During the past year we have been conducting a quantitative onomastic (the study of the history and origin of proper names) study of aboriginal Ecuadorian names listed in a 1559 colonial census. The names are valuable both as data on one of the few extant samples of languages later replaced by Quechua and Spanish, and as a clue to precolonial social organization. The census lists the names of 3258 native people from six villages near Quito, together with data on their residence, rank, politi- cal status, civil status, sex, age, and standing in household. In order to find relations between social variables and names or parts of names, co-investigator Grosboll devised some techniques for using statistical and computer software on nominal data. These may be worth the attention of CAAN readers, especially mainframe and minicomputer users.
The data was extracted from the colonial document, creating one record per person that included fields for first and second native name, christian name, and the social variables mentioned above. To facilitate data entry, a program was written for the Datapoint ARC System in a time-sharing language (DATASHAR/DATABUS, a form of BASIC) which created a screen format for rapid CRT entry and data verification. This program and others for formatted file reports and file updating were designed to handle the addition of new data fields; additions which were anticipated in the initial study design. The original programs were written in 15 hours, data entry for 3258 records took 20 hours, and file design modifications for 11 field additions took only 3 hours. The Datapoint System was used both for its availability to us, but also because of the ease of file handling made possible by DATASHAR/DATABUS.
Before beginning the statistical analysis, a number of sorts and reports were run on the file to give us clues as to basic data relationships and to create references for later analysis of test results. Though this type of file referencing can be handled by screen lookup and scrolling of the file, this is not a cost-effective procedure unless testing and analysis goes on continuously, making an on-line data file feasible and desirable.
These sorts led to the creation of several new fields: common name, Inca name, and initial morpheme. Because of variations in the spelling of names, codes were assigned to represent these fields. (Alternate spellings of names occurred as a result of variations by the original scribe, illegible handwriting, and/or splitting of names.) At this time, the data was transferred to the UNIVAC at the University of Wisconsin's Academic Computing Center for the statistical analysis. This and subsequent transfers were accomplished through telecommunications or magnetic tape backup.
The name or name element codes were then tested against other social variables using the SPSS subprogram CROSSTABS and the statistics of chi-squared, Cramer's V, Lambda, and the correlation coefficient. While the Madison Academic Computing Center offers both SPSS and BMDP to its users SPSS was selected for this project because of its ability to handle nominal data, which comprised the major portion of our data file. (Though BMDP was recommended for its smoother data entry, its utility seems to lie more with interval and ratio data. During the coming months, I will be testing out this package in another package, and would appreciate any comments on it from readers who have experience with it.) [Editor's note: The Loglinear Analysis Package in BMDP (4F) is an extremely powerful and modern way to handle multivariate nominal data.]
The creation of UNIVAC and SPSS control statements and their subsequent execution is facilitated at the Academic Computing Center by procedures called KEY and UDH. KEY is an interactive text editor used for rapid data entry and storage that allows modification and/or duplication of lengthy SPSS statements without time-consuming reentry. UDH (Unified Data Handler) is an easy to use program for file printouts, reorganization, subfile creations, and field changes. It also provides a means of creating an SPSS system file without the expense of running SPSS. Though the KEY and UDH packages are specific to the University of Wisconsin, most computing centers have similar utilities, and they tend to be unknown or underutilized by many researchers. Their use can make computer use less painful, particularly on the project budget. For instance, UDH was used to 'strip' the test file down to only the needed elements for this particular analysis, thereby reducing file size, storage space, and daily disk storage charges. It was also used to off-load files when not in use for periods of 5 days or more, again to reduce computer storage charges.
The use of the computer for simple data handling and statistical correlations made it possible to shorten possibly 3 years of manual sorting to less than 3 months by computer. But several other techniques were used that manually would have been extremely laborious. Just as we compiled a list of common initial morphemes, we also wished to compile a list of final morphemes. While initial morphemes could easily be grouped through an alphabetic sort, final morphemes are not as readily available. To do this, first and second names are concatenated (after a linguistic analysis proved the separation unnecessary), the name was reversed to read from back to front, and then sorted in alphabetic sequence. This was easily done using CHAIN (a boolean-process control language on Datapoint), and was completed in 15 minutes. Using the subprogram FREQUENCIES in SPSS, a list was made of each occurrence of a 2-letter to 6-letter name endings. From these counts, a list of statistically testable final morphemes was drawn up, codes assigned, and tests run.
