P L o f fA 'LrlDWV aJlb 5lBrnWVU aJ $ DlB 13lDW 13 '1 OO13aJfil13 J' OOUJW I1 iimfil I I I o YES DON THERE IS AN ELINT oo o oo o o 186 INITIATIVES IN SlGINT REPORTING 0o ooo 4 36 f T I O $ YTic' DATA 1 NSA'S SYSTEM FOR GRADING TRANSLAT IONS o o ooooo 9 LANGUAGE IN THE 'NEWS FRENCH ooooo o r 12 INTEGRATED ANALYSTS FOR ASIA r 13 RESEARCH IN SPEECH PERCEPTION o ooooo oo Dr Ruth Day ooo o o o o 15 NSA CRYPTOLOGIC COLLECTION oo oo oo o oooooo oo ooo ooooo o oo 17 LETTERS TO THE EDITOR oooooooo o o ooooooooooooooooooooooo ooo 19 'fillS BOEURtJMEN'f EURON'fAINS EUROBEWORB MA'fERIAb -TOP-SEEREf- 81waif1ed b BIIL fSA 8ness NSAj eSSM lz t 02 Ikemp' fa 0111 eBB HaIIH tigu lJfH elil oolie la Ille Ori ler he ' Declassified and Approved for Release by NSA on '10-'1 '1- 20'1 2 pursuant to E O '135 26 vl DR Case # 54778 8 DOCID 4019632 TOP 8BCRET Published Monthly by PI Techniques and Standards for the Personnel of Operations AUGUST 1976 VOL III No 8 WILLIAM LUTWINIAK PUBLISHER BOARD OF EDITORS Editor in Chief Co11 e c t i on Cryptanalysis Language Arthur J Salemme 5642s 1'-I P L IC89SSS 80Z5i Emery W Tetrault 5236s Machine Support j 1 3 21S Ma thema tic s Reed Dawson 3957s Special Research Vera R Filby 7119s Traffic Analysis Frederic O Mason Jr 4l42s For individual subscriptions send name and organizational designator to CRYPTOLOG PI TOP 8BCRBT 86-36 DOCID 4019632 SBCRt3'f' YES DON TH IS AN ELINT L ----_I Chief VV 011 i_ce otfriT W2 4 Ol 4 c P L 86 36 Yes Don there is an ELINT Because you don't recognizeif dQE sn't mean it doesn't exist Your NSA co league ho plant its seeds who nurture i tsgrowth wllO channel its energies and who harvestit fruits resemble in many ways the more normal N$A employee You might therefore have as muchd ifficul ty recognizing an ELINTer as you do ELINT bl lt believe me they exist too are devoted primarily to ELINT NSA partitions ELINT into two classe Operational ELINT and Technical ELINT In the April issue of CRYPTOLOG you asked if the real ELINT would stand up You indicated that it appears to be a shadowy operation and that many COMINTers would like to know what it is all about Happy to oblige We ELINTers have been so busy practicing our science art that we didn't notice all of you waiting to be enlighted in the ways of ELINT -- perhaps to join the fun We're a proud bunch who know we are an NSA minority specialty group with an important job to do and we think that we're doing it rather well We can do it even better and we are working at that We will attempt to throw light into the shadows remove mystery and minimize jargon r Formal definitions of ELINT do in fact include all non-communications electronic emission intelligence except lightning and nuclear emissions As practiced at NSA our energies 1 I Will the Real ELINT Please Stand Up CRYPTOLOG April 1976 p 5 August 76 CRYPTOLOG Page I P L 86-36 SECRE'f' IlA nI3bl V1A 68 1lllT 61bUHlHS 81lhY P L 86-36 DOCID _4019632 SECRE'f EO 1 4 c P L 86-36 August 76 CRYPTOLOG Page 2 8ESRE'f Ih tJ I'lLE rnA eelolIWf elll'rliNEL5 81lbY DOCID 4019632 1 4 c SECRET 86-36 August 76 CRYPTOLOG Page 3 SEeRE'f Ih'tlll3bE YIA E811IIlT E1WINEbS allbY DOCID 401E 16- c P L -36 P L 86-36 SI3CRI3T In the last few years the U S SIGINT System has been pressured from many directions to improve the utility access of SIGINT and at the same time to respond to oft-expressed concerns that SIGINT security needs tightening Thus on one hand we must provide SIGINT quickly and directly to the levels where it will be useful in some cases to non-indoctrina ted recipients on the other we must scrupulously avoid distribution of product to places where there is no need for it and above all we must not gratuitously reveal the SIGINT secrets sources methods techniques So by varying degrees of loosening and binding the reporting and distribution procedures we have made significant modific ations which will work toward increasing both the usefulness and security of SIGINT To highlight the key changes which have occurred in the last year or so let's start with Subcategory II X COMINT reporting P S6rne86-36 time ago NSA invented the detach line format as an aid to recipients in implementing the less restrictive Subcategory II X COMINTusage provisions of the DCI's Communications Intelligence Security Regulations CISR The idea was that the producer would issue this less sensitive COMINT at a straight SECRET noncodeword non-COMINT channels classification between two detach linesl August 76 CRYPTOLOG Page 4 S CR 'f EO 1 4 c P L 86-36 1 'rliBbE VIA eemNT eltA1414f L l JNLi DOCID EO 1 4 c 4019632 P L 86-36 8HEURRI3'f EO 1 4 c EO 1 4 d P L 86-36 August 76 CRYPTOLOG Page 5 SI3CRI3'f II1d4BLE VIA e6 1IJff elI1tUNELS 6NLY DOCIO EO 1 4 c P L 86-36 4019632 SECRB'F There are many other areas in the rg o t tiJ1g field that need to be fixed up We mu 'f fuel' c at the proliferous secondary regurgitati nl 4 d read republication of SIGINT We be fleh 86-36 the current distribution lists of individual product are excessive and we want to trim them down The sanitization manual now being written for what we hope to be an annex of the CISR should be hurried along The rewrite of S-5200 17 also underway should be expedited We are moving on these things and we hope there will be something positive to report soon eSE6RE'f FPG LIVES Barbara Dudley File Executive of the fRANCOfHONEQLOS computerized dictionary of the French Language for details see CRYPTOLOG October 1975 p 10 has returned to NSA on a consultant basis She is training her successor who will also be available for researching questions and for giving instructions in on-line