Lexical Segmentation PSY/LING 34
Popular in Language Development
Popular in Linguistics and Speech Pathology
This 27 page Class Notes was uploaded by Texana Sonnefeld on Monday February 23, 2015. The Class Notes belongs to PSY/LING 34 at University of Arizona taught by Louann Gerken in Spring2015. Since its upload, it has received 53 views. For similar materials see Language Development in Linguistics and Speech Pathology at University of Arizona.
Reviews for Lexical Segmentation
Report this Material
What is Karma?
Karma is the currency of StudySoup.
You can buy or earn more Karma at anytime and redeem it for class notes, study guides, flashcards, and more!
Date Created: 02/23/15
DISCLAIMER This note outline is made by LouAnn Gerken All text added in blue is mine Gerken L 2015 Feb 17 Lexical Segmentation University of Arizona Tucson AZ What is the Mental Lexicon A stored representation of the words we know pmiowordmadeup Word an arbitrary sound pattern that is consistently associated with a particular meaning Words in the mental lexicon have at least three types oF information stored grammatical information 89 syntactic category gender num er 9 c The mental lexicon within the larger picture Sentence interpretation Semantics Sentence formation Syntax Word creation Morphology Grammatical Information syntactic category gender number etc Word meaning Lemma level Word form Lexeme level Abstract Word Form Phonology Mental Lexicon Some basic Facts about lexical development how we learn a bunch of words children are starting to 6montholds associate mommy and daddy with build a lexicon very ear39yon videos of the appropriate parent and things around them even though they do not yet pronounce these words This represents one of the earliest Form of referential knowledge expressed by infants Babies at 12 months of age respond appropriately to many words Parents estimate that their 12month olds know between 50100 words slowly 12montholds may only produce 510 words if any big gap between your lexicon and what you can produce in the first year of life recep ve Some basic Facls abou r lexical developmenl From 1218 monfhs fhere is an increase From 10 or so words lo around 40 The word producfion explosion lakes place af abouf 1824 monfhs oF age during which lime children go From being able lo produce 40 words lo quot300 words By fhe lime children are 6 years old fhey have quot14000 words This works ouf lo roughly 9 new words a day From 18 monlhs on How children do fhis is sfill a mysfery 3 problems For the child acguiring a lexicon 1 Finding where words begin and end People of second language seem to talk so fast 2 Learning the meanings of these words Seems simple but its harder 3 Learning the syntactic category oF these words to be discussed in section on syntax finding the category is difficult is it a noun or a verb The segmen ralion problem What a nice doggy spoken to a Child Whmamnt a n i woe dun0 uglgl y words blend From lhe signal how can one fell where lhe words begin or end An example From anolher language Farsi P Some solutions to the segmentation problem NO LEARNING LEARNING REQUIRED more ep U UI reqUIre earning IHI e anguage e ore Adj acency to Singleword grammatical utterances morphemes doggy the you know a word is following Monosyllobic words Languagespeci c at the ends of stress patterns every language differs utterances lengthen words at end of sentence C Statistical information phonotactics which syllables occur with other syllables Which can occur consonant and vowel placement No Learning Required Segmentation Cues Singleword ullerances Paren r migh r say Dog or Run as single word u39l39fer ances their name no dog name A recenl sludy has shown lhal aboul 9 of parenlal ullerances conlain a single word Furlhermore lhose words rha r paren rs produce as single word ul rerances are among lhe earlies r words produced by lheir children However we slill need a solulion lo lhe segmenlalion problem For lhe words lhaf are nol produced lhis way Words at the ends of utterances Note that in an utterance like Look at the dogquot the end of the word dog is easy to nd Middle class Western parents do put words that they want to highlight in utterance nal position And children are better able to identify words in this position However this may not be a universal Feature oF childdirected Speech Can identify where is the doggy at 18 months in a picture Less better at finding the dog when it is phrased as where is the doggy in the picture in the middle S ra risfical informa rion 0 Listen fo the Following sfimulus and fry 0 nd fh fh 3 b d e ree sylla le war 5 Which of fhese words were in fhe recording pado do go kupado bubida bupado ku bu labupa dakugo golabu bipago do bi f da bidaku kugola dadola dakupa labubi gokl Experimenf chu heard fhree 3syllable words sfuck fogefher bidakL190ab padotgolab bidaku cda Follows bi 100 of fhe time whereas go Follows ku 33 of the fime 0 8monfhold infanfs can use fhis informafion and prefer new words creafed From word parfs such as kugola