TOP 11e• lf WuJV Wm 0C5BIWC ll alll Dlil Ul IW 1 i l DWU B l Wl 41l ll f W41W lkjGJtl CBCE 'Ul lbl rn aJl lBl l 'Ul500CBl5CD 0 EO 3 3b 3 PL 86-36 50 use 3605 io YEARS QE -PROCESSING • COMMUN I SPOSITION - Jane Dunn I IT--I N ••••••••••••••••••••• • Robert Stur ess 7 • ••• 10 • •• • 12 ____ ______ v oL c ---- l ' --- --om as Jr • • 13 ABDUL AND 615 40 TANKS •••••••• •••••••• Frederic 0 Mason Jr • 17 LINGUISTS·- WE NEED AN EXPERTS YELLOW PAGES •• •••• •• ••••• •••••• Jane Merlin •• •• •• 19 1· 2 o -----------e1w•• It IIIIHM18111188 m e111r 1 • •• • 111• •••• QN 11811111 Raee1w I f Hr N tf 9 ft 1 9 I a rs n llillli NSIII ••• B I a pll-88811911111 8 1 I J WpH •• I ••••fll Declassified and Approved for Release by NSA on 03 - 01-2021 pursuant to E O 13526 MDR-109385 EO 3 3b 3 PL 86-36 50 USC 3605 --- 20 YEARS QF TRANSPOSITION Jane Dunn El -- Aug-Seot 75 CRYPTOLOG • PaJe 1 FfOP 8H01NH' tlllBM -------------- • - - ----·- -··- _____ -· - _ ·--·- EO 3 3b 3 PL 86-36 50 USC 3605 '18P 808RM IJPulDRt'l Aug-Sept 75 CRYPTOLOG Page 2 T8P 8D@IHH l lDR 'J EO 3 3b 3 'P8P 8R8BIW ti'1-IBM PL 86-36 50 USC 3605 Aug-Sept 7S • CRYPTOLOG Page 3 'llOP II0DIR' lH'11Blli t or 81100119 IJfllBR5' AUR-Sept 7S CRYPTOLOG Paie 4 Y8P OD@BIII tJP IBRA EO 3 3b 3 PL 86-36 50 USC 3605 T8P 8B8Rllf 8 CIBIIII EO 3 3b 3 EO 3 3b 6 PL 86-36 50 use 3605 r Aug-Sept 7S • CRYPTOLOG • Page S •or HIOMI · • - - - - __-----1 EO 3 3b 3 PL 86-36 50 USC 3605 •8P 810RM IIUIR11 Aug-Sept 75 • CRYPTOLOG • PaRe 6 a 811 181• n11a _ I Non - Responsive f or illRDf PROCESs1 c I DflllR 1 COMMUNICATIONS I 11 BERT STURGESS A613 · • • ' •' '• • • Aug-Sept 75 • CRYPTOLOG • Pa2e 7 EO 3 3b 3 EO 3 3b 6 PL 86-36 50 USC 3605 EO 3 3b 3 PL 86-36 50 USC 3605 TOD filliQllllli IH 11111 z · ··· _ - ·· - - - - ---'-----I EO 3 3b 3 Ter e1neu Otl• M Aug-Sept 7S CRYPTOLOG Page 8 98P 81UUIM tH1IIM PL 86-36 50 use 3605 •or EO 3 3b 3 PL 86-36 50 USC 3605 fJB@• DI t l'IIB• A I I Aug-Sept 7S • CRYPTOLOG • Page 9 • T8P 8D8 M iHIR'1 - - - EO 3 3b 3 EO 3 3b 6 PL 86-36 50 USC 3605 I 8H8BBT I AFIX FOR THE LANGUAGE PROBLEM m- JOHN B THOMAS JR Special Assistant Language A9 An old hand proposes a managf1fll8nt tool that ll ight be of hsl p to dsal r ri th • - ·- - ----- - -- _ - a pezts 111i al buokst of r l01'fflB • Doris Miller's article Language and the or in the original foreign language or writes COt IINT Production Process NSA Technical Jo ao down word-for-word everything that was said in nai Swnmer 1974 falls gracefully into Alexan- English or in the original foreign language der Pope's category of lllhat oft was thought This process is rarely one-for-one in te 'llls of but ne'er so well express'd It is for good transcription man-hours required per chaMel reason that that article won the First Prize hour The best ratio might be 20 minutes per in the Crypto-Linguistic Association's 197S channel hour for selecting and gisting the Essay Contest easiest material the average might be 3•4hours I would like to expand this subject in the to transcribe o e h ur of voice intercept But same general vein by would most nonl1ngu1st 11 11agers believe that it can take 10 20 or 30 hours to transcribe just supplying a ki d of p stscr1pt dea ing one chaMel hour of voice material with poor with some specific points about 011 e audibHity and high linguistic complexity but lan Uage work and the role of the military also of high intelligence interest e examining the paradox If11these ideas We need all the help we can get Channel were indeed oft thought why hasn't identification dial recognition and any other more been done about them possible selection processes need to be exploited suggesting a methodology and management to the fullest to put the best possible interposture by which the working linguist's cept the richest of all the ore before the transcriber and the working linguist-manager's wisdom and observations can be converted Does everyone fully realize that until the fr0111 a still small voice into a real magical machine transcriber appears the help for higher management voice language processor is the valve through Looking first to the voice problem I would which the whole production stream must flow caution that we are on shifting ground in this Some realize it and apply the unfair pejorative area Miss Miller discusses things that speterm bottleneck to hill But the term is pascialists al1UOst unanioously feel should be done sive whereas the transcriber is active In She considers the problem to be one of enlight- ·fact only in the degree to which he is active ening and persuading 111anage111ent to put some is any intelligence possi le And a person carryforce behind the ideas she projects The voice ing out a key that is critical process deserves problem on the other hand I think finds even to be recognized with at least a positive-sounding specialists with more uncertainty about stanterm Key processor sounds awkward but it is dards and procedures This is no wonder certainly accurate The Voice Explosion is terrifying Voice tape is the 111ost anonymous most unsortable Key processor though he is the Voice Language unfileable indigestible ungistable stuff in AnalyS t or Technician suffers from the fact that