Download Predictability Norms: Analyzing Letter and Word Changes in Three- and Five-Letter Words and more Summaries Construction in PDF only on Docsity!
(
DOCUMENT RESUME
ED 220 813 CS 006 802
AUTHOR Blanchard, Harry E.; And Others
TITLE (^) Contextual Predictability Normt for Pairs of Words.
Differing in a Single Letter. Technical Report No.
INSTITUTION Bolt, Beranek and Newman, Inc., Cambridge, Mass.;
Illinoi Univ., Urbana. Center for the Study of
Reading.
SPONS AGENCY Nat Inst. of Education (ED), Washington, DC.;
National Inst. of Mental Health (DHHS), Rockville,
Md. (^). N\
PUB DATE Aug 82
CONTRACT 400-76-
GRANT NIMH-MH-32884; NIMH-MH-
NOTE 192p.
EDRS PRICE
DESCRIPTORS MF01/PC08 *Norms; *Perception^ Plus^ Postage. Tests; *Predictive Validity;
*Reading Research; *Research Tools; Research
Utilization; *Resource Materials
ABSTRACT
researchers^ A^ withTo texts^ conduct that^ a mayseries prove^ of usefulstudies in^ and answering^ to^ providea variety.other
ofIlquestions about perception in reading, it was.necessary for two
researchers to create pairs of texts that were different in meaning but were physically different in only one letter. These texts were
created by first identifying pairs of words thit differed, in a single
letter and then writing a context of one or more sentences in which either member of the pair would fit and make sense at the same word
position in the text, (For example, "leaks/leans" in."The shed...ge...
much it is unusable.") After the introduction, the first section of the document describes the construction.of the materials, the two. questionnaires used to determine the predictability of the critical word in the texts, the organization of the norms, and the potential uses of the norms in research. The bulk of the document then presents
the words-themselves, organized by length, from three to seven letters
and by 3,arts of speech and shape of the pair. (JL)
..
- Reproductions^ supplied^ by^ EDRS^ are^ the^ best^ that^ can^ be^ made^ * ***********************************************************************from^ the^ original^ document:^ *
CENTER FOR THE STUDY CPI' READING
Technical Report No. 260 CONTEXTUAL PREDICTABILIT (^) NORMS FOR PAIRS OF WORDS DIFFERING IN A SINGLE LE TER Harry E. Blanchard, George W. McConkie and David Zola Univer'sity of Illin is at Urbana-Champaign
University of Illinois at Urbana-Shampaign 51 Gerty Drive Champaigrt, Illinois (^61820)
August 1982
NATIONALINSTITUTEOFEDUCATION^ USDEPARTMENTOFEDUCATION EDUCATIONALRESOURCESINFORMATION CENTER(ERIC) stewed^ )(is^ documenthasbeenrproducedasfromthepersonororganization originatingitMinotshantieshavbeenmadetoimprove reproductionOwlity memdonotnekessattlyrepresent(dimwINIE^ Pointsofvieworopinionsstatedinthisdocu positionoipolicy
_Bolt Beranek and Newman Inc. 50 Moulton Street Cambridge, Massachusetts (^02238)
This research was conducted under grants MH 32884 and MH 33408 fromthe National Institute of Mental Healthto the first author, and National Institute of Education coritract 4WAG-4-400-76-0116 to the Center for the Study of Reading. (^) Copies of this report can be obtained bfwriting to George W. McConkie, Centerfor the Study of Reading, 51 Gerty Drive, Champaign, Illinois (^) 61820.
Predictability Norms.
Contextual Predictability Norms for Pairs of Words Differing in a Single Letter
In order to conduct a series of studies in,our laboratory, it was rieces- sary to create pafrs of short texts which were different in meaning but were physically different in only one letter. These texts were created by first identifying pairs of words which differed in a single letter ahd, then.writing a context of one or more sentences in which either member of the pair .would fit and make sense at the same word position in the text. For example, either member of the word pair, leaks/leans fits into the context, "The .% shed (^) so' much it is unusable." (^) As a result, there are two different versions of each text distinguished only by a single letter in one word. In addition, it was necessary to obtain predictability norms for these yoi"ds, given the text prior to that word position. (^) Having created the sentences and obtained the norming
/knformation, it appeared likely that these materials could be useful to other
researchers interested in studying perceptual or languageprocesses in. read- ing. (^) This led us to compile the information in the form presented here io it could be available to others. The following sections describe the set of text,^0 pairs, the process of norming the materials for pVedictability, the nature of the normative data presented here, and some possibleuses for these materials. In the following discussion, the letter that distinguishes the two iersions of / a tex; will be referred to as the critical letleri and the word which cont (^) ns T critical letter will be referred to as the critiDal mord.
, IA
2
Construction of the_ matgrials
The word paits used to construct the texts were classified along f6ur dimensions which pro ed 52 groups of word pairs, as summarized in Table 1. The first dimen (^) on wad word J.egth: three, five, and seven letter words were chosen for these texts. (^) Second is the letter position of the critical letter in the critical word. (^) Three and five letter words could differ in any of the letter positions of the word. (^) Seven letter words could differ in the first, second, fourth, sixth, or seventh letter positions. (^) The third dimension is
Dart 91 speech. The word paird were didhotomousiy classified. TwAategories
were used, one containing nouns and the other verbs and adjectives. (^) Finally, the 'word pairs were distinguished by tiieir contrast in word shape. (^) The members of a,word pair had either the same or different outline word shape or
envelope (cf. Haber &_Haber, 1981). For example, the words cling and _wank,
differ in shape, because one critical letter, z, is a descender and the other,
k, is an ascender, while the words srice and _prize do not differ in shape,
because both critical letter; g. and z are short letters.
TABLE 1. Classification hierarchy for word pairs. 1 Word^ length^11 3 11 5 I (^7). I ILetter position (^11) 1,2,3 I 1,2,3,4,5 (^1) 1,2,4,6,7' 1 1 Part of speech (^) II noun,verb/adj. i noun,verb/adj. I (^) noun,verb/adj. I (^1) I (^) TotalWord^ shapegroups^ 11.II same,diff.^ I^ same,diff.^ L,^ same,diff.^ I
. 12. I^20 I^20
v. Predictability^ Norms
the appropriate category. (^) Many of the other 51 categories contain more than six texts. (^) With the exception of five letter words contrasting in the first. and fourth letters, only the first six texts listed under each categoryhave been used in our studies, and these are typically the most natural-sounding members of the category. There are 456 texts overall. (^) An additional 21 texts containing five letter word pairs contrasting in thefirst or-fourth letters were written to provide extra materials for one experiment. Since complete predictability data were not.collected on these extra 21 sentences, theyhave. been labeled differently.
Data on Predictability
. (^) Two questionnaires were prepared in order to determine the predictability of the critical word in the texts. (^) eIn the firet one, the word questionnaire, subjects were given the texts up to and including theword before the critical
word, asip Fig, 1.
PASSAGE 6 On their first trip to Yellowstone, the noisy young campers were in for a nasty surprise. A marauding black bear apparently heard a Figure 1.- Example of text as given in word questionn4re.
They were instrOcted to,complete thqksentence by writing Wee words or more to finish the thought: (^) Supjects wew told to give at least three words: if they completed the sentence in less they were to begin another sentence. (^) The
5
results were taken to indicate how predictable the criticalwords were from litheir preceding cont xt. (^) The second questionnaire, the letter questionnaire, provided the texts up to and including the critical word. Tha critical letter position, however, was replaced by a blank as in Figure 2.
'*) PASSAGE 6
On their first trip to Yellowstone, the noisy young campert were in for a nasty surprise. A marauding bladk bear apparently heard a q_out gure 2. (^) Example of text as.given in letter questionnaire.
The subjects' task was to supply the missing letter. They were also told that sometimes more than one letter could fit to makea word appropriate to the context. (^) If more than one word occurred to them, they were instructed to write down only the word that occurred to them first. The results were taken
o indicate the relative probability of the two critical words, given the con- text and letters common to both.^4
Summary statistics for the word /nd letter questionnaire responses are
given in Tables 2 and 3, reqpectively. (^) Statistics on the word questionnaire refer only to the first word given to complete the sentences.
A
0
TABLE 3. (^) Summary statistics for responses on the letter questionnaire* from
30 subjects.
- Number of different letters given by subjects. Mean (^) 2.9 Median (^) 3.
- Number of subjects giving most frequent letter. Mean (^) 2W, (^) Median (^) 22.
- Number of passages in which one of the critical letters was the mol frequent letter given (^430) (94 %)
- Number of passages in which one of the critical letters was the most or second most frequent letter given (^451) (98 %)
- Number of passages in which the two critical letters were the most (and second most frequent (^347) (76 %)
- Mean percent of subjects^ A giving the most frequent critical letter as their response (^) 71.2 %
- Mean perctnt of subjects giving the second most frequent critical letter as their response (^) 18.8 %
8
Each,questionnaire was given to thirty subjects. Different subjects filled out the word and letter questionnaires. All subjects were University of Illinois students who were paid for completing the questionnaire. (^) Each questionnaire contained the main body of 456 texts. (^) For the additional 21 sentences added later, however, data were only collected from the letter ques- tionnaire. This ad4itional questionnaire was separate and also was given to thirty subjects, a few of whom were different from the subjects who completed the larger questionnaires.
prganization of the norms
In the following norms, the texts are arranged according to their category. (^) Accompanying each text are the corresponding critical words, non-
critical word pairs, and responses from the word and letter predictability questionnaires. (^) Figure 3 gives an example of the presentation format of this
information, which is explained below. c,
10
pairs, both words may make sense in the text, although onlyone actually appears in it.
The identifying number
0 The^ four^ digits^ to^ thelett^ of^ the^ decimal^ point^ in^ the^ Identifying number serves as a code for the category to which this text-belOngs. (^) The first digit identifies the word length. (^) The second identifies which letter pOsition contains the c'ritical letter. (^) The next digit codes the part % of speech. (^) If this digit^4 is a zero, the word is a noun; if it ins a.one, the word is a verb or adjective. (^) The fourth digit codes the word shape difference. (^) If this digit is a zero,-the outline shapes of the critical words are the site; fo, it it is a one., the outline shapes of the critical words are different. The two digits to the right of the decill uniquelylabel the teXt within the category to which it belongs. (^) That is, each text within a category recOves,a different set of digits to the right of the deciial. The combinalion of these two digits and the four digits identifying' the category unicwely labels the textwithin the entire set of materials.
In order to distinguish the 21 additional texs which only'have norms from the letter quesionnnaire, their identification numbers contain a 5 in the tenth's position of the decimal. (^) These sentences have been placed at the end of the lists in the categories to which they belong.
7
(
Ile text
Directly below the first line is the text. (^) The critical word appears
11
'underlined in re toXt.14th an asterisk in the position of the critical letter. (^) The tim,deltie-al letters appear in the critical words given on the first line. (^) When one of the foil words appears in the text, as is Avsually the case, that word is also underlined in the text:
Word auestionnaire responses
The responses to the word questionnaire are listed below the text itself. Only the first word rthich subjects gave in their sentence completions are con- sidered here. (^) The number in parentheses is the number of unique words given for this text. (^) Those words are then listed in order of decreasing frequency. The number to the left of each word refers to the number of subjects thatgave it as a response. (^) Cases where a subject gave no response were also counted. If two responses have the same frequency, theyappear after that frequency, separated by commas. (^) In the example, four people responded with the word a and four people with the word wanderer. (^) When the response listed is also one of the critical words, it is underlined
Letter questionnaire responses
The lettei,s given by subjects in the letter questionnaire are listed in the same way as the word questionnaire responses. (^) The number in parentheses is the number of unique responses given. (^) The letter responses are listed after their frequencies. (^) Those letters given in response which are also one,
"1r
Since word shape similarity is a.variable in these materials, theymay be useful in studying the role of word shape in word perception,especially with respect to the use of peripheral vision during reading.
Another issue for which these materials may be useful concerns the rela- tive perceptibility of letters at different positions ina word, and whether an important factor in reading. (^) The effects of lateral inhibition on letter perceptibility is well documen'ted (Bouma, 1978): (^) The question is whether, during reading, letters bounded by spaces'are perceived from a wider area than letters internal to the word.
ft^ Thus,^ these.materials^ may^ be^ useful^ to^ study^ a^ variety^ of^ questions^ about' perception in reading, dealing with effects of word shape, word length, and letter position. At the same time, the predictability norms add the-capabil- ity of taking languagetconstraint into consideration. (^) We hope that having these materials available will stimulate and facilitate research on the per- ceptual processes taking ylace as people read.
4
References 40 Houma, H. (^) Visual search and reading: Eye movements and functional visual field: A tutorial review. (^) In J. Requin (Ed.), Attention and Performance VII,. (^) New York: Erlbaum, 1978.
Haber, R. N., & (^) Haber, L. R. (^) The shape on,a word can specify its meaning Readinsr Yesearen Quarterly, 1981, _a, 334-345.
_
100
Noun: Same _Shape
THREE LETTER WORDS
First letter changes A
16
3100.01 <can/man> (^) picking/packing (rat,cat trash,,crash) I trash. was walkingSuddenly, down from in theout alleybehind yesterday the garbage when ilsin they camewere the jaaking biggest up ratthe you've ever seen. ,(not included)
3100.02 (^) <cat/rat> adopted/adapted (brain,train (^) monkey,donkey) The for neuraltheir new physiology brain monitoring lab has adoptedresearch. the rhesus monkey insteid of the *at Words(19); (^6) chimpanzee 32 rat,other, white (no response) (^1) African, guinea, huMan,ape, black, lab, marmoset,chip, common, mice, dog, rabbits, gorilla, usual Letter3(3): 24 .c 3 b,
3100.03 (^) <bat/hat> pests/posts (grandmh,grandpa quiet,quilt) Billy dumb kidwouldn't is one beof quietthe worst until Pests his grandmaI've seen. bought him a baseball *at. (^) That
Wards(13): 10 and
53 mittNIL, glove
1 a, with ball, for, from, that, to, uniform, which, Letters(2): 27
. 3h
17
3100.04 <gal/pal> (^) cube/cure (^) (arm,art slipped,shipPs8)
,standing^ Bill,^ the besidehired him.hand, put"You're^ his quitearm^ around a the^ heneck whispered,^ of^ the^ tiredand slippedmare her a cube of sugar. Words(11):g horse (^32) Agood,tfil,"1 worker (^1) brave, plucky, pony, race, remarkable, woman Letters(2): 24 g 6 ja
4
4 3100.05 death/depth (^) (left,lent bad,sad) The bad (^) financialdeath of pa's condition. blue ribbon ilow left thq family depressed as well in a Words(15):6. was
1 rwboar,eifer calf, caused, dog, h spaniel, Steer, turkey. Letters(4): 22 a 4 rno response)
holstein, horse,
3100.06 <dog/hog>^ chimp/champ^ (train,trail (^) willing,killing) It's the l'ag. common (^) They're knowledge very thatfriendly, the smartest easy to animal,train, andexcepting willing the to chimp,learn. is
Words(18): 6 dolphin 43 porpoise 2 ADScommon, great 1 Homo, orangutan, cat, duck-billed,ranchers, red, go, sea, groundhog, silver, monkey, well-known, wonderful Letters(2): 28 si 2