querying Phone 4814s or 4707s to make known your critical needs in French-language subject matter so that the most needed source material can be fed into the computer August 76 CRYPTOLOG SECRE Page 6 lI' 6E8' L DOCID 86-36 4019632 SECRET SPOKE oo ooo ooo oo ooo oo oo oo oo ooo o August 76 CRYPTOLOG Page 7 EO 1 4 c P L 86-36 SECRET SPOKE - - r ' '4'W' -- Docro 4019632 SRCRR SPOKE L c 86-3E ----- l1I' O e It I' T jIl K E-----_ OR REALLY John Kenneth Galbraith then United States ambassador to India was in Toronto to receive an honorary degree As he recalls it in Ambassador's Journal Houghton Mifflin I had asked that a certain cable from Washington to New Delhi be relayed to me through our consulate in Toronto Consular aides brought the coded message out to me at the airport -- a mass of numbers There were no facilities for decoding so I asked how they managed They said when something arrived in code they phoned Washington and had the original message read to them UNCLASSIFIED August 76 CRYPTOLOG Page 8 8 Cln T 8POK DOCID 4019632 UNCLASSIFIED NSA'S SYSTEM FOR GRADING TRANSLATIONS - t r f4J 'S ' P L 1 Introductory Note A great deal of language esting in the U S government and elsewhere involves setting a passage of some arbitrary length for translation and applying equally arbitrary and often unreliable grading procedures The wisdom of using the translation vehicle as a method for assessing foreign-language competence is probably questionable but if this procedure is adopted it should be applied as consistently and reliably as possible I describe below the criteria for evaluating translations which we have applied at NSA since early in 1972 2 Scoring System Our testing system requires that subjects translate material of nontechnical nature which will generate 500-600 English words My colleagues and I have set up a scoring system based on 100 and have set 70 as a passing grade At least two graders and most often three or four grade each paper independently assessing errors according to the system described below Each paper is studied with great care by each grader in the interest of absolute fairness Even so the tendency to read into the text what one would like to see there plus inevitable disagreement over the exact nature of translation errors require that the graders compare their results Agreement on every point is often impossible since 86-36 no two people view the texts of natural languages in exactly the same way However final pass fail judgments are usually made against a spread of no more than 5 points Thus it is possible for two graders to award a 70 and a third 65 As unanimity is not essential such a paper could pass Cases of this sort have been infrequent 2 1 Types of Errors We have made a three-way distinction among linguistic phenomena which we have found valid and reliable in assessing the errors of beginning translators this scheme is of little or no value in evaluating the work of experts The three categories we have distinguished are o foreign-language syntax o lexicon plus some grammar features of the foreign language and o English-language usage and convention The first two involve the translator's comprehension of the foreign language and his ability to show his understanding in correct if unidiomatic English rendering The last reflects his skill in not only conveying the meaning of the passage but doing so in idiomatic English Four points are taken off for each syntactic error two points for each lexical error and one point for poor English usage and violation of convention August 76 CRYPTOLOG UNCLASSIFIED Pa e 9 DOCID 4019632 UNCLASSIFIED Below I attempt fuller definitions and examples of the three possible sources of error and follow with some actual instances of poor translation from foreign languages to English I also explain the rationale for specifying the type of error in cases where analysis may be difficult other citizens as the object of risking the second is the omission of the citizens as subject of cross the bridge is of course omitted here too but our grading practice is to assess no more than 8 points per sentence This and other general grading procedures will be discussed under General Observations below 2 2 Syntactic errors 4 points each In the English sentences The dog bit the man and The man bit the dog the words are identlcai but the meanings are exactly opposi te The element which determines the dltterence is the positioning or order of the words A native speaker of another language trying to learning English would have to be aware that in this sentence as in most others word order is all-important and that failure to appreciate this fact will lead to major error In theNSA translation-grading system a test subject making an equivalent error in translating from another language into English would be docked 8 points for two 4-point errors subject and object incorrectly identified in relation to the verb A second example somewhat more difficult to judge in respect to verb-noun relationships would be Air filled the balloon vs The balloon was filled with air If it is clear from context that someone or something was involved in the act of filling the balloon a mechanic a pump etc then the two sentences are vastly different and failure to bring the fact out would cost the translator 4 points If however the phrase was filled with is roughly synonymous with was full of we would at most take off one point see below The two examples above invoive what are called case errors errors in establishing the true relationship of noun to verb Below I give some actual examples of case error in translating from Russian to English The following two examples are taken from a passage dealing with the destruction of a bridge and the need for a new one Wrong Correct translation translation In 1943 the bridge In 1943 the Germans was blown up during blew up the bridge the German retreat while retreating They the youngsters The youngsters cross cross a makeshift a makeshift suspensuspension bridge risksion bridge Other their necks and those citizens also use it risking life and limb of other citizens The two examples above are instances of mistranslation within the framework of simple nounverb relations We also assess 4 points for interclause errors which usually involve misplacement of relative i e adjectival clauses or adverbial clauses and which impair relations in meaning between clauses or sentences One example from a Russian text concerning a method of chemical analysis Wrong Correct translation translation It is true that it is an expensive but proven method However I The method is expensive but in truth well tested But It is true must govern the entire sentence especially as the sentence is to be contrasted with the following sentence beginning with However The misplacement of the initial phrase is thus a 4-point error A second example is more complex In this case also involving Russian the original text had a somewhat unusual conditional construction which covered two sentences Correct translation If the ore is good a mine can be constructed If it is worthless you can search further Wrong translation Good-you can build a mine but it isn't suitable to look for another Here we took off 4 points for the wrong relationship between good and you can resulting from the failure to translate the if clause We did not give a double penalty for the omission of the second if but we did deduct another 4 points for an incorrect collapsing of it isn't good you can into one clause A final example is a case of appositional agreement in Japanese Correct Wrong translation translation The North Korean For- The North Korean Foreign Minister met eign Minister met again with the Forei n again with Peru's ForMinister of Peru eign Minister who is In the first example there is no mention in which is the sponsaid to be close to the wrong translation of the agent the doer soring country whose South Korea's side of the act of blowing up the bridge i e the position is said in the host country 'Germans whereas in the Russian original this to be close to that was clearly specified albeit in an inverted of South Korea construction In the second example there Failure to put the clause in the right slot are actually two case errors in the wrong transcost the person taking the test 4 points lation The first is the misrepresentation of August 76 CRYPTOLOG Page 10 UNCLASSIFIED DOCID 4019632 UNCLASSIFIED Dozens of other examples could be easily cited but the above should give some idea of syntactic error I might note in passing that in our experience few people pass who make more than two such errors 2 3 Lexical errors 2 points each The term lexical is something of a misnomer here because the error type covers not only incorrect vocabulary items but also certain prefixes and suffixes which are usually associated with syntax but are in fact more easily treated independently of syntactic relationships Mistranslating say the German die Strasse the street as the apple would obviously be a glaring error but it would not as such affect the syntax of the sentence The same can be said of mistranslating the plural form die Strassen as the street or of rendering the future tense of a verb as a past We have coined an admittedly awkward phrase punctilinear error to cover both kinds of 2-point errors Such an error can be discerned at a given point in a line or string hence the name without the need for syntactic analysis Below are some examples of punctilinear error Correct translation There were only suppositions no one knew for sure whether there was any ore in the prospecting area The mine is located near the city of Wrong translation There were only no one knew suggestions The mine is located next to the city of It is a very rich It is the richest mine mine The first two errors clearly involve vocabulary errors while the third reflects a misunderstanding of a point of grammar relative vs absolute degree of superlative in none of these cases is it very difficult to determine that a 2-point error has been made In another instance the judgment is more difficult a phrase which should have been translated from Russian as the beloved home team was rendered as the favorite local team The latter phrase can be understood as synonymous with the former or it can be read as the favorite of several local teams Although 2 points could be taken off here it would seem more reasonable to take off one the penalty for a violation of English usage which I will now discuss 2 4 Errors in English usage l point Admittedly questions of usage and convention are to a degree subjective and it has been our experience that what is awkward or strange to one person may not offend another in any way Hence we usually assess one-point errors only when there is consensus among the graders There are enough such cases however to make the one-point error a major factor in the final grade Some examples appear below The first example is a sentence concerning repair of a bridge Idiomatic translation No one has put his hand to restoring the bridge I Stilted translation By no means have somebody's hands touched the bridge The English in this case is so poor that any grader would deduct at least one point