A second example of tedious work made easy by the computer, was the search for consonant clusters found anywhere within the aboriginal names. A list was drawn up of 49 consonant clusters thought to be useful for detecting dialectal differences within the study area. In one day, a DATABUS program was written to do a character-string-search on each name for the presence of any of the consonant clusters. These 159,642 searches against the 35 character names were accomplished in 20 minutes on a mini- computer. From this, cluster frequencies were produced, and the statistically testable clusters were selected and assigned codes. These codes were dynamically entered into the record fields by the program, thereby avoiding tedious data entry for each new field. Surprisingly, the UW Computing Center did not have any readily available software to do this type of search. Though several utilities had a pattern-search capability, these were not intended for the use we proposed and would have proven prohi- bitive in time and/or computer costs. If any reader is aware of sophisticated character-string-search software suitable to this type of research, please send information to the address below. Contact person: Sue Grosboll, Dept. of Anthropology, University of Wisconsin, Madison, Wisc. 53706.
Brock University
Looking through the first number of CAAN I was struck by a significant lacuna -- Anthropology! Let me hasten to add that I am not referring to that methodological strawman of some of my innumerable colleagues, qualitative versus quantitative, nor will I harangue you with an ethnomethodology sermon. However, when anthropologists discuss the development or introduction of a technological artifact, the conversation inevitably swings to a discussion of the cultural beliefs and values that; underlie the item, facilitated or blocked its development/ adoption, and its impact within the culture.
Anthropology's discussion of computers has largely centered upon the physical characteristics of the technology and specifics for implementation; "find the technical information that can help us achieve our goals ..... help up with our work." CAAN; 1:1,1) This approach is fine for beginners. But, it seems to me that the group addressed by CAAN should be moving beyond this stage; discussing computers and our use as Sharp discussed steel axes and stone-age aborigines. In computer jargon -- the study of LOGIC (the beliefs and values that underlie the construction and internal operation of the technology) and INTERFACE (the group's perception and understanding of the technology and the effect of this upon the culture).
LOGIC -- Computers result from centuries of Western inquiry into the processes of 'reasoning' and 'knowing'. A good starting point is Aristotle's assertion that information (knowledge) can be assigned to fixed categories. If information can be categorized then it can be measured -- and the search for 'logic machines' began (c.f. Gardner, Logic Machines and Diagrams; U of Chicago, 1982). The basis of computers and their internal operation (both hardware circuits and software programs) is Aristotelian (Bolean, set) logic. This logic has as significant an impact upon a computer (input, processing and output) as the beliefs and values surrounding the construction of a stone axe have in its shape, composition, use and other properties. The significance of logic in determining the properties of a computer cannot be underestimated -- particularly as our 'steel axe', FUZZY LOGIC, has just begun to be introduced. Fuzzy logic attacks the Aristotelian underpinnings of our technology by refuting the assertion that information can be categorized with any degree of specificity. McKean and Dworetsky (Fuzzy Means to Logical Ends; Discovery 6:2;70, Feb. 1985) have one of the more informative discussions of this 'new technology' -- How many hairs does Fuzzy Wuzzy 'need' to be 'categorized' 'fuzzy'? Can you build a machine, write a program, process data, do a multiple regression or interpret results without fixed categories? Anthropologists have been asking themselves questions about the nature of their data, its processing and interpretation since the field's inception. It should be one of the first questions we ask of the technology we use.
INTERFACE -- How do we perceive and (ab)use the machine? Those of us who are SF buffs (some anthros and their children even write SF) will remember CALLOSUS, HAL, Positronic Brains and other favorite computers. The basis of all of these stories is the perception of machines by humans and the impact of this upon entire institutional and social systems. This is the meat of anthropology. Few, such as Gerry Gold (York Univ., Toronto) whose class studied a computer user's group, partake. Furthermore, as Alvin Wolfe (U of South Florida) pointed out in a paper delivered to the SFAA ("Electronic Ethnography"), we cannot expect traditional ethnographic methods to operate effectively in an electronic society. How can you study a society where millions watched "Roots" and made it part of their 'folklore' using methods designed to study a few thousand people in face-to- face contact? The computer opens up an unlimited variety of potential ethnographic data ranging from analyzing patient records to determine a medical staff's perceptions (Watson, 1985) to Alvin Wolfe's proposal to link an area's human services data banks and through different retrieval algorithms construct a community ethnography.
CAANews readers, as the computer 'experts' of anthropology, must begin to deal with some of these issues. As anthropologists we have the tradition -- the questions. As computer literates we have the knowledge and skills to discuss the questions. Finally (lest some accuse me of creating a new "Anthropology"), we have a variety of fields of interest; technology, development/change, cognitive, methodology, and theory. All of which we can use to do the anthropology of computers.
U. C. Irvine
Doug White and associates Mike Burton, Karl Reitz, and David Gregory (Irvine, Chapman, Dartmouth) are publishing an electronic journal on microcomputer diskettes in order to distribute various components of the cross-cultural database -- codes, codebooks, bibliography, sampling frames, atlas information, -- along with review of comparative research problems and computer programs. They have enough on hand for a quarterly diskette's worth of research materials for the next two or three years. The want to extend the journal format for other authors to contribute new research materials, guided instructional programs, programs for data analysis, and anything else that might be of benefit to the comparative research community.