the world What do you do conte111platina racks the Agency both infol'llllllly in the minds of superof tape containing a million or five million visors and formally in the language of personchannel hours of voice Obviously you panic nel handbooks is uncertain and vacillating in You are lucky if you don't babble of green fields its concepts about his career At hiring time but you do IIIWllble for a few years about machi1tt his transcription skills are at a premium Once trarisi ription you speed up or slow down the tapes hired however he will probably best serve his a bit and you half-way persuade yourself that you own interests if he gets out of transcription have mechanized the problem when you have just as quickly as possible Transcription is a mechanized the periphery You zip voice tracks calling in which there are extraordinary physiback and forth in fancy 111achines and then you get cal and even emotional d-eu nds particularly more fancy machines to zip printed transcripts tolerance of noise -probably not amounting to back a nd forth But no machine can produae a hazu·dous duty but beyond the usual threshold transcript Finally you bite the bullet and of iTritations The best and 1110st assiduous admit that the human transcriber remains transcriber tends to be isolated from lllOSt of the heart of the business The transcriber -- his organization•s activities by his attachment not a machine-· listens to the recorded voice literal and figurative to his recorder -- a intercept and discards it gists it in English beautiful for11111la for not gettina ahead in the e e I I l Aug-Sept 75 CRYPTOLOG Page 13 8811111T 512i 57 PP SSH sr m s a mnx 811J@ll• t world Aside from these obvious points there are lllllny questions of the predictability of success in transcription the extent to which a person can overcome his initial reluctance to engage in it the proper division of time for transcribers do they analyze and report on the lllllterials that they transcribe and the question of what carrots can be held out to them There is generous rooa here for the specialist to analyze and institutionalize his experience and for the manager to apply it Ro Z • of ths Ni U ta ry Another thorny problem is the place of the military component in language operations Part of the answer rests on the fact that Jhat needs doing is only a part of the question J hsre it needs doing and 1um are equally important Does the job have to be done in an office close to all collateral resources and computer services 11Fro11 eight to five Or 24 hours a day On a mountain on the border somewhere Jn an airplane or on a ship n general the military man cannot have the particular kind of continuity and depth on language problems that a civilian specialist can have But he does have his own special qualities -- mobility adaptability awareness of what the collllllllnd needs to know right now and most of all the quality of being there and staying there Add the qualities of savvy and a good ear and you have an indispensable member of the team It is an extremely difficult thing to train use supervise and motivate military people in a field as c0111plex as the pl'ocessing of nWDerous languages It would be a miracle if it were always done propel'ly everywhere for every language It will certainly be done better if linguistic unagement and military know-how are continuously focused on the challenge that each language poses to the operations of the entire S GINT COIJIIIIUnity The proper division between •ilitary and civilian resources on a language problem has to be established by weighing the what where and when factors but not completely independently of the language itself ls the language so difficult for instance that the military cannot assign the highly selected personnel and spend the tra·ining time and money to keep competent linguists in the front end of the system ln such a language as Arabic for instance l would suspect that this is more often the case than is fully accepted throughout the collllllUnity all that sounds possible in theory is not done in practice never has been and never will be E a1nirzinq the Paradoz As I have said -- or let Alexander Pope say -i iss Miller has written an excellent SUIRIDary of the consensus of allllOSt every gathering of language specialists that has occurred in the Agency for the past 10 years If we k wz what should be done why do we still have a language problem Well for one thing as Miss Miller cannily observes it is hard for language processing to go bad with enough of a bang to get attention clear up the line For another we don't really have a monolithic Language Problem ColllJ lete with capital letters Instead we have a Chinese Language Problem an Arabic Language Problem a Korean Language Problem a Mongolian Language Problem and a Problem for every other language that we work or should work This fractionation strains Jn nagement's attention But Coaimand Attention in the military or ManaJement Attention in the civilian world are the keys to improvement v lllfP Qg8 ChsokZ iBt for Hanagement For clarity and impact let us take several principles about the language business that Miss Miller and J have dealt with and