regardless of possible major errors in the string In most cases though the strings are not as long Let us consider a few less-extended cases A test taker wr t ng about mine excavation transla-ted a phrase as fl o o o dug one drill hole an action impossible by definition although the general meaning is clear Poor usage of this kind is often the result of a translator's adhering too closely to the original language forms Another used the nonexistent phrase to his luck in the sense of luckily for him a third came up with It completely won't do for It won't do at all The list could from our experience be extended indefini tely Finally we take one point off for English words and phrases which while not in themselves awkward or peculiar violate conventional usage For example translators who stick closely to original language forms will render Secretary of State Kissinger as Foreign Minister Kissinger or the Pacific Ocean as the Quiet Ocean When we feel that a personal or geographic name should be known to any literate person we will deduct a point if a test subject misrepresents it 2 5 Spelling and punctuation Nothing is taken off for orthographic mistakes unless they will likely lead to misunderstanding in which case the appropriate points are deducted 3 General guidelines and caveats 3 1 Recurring errors We try as hard as possible to avoid taking off points each time the same error is made This is not too difficult in the case of lexical items for once a translator decides that a word meaning general should be translated field marshal he will usually be consistent Much the same is true of violations of English convention such as Foreign Minister for Secretary of State Repetitions of syntactic errors are much more difficult to spot in cases where we are unsure whether the identical error has been repeated w assess an additional 4 points August 76 CRYPTOLOG UNCLASSIFIED Page 11 DOClD 4019632 UNCLASSIFIED 3 2 Double jeopardy A double error over an identical stretch of text is penalized only at the higher rate Thus if the translator not only incorrectly reverses the order of dog bites man but in addition misidentifies the animal man bites cat he loses 8 points for the two syntactic errors but nothing for the vocabulary mistake 3 3 Importance of context We are careful in judging errors particularly of the 2-point variety in terms of context and we try to avoid relying on dictionaries as final arbiters in every case Thus the sentences He doesn't have the money to go and He doesn't have enough money to go mean approximately the same thing but no dictionary will ever inform us that the means enough -or vice versa Similarly a bilingual dictionary particularly a limited one cannot give enough translation choices to cover every situation Hence the literal translation monstrosi ties that are docked one point and sometimes 2 or even 4 points 3 4 Need to distinguish between language terion for grading is syntax which is to say the relationships among the elements of the sentence what things go with what other things and in what way and the minor criteria relate to word choice and usage This division is admittedly reminiscent of training and the classroom rather than language used in everyday life Yet these are the criteria which have been used in the teaching of foreign languages and with which subjects and graders alike are most familiar Furthermore it would be extremely difficult perhaps impossible to devise a grading system according to which certain kinds and strings of information are considered more important than others nor is such an approach needed for beginning translators We have found that natural texts not contrived ones are best used but the criteria by which the translations are judged should be formal of the kind described above They cannot be gut reactions to mis information 3 5 Procedures to aid reZiable and valid grading As a final observation we have found a tendency among some graders to temporize forms and language as used especially in docking a paper 4 points if they One of our major problems from the time the feel that the underlying cause of a mistake was grading system was put in effect has been the a simple misunderstanding of a single word or tendency on the part of some graders to consider grammar form I can see how he arrived at the forms of language as somehow equivalent to that This is unfortunately a kind of mindthe effects of utterances That is they will reading that may be invalid and is certainly protest for example that the omission of a unreliable since no two people draw the same negative particle in a translation can as seri- kinds and numbers of inferences The only ously affect the meaning of the passage as say thing we can judge is the syntactic and lexical an error in verb-noun relationships and will rightness or wrongness and the English usage insist that the same number of points should be present in each paper docked for each UNCLASSIFIED __ This attitude is of course very understandable but I should observe here that our major cri- f IUII II J _J J J lIJ JlIJ J 1i 1i'il J i '0 'e l III Starting next renoh o lor _ It provides a series of year Frenchfines for first offenders men using English words in who employ foreign words the wrong French places in advertisements sales will be fined up to $35 contracts job offers emeach time they are conployment contracts and victed of Franglaie operating instructions In 1977 the duty free shop at French airports will be known as La Boutique Franche Tourists I expecting Ie discount for paying in travelers I checks will have to ask instead for Ie miniI marge