Cost of subscription is $60.00 annually, plus a one-time database entry fee of $95.00. The latter fee provides the subscriber with a diskette that acts as a key to unlock (for any purpose: reading, printing, analyzing data. etc.) those files which are protected against copying (most are not).
The first four issues for 1985 will include the Standard Cross-Cultural Codebook, including over 700 variables from the various contributed studies using Murdock and White's Standard Sample, and over twenty sets of codes by different authors which make up the 700 variables, coded on 186 societies. This consti- tutes the central core of the cross-cultural instructional system which has been in place for eight years at UC Irvine, earlier in attenuated form at Dartmouth. Various computer programs for statistical analysis will also be reviewed (Leland Wilkinson's SYSTAT, retailing out of Evanston, Ill at about $400 is currently the hot program.)
The 1985 issues provisionally include the Standard Sample ethnographic bibliography, the new SCCS sampling frame (inventory and classification of well described world societies) prepared by White, materials on and discussion of other sampling frames, a special selection of SCCS codes of high reliability or demonstrated construct validity, and an abundance of new materials for the treatment of Galton's problem.
The 1986 issues are even more provisional as it is hoped that author contributions will begin to pour in (if so, additional diskettes may be added to our quarterly distribution). Plans now inlcude the Russian Atlas Narodi Mir (Peoples of the World as of about 1960, cartographically specific to the ethno- graphic pinpointing level), Joe Jorgensen's Westen North American Indians data, materials on comparative social change, and further material on cross-cultural sampling.
For subscriptions, write to: World Cultures, P.O. Box 12524, La Lolla 92037 - 0650. Diskettes will be mailed for nearly any PC or microcomputer format.
One of the exciting things about cross-cultural research, given its inherent richness in the opening of perspectives, is the cumulative nature of its databases. As new problems open up in this tradition, they can be related back to earlier theories, hypotheses, and concerns. What makes this possible is the limited number of ethnographically well-described societies. Whether by design (as in the SCCS or HRAF standard samples), or accident (any two large cross-cultural samples), the codes contributed by different authors offer cumulative information on common or intersecting sets of societies.
The electronic journal offers the vehicle for easy distribution (on 5 l/4" diskettes for nearly any PC) of the components of a cumulative database. Sample composition, bibliography, cumulative codes, cumulative codebooks, and sampling frames will be published for a variety of samples. The scholarly material that Murdock, I and other authors assembled on our standard sample are distributed by this means. Contributors of coded data and authors of other samples are invited to do the same. Several authors have already volunteered unpublished codes on which published studies were based. John Whiting is providing his standard-sample weather station data. Joe Jorgensen is making available his Western North American Indians material. I would like to see materials published in World Cultures from the 60-society HRAF sample and other standard samples. We encourage those authors to submit materials.
Biases of all kinds (sampling, ethnographer, informant) are a problem in cumulative as well as newly chosen samples. Cumulative databases can provide data quality control variables to check for possible sources of bias. Naroll's work remains a classic statement of the appropriate methods for doing so. Thanks to Ron and Evelyn Rohner, and others, we have such codes for the standard cross-cultural sample.
A micro-computer database of scholarly cross-cultural materials has other advantages, such as word processing applications. Since the cumulative SCCS codebook is provided as a text file, portions can be extracted and re-edited for special purposes. One might organize a project using a particular set of variables, or write an article using these variables, and want to make up your own codebook by modifying the original text file. The SCCS bibliography is also provided as a text file. With appropriate modification, it can be converted to a datafile, and sorted alphabetically by author for an efficient search of sources in your library. Other wordprocessing and database applications are easily envisioned.
Standard samples and cumulative databases are of great benefit in cross-cultural research. They are also of obvious pedagogical value. World Cultures is intended to provide ample materials to instructors for students to use in a variety of classroom applications. Personal computers provide access to data as well as cross-tabulations and statistical analysis.
We have found Lee Wilkinson's SYSTAT, a statistical and database management package available for most PC's from software dealers or from SYSTAT's office in Evanston, Illinois, to be nearly ideal for cross-cultural research. Its use is fairly easily taught for student use. Special purpose programmed instructional applications, however, remain to be developed. Authors of such programs are also encouraged to publish through the electronic journal.
I think the cross-cultural research community will find the electronic journal to be an exciting new means of networking our scientific discipline, providing broader dissemination of our research and instructional materials.
Oakland University
Recently Lee Sailer (Pittsburgh), Jerry Smith (U. of South Florida), and I were asked to look into the potential for computer communication to enhance the work of the Society for Applied Anthropology. If this request was made a year ago, I probably would have said, "All you need is money. Anthropologists don't have it, so forget about computer communication." However, microchip technology is moving so fast that the rules of the game are changing again. I suppose one should get used to this. Thus we find ourselves reevaluating computer use in general, and computer-based communication in particular, for the anthropological sciences. I am sure there will be people with other perspectives than ours, and if they would like to reply to these comments, we would be glad to print them. The general rule, "All you need is money." still seems to be the most important one, but what you can get for your money is changing fast.
