BYU Studes Quarterly Volume 5 Issue Artcle 7 --975 A Computer Analyss of the Isaah Authorshp Problem L. Lamar Adams Alvn C. Rencher Follow ths and addtonal works at: https://scholarsarchve.byu.edu/byusq Recommended Ctaton Adams, L. Lamar and Rencher, Alvn C. (975) "A Computer Analyss of the Isaah Authorshp Problem," BYU Studes Quarterly: Vol. 5 : Iss., Artcle 7. Avalable at: https://scholarsarchve.byu.edu/byusq/vol5/ss/7 Ths Artcle s brought to you for free and open access by the All Journals at BYU ScholarsArchve. It has been accepted for ncluson n BYU Studes Quarterly by an authorzed edtor of BYU ScholarsArchve. For more nformaton, please contact scholarsarchve@byu.edu, ellen_amatangelo@byu.edu.
Adams and Rencher: A Computer Analyss of the Isaah Authorshp Problem A computer analyss of the saah authorshp problem L lamar adams and alvn C rencher bblcal scholars have been argung the saah problem whether or not the same man wrote all 66 chapters of the book of saah as t stands n the old testament at least snce the twelfth century the dvsonsts separate the authorshp nto st saah chapters 39 and 2nd and or deotero deutero saah chapters 466 some crtcs have proposed that all evdence ponts to more than two dvsons wrtten by dfferent authors the greater part of the crtcal attack on the unty of saah has concerned tself wth the lterary unty of the book wth the multple author theorsts clamng that varables such as vocabulary syntax textual forms contents and poetc style dffer from secton to secton they generally attrbute smlartes n the varous sectons cted n defense of the sngle author clam as the work of saah s dscples who patterned ther work after hs or even coped the prophet the result s that n many cases both sdes cte the same sources to support opposng conclusons 2 snce the stylstc elements frequently cted by crtcs are amenable to statstcal analyss many lngusts have suggested that statstcal research methodology be appled to problems of authorshp usng subject matter and contextually orented word famles as a base two such studes have been done 3 dr adams s a research analyst n the offce of nsttutonal research at brgham young unversty dr rencher s assocate professor of statstcs at brgham young unversty R K harrson ntroducton to the old testament grand rapds wllam B eerdmans 969 PP 77576 76 2L lamar adams A statstcal analyss of the book of saah n relaton to the saah problem phd dss brgham young unversty 972 pp ap 2829 29 37 see asa kasher the book of saah characterzaton of authors by morphologcal data processng essay department of mathematcs barlan lan unversty 97 pp ap 54 yehuda radday saah and the computer A prelmnary report computers and the humantes 5 november 976572 95 976572 Publshed by BYU ScholarsArchve, 975
BYU Studes Quarterly, Vol. 5, Iss. [975], Art. 7 96 but they faled to solve the problem because a dfferent researcher applyng the same assumptons varables and statstcal methods to dfferent texts by known authors wll get wdely dvergent fndngs 4 As a matter of fact the studes noted were nter and even ntra contradctory procedures USED IN THE PRESENT STUDY n the nvestgaton of authorshp style of lterary works varables should be sought that are consstent from work to work for a gven author and at the same tme vary from author to author ths type of varable s referred to as a marker varable and may be used to dentfy the lterary style that s vatz varl unque to a gven author compared to other authors many characterstcs of the hebrew language provde excellent sources of pertnent stylstc elements n authorshp dentfcaton 5 these nclude functon prefxes certan parts of speech that reman constant from text to text for a gven author and specal vocabulary even word famles provdng there s consstency and relablty n usage rates for the gven author functon prefxes nclude all prefxes except those whch are pronomnal verbal and partcpal the latter three types were consdered to be too contextual to serve as relable lterary elements for authorshp determnaton vowels n the hebrew language are referred to as ponts ancent hebrew scrptures were wrtten wthout these ponts whch were later added by scrbes and other transcrbers ths study s based on the unponted hebrew texts morphology of the hebrew word ncludes the basc word root plus suffxes and prefxes one phase of ths study s based on the hebrew root to avod effects of later changes by scrbes and transcrbers functon prefxes are perhaps the most useful stylstc element n determnng authorshp from a hebrew text n hebrew these prefxes consttute a major stylstc element correspondng to the habt prone parts of speech n englsh the major dvson n ths study for analyzng style n the adams statstcal analyss pp ap 86 n any language some stylstc elements are less relable for determnng authorshp than others some of the less relable elements n hebrew nclude word famles whch are subject to contextual nfluences parts of speech that are subject to contextual nfluences and certan prefxes aes and suffxes that are n- fluenced by contextual orentaton and by the audence to whch the text s addressed https://scholarsarchve.byu.edu/byusq/vol5/ss/7 2
Adams and Rencher: A Computer Analyss of the Isaah Authorshp Problem COMPUTER ANALYSIS OF ISAIAH 97 book of saah conssts of the commonly used twofold dvson chapters 39 and 466 n addton stylstc elements are examned usng several dfferent types of subdvsons n the book of saah for ntratext analyses some of these smaller dvsons are a subdvson of the second half of the book chapters 455 and chapters 5666 2 dvsons used by radday plus the secton he omtted chapters 2 2 323 2435 36-39 the omtted secton 448 48 4957 and 5866 and 3 a combnaton of the dvsons postulated by S R Dr drver6 drverf drvere verg and alos barta 7 2 2 323 2427 27 2833 3435 35 3639 39 448 444 48 49-55 5662 62 and 6366 66 perhaps one of the most determnng factors n authorshp dentfcaton s the comparson of ntertext text wth ntratext text varaton changes between the two saah dvsons take on more meanng when compared to varatons between control texts and wthn control texts t must be determned whether or not ntraauthor author varaton s smaller than nterauthor varaton for a gven element f that element s to be used as a stylstc marker varable n determnng authorshp the lterary style n the complete book of saah s compared wth the style n random samples from the followng old testament books amos jeremah ezekel hosea mcah habakkuk zecharah danel ezra malach and nehemah these eleven samplngs serve as hebrew control texts for comparsons wth the book of saah ths study s the most extensve to date specalsts n the areas of semtc languages statstcs and computer scence were nvolved over seventy dfferent types of stylstc elements were examned and several hundred lngustc varables were analyzed statstcal RESULTS A number of stylstc elements from the unponted hebrew text of saah ndcate a lterary style n both parts of the book that s characterstc of that book n contrast to the other books of the old testament examned stylstc elements that support sngle author unty nclude functon prefxes hebrew marker roots specal vocabulary certan parts of speech rep- GS s R drver an ntroducton nto the lterature of the old testament new york charles scrbner s sons 92 p 26 alos barta syntax of the sentence n saah 466 chcago the unversty of chcago press 9 pp 3 4 ap 34 Publshed by BYU ScholarsArchve, 975 3
BYU Studes Quarterly, Vol. 5, Iss. [975], Art. 7 98 etton edton of phrases frst letter and last consonantal letter of the hebrew word these letters of the hebrew word show ndvdual habts of speech because of morphologcal characterstcs of the hebrew language among all stylstc elements examned the functon prefx appears to be the most salent the book of saah has a surprsngly large number of functon prefxes whch ndcate sngle authorshp out of 36 dfferent prefxes and types of prefx combnatons examned n the hebrew texts 24 occurred n the book of saah table contans frequences for IS 8 of these 24 although each of the 8 prefxes s used n both sectons chapters 39 and 466 n the book of saah some have zero frequences n a number of the control texts t s evdent from the last two columns saa A and sab B n table I that for a number of prefxes the rates of usage ex hbt a smlarty between the two saah texts at a rate pecular to the book of saah contrasted to the control texts for example both saah sectons have a rate of two equvalent to.2 per 5 prefxes for the prefx nn D n compared to oo. for each of the control texts A correlaton measure was devsed to compare rates of usage of these s 8 prefxes between any two texts t can vary from zero to one wth zero ndcatng no relatonshp and one showng perfect relatonshp all gradatons between these two extremes are possble among the eleven control books and two sectons of saah there are 78 possble parngs of texts the correlaton between the two saah sectons s.98 ndcatng a very hgh degree of overall smlarty n rates of prefx usage only three of the