Frenchmen who want I to book on le hovercraft across the English Chan- o moe' d u ed throughout France Frenchmen will have to use savoir faire an expression Americans and Englishmen adopted from the French centuries ago without bastardizing their own language French officials insist that the new regulations are not designed to retain the purity of their mother tongue as much as they are to protect consumers and employers who do not understand the true meanings of the foreign idioms lllllllld Parade August 76 CRYPTOLOG Page 12 UNCLASSIFIED 13 Apr i 1 1976 DOCID 4019632 SECRET SEPTErdBER p III I II U 1 II 8 9 3 f 10 11 111 18 17 18 19 110 III 1111 113 IIf Ill 118 117 lIB 119 30 OCTOBER P L 7 8 III 13 B if 111 19 110 IS 16 17 18 III 1111 118 117 1111 113 IIf Ill 118 119 'G- 9 10 11 119 30 31 On the other hand in spite of the resource decrement the intelligence requirement continues SIGINT remains our first line of defense and it is incumbent upon us to respond to this charge with all the resourcefulness we can muster ------J L ---_------JL - 1 0 EO 1 4 c P L 86-36 August 76 CRYPTOLOG Page 13 SECRET EO 1 4 c EO 1 4 d P L 86-36 HAND' E I' r QUINT bIWl IEbS 8t1 Y 86-36 EO 1 4 c cE0 1 -4 d P L 86-36 DOCID 4019632 SEEURREf If you are still with me to this point please don't send me any letters saying that we are already doing all of this For each case you cite wherein the full objective has been achieved I can cite two or three wherein the objective was either ignored or effectively undermined and that is not my purpose in any case What I am earnestly requesting is that this Agency recognize an area which will be of vital importance in the future and take every reasonable step to insure we are up to the challenge If we are to succeed as succeed h'e must we need to be innovative thorough deliberate and above all cohesive in our integrated analyst program management and it is within our capabilities to do so We as an Agency should not settle for anything less SEERET L- ----I - -- - ---- -- August 76 CRYPTOLOG Page 14 SECRBT 1I'1EE8 -- -- IIJdII5U VIA CUMiN I C 'li' 141U t 3 81lb'i DOCID 4019632 UNCLASSIFIED Summary by Emery Tetrault PI6 of a talk EJ DR RU'rH DAY AS The subject of individual differences ln speech perception has not been a burning issue either in cognitive psychology or in psycholinguistics but it could be of intense interest to anyone who is responsible for selecting and training voice linguists This was the main inference drawn from a talk given by Dr Ruth Day of Yale University and the Haskins Laboratory to SIGVOICE theCrypto-Linguistic Association's Special Interest Group on Voice Her talk was enthusiastically received and considerable support was given at that time to th notion of following up on her findings Durlng the time that has passed since that talk a few actions have been taken in the direction suggested While NSA has not funded Dr Day directly R54 has been supporting research in individual differences in speech perception at Haskins Laboratory for the past year In addition the Office of Naval Research has let out two contracts studying individual differences in speech perception one with Dr Day and one with the University of Oregon R54 has been receiving the progress reports from these ONR projects Dr Day began by introducing the concepts of rivalry and fusion as demonstrated in experiments with visual stimuli through the use of a tachistoscope a device which makes it possible to present different stimuli to each eye and with auditory stimuli through the use of the dichotic listening technique One of the dichotic listening tasks was simply identifying input did the subject hear banket in one ear and lanket in the other ear or did he hear the single word blanket It turned out that subjects thus tested were either very good fusers that is they consistently heard a single word or very good separators There was almost no middle ground This was particularly so when the outcome of fusion proved to be an actual English word although the pattern was maintained as long as the result of fusion violated no English canoniAugust 76 cal form sc Butcher in NSA Technical Journal Special Linguistics Issue No II Another task consisted of identifying the first sound heard when the two stimuli are presented with a slight time lag in onset from 25 to 150 milliseconds Everyone heard b when banket was given before lanket but part of the test population continued to hear Ibl first even when the onset order was reversed Dr Day advanced the hypothesis that this group was hearing only that which English permits -- Ibl followed by 11 She called this group of people language-bound The people who were capable of receiving and holding auditory stimuli in some kind of raw storage without immediately activating their grammar machines she called stimulus-bound This bimodal or bactrian distribution of scores has been replicated with more or less randomly selected sample populations In studies involving 100 Yale undergraduates more than 90 were firmly identified with either one or the other group and only five people showed no marked tendency toward membership in either the language-bouad henceforth LB or stimulus-bound SB camp Moreover group membership appears to be stable LB people do improve their performance by learning the task but they never get to be as proficient at it as the SB types Dr Day noted that the poor performance on this second task by LB people could be attributed to the fact that they simply can't make temporal-order judgments However when fusion of the two stimuli produces strings which are not canonically possible