Let us first consider why one should bother with computers in one of the most humanistic of the social sciences. The basic work of anthropology is discovering, disseminating, and applying knowledge developed within our particular field. Computers, machines for the logical manipulation of numbers and other symbols, are merely tools that may be used to assist this work. The significance of computer to the future of this work lies in the ways that they can be used creatively to enhance it. There are several basic areas of application at the moment: (1) improving the methodology of discovery, (2) increasing the dissemination of knowledge (3) making teaching more effective, and (4) improving the end applications of anthropological knowledge.
Perhaps the most fascinating area to me is (1) improving the methodology of discovery. Anthropology is a good social science, but it does badly need methodological improvements. We can see applications for computers in the area of (1) data collection, (2) data archiving and retrieval, (3) numerical analysis using a host of quantitative methods, (4) faster better qualitative analysis, and (5) systems simulation. There are others, and each one of the above areas has many existing and potential appli- cations.
The use of computers in teaching is a world in itself. We are beginning to see a link up between research and teaching, as students are given their first contact with the data of anthropology, qualitative and quantitative, through computers. Computers give them the chance to analyze the data immediately. Also the use of computers in applied anthropology is beginning and should be recognized. For example, Al Wolf's metropolitan data base puts social agencies in contact with each other, and Jeffrey Backstrand is making data available to community workers with microcomputers.
What we are concerned with right now is Area 2, increasing the dissemination of knowledge, in particular to other anthropologists. It seems to be the biggest problem area. It also seems to attract great enthusiasm and create deep pessimism. We anthropologists are not alone in facing the unnerving potential of electronic technology. One large problem in bringing the benefits of electronic communication to the poor academic sciences is that the potential benefits of electronic communication to business is so great that capitalist forces have move in to develop, and to charge handsomely for, the communi- cation services. Businesses benefit from speed factors much more than academic sciences. Academic sciences have to move slowly and accurately as they break new grounds of knowledge. News media often care more about how fast they can print something. The academic sciences are the most committed to the accuracy of what they communicate, and perhaps they have the least to gain from the speed factor of electronic communication.
For a moment let's look back on penned epistles in the days of horse and sail. People like Abigail Adams who wrote letters to her husband John, were rare in those days. Those who learned and who became involved in the creation of knowledge about the world knew how to communicate well. In a sense we social scientists of this generation are like those scholars and politicians of two centuries ago. We don't just want to talk as much as we can. We want to communicate clearly what we feel is important. Why then, do we want to fool around with electronic communication, packet switching networks, electronic mail, computer conferencing, etc. Why not the quill pen? My feeling is that one of the problems that Abigail Adams faced has gotten worse, not better since then. This is the problem of community.
The benefit of computer communication to the social sciences is different than its benefit to the news media or to corporations. It helps the social sciences create and maintain communities of people with like interests. Regular mail is good for sending scientific papers and letters that one might muse over for a month or so. However, regular mail is too slow for the kind of interchange that creates a community of interest. On the other hand telephone communication is too fast. Face to face meetings are a place where communities can be formed too, but they have to be organized, they last only a short time, and the best fit between interests, time, and temperament does not always occur. Computer communication does have a place in scientific group work. It can aid the leisurely formation of cooperating communities and can maintain that cooperation at a pace that can be adjusted to differing work loads.
Are communities, groups, of people really important to have in anthropology? Perhaps for the most traditional field-oriented anthropology of our forefathers they are not important, but for modern anthropology they can be very important. Today most of the significant discoveries in the other sciences are being made by groups. Anthropology is just learning the importance of team work, but it is very important. With team work average scientists can make outstanding discoveries.
The importance of computer communication in academic work is attested to by the recent rapid increase of computer networks on U.S. college campuses. Campus electronic mail and conferencing systems are being linked together by academically-controlled nationwide networks. The technology for doing this has existed for some time. The DOD ARPA network was the first national packet-switching network, but in it's early history it served only computer scientists, people with high technology R and D contracts, and scientists working in the more sinister areas of weapons technology. What is happening now is that the rest of the academic world is rapidly waking up to the value of computer networks, electronic mail and conferencing through computers. A cost barrier has been crossed. Technology previously used only by businesses can now be installed on campuses at reasonable prices and made available to many people there.
Developments such as FIDO for personal computers, BITNET, and MAILNET for university computers have lowered costs and increased the utility of computer communication in the last year. These developments call for a new look at computer communications in anthropology. For example a local microcomputer can run FIDO as a conference bulletin board during the day. At night when the phone rates are low, the FIDO machines known to each other as a network, call each other up and exchange messages for their areas. The costs can be quite low.