other 77 comparsons are as hgh a lower number than would be expected from the laws of chance when ths many comparsons are made these results show a closer unty between texts n the book of saah than there s n accepted sngle author texts the analyss was repeated wth prefxes elmnated whch were not characterstc of the style n saah correlatons were obtaned based upon the remanng sx prefxes D an n a 7OHM no M 7 3 n I and the I average correlaton for the I control texts was.46 wth a range from to sg.89 the correlaton between the two sectons of saah s 997.997 whch can these results are ndeed strong ev- easly be rounded to. loo dence for authorshp unty n the book of saah an nvestgaton of marker roots dscovered approxmately https://scholarsarchve.byu.edu/byusq/vol5/ss/7 4
Adams and Rencher: A Computer Analyss of the Isaah Authorshp Problem ose oae 4 COMPUTER M ANALYSIS OF ISAIAH 7 CQB vo G r 2 sab me 676 ee r- weg 467 heg o 29 mem 863 meh O hbl 52 M 884 oo rr amm sa aga o y M oo A oo 2 s o6 6 o rm 2 8 O 56 2 n oo 47y 67 73 A saa 688 lomm zoma sa o r 2 4 2 oo M TIT I oo 652 752 658 T 23 co 7 446 r 834 rn 5 68 76 5 68 56 zoa o r M y oo ZK 85 5 75 5 95 8 f 375 mgo 375 65 r 5 5 C ac rc rt FIF I r r o 66 o J r u ac zc 25 u N 625N 475 r 75 r 4 5 u 95 N 25 N 5 4 o r r r G sample I block wable testament saah of ceh old qed boom book moom glom from gloa lao the kao of prefxes functon rt saah and 39 eee tee NH 2 2 225 25 5 25 5 2 5 475 65 75 2 2 C n rt SO O O S 6 8 33 o 7 aq rq jem 5 gem 533 r O t 2266 N lee 66 O lee 66 O 33 cd I ML saah 3 u texts e E e L pam U VO O O VS cr ro rf cr O mc t O pee 566 YO 666 33 S 966 O 3 m 666 O 633 75 u r 33 cr 66 O 33 rt Q sectons ohep samp ohea e JR 65 75 5 25 n 85 3 65 r 5 ons lons canons catons prefx r oo r r o oat cat oal mog per nol hwo two emo the zao uzl for certan lmo the f m 66 O 66 vo 33 9 N mam 233 bam r emm 633 r 9 333 m 833 m a mol for mog r HS C m CO appl ca frequences mcd and mcm texts on 3 fal ffl abbrevate m B S H 733 cl l 6 7 gee 833 233 c r r t r r 833 oo jee 5 r 8 co 33 m lee r gog for log r statst average EZ 4 825 925 5 825 325 425 25 25 25 25 75 5 25 by ed alph DN u r 25 5 5 r 5 75 5 325 fsr 425 r M 625 7 25 m o r N v r mu rogo were 2 8 frequences 98 92 N 2 C 2 M CT T I 5 36 OS OO AM 4 4 8 TI m 7 x n r r j nl mad r n Q n 73 rr fa f3 n - n j C7 pref 7 73 mal average Publshed by BYU ScholarsArchve, 975 5
BYU Studes Quarterly, Vol. 5, Iss. [975], Art. 7 35 whch have rates of usage characterstc of the two saah texts n contrast to the control texts correlatons for varous groups from the 35 marker roots s.95 95 ths s an extremely hgh ndex of style smlarty compared to the control text ndces whch range from to 74.74 for the same varables ntertext Inter text varaton was compared wth ntratext varaton for prefxes marker roots and other stylstc elements to determne the degree of smlarty between the varous texts statstcal comparsons of ntratext text wth ntertext text varatons ndcate that a hgh degree of smlarty exsts throughout the book of saah contrasted wth the control texts the correlatons between pars of sectons n saah range from 97.97 to. loo for the followng sectons chapters 2 2 323 2435 3639 39 448 48 4957 and 5866 ntratext Intra text correlatons for the other old testament books sampled range from 85.83 83 to 95.93 thus saah shows greater nternal consstency than any of the other books examned another example of unqueness n the book of saah for the usage rate of the prefx n table may be observed n the englsh text saah 924 concludes wth a descrpton of a condton that wll exst n the mllennal era n that day shall srael be the thrd wth egypt and wth assyra n ths phrase the hebrew prefx s rendered and wth ths prefx combnaton may also be translated as follows I but for then oretc oreta and to unto at nto jor for etc thus the same prefx s and to kng james n saah 69 from the phrase to brng thy sons unto the name of the lord thy god and to the holy one of srael ths has reference to the gatherng of exled srael n the last days the rate usage for I l s unque to the book of saah occurrng approxmately 68 tmes for every functon prefxes n saah 39 and 67 n saah 466 compared to a zero rate e almost never for such books as amos mcah and ezekel usage occurs at a hgh rate of 2 n the book of nehemah as exemplfed n nehemah 26 nether had I as yet told t to the jews nor to the prests nor to the nobles nor to the rulers nor to the rest that dd the work A number of other stylstc elements also support authorshp unty table 2 contans a representatve sample of correlatons for a number of lterary elements that are character https://scholarsarchve.byu.edu/byusq/vol5/ss/7 6
