representation capabilities that fit the words and structures of the input.
Computational Semantics
is the interaction of the “real world” with the language system.
e.g. He saw the boy with a bicycle. He saw the boy with a telescope.
Computational Pragmatics
• Linguists are working on computer algorithms that will recognize sign language much in the same way that speech may be recognized.
Purpose:1. To produce a dictionary of signs2. To enable a computer to search through ASL videos
for a particular sign, just as search engine like Google searches for certain key words in text documents.
Computational Sign Language
• computer models of grammar• Frequency analysis, concordances, and collocations• Computational lexicography• Information retrieval and summarization• Spell checkers• Machine translation• Computational forensic linguistics: trademarks interpreting legal termsspeaker identification
Application of Computational Linguistics