As part of this effort to utilize electronic communication the editors of CAAN would like their readers to find out what local and world-wide computer communication facilities they have around them. If you have an electronic mailbox, an anthro- pological bulletin board, or an anthropological conference which you would like to be known, write us. We will set up a directory of electronic mailboxes, etc. and publish it. For example you may work on a college campus with a mainframe computer that has a mail facility connected to MAILNET, ARPANET, CCNET, BITNET, JANET, or MTSNET. Get an account and open a mailbox. Then let us know the address. If you have a mailbox on EASY-LINK, TELEMAIL, MCI, CompuServe, the Source, etc. let us know, so that other colleagues can utilize these media.
By the way, our cost analysis has shown that MCI mail is currently the cheapest thing around and is approaching the cost of regular mail. I would like to thank Russ Bernard for first directing our attention to it. Their latest charge is 45 cents for a 500 character letter. The annual membership is $18.00 and larger volumes of data can be sent at $1.00 for 7500 characters. Their information phone number is 1-800-624-2255.
Department of Anthropology Southern Illinois University - Carbondale
In "Hard Comments on Hardware" in the last CAAN, it was said that "the 6500 family of CPUs is different from the 6800 family. They use different instructions and register banks and were developed by different companies." While this is true in detail, it does not accurately reflect the close relationship between the 6502 and the 6800 family that I had commented on in "Apples and Archaeology: An Amplification" in the same issue of CAAN. The relationship between the two CPUs lies in the fact that persons from the Motorola 6800 design team were among the founders of MOS Technology, the developer of the 6502. As Rodnay Zaks points out in From Chips to Systems: An Introduction to Microprocessors, "The 6502's bus organization, internal registers, and instruction set are all very close to that of the 6800. In fact, the 6502 support devices can be used with either the 6502 or the 6800" (1981:226).
As to 8-bit and so on, it is correct to say that the terms are used, especially by IBM, to refer to the internal data registers. This is what justified calling the IBM-PC a "16-bit" computer, or calling the Macintosh a "32 bit" computer. However, the term "full 16-bit" or the like is commonly used, more-or-less correctly, to refer to the total package. Many (see for example, the editorial "Chips as Runes" in the 8 April 1985 issue of InfoWorld) would describe the IBM-PC as "8-bit" and the Macintosh as "16-bit." This usage is more accurate, strictly speaking, than the inflationary advertising claims of the respective companies. Those who are new to microcomputers should not worry too much about these details. What is important is the actual speed of the machine which, as Dow points out, depends on clock speed, bus structure, efficiency of operating system code, and myriad other variables. Memory management tricks and better, tighter code for operating systems can give a 6502 machine like the Apple /// or Z80 machines the ability to address (indirectly) large memories and work as fast as some supposedly more advanced "16-bit" machines. However, even Jerry Pournelle of Byte, a long-time Z80 supporter, says that the market base for Z80 machines "isn't large enough, not compared to the IBM PC or Apple II" for much new software to be developed for that family (Byte April 1985, pp. 370-371). This should not deter persons from investing in, say, a Kaypro if it will do what they need with its current, very large body of software. Most people only use a few programs intensively anyway, just be sure what is needed is available now (not promised in the "near future").
The statement that "The speed of a program does not depend on the level of the language in which it is written if that language is compiled into relatively fast machine code" can be disputed. Compiled code is almost never as compact as programs written in assembly language, nor is the code usually optimized by a compiler in the same way that code written directly by a skilled programmer can be. Developers have sometimes used compiled high-level languages, but a word processor or operating system developed in this way is likely to be slow, despite the fact that the program that is actually executed is compiled machine code. The efficiency of the compiler is a key variable for the speed of the final product, of course. One high level language, C, does provide relatively fast compiled code in some versions, but even there careful assembly language optimization of bottleneck areas of the compiled code is critical in making programs run at commercially viable speeds.
Since I seem to becoming the de facto representative in CAAN of the "software" school, I will remark that a prospective microcomputer buyer would be advised to determine their software and programming needs and then find a machine that will run the required programs and languages well. I doubt that the average computer salesperson will be of much use in getting information on either software or hardware. I strongly recommend getting in touch with user groups of any computer that is being considered for purchase. If you can find users of a word-processor or data base, for example, be prepared for partisanship; but these people will be able to tell you better than salespersons can whether your needs can be met by a particular package or language. Most clerks simply don't have the time to learn more than a few general interest applications. Moreover, the salespersons will all too often push the buyer toward what they have in stock.
This section of the CAANews is devoted to reports of what anthropologists are doing with computers. Please send the information on what you are doing, your successes and failures, and what you are interested in to the editors. We will publish your reports with your name and address so our other readers can get in touch with you if there is a possibility of mutual assistance. The reports are going into a data-base that can be searched in the future if someone needs to find a person to help with a CAA problem. We will also publish comments here.
Please let us know if you would also like to participate in on-line conference to exchange this information. For this you need a terminal, a local network port, and at least $6.00 per hour.