Adams and Rencher: A Computer Analyss of the Isaah Authorshp Problem COMPUTER ANALYSIS OF ISAIAH table 2 correlatons for old testament control texts and the two sectons of the book of saah for three stylstc elements marker parts of repetton conjunctons the body of phrases old testament control texts.65 65.8 8.42 42 sectons of the book of saah.99 99 ogg.99 99 ogg.7 99 99 ndces for control texts are means for the ndces of several texts cstc of the style found throughout the book of saah wllam harper lsts hebrew roots for a number of parts of speech and word famles 8 frequences from the harper prepostons and conjunctons were analyzed usng the correlaton as well as other statstcal measures the two sectons of saah are more closely assocated to each other n stylstc usage of prepostons and conjunctons than are the control texts for example the correlaton for the saah texts s.99 99 ogg whle the average ndex for the control texts s.65 65 7 65 see tabe table 2 harper lsts roots for a number of word famles found n the old testament one word famly found to be an excep- tonally strong marker famly conssts of words for dfferent parts of the body approxmately one dozen words from ths word famly have a more consstent rate of usage n the two snce some conservatve scholars argue that n comparson to the rest of the old testament wrters saah had a greater tendency to repeat or quote hmself repetton of phrases was examned to test the valdty of ths clam roots whch have rep- hgh frequences n the old testament texts were submtted to the computer program to obtan phrase frequences ettons were counted for nne word phrases that had four or more dentcal roots or roots and prefxes correlatons were obtaned for comparsons between the two saah texts and the saah texts than n any of the other old testament texts examned correlatons for the rates of usage of the roots for these words are also lsted n table 2 the ndex for the two saah texts s agan hgher than any other set of comparsons these results also tend to support unty of the book of saah wth.99 ogg 99 contrasted to a mean of.8 8 for the control texts wllam R harper hebrew vocabulares 8th ath ed new york charles scrbner s sons 92 pp ap 32 bd Publshed by BYU ScholarsArchve, 975 7
BYU Studes Quarterly, Vol. 5, Iss. [975], Art. 7 2 combned sample texts n repetton rate the two saah texts have a hgher degree of smlarty to each other correlaton of.7 than to the combned sample texts 42.42 addtonal stylstc elements were examned and found to support the unty of the book of saah these nclude the frst and last consonantal letter of the hebrew word both of whch contrbute sgnfcantly to morphologcal characterstcs of the hebrew language the majorty of stylstc elements examned n ths study were also analyzed usng other types of statstcal procedures the results confrm the nferences drawn from the correlaton measure and support authorshp unty of the book of saah SUMMARY AND conclusons the statstcal results n ths study do not support the dvsonst clam that lttle or no evdence exsts for unty of the book of saah to the contrary results from the statstcal analyses over a wde range of types and numbers of stylstc varables strongly support the authorshp unty of the book several dfferent types of stylstc elements were found to have marker varables unque to saah chapters 59 39 and chapters 466 these elements nclude functon prefxes marker words prepostons and conjunctons certan word famles frst letter and last consonantal letter of the hebrew words and repetton rates of certan types of phrases the two parts of saah most often clamed to have been wrtten by dfferent authors chapters 39 and 466 were found to be more smlar to each other n style than to any of the control group of II other old testament books the book of saah also exhbts greater nternal consstency of the other II books than any these computerzed results do not exclude the possblty that mnor changes n the text have been made by scrbes and edtors snce the tme of ts orgn however the evdence ndcates that n spte of such possble changes deletons or ad- dtons an overall style has been retaned as measured by the lterary varables examned the results of our research bear wtness that the book of saah has a lterary unty characterstc of a sngle author https://scholarsarchve.byu.edu/byusq/vol5/ss/7 8