recast them as questions that nianagement might ask -- in effect as a language checklist for management For the time being let's avoid the question of who in 1111nagement or even what level might ask these questions J think it uy develop that this is the biggest question of all • Questions on 8asic Principles Are resources in the language solidly related to the national priorities on the one hand and the possibilities for their satisfaction on the other Are tactical direct-support military requirements for language skill realistically provided for By the most economical and most realistic means Is there a balance of cost and probable usefulness to U S security between National and tactical efforts more solidly reasoned than the squeaky wheel system Do the analyses on which the above decisions are based cut sternly across what we feel can occur as the four built-in temptations toward bias of Couunity elements It's lllOre equal than others if it a rolls on wheels fb floats c flies or d can be called 'National 1 It seems unnecessary to specify who is tempted to hold the respect ve opinions Facing the facts can the planners md opera•Questions on Technical Skills tional and language experts come up with the Is the civilian language-operational work best compromise -- in particular with the force skillful enough to work materials in this best mix of civilian and military language relanguage with a high degree of proficiency sources worldwide to spend the least and get the most Aug-Sept 75 • CRYPTOLOG • Page 14 iliODll1f IUIIIII 'I llllllff IIL HIIIIII 11111£ 811@Rlllf Is the military force proficient at what Do we deund the highest professional Lt must do If not should preparation be standards changed elements of the job shifted or should certain limitations be lived with as eQuestions on Techniques and Procedures acceptable balanced against the cost of Are money skill and attention given to eliminating them the development of language working aids Does somebody fight for machine time Are some people overtrained for their actual job with unnecessary costs and personal Are linguists involved at the R5D stage frustration as a result or as soon after as possible in working eQuestions on Management with intercept and processing systems that will affect that language At what point do language expertise operational experience and persoMel and training • Questions on Careers and Morale experience in language converge to provide Is there at least one visible civilian higher management with the best possible adwhose skill and professionalism in this lanvice on allocation develop1110nt and use of guage have clearly Jed to the senior grade resources in this language On second he holds and the respect he is accorded Is thought in s0111e instances do they converge there a similar military man at all Is a young specialist's skill and productiWho 11akes the hard decisions that affect vity in this language directly and obviously future capabilities in this language esa central factor in his advancement and the pecially the decisions with distant pay-offs regard in which he is held Examples To spend money for long-term training for contingencies or conversely to say Are many if not necessarily all linguists that while a certain contingency is possible trained toward the attai11111ent of desk ofit does not on balance justify the use of ficer status with an analysis and reportscarce resources to prepare for it Who ing capability -- toward beco•ing an expert takes and lives with the responsibility for or potential expert in every sense of the the decision word Does 11B11agement training include e111phasis on language processing To be specific Need foY- I anqtuzqe Studua is Doris Miller's article required reading The previous checklist might see111 overwhelmr for 1111111agers ing especially when one considers that it If this language is in a 'llllrginal or careshould be applied separately to each language taker status have full ·measures been taken of Agency interest Is there a way to get a to preserve continuity handle on the problem to codify and instituIs fragmentation avoided in that even if the tionalize the principles il IJllied in the list language must be handled in various divisions One way working toward Co11111and and Management or offices the training professionalization Attention might be to prepare a written study and collateral support are coordinated By of every language of current and potential Coinwhom munity concern Some languages can be discussed Is coeducation for linguists managers rather SUlblllarily Ho Bul-Bul traffic has ever and systems analysts being conducted been seen and none is anticipated No work judged to be required Dictionary 190S and • Questions on Hiring and Training gramnars 1887 and 1933 available in CryptoDoes hiring policy respect the long-range logic Library Agency multilinguist Dr Doe requirements of the language could identify texts in language if ever required Bul-Bul is mythical but there are rs hiring flexible enough to get the best languages of this sort At the other extrne people when they are available a full study of Russian-language requirements Do we maintain and develop