WALTER WATSON made several interesting points in a recent letter. First, he joins the chorus that nags CAAN about being anti-apple. The editors both use other machines, so that we have no easy way to accommodate Apple's disk format. Sorry. We are waiting for some Apple-using anthropologist to come forth in support of Apple users.
Watson also says about SPSS-PC, "The PC version costs too much to operate in the micro setting. PC and SPSS data files are not even the same. It's making money because anything "micro" sells today." He goes on to say, "The loss in processing capacity and statistical variety in the PC is unacceptable to most researchers."
RICHARD RANDOLPH, at UC Davis, mentions in a letter that he has a Pascal program that computes the age at death of "cud-chewing critters" based on the extent of wear of the molars. He says it would be easy to modify the program to change from, say, cows to sheep or goats.
MIKE EVANS, of U. Florida writes from Micronesia. He reports in detail on an interesting dictionary project, plus some other stuff. More next issue.
LEE SAILER has new programs to offer. GROUPS is an enhanced version of a data analysis program used in Sailer & Gaulin, Amer. Anthro., March 1984. You all naturally recall their technique for quantitatively comparing an arbitrary set of group assignments to an interaction matrix.
B-EQ is a program that computes "structural equivalence", a concept now popular in the social networks literature (see Sailer, 1978, Social Networks). It only computes a subset of the concepts that programs from UC Irvine compute, but computes asymmetric versions.
These two programs are written in Turbo Pascal, come with examples and written documentation, are free, and are available on many common CP/M, IBM, and MS-DOS formats.
LEE SAILER has acquired an AT&T 3B2 computer for the Anthropology Department at Pitt, which will be used for database management and graduate student training. He craves contact with any experienced Unix users.
As far as we know, JANE STONE is the first Anthropologist to obtain academic employment in a Computer Science Department. Last year, KATHY GREGORY, PhD Northwestern Anthropology, was hired by the Computer Science Department at UC Irvine. Just out of curiosity, does anyone know of earlier such cases?
Some anthropologists already communicate one way or another via some form of Computer network. Read the above article on computer networking by Jim Dow for a discussion of this. We would like to know if you are already communicating this way. We plan to publish these numbers in an anthropology directory of computer mailboxes. Please drop us a message, or at least a post card. A brief list of those we know about:
Lee Sailer sailer@cmu-cs-c.arpa
74756,3514 CompuServe
Jim Dow 70150,266 CompuServe
Jerry Smith 70025,1674 CompuServe
Mary Douglas EIES
Russ Bernard R.Bernard ScienceNet
Dwight Read ibh7cab@uclamvs BITNET
Kathy Gregory gregory@UCI.arpa
Jeff Johnson B.Copeland ScienceNet
Doug White 70357,3251 CompuServe
Note that BITNet and USENet are often available at university computer centers. Check with yours!
Kansas State University
The World of Kinship, is a simulation of kinship systems in Microsoft BASIC, consisting of four separate programs plus a main menu that takes about 64K of disk space. The first program is an introduction to the kinship diagrams. It illustrates how the patterns of marriage, descent, and residence are displayed in the remaining programs.
The second program displays systems of first cousin marriage and enables the user to combine any one of the systems with different descent and residence rules. The user can display any combination of four marriage systems, five descent patterns, and four residence rules.
The third program permits the user to combine descent and residence with second cousin marriage systems, illustrating for students the results of combining different marriage, descent, and residence patterns. The fourth program, in contrast, designed primarily for the researcher, enables one to vary the input to the diagrams of cross-cousin marriage. One can vary the number of descent groups and the number of generations in the displays, giving the researcher a flexibility in analyzing the effects of combining the various patterns of kinship and marriage not available in the other programs. The researcher can simulate ten section systems, for example, and see what are the results of utilizing different residence rules.
The programs are written, at present, with a subroutine enabling the diagrams to be printed on an Epson FX-80. The World of Kinship is available in versions for the CBM with BASIC 4.0 and the IBM with BASICA. Now being tested at several universities, it is available on a trial basis for the price of a diskette plus shipping. Contact person: Martin Ottenheimer, Dept. of Soc., Anth., and S.W., Waters Hall, Kansas State Univ, Manhattan, KS 66506
David Howard Day compiled a bibliography of computer uses in anthropology dating from 1963 to 1983. It contains 66 entries with annotations and includes archaeological applications. Contact person: David H. Day, Monroe Community College, P.O. Box 9720, Rochester, NY 14623-0720.
My interest in the computer field concerns the computer assisted manipulation of unstructured qualitative data. More specifically, it concerns the use of computers to help anthropologists organize and access the data they collect in the form of unstructured responses to open-ended questions.
I have identified a program that can be used to this purpose. The ARRAS program written by Dr. John Smith at the University of North Carolina allows the anthropologist to search and organize very large bodies of verbal testimony.
