Bryn Mawr Classical Review 95.03.05


Response: Neuberg on Golemo on Neuberg


Of course Karl Golemo (BMCR 94.2.11) is entitled to his opinions of my JACT Greek hypercard exercise stack system, but promulgation of incorrect information about it, without consulting me nor, apparently, having read the documentation, does no one any service.

Matt Neuburg's JACT Greek software (which is distributed without fee over the Internet) provides exercises designed to accompany the first sixteen sections of the textbook Reading Greek: the Joint Association of Classical Teachers' Greek Course.

What is provided is really an authoring tool whereby teachers can easily make their own exercises to go with the textbook; but as a courtesy, to save time and to get you started, I have also included the exercises that I have written with that tool and which I currently use. They can quickly be modified or replaced at will, to suit your own tastes or needs.

Each exercise includes...

This description treats a purely contingent fact as a given of the system. Each exercise includes whatever you want it to include. It happens that for my purposes I have found it best to computerise all the (b) and (c) exercises (Mr Golemo does not make this fact clear) from sections 1-16 of the JACT, plus all the vocabulary and paradigms the student is asked to memorise, plus some extra exercises where I felt the JACT did not provide sufficient practice; and I have included all this in my online release. But you are free to accept, reject, modify, replace any or all included exercises.

Using the Macintalk Pro system extension, the program can pronounce all of the Greek words and sentences in a variety of voices.... [But] certain syllables (especially where vowels adjoin) are wrongly pronounced and stress accent is entirely haphazard. This ... limits the useability.

These problems are due to the nature of the phoneme repertoire and encoding system of Apple's Macintalk Pro, which is oriented to English only; for example, no French -u- can be produced. If you look at the stack script you will see that (although it slows down the program) I have included the pronunciation routine, which translates Greek letters to Macintalk Pro phonemes, as HyperTalk rather than as an XCMD. This, as a comment in the stack script makes clear, leaves you free to study my routine and write a better one if you think you can. Instead of carping, let Mr Golemo set his hand to it.

There are also limitations within the main program. First, the responses judged correct by the program are too restrictive in some cases. For example, for the word A)POFE/RW, the program will not accept "carry off" or "carry away" but only "carry back" as a reply.

An odd use of "will not"; it will if you wish it. Mr Golemo scarcely notices that you (the teacher) determine what any exercise "will accept". Further, the correct answers which you have entered may incorporate a syntax permitting alternate student responses to be seen by the computer as right. In this instance, the teacher's entering "[I carry off, I carry back, I carry away]" would cause the computer to find any one of the three acceptable. I didn't do this myself merely because the JACT defines APOFERW only as "carry back".

Similarly in many instances the program does not allow for alternate spellings. For example, the program will accept only "saviour" for the word SWTH/R, which will be confusing to American students who are, of course, used to writing "savior".

Ditto: just replace "saviour" with "[saviour, savior]". My students happen not to be American, and the JACT says "saviour", so no problem has arisen locally.

This flexibility, of which Mr Golemo seems ignorant, is one of the system's salient features, and revolves around an XCMD, written in C, which evaluates the student's responses. The available syntax for the teacher's correct answer is simple but quite powerful: one may specify the answer as a single word, a list separated by commas, or a list separated by spaces (e.g. a sentence); may require fixed order or allow free order; may be strict or lax about accentuation; may supply equally acceptable alternates; and may even include optional items. Thus, for example, "[position, office], [start, }], [rule, }]" (as a definition of ARXH) tells the computer to accept such replies as "position" or "office" or "position, start" or "rule, position" -- but the reply "start" by itself will be regarded as incomplete. Further, the evaluation routine distinguishes between student responses that are right, wrong, or incomplete; it highlights the particular word that is wrong; and distinguishes between wrong absolutely and wrong in accentuation alone.