in English then the LB subject does as well as anyone else in determining that 1 there are two stimuli and 2 one of them clearly precedes the other in onset time Thus when the stimuli are back and dack he has no problem making the temporal-order judgment because neither bdack nor dback is a good candidate for an English morpheme In other words he can do the task in those situations where higher-order linguistic processing is ruled out CRYPTOLOG Page 15 UNCLASSIFIED DOCIO UNCLASSIFIED How do these two groups once identified fare on other kinds of tests Do they retain their group identity Dr Day described some other tests which showed similar distributions Digi t Memory This warhorse of psychological testing usually produces a serial position curve Test subjects have to remember sequences of digits and write them down in the same order as they were dictated Items presented early and items which were last in the sequence tend to be remembered but items in the middle are usually lost This is what everyone expects and this is what all the methods of averaging and other obscuring individual differences give us This is not however what SB subjects do They make relatively fewer errors on this task than the population at large and the errors made occurred more or less evenly throughout the entire sequence To get an SB subject to show the serial position effect the task has to be made so difficult that the rest of the population finds it nearly impossible to do so Digit Memory wi th Zero Reaa U Cue This task is the same as the previous one with one difference each sequence ends with a zero which is the cue for the subject to start recalling items This procedure neutralizes the recency effect at least for the LB types but for SB subjects it affects not just the last items in the string but all the items in the sequence This according to Dr Day suggests that there is something more than a quantitative difference between the performance of LB and SB subjects on this test It supports the notion that the LB person has begun some immediate coding and longer-term storage of incoming auditory stimuli while the SB type is holding the whole string in a kind of buffer memory This is why the zero recall cue tends to affect the entire sequence for SB subjects publication runs a puzzle consisting of words running every which way actually eight directions in a matrix of letters SB subjects did better than LB people when left-to-right directionality was violated Dr Day advanced the notion that SB types have the option of using both left- and right-hemisphere capability to solve tasks while LB subjects are more or less locked into left-hemisphere -- i e linguistic -- processing Searet Languages Dr Day used a simple encode device to come up with one secret language she transposed all 1 and r phonemes Maly had a ritter ram o Test subjects were asked to transform standard English discrete words and then stereotyped connected discourse into the secret language She played some tapes of subjects performing this task SB subjects were q ite fluent but the LB people had problems One subject was almost pathetic as he tried valiantly to transform the word bramble and Dr Day said she was going to spare us the pain of listening to his efforts against the name Nelson Rockefeller Delayed Auditory Feedbaak DAF The test subject wears earphones and talks into a microphone His speech is fed back to him with a delay of about a fifth of a second Again Dr Day played some demonstration tapes for us and again some subjects had little or no trouble with this task and others sounded like drunks in a nearly comatose state Dr Day herself was one of the subjects on the tape and we heard her trying not altogether successfully to recite the English alphabet something which she assured us she knows She noted that OAF performance is related to such tasks as foreign-language mimicry and phonetic transcription in IPA or some comparable system of unknown language forms Jotto Given a target word -- say charm -- and a list of comparison words subjects are asked to find the one -comparison word that shares the greatest number of letters with the target word According to Dr Day the LB type will have a harder time with this task than the SB subject He frequently underestimates the number of shared letters between target words and comparison words He appears to be operating on the whole-word level and not seeing lower-order elements With a target word like sylph and a comparison word like prone he would tend not to count the letter p perhaps because he is going immediately from visual stimuli to phonemic representation from ph to f Word Searah NSA Newsletter devotees immediately recog- Summary So we know that some people are languagebound and others are stimulus-bound and we have some notion of how to identify members of one or the other group The immediate executive response ought to be ' ll right dammit which one is the better one for a voice job Is it the LB who operates on a strong sense of linguistic expectancy the anticipation of what can follow at every structural level from sound to meaning Or is it the SB who can jltore raw auditory data in a short-term memory and more or less call things as he hears them The answer to this question may not be as obvious as it seems on the surface but it would certainly be worth the effort to push this research a little bit farthere We are in no position to pass up any help in the field of language processing nized this task since every month this esteemed August 76 CRYPTOLOG Page 16 UNCLASSIFIED DOCID 4019632 UNCLASSIFIED NSA CRYPTOLOGIC COLLECTION Have you ever visited the NSA Cryptologic Collection Have you ever heard of it If not why don't you visit it You'll be fascinated by what is contained in it The collection is in Room 3W076 Operations Building and is administered byl x4017s other data derived from a broad spectrum of languages A few samples are shown on the opposite side of this page Relative newcomers to the Agency who started late to save their back issl ll JS offhe NSA Technical Journal can find a complete set here back to Volume I Number 1 Incidentally The collection is in a small room but the have you seen the special Twentieth Anniversary room is crammed full of cryptologic Ii terature of Issue of the Journal yet historical as well as completely up-to-the-minute The collection also includes a large number value Much of the collection is shelved in of open-source publications dealing with more than 1300 Shinn boxes plus additional file-cabinet and shelf space Much of the cryptologic subjects including several rare literature is unique and not represented in any books And of course it contains a facsimile other Agency collections although the collecof the Voynich Manuscript You can't say that you haven't heard of the Voynich manution holdings are indexed in the NSA Technical Library C5 records script With the exception of the rare books most The collection contains historical documents pertaining to the use of cryptology in World of the articles in the Cryptologic Collection War I and World II In addition it contains can be borrowed All the holdin'rg s a r e _ _ 1 working aids of all vintages pertaining to carefully cross referenced mandl I all aspects of cryptology including a large I 6salways pleased to help you find what number of cryptolinguistic working aids conyou are looking for taining letter-frequency word-pattern and UNCLASSIFIED I August 76 CRYPTOLOG Page 17 UNCLASSIFIED P L 86-36 P L 86-36 DOCID 4019632 _ _ UG' _ U UL und the patterns ir the book are alphabetized FROM RIGHI' TO LEPI' HORDS AND PHRASES pp 1 - 130 This section contains individual words and a few commonphrl1ses '-- Th l EO 1 4 c P L 86-36 86-36 1 4 e A UguS t 76 -4'8P CR' PTOLOG Pa ge 18 SBettE'f tlMBlt1 - DOCID 4019632 SECREf 8110 K E O 1 4 c P L 86-36 LETTERS TO THE EDITOR ARE COMING Yes letters to the editor and artieZes too are still coming in response to Vera Filby's article How Do We Know It's True CRYPTOLOG February 1976 Take for example the following letter which is irt respon to an article that was written in response to Mrs Filby's article To the Editor CRYPTOLOG There is a disquieting bias in Mr David Gaddy's On Being Truthful CRYPTOLOG April 1976 which seems to want to pull SIGINT equals truth to the other side of the seesaw To balance SIGINT's unsophisticated view of itself as being on the side of the angels Mr Gaddy points out intrinsic flaws in the SIGINT production process sources might lie some SIGINTers are more competent than others the SIGINT facts suffer from interpretation sanitization and suppression cause obfuscations communications are susceptible to deception He piles up the evidence until SIGINT equals truth slides to the other side of the fulcrum where SIGINT is presented as inherently false What should we think then we SIGINTers Is a SIGINT fact true Yes Emphatically yes Certain basic principles underlie and guide any enterprise In the SIGINT business one of these is that the lines of communications follow the chains of command This means that organizations use electrical communications to systematically pass orders requests answers plans instructions etc and these communications speak for the organizations Those lines of communications are not established to deceive us That the SIGINT derived therefrom is true is really the only way to approach our profession I am convinced of this yet I would join Mrs Filby and Mr Gaddy in encouraging a deception study to determine its frequency and the conditions under which it would likely occur The results are sure to reinforce the value of SIGINT and to allow quick reply to inquiring non-SIGINTers who understanding little of the nature of the SIGINT beast are prone to worry about that aspect of it they do understand -- its potential for deception In all fairness though deception is not the main point of Mr Gaddy's article But what is the main point Is it his summary idea encouraging continuing education indoctrination and training about the limitations of SIGINT I do not think so because that is certainly an eas lyacceptable idea that scarcely requires much rationale in its support Main point aside I think Mr Gaddy's most important point is his allegation that we add falseness to August 76 CRYPTOLOG Page 19 EO 1 4 c P L 86-36 SECRE'f SPO KB DOCID 4019632 SECRET SPOI E it not the professional SIGINTer To use the bicentennial VISINT example now in vogue should Paul Revere have dashed from house to house giving his fellow-citizens a flee no-flee option thus Prearranged signals indicate British marching on a path that leads to our town then in an aside under this breath but don't blame me if you sacrifice a night's sleep and they never get here -- I'm just telling you what is reflected by the one-if-by land-two-if-by-sea signal No Paul Revere said To arms The British are coming A cyn i calto wns lllll n- -at leastone--probably said What does that turkey know I'm going back to bed But as we now know there were few practicing cynics in those days and if any they were probably doomed to obscurity because of it History shows that Paul was believed But why was he believed Because he was trusted He was a hard-working technician deeply involved in his pursuits and well known for his competence -- he often provided fellowcitizens good counsel The townspeople trusted him to understand what was going on and relied on his judgment P L 86-36 P L 86-36 Also Paul was believed because his message was direct unequivocal forceful Its point was not begged by passiveconstructionS Ondr 4 c cheapened by apologetkphrases nor c lJOuaLed 86- 3 6 over by stereotyPed redundancies His message neither required nor encouraged further interpretation The townsfolk heard believed and reacted Likewise SIGINT product would become more effective if our reporting faithfUlly adhered to these same principles The suppression of' evidence incidental to the SIGINT fact -- evidence whose lesser lights do more to divert attention than they do to illuminate truth -- rather than obfuscating the SIGINT fact would help us present it in a professional manner The suppression of nonessential data is not a burden to the reporter It is rather a precept fundamental to the reporting discipline It is good reporting _ _ _ _----I VJ2 SEERE'f SP8IEE Mr Gaddy replies To the Editor CRYPTOLOG Thank you f Cl rthe6pportunity to comment on ________ Iletter After all who is in the best position to judge the accuracy and validity of SIGINT Is I am afraid that his reaction demonstrates that the attitude I deplored is so ingrained in some of us that an attempt at redress -- moving the pendulum back to a neutral midpoint by reminding ourselves just what we're about -- is instantly construed as a swing to the opposite extreme I intended to be disquieting My bias on the other hand was toward objectivity I was attacking attitudes not SIGINT The attitude I was specifically after is cocki- August 76 CRYPTOLOG Page 20 CR T POK DOCID 4019632 SECRECP SPOKE 'P L 86-36 meat 'ithat point Does the horse equate to ness an unseemly air of intellectual arroCRrrlCOMM gance That was my main point stated in the second paragraph To develop it I recalled the components of COMINT not to illustrate' David W Gaddy 05 intrinsic flaws in the SIGINT production proce0l41l101 lQTIAL - MttO cess but as realities as givens Total balanced recognition of these givens marks the SIGINT professional not the sign over his dl lSk saying SIGINT doesn't lie And it was the To the Editor CRYPTOLOG SIGINT pro's attitude that I so extravagantly praised You have asked me to comment onl response to Dave Gaddy's response to my article Rather than rebut the strawman precis of my on deception My comment is great This is v ews advanced byl Imay I tak e issue just the kind of discussion I wanted to start WI th what he says about deception which some And as in all good lively informed discussions readers may recall was Mrs Fil by's original point A mind closed to the likelihood of de- not only is the main theme developed but also sidelights and subtleties and related aspects P L 86-36 ception is a likely candidate of dec ption emerge This happened in both these responses The now universally recognized tlbasiciprincias well as in those from y pIes on which SIGINT depends are at the same CRYPTOLOG and Junet me the cornerstones for a program of decepJuly CRYPTOLOG The next stage t at we appartIon We generally posses'S a highi level of confidence that in partbecauseiit is so diffi- ently all want to get to is the serious research cult to pull off deception wil l inot be used or into the facts of communications deception to the degree that we can arcertain them We know i f it is that we will be smartienough to dethere has been deceptionL lin North tect it As long as4bat b1 h level is not Vietnamese communications in recent years absolute I sharel jittitude I do Where else Can't we get everything we know not expect deceptlon In the mass of normal assembled for study Then we can proceed to an peacetime intercept for example On the evaluation of the extent and circumstances of other hand leicept a foe to be as wily as I deception the possibilities of recognizing it would proposl lto be Without citing October and the chances of missing it Granted to 1864 as an early example of enemy use of decepsome extent this is like research into psychic tion against the U S Army I would commend to phenomena in that it may be an effort to know kand other readers who have not the unknowable -- but at least we can learn delved into it Brown's recent Bodyguard of something in trying Lies The test of camouflage is not whether you see it but whether you don't Vera R Filby El2 Finally I must say that I am as pUZZled ovel C EeRE'F SPSKE the Paul Revere story as when I first heard it EO 1 4 c used to construct some sort of analogy It's a P L 86-36 good example of signal communication but loses So doe s Mrs Fil by I I SORUTION TO NSA-CROSTIC No 4 CRYPTOROG June-Jury 1976 Jacob Gurin -san Employment of Mil i tary Lingui st s NSA Technical Journal Vol XIII No 4 Fall 1968 Also reprinted in NSA Technical JournaZ Special Linguistics Issue II During the showing of a Japanese movie used as a training aid the students cheered mightily when after about one hour of total lack of comprehension they understood the maid when she knocked on the door and said 'Excuse me ' UNCLASS I FIED August 76 CRYPTOLOG Page 21 Pl-Jun 76-53-24723 SECRET SPOKE DOCID -- 4 01 9632-- - - - - - - -- - -----V- CJ_------I-__ - - - - - - - 'fIllS BOCUMFlN'f CON'fl' INS COBEWORB Ml TERlhb This document is from the holdings of The National Security Archive Suite 701 Gelman Library The George Washington University 2130 H Street NW Washington D C 20037 Phone 202 994-7000 Fax 202 994-7005 nsarchiv@gwu edu