Alexander Randall (Ph.D. Columbia 1977), anthropologist and former student of M. Mead has started a computer brokerage in Boston. One of the aims of the firm is to sell used computers to people in other countries. Randall is also interested in the problems of making computer technology useful in other cultures. This involves translation of software into an new cultural milieu, a problem for which anthropological knowledge is quite useful. Language translation may require a language specialist, but what does software translation require, perhaps an anthropologist. He has supplied us with a nice credo.
We believe that computers extend the mind the way telescopes extend the eye. We believe computers enhance productivity, giving each user the power of many hands. We believe computers help reduce tedium in the workplace. We believe computers give people freedom to express themselves, examine ideas, consider alternatives, and analyze possibilities. We believe the cursor is mightier than the sword, and that peace will prosper in a computerized world.
Randall's firm is called the Boston Computer Exchange. They provide a brokerage service for ordinary people selling and buying used computers. The address is: Box 1177, Boston, MA 02103. Telephone: (617) 542-4414
This new section of CAAN will present information received from people large and small, about items for sale. These announcements may not have the objectivity of a review. The editors will be asking the contributors for a small fee, presently $20.00 US. to help the non-profit work of CANN to continue. All of the products will be of specific interest to anthropologists. Announcements of non- profit publications, information sources, and data banks will not be included in this section. Computer programs that are being distributed at cost may also be described elsewhere.
In the issue of Practicing Anthropology devoted to microcomputer use, I described something of the history that led up to developing a computer program for entailment analysis -- studies seeking to describe networks of implicative or entailment relationships among variable features in social behavior, division of labor, attitudes and decision- making, cognition, language, or educational testing. Anthropologists typically describe behavior in terms of rule-governed systems, but other fields have not developed methods of analysis to match our descriptions. Since 1972 off and on, I have been tinkering with a program and method of statistical entailment analysis that would serve all of these purposes.
I should like to report now the work I have done over the past year in making entailment analysis available on an IBM-PC. I took the mainframe version, in FORTRAN, converted it to Microsoft's FORTRAN for the PC, and worked the program over to make it interactive. In the process I made major overhauls of the program, which now make it amenable to use by a wider set of researchers. Working with the microcomputer gave me a far better capability (at no cost other than my labor) of adding new modifications and finding bugs.
With the finished product in hand at the end of last summer, I was in a mood to retail the program to social scientists, so I set up the means of doing so by filing my own fictitious name statement and getting a business license from the state of California. So the ENTAILMENT Manual and IBM-PC Diskette are now available from CALIFORNIA DATA-LOGIC, P.O. Box 12524, La Jolla CA 92037 - 0650, for $195 (Calif.: Add 6% Tax). For CAAN subscribers, a 20% discount is offered.
To give an idea of what the SEA program does: given a dataset of dichotomous variables, it analyzes contingency tables for all pairs of variables and assembles the conditional relationship of type "if x then y" into entailment chains or structures with different levels of exceptions. Items are partially ordered by inclusion relations, separated by relations of exclusion or coexhaustion, or bound together by equivalence (mutual inclusion). The entailment network is equivalent to a Venn diagram of bivariate relationships. Cumulative Guttman scales appear as chains of (transitive) inclusion relationships. However, even in treatment of uni- dimensional scales, SEA does not resemble Guttman scaling in determining scalability or scale errors, but relies instead on pattern-detection methods, tests of transitivity, and comparison to a random baseline.
The final entailment structures, which result from the analysis, are optimal representations of the strongest transitive entailments, not attributable to chance, decomposed into nested substructures at various levels of exceptions. The user chooses a level of exceptions (and measure of transitivity) appropriate to the measurement considerations of the particular problem area. -- Doug White, Univ. of California, Irvine 92717.
ANTANA is a package for interactive multivariate data analysis. There are more than 15 modules for data and matrix manipulation, simple statistics, multivariate procedures, and graphics. Data input is optionally from any file or from the keyboard in a free-form format, and disk files, the keyboard, or the printer may be specified interchangeably for output.
Several utilities perform housekeeping chores like stripping labels, identifying and counting missing values, pruning and merging data matrices, various normalizations, reporting simple statistics, box plots, and ranking. These routines are useful for sorting and editing data, and they also will edit the (optional) label files used for formatted output of results. There is also a general and powerful cross-tabulation facility.
Matrix utilities such as multiplication, transposition, matrix-invert, and matrix-divide are provided for those who wish to perform or to teach fancier analyses. We also include a utility which will compute all eigenvalues and eigenvectors of a general real matrix. These matrix programs are useful in many scientific contexts outside of statistics.
Standard statistical routines include t- tests, contingency tables, multiple regression and polynomial regression, and comparative box plots. Other procedures are principal components and coordinates using the singular value decomposition (simultaneous "q-mode" and "r-mode" factor analysis), non-metric multidimensional scaling using the Kruskal algorithm, and cluster analysis. There are also several specialized programs for use in population genetics.