Mr Golemo mentions none of this, so the reader never hears about it -- and neither does Mr Johnson, who believes Mr Golemo, and so might be wrongly including my program when he writes: "Even more deplorable is the general ability of most programs to understand the most rudimentary synonyms in processing the user response. Our reviews are sprinkled with complaints on these sorts of inflexibility." But are those complaints justified? My program is quite flexible in evaluating the correctness of the student's response, and yet produces that evaluation with lightning speed.

Secondly, some will find the vocabulary offerings meager, since the program only uses the words from the Vocabulary to be Learnt section of the textbook rather than from the entire vocabulary of a given chapter. An option to include the wider base of vocabulary would be helpful.

This makes no sense to me personally; why train the student to learn words the book does not say to learn? However, de gustibus, and you can create or modify any exercise you like, so it seems to me this "option" is very much present.

These limitations, admittedly, are not very great.

Nor particularly real. What is "limited" is Mr Golemo's knowledge of the stacks.

One ... wants some scoring mechanism so that the students can get a sense for how well they are doing and so that the instructor can properly monitor the students' progress. (The advertising is somewhat misleading when it says, "Why waste time checking the results of the student's exercises when the computer can do it for you?")

The first sentence is personal and philosophical. As for me, I monitor my students' progress through quizzes, and feel that scoring mechanisms tend to be Big-Brotherish and off-putting to the student. If Mr Golemo asked me for a version of the stacks that incorporated scoring, though, I would write him one.

Second, I do not promulgate "advertising"; I'm not selling anything. I do occasionally distribute by email a notice reminding folks of the stacks' free availability, it does include that sentence, and the sentence seems to me true: the stacks do check the students' answers, and they have freed considerably my time as a teacher without loss in student achievement. If Mr Golemo has taught from the stacks and his experience differs, let him say so; but to speak of misleading advertising is inappropriate and offensive.

Mr Golemo has appreciated very little of what these stacks are up to, in fact or in spirit. He shows familiarity neither with the JACT textbook, nor with HyperCard, nor with the documentation. He has not appreciated the nature of my stack system, the extent of its customisability by users and, on request, by me, the nature and power of the evaluation routine, the fact that the stacks are free, or even, it seems to me, the fact that it is a computer program.

For example, his suggestion that I make the stack's ability to pronounce Greek more obvious and available to the student (a button rather than a menu) is a good one. But instead of trumpeting this as a shortcoming of the stacks, why not make the suggestion directly to me? I could implement it, for him if he is serious or in the online version, in a few moments. This isn't a frozen entity like a book; it is entirely mutable, revisable, and in-progress -- and always will be. That is part of the wonder of this new computer age.

It is right and valuable to describe to readers the available CAI packages; but to describe them so very misleadingly is dreadful, and to punish them is gratuitous. I know, much more accurately than Mr Golemo, that my stacks have shortcomings and what they are, and I am not averse to criticism; but such criticism needs to be appropriate to the spirit and nature of what is under discussion. I develop a teaching tool over many years, then share it at no cost with any wishing to avail themselves of my labour, and Mr Golemo greets it as he might the publication of a pompous $80 book claiming to solve the Homeric Question. Yet these stacks, which I use constantly, I have placed online, in their present state, and have continued to support and revise, neither for glory, nor for tenure, nor for financial remuneration, but from a motive Mr Golemo seems unable to conceive or respond to -- to be helpful. I really wish now I had not done so.

Still, those wishing to discover what the stacks are really like may wish to know how really to obtain them -- since Mr Golemo gets this wrong too. For ftp, use Fetch or a mainframe (not Anarchie); Fetch is best. Ftp to cantva.canterbury.ac.nz, and login anonymous. You will enter at root level /public; now, cd mac; then, cd classics (that is, get to /public/mac/classics). They are called [.MAC] and [.CLASSICS] in Fetch's listing. Then get jactgreekpt1.hqx and jactgreekpt2.hqx. For gopher, set your gopher to delphi.dur.ac.uk, and take the following path in: Academic Departments and Faculties; Departments Listed P-T; Theology; Theology and Computers; Software for Theologians; An Archive of a Selection of Software. You will then see appