You are on page 1of 7
vain sgnaing pening clision Neurophse ", 86-295 60 ito, (891 computstonanevoetholgy: 4 provsanal ante, In From Animals to Animats Meyer, HA. and Wien, 5M, ed 161 hie #1. and See, RD. (1997) The bran Mas a boa: adaptve oir emerges fom Ineacons of nervous stem, Body ad evieonment ends Neuos 20, 53-557 2 dean, 1. 1990) Cocing. sropioceptie Information te contol Gyoum 62, 115-190 63 urn DE and Dean. (1988 Iteseprental an oanternearons 8 tan, WB, Lekey SR and Lens, 16 (1988 Using rallave tir of the redial ech tty information proening (6 Harvey Husbands Pan Clit, 0 (193) tue in eoiaionsy ‘ote From Arima to Animate 2 (Mayr, A, Rota, Lane risor, Sk oe 364373, ne Pes (67 Zora, W. Gill. and Brter, J. (886) (et) evohng colecve ‘sharon yh an from Ania to anit 4 oe Pt 2 es) p35 IT Pee (68 Hetgenbers (991 Neural Metin let Fah MT Pes {69 Gry, RO. (987 Fath and foraging: = eique ofthe ‘raion argument ram des n Foraging Behavior Kal AC, Krebs JR iam, MR, ee 69-120, Plenum ree, 10 von Eckard 8 (1290 What Cognive sence? MI Press 171 Mas, (981) & bottom-up mechan for bahar lection in 30 ri ceture,ntrom Anal Animas eyes, 2nd Won ‘Swe op 238-286 MT Pree Capacities underlying word learning Paul Bloom and Lori Markson Children are strikingly good at learning the meanings of words. Current controversy focuses on the relative importance of different capacities in this learning process including principles of association, low-level attentional mechanisms, special word fearing constraints, syntactic cues and theory of mind. We argue that children succeed ‘at word learning because they possess certain conceptual biases about the external ‘world, the ability to infer the referential intentions of others and an appreciation of syntactic cues to word meaning. Support for this view comes from studies exploring the phenomena of fast mapping, the whole object bias, the acquisition of names for entities belonging to different ontological kinds and the effect of lexical contrast. Word learning is not the result of a general associative learning process, nor does it involve ‘specialized constraints. The ability to learn the meanings of words depends on a number of capacities, some of which are specific to language and unique to humans, ‘others of which are potentially shared with other species. "Tio and terres poor noo cn olan nd manus yee aa and nee dine paces Howe hy ae sting ed ea inthe mann of worl Chien at fo wore Dy 12 months of age tele rfene a werd ere bg by Te 18 ment and ental come to an a ye Boe) Thi ely vcr nde pera pronoun fe you proper mans Fd, Mom: preposto Co) Tien good bp wes ean ad many css touminduting se ering solos ng) words at a rate of over ten new words per substances (milk, wate), parts (eye, finger) habitual ativi- ties (bath, nap, periods of time (minute, day) and abstract notions (story, game)'*, While children sometimes get che precise meaning ofa word wrong — for instarce, sometimes calling a cat 'a dog’ - serious mistakes are eate: children log’ of congise proper names with never call a ch ‘common nouns, object names with substance names or adjectives with verbs (One perspective on word learning is that parents do much of the work, carefully tailoring cher speech eo make the connection between words and what they describe pat ticularly clear to her children, Such tutelage does occa sionally occur in many cultures, including the middle-class Western cule chat is the facus of most language acquis tion research, But itis not universal: there ate societies in which patents make 90 effort co teach words to children, leaving them to learn words on the basis of overheard speech’, Nevertheless, such children have no preblem in de- veloping a ich vocabulary. Furthermore children raise in Western cultures learn a least some words, such 28 the pet- tonal pronouns, by overhearing them in che conversations of others, and even the most pampered child will leatn many words that are sed when the relevanc objector event opyaht © 1866 Ehaer Sconce All righ reseved.THSE139BS1800 PH: SII6A-TASEIOTTZT 8 PR and 1, Marko ae at the Dpprenent afr. Unie of Arians, Tae, ‘AZ 85721, USA, teh 91 20.28 7467 Foe 115206219506 mui team ‘itionate Bloom ang Markson ~ Word learning Box 1. The rate of word learning lets fen sid ha children ar of ening words very omy, about one or rwo 2 week and then erlang vorewhete be- tween 20 and 100 words, there i den acceleration in the ate of wod leaning ~ sometimes cle & wacabulisy spurt, naming aplovon, or word burst ta peopel cident earn word at ‘the ate of five, tenor even twenty new wordes day, But vocab lary got is acaly les cramatic than this, While some hi icn might show a sudden increase in che tte at which they lean ‘word, many do noc. exhibiting instead a series of all ues 2 slo monotonic rie ors smowthly ncrewingesponesial anc 2, Moreoer a vacabulary spurt wl at mest ring child to the pine where se ars five co en words wee, not a dy To see how the rat of word leening changes, consider the estimates (ce Table) from Fenton al, bated on parental ports f the yocabulaies of -hilden from 124030 months, and fiom Angle, based on comprehension sis wit sic, eight and tenyearold children, (Anglin included only chose words ‘whose meanings could not be worked out using "problem sl: ing’ sues and ence mast be leaned) Estimating vocab lary sizes ricky for sever asons so these numbers should be taken oly a8 very rough estimates Age ‘Average number of words per day || 1216 months 03 16-23 months os 23-30 months 18. 30 months-6 years 36 6-8 years 66 B10 years 1 is nor being attended 10. ‘These considerations suggest that word leaning isa robust process, requiring ech infer- ental capacities on che pat of the child. But which capacities? The proposal that we wil explore hove is thae children succeed at word learning because they possess rich understanding ofthe external world, the able ity eo infer the referential “ntentions of athers (theory of mind’) and, by the time the child i abou ewo yeas of age, an appreciation of syntactic cues ro word meaning, ‘This Position contrasts with the theory that word learning accomplished through an associationist learning, mecha- nism that i sensitive to statistical properties ofthe input", pethaps asisted by ‘dumb acentional mechanisms”. I is also diferent from the proposal that children possess special constraints dedicated to the process of word learning’ ‘We suggest that children’s conceprual biases, intentional understanding and syntactic knowledge ate nor only neces- ary for word learning (nothing simpler would suffice) but cha chy ae aso sufficient (nothing ese i needed). Fast mapping Given the high numberof words thc children come to un- derstand (about 10000 bythe age of sx abous 60000 for the average high school graduato”™ one would expect them to be proficient at luring and storing word mean- ings on che Basis of minima experience, They ae. Ina clas- Trends in Cognitive Scianees ~ Vol. 2, No. 2, Why docile gaally improve at word ering? There are several explanations, including the maturation of memory and attentional capacities, a growing sensitivity to dient cues (0 4 word’ meaning (ch ais sya category and f the discourse conten in which ici wed) and an inctessing narenes of differnt enties hat can be named, Another factor is aces 1 new words in particular, tracy exposes children 1 many mote words than they would encounter ‘dough social interaction or watching television, and cis ikely thatthe gargantuan vocabularies of some English speakers (wel over 100000 word’ could not cccur without the abily 0 read ‘Mot aduts do not learn several new words day, This mighe bebecause we are nota good a woed leaning as We once wes, bu simpler possibiny i thar we already know mau of the won hat ou immediate environment has oof. Unless we lear anew language ou aly oppor for word lang ae oper names archaic or techni erms, ev words hat ‘emer the lngusge sacha ‘inteane'dsig’ and “karaoke 1 rom (887) Ear ese Deepen, Carbide Urverity 1 eon, K- (97H) Stature and sratpy in leaning to tlk Monogr So A. hil ev 3, era No.8 Golde 8A. and Resnick, 15. (1990) Fal el aciton: ae, content. and the cable spurt J Chil tang. 7, 4 Fersen, . et ak (988) Varabiy In eary communicative eveloperant Menage Res Cid Oe 59, Sere No. 22 Baglin. 1 (953) VorsbuBry deelopmant: morphological sna Monogr. Sc Res Cid De 58, Seal No. 238 aie. (1998 The science oF Wor (ned WH. Freeman siestudy, young children who were involved in an unrelated activity wer asked 10 walk over to wo trays blue one and an olive one, and to: "Bring me the chromium tray, not the ue one, the chromium one". All ofthe children retrieved the olive eray, correctly inferting thatthe experimenter in- tended ‘chromium’ to refer to this new color. Furthermore, _most ofthe childeen still remembered some of the meaning ofthis word when tested six weoks late. ‘This finding raises the question of whether this rapid learning ~ sometimes called fast mapping ~ only happens for words. To test this, three- and four-year-old childeen were given ren objects to play with as part of & measuring ‘game and were casually introduced to a novel name for one of the objects: ‘Let's use the koba to measure which is longer. We can pur the koba away now’. Fora diferent ob- ject, the children were told: ‘We ean use the thing my uncle gave 10 me co measure which is longer. We can put the thing my uncle gave co me away now’, During the rest phase, participants were shown the same objects and asked tw: ‘Find the koba’ and ‘Find the one my uncle gave co Even after a month, the childten remembered the new ‘word, providing further evidence forthe robustness of fast mapping. A group of aduls did just as well ac the same task, which suggests that word learning difers in an inter- ‘esting way from other aspects of language leaning, such as February 1998 morphology and syntax, in which childsen are clearly s petior®. Children and adults were equally good, however, at remembering the arbiteary linguistically presented fact thar chey had learned about a novel object given to the experimenter by her uncle. A further experi ment found that fast mapping has its limits. When eaught the locaton ofa sticker: "This goes here. ‘was placed on t0 one of the novel objects, both children and adults did relatively poosly when asked a month later 0 place a small sticker on the appropriate abject (se Fig. 1. ‘The question of what can and cannot be fst mapped is as yet unresolved, but i is now clear thae fast mapping is not restricted to language learning thae ie was asasmill sticker Object names ‘When contionted with the unfamiliar word ob inthe above seudy, children immediately understood it as name for one of the objects in front of chem. This lead to rather vexing puzzle. Suppose a child heats a new word and determines thacit describes a certain object in the world, for instance, a rabbi, Most likely, children learn this through attending to the referential focus ofthe speaker as indicaced by cues sch as ditecion of gaze". But the problem now azises that there are an infinie number of logically possible meanings for the word. Ie could refer to he color of che rabbis its shape its sufice, the tail, che eats, the rabbit and the groand itisstand- ing on is motion, even ‘undetached rabbit parts™. Childeen do not entertain such possbilites, however. n his situation, they will immediatly take the word as naming the whole abit. More generally, shere isa wealth of evidence show ing chat children and adults tend to interpret now words as referring to whole objects, not o pars of objets, properties of objects or the stuff that objects are made of How is this bias best explained? One theory is that chil dren possess a special constaine that guides them co view new words as names for objects!™™, An alernatve i chat this bias towards objects is the result of a mete general fact Box 2. What is an object? Whar precisely do we mean when we say chat children have an objet bin? We have an instvesente of what we mean by ‘he erm, so that a abbit isan abject, but he rabbi’ foot not (atleast, not if iis atached to the cabbie, nor is the rabbie and the te tha cis next to, But what are precise eiteia that guide children to view some thing at objects and others as (One promising theory has boon elaborated by Elizabeth Speke on the bass of infant rexerch“. The most important teion shat objecs follow te principe of cohesion; an ob ject is connected and bounded region of mater hat mainains is connectedness and boundaries when in motion. If young infans se a portion of mater obey the principle of eobesion, hoy tea it a an abject and expect other object principles to apply. (These other principle state thar objec ae slid and do ot passthrough each othe, they Follow continuous paths through spac and thy mave if and only if they touch) OF course, ts noe necessary actually 1 observe cohesive and bounded movement before concluding hat something isan ob Proportion recalling object 00 | Koba Uncle about how people reason about the world. We ae prone to think about che word in terms of whole objecs and hence, ‘when searching for the meaning of 2 word, are driven to favor the object interpreration®™ (see Box 2) The claim here is not merely thac children parse the ‘world into objects and, therefore, tend to take new words as ‘object names. Afer all, children are also sensitive to mocion and color bur show no bias to interpret new words 8 naming motion and color. The proposal is instead that objects are highly salen, both linguistically and non-linguisticaly. In support of ths, note that the very same focus on objects shows up in domains othee than word learning, When pre- schoolers are shown an array of objects and asked to count, they show a strong tendency 10 count the objects, even ject: it is enough to infer hat there could be sick movement Hence, we can pase atonary scenes nto disint objects when the gape between enies imply that they will move indepen erly, when the seme continents that we know rom pre vious experience exist a separate objects (swith a abi char ism contac with a ees) of when Gesale cues, sich a good continuity and samenes of color and vexture, sgt that di feten entices have the potential fr independent movement (at when we sea shiny ed jot esting om a At geen suri Prelingustic infin atest to all ofthese cues, but only sometime afer they ate able xo parse objects on the basis of 1 Spit, C598 nal owed si suggestions Copnitin $0, Sorte Ee a (1982) Origins of knowdge ye Ae $8 | Needham, A and Bailargeon,R (1997 Obie segregation in ‘Sticker Fig. 1 Fast mapping i not limited to word learning. The proportion of thee and four year-old chiro (open bars and adults ied bars} whe, ater a ane month sly, ealed {he object to which the navel word referred (Kab. the obec that had the property of being given tothe expcimenter by her uncle (Uncle, and the object tat had licker sived tt (ticker. There as no ference in performance between chidren and ait In both the Kaba’ and Uncle" contions suggesting that there sno

You might also like