Graphics include simple character-graphics, which may be used like text in a document, and interactive high-density graphics for dot-matrix printers. We currently have drivers for the Nec 8023 (ProWriter), the Epson FX-80, and the Okidata 92. In addition, an interactive screen-graphics utility with screen dump is provided for the Zenith Z100 microcomputer. Source code is provided (in C and Pascal) for the graphics programs so that they may be customized for other printers and terminals. A scientific desk calculator and a technical text formatter are also included in the package.
Output may be unformatted, it may be formatted with no labels, with default labels, or with user-supplied labels. Our data-editing utilities allow simultaneous maintenance of both case and variable labels.
The package is written to maximize the computational power of microcomputers. We have avoided unnecessary overhead wherever possible. For example, nicely formatted output is easy, but it is not necessary and we do not impose it on users. We have been surprised at the capacities of this package. Our most complex algorithm is nonmetric multidimensional scaling, and we can do up to forty cases on a Kaypro II with 64K of RAM. Many routines, such as correlation, covariance, and others, process one case at a time, so an arbitrarily large data matrix may be used.
The price is $200 ($120 for payment in personal funds) and includes a tutorial manual of approximately 100 pages. The package is available for CP/M- 80 microcomputers in many diskette formats and for MS-DOS/PC-DOS versions 1.x and 2.x in the standard 5.25" format. The price for the manual only is $25, refundable with purchase of the software. Contact Henry Harpending, 1080 Cypress Rd, Bosque Farms, NM 87068, 505-869-3595.
It has occurred to us that some of the things published in CAAN, in particular computer programs, may have commercial value that the authors want to protect. Authors may retain copyrights to their articles if they feel that there is some valuable property in them. If you wish to submit copyrighted material, please include the form printed at the end of each issue of CAAN granting CAAN the right to publish it . If you would like to let certain people copy the material without contacting you, please include a publishable notice to that effect. This we also encourage. For example a title might look like this.
[Copyright 1985 Bronislaw Malinowski III]
[The author grants rights to reproduce this article and sections of it including computer programs for non-commercial use to persons holding degrees (BS BA MA PhD) in anthropology. All other rights are retained by the author]
[Please copy this form.] <Place> <Date> I <name> of <address> grant to the publishers of the Computer Assisted Anthropology News the right to print my original written work entitled <title>, including computer programs, if any, in their newsletter entitled the Computer Assisted Anthropology News, and I also grant them the right to copy and distribute all parts of the aforesaid work not containing computer programs in electronic form. I also grant them the right to distribute in electronic form up to <number> lines in total of computer programs contained in the work. I hereby declare that no part of this original work has been copied from any previous copyrighted source. The granting of no other rights is implied herein.
Signed: <signature> on <date>.
Material should be submitted to the editors at the addresses which follow. News, letters, notes, and articles are welcome. We prefer material submitted electronically on 5 1/4 diskettes or sent to User 70150,266 on CompuServe. Send short items for "Hunting and Gathering Tales" to Jim Dow. Subscriptions should be sent to Lee Sailer. Anything else can be sent to either co-editor.
Lee Sailer James Dow CAANews CAANew Dept. of Anthropology Dept. of Soc. and Anth. Univ. of Pittsburgh Oakland Univ. Pittsburgh, Pa. 15260 Rochester, Mich. 48063 Tel: (412) 624-3388 Tel: (313) 370-2430
We can accept diskettes in the following formats: A.B. Dick Magna II, Actrix (SSDD, DSDD), ATT 3B2, Chameleon CP/M-80 (SSDD, DSDD), Chameleon CP/M-86 (SSDD, DSDD), Cromenco CDOS (SSSD, SSDD, DSSD, DSDD), Cromenco IT CP/M (SSDD, DSDD), Datavue, DEC VT-80, DEC (Rainbow, Pro), Epson QX-10, Heath/Zenith Magnolia CP/M, HP-125, IBM-PC CPM- 86(SSDD, DSDD), IBM PC-DOS(V1, V2, SSDD, DSDD), IMS 5000, Kaypro (2, 4, 10), LOBO MAX-80, MAGIC, Morrow (MD-2, MD-3), NCR Decision Mate V, NEC PC-8001 (SSDD, DSDD), any Osborne, Otrona, any PC clone with MS-DOS, Sanyo MBC-1000, Superbrain (Jr. 35 track, 40 track, QD) Televidio (TS802, TS803, TS806, TPC-I), TI Prof. CP/M-86, Toshiba T-100, TRS-80 I Omikron CP/M, TRS-80 III Memory Merchants CP/M, TRS- 80 IV Montezuma CP/M, Xerox (820, 820-II (SS, DS)), Zenith (Z-90, Z-100 (SS DS)).
Those with formats not listed here are invited to call: we can probably figure out some special arrangement. This offer even applies to Apple users.