over old words such as bidaku Learning Required Segmentation Cues Adjacency to grammatical morphemes Grammatical morphemes in English are items like fhe a was can e39l39C the isthe mostfrequentword in English They occur before many words of English And can help us identify known as well as new39 words can guess word beginning and end and what category the word belongs in good for segmentation and category as briialig and the sithy taovems dict gire anci gimEble in the waabe Adjacency lo grammalical morphemes Compu rer simula rions of word nding rha r use rhese morphemes do very well segmentation Bu r These usually use wri r ren sen rences wi rhou r spaces and we don39f know as much abou r acous ric sen rences eliminates sloppiness Children oF ren Fail ro produce grammatical morphemes so we need ro ask if rhey can use rhem Adjacency to grammatical morphemes 24 monthsold Studies of children39s ability to associate a word with an appropriate picture depending on the preceding grammatical morpheme Find the dog For me Best performance using the the to help them find what it is they are looking for Find dog For me Find was dog For me Find gub dog For me Is the eFFect due to segmentation or syntactic processing Adjacency lo grammalical morphemes Using rhe Headlurn Preference Procedure 8mon rhold French learners were Familiarized wilh passages wi rh rwo inFrequenl nouns The nouns were preceded by eilher a real French de rerminer des or a nonsense de rerminer kes some doesn t mean anything Al res r infan rs heard rhe 2 nouns in isola rion They lis rened longer ro rhe nouns rha r had occurred aF rer rhe real grammalical morpheme desquot sugges ring rha r rhey perceived rhe o rher noun as par r of a longer word kesnoun using grammatical morphemes are helpful in finding words Languagespeci c slress pallerns In mosl languages words exhibil a rypical slress pallern In English lhe vasl majorily of words begin wilh a s rressed syllable In Hungarian mos r words end wilh a s rressed syllable Children mighl use slress as an indicalor of where words begin and end in lheir parlicular language bu r rhey mus r already know some words Languagespeci c s rress pa r rerns 9mon rh olds learning English prefer rhe Sw s rress pa r rerns over wS s rress pa r rerns 6month olds DO NOT do that Bu r do rhey use s rress in segmen ra rion 75mon rholds Familiarized using HPP wi rh passages con raining w kingdom hamle r vs w words gui rar deLce could pick out Sw words couldn t tell wS words apart from each other A r res r inFan rs heard lis rs oF words rha r did or did no r occur in Familiariza rion InFan rs lis rened longer ro Familiarized Sw words By 10 mon rhs inFan rs lis rened longer ro Familiarized words regardless oF s rress pa r rern Phonotactic information Some sequences of phones occur within words while others occur between words Computer simulations of word segmentation with IPA transcribed text without word boundaries show that phonotactic information is a good cue What about For infants Phono raclic informalion Infrequent Frequent 9mon rholds familiarized using pr with passages withinword withinword con raining gaffe and love fanggaffe Con rex rs crea red differen r beangaffe fine amoun rs of evidence for hOld l segmenfing fhe words infrequent found the words 99 8 infrequent ntog and etoh were not able to find the words listened for 2 minutes A r res r infan rs heard isola red words rha r were ei rher from rhe familiariza rion passages or no r Only infan rs who had good evidence for segmen ra rion lis rened longer ro familiarized words Pitting the was against QOCh ther Johnson amp Jusczyk Sntontlt olds 9 9 Stt39t391 l Stt39t39nl 8 I D Ptlitli ldrld 8 l D Paitlivlotrld 7 739 St39 t39 t39 l j Stat39st39nl E 21 It j E World1L original 5 33 I 33 Ealj E4 5 j 5 I 53 E3 3 5 against 4 2 2 O 0 8montholds bidakugolabupadotigolabubidaku bidaKUgolaBUpadoTIgolaBUbidaKU listened longer to part words when stressed information was put in they think that the words start with the stressed syllable Conclusions From con icfing cues S raiis rical cues may play a role in segmenafion very early before infants know much abou r the word s rrucfure of rheir language Once infanfs know a Few words fhey begin ro depend on languaqespeci c sfress phonofacfic information Summary of lhe segmenlalion problem Words are no r separaled From each olher in lhe speech s rream Infan rs and adul rs have available a varie ry of cues For ndlng words languagespeCIfIC cues bables use cues to help Identify words some words reqUIre learning In the language Some cues are languageneulral eg singleword ullerances O rhers require experience wilh lhe parlicular language eg slress Learners change rheir weigh ring of cues over development as rhey discover which ones are mos r reliable
Are you sure you want to buy this material for
You're already Subscribed!
Looks like you've already subscribed to StudySoup, you won't need to purchase another subscription to get this material. To access this material simply click 'View Full Document'