linguistic com• would have to cover a lot of ground indeed petence once acquired Can we identify for Both types of studies are needed but the this language the individual who bears that studies that are probably 110st needed are those responsibility and carries it out for the in- between languages which are neither Do we cross-train toward 111Ultilingualism so insignificant as Bul-Bul nor have had the for breadth and flexibility where appropricontinuing high-level attention that Russian has had ate Or for Voice Language Analysts Do we develop the best voice linguists toward Whether one para lr Ph or a hundred pages near-native aural comprehension fully realong each language study might well have the lizing the difficulty and cost of such defollowing characteristics velopment Aug-Sept 7S • CRYPTOLOG • Page 1S fillEJR- lldldllH Ill 1111111 IIUIIIIMS Ubl 8R80Rrt' •It should represent expert opinion based on linguistic and language-operational experience that is backed up by factual and statistical infonnation • The body of each study should describe and quantify the operati ons requiring the language show the intelligence requirements that steer the syste111 exa11ine personnel career and training adequacy both for civilians and for military examine the present civilian-military balance in the language and decide whether it is in accordance with cost and production efficiency studies and with policy considerations and finally examine as objectively as possible the extent to which increased language resources including quality upgrading could increase U S intelligence production and conversely the extent to which a decrease or cutback would reduce that production The study should ask and try to answer all the questions on the checklist eEach language study should be prefaced with a clear and skillfully written sunnary of the principal problems or issues pertaining to operations in that language lf there are unsettled policy issues decisions that will have to be made in the fu-ture shortfalls of quality or quantity those points should be fairly summariied Those statements brought together perhaps with an analytic prologue and summary of the sunmaries would let any reader including the decision-maker know more about the Community's language posture than has ever been known before only be done by long-range policies and longrange supervision of their application Who can ask the questions direct the language studies pinpoint the problems and follow up with corrective action In a sense of course good managers are asking and trying to answer the key questions every day -- not consciously perhaps but as a part of the good manager's instinct for better operations It is useful to have nice co111petent people whose good instincts tell them to d9 proper things for the language business along with their other duties and within the limitations of how they rank and where they are placed organizationally We have such people But if many good things are really to happen and keep happening perhaps we also need a senior individual placed where he can swing weight in 11 tters of plans 110ney careers and policies throughout the C0111111Unity not just within NSA a person whose business -- preferably his sole business -- is to see that things go right with SIGINT language operations If such a solution is hopelessly unacceptable organizationally maybe there are clever people around who can see a way to achieve the s 11e results through other management devices The Language Problem is big enough to justify the best thinking by the best people to arrive at the best strategy The trouble is that while big enough it 11111y not be visible enough Meanwhile things in the language world continue to go wrong from time to time -- not with a bang hut a whilllJ er Who Will Ask the Questions If these are the right procedures one thing is cpnspicuous about them No stroke of the pen no adopting of new recomendations by manpower c01111Dittees no single directive is going to establish that the right questions can be asked and the right steps taken to see that they are eventually answered affil'lllatively That an Aug-Sept 75 • CRYPTOLOG Page 16 i IRORIJI' IL t 1111111 1111 881111 i lbcliil IJO 811111 • I Non Responsive EO 3 3b 3 EO 3 3b 6 PL 86 - 36 50 use 3605 One of our major problems with overage of the Near East is that nol'lllally we•re not close enough to get line-of-sight VHF tr lffic directly and without covering such a large area that separation of target groupings into h geneous sets is difficult • In the following traffic • is assumed that the VHF traffic is copi@d a very short daily window fro111 '______ -and that the HF voice is copied on a matching schedule from some site in the Mediterranean While there is ty no pretense that this resembles actual traffic still the problem does illustrate that unit idehtifications whether real or by using targe covernames or covernU111bers or by using T A ar itraries can be carried forward in time by T A solutions Also illustrated is the co11U10n T A problem oi having to keep track of a wide variety of classes of terms some of which are synon 'llls or n arly so and others of which are related but cUfferent Aug-Sept 75 CRYPTOLOG Page 17 HI Jllltil 1'1 I 1111111• --------·· • UHlllii ctlLM • APIB ltlrs will appear in next issue of CRYPTOLOG Aug-Sept 7S • CRYPTOLOG • Page 18 EO 3 3b 3 EO 3 3b 6 PL 86-36 50 USC 3605 IWllilLi 1 111 1 liliHiDR' QIWltl Llil 811L¥ EO 3 3b 3 PL 86- 3 6 50 USC 3605 ' LINGUJSTS--WE NEEll IN EXPERTS YELLOW PAGES Jane Me lin G522 Time was when a translator 'Jad to be extra • careful when dealing wi th the difference between request• and demand oj - with tt e exact ren• dition in to English to indicate l' ist how miffed or 11111d OP scared the original drafter of the • message ltad been Life as simple then or so • it seems •now The subiBct matte r '1 n the mes• sages wa' eneral iberil arts a_lthough the • language ______ wac always IIIU 110re precise than t e anguage appearing in he daily • press But u S intere ts are a bit 1110re com plex and technical now ind so are t subjects that our message texts cuial with Out ' generalpurpose dictionaries or-even our suec •Iized • dictionaries are neither up to date nor ufl•to it • Simultaneous interpre ers advise us f study in advance English-langflage reference a dS • dealing in general with the subject mattel that is likely to be discussed in the foreign an• guage But we NSA-ers 11an't always do th $ • Not when a single transl tor might have to 4eal with B eaifica of • • gree in widget design or used to work in a widget factory Maybe he's just a widget nut who is interested in them as a hobby -- 111 ybe he's even the president of the American Amateur Widget Nuts AsSU111ing that that expert does exist how does the translator find him in time to fill the hole in his translation By trial and error the translator might 111 nage after 15 phone calls tolocatesomeone almostasgood lt's a guy in the basement who thinks you ROLLEX widgets but isn't sure of the spelling Sn with the production deadline still approaching the translator still hasn't found ths expert who can provide the precise answer and he is forced to cop out with a literal translation a weasel-worded footnote and a generally dissatisfied feeling What we linguists need is a book of Yellow Pages of expertise a list of individuals by name not a list of branches since we do not have any Widget Problem who are likely to know the English word for things and processes in every conceivable field from industrial sanitation to beekeeping That list could be made available to every translator If he· had a question about the precise English word in a technical field he could let his fingers do ·the walking Do I hear someone say that this would put a large burden on the experts and is only a translator's scheme to get out of reading stacks of collateral Actually however a translator could read The New York Tiaes from front to back if he were a speed reader and had no other a 1 ili il J J i 1 ilil J WUW JiJl l •ews items on l•l 'l o 11o r- ' 1 1 a i 1 a - • - ' 'il ill il 1 1 i 1 and could an never onces e ec n ca --------- _ _ _ r-• -for gyrating widgets So wouldn't it be ets 1110re efficient more econo111ical more profes- So the translator is confronted sa 1-- - -- _ _' _ _- _ _ ' -- J•i • H - lei -lc•n•o•w•s•h - s t a r g e t-c ou n t r y _ _ sional to try to locate the person at NSA CIA 1 or wherever who can solve the translator's and its language Let us also asswae that he has pretty good general and specialized diction- dile111111 immediately arics But he knows absolutely nothing about With the Yellow Pages at hand the t ranswidgcts except possibly that they don't bite lator would not havetomake dowitha general None of his English-language reference aids even or literal translation which could be wrong mention widgets or any process that sounds like or misleading -- remember epi n vs turrrbl e for GRFULLing So he checks ''GRFULL v in all his gyroscopes He could avoid that unsatisfied dictionaries He finds lots of eanings tum feeling He could call the expert and ask a key J'Otate a prism roZt a missile spin but What do you call it when a widget goes round not twnble a gyroscope etc He gets the idea and round ••• Oh ROLLEX How do you spell of rotary motion but still he does not see al y it • • • R-0-L-E-X with one L Thanks a lotl reference to the word in connection with widgets One phone call with the results recorded in He knows perfectly what the word means in general the urgin of his dictionary and it's done but not the specific English tern as applied to The recipient of the translation for whom widwidgets At this point he doesn't need a better gets are a burning issue is given completely expert in the target language Nor does he need accurate information instead of vague misa more up-to-date dictionary there probably isn't leading or even erroneous information from an one anyway What he needs is a i n dget e pert almost right translation Now somewhere in this Agency or in the comHow about it Don't you agree that translators nmnity in general there is probably someone who need an Experts Yellow Pages Come to think of is an expert on widgets Perhaps he has a de• it a lot of nonlinguists could probably use_it tool Aug-Sept 7S CRYP1'0LOG • Page 19 IUllltl ilh W lllh 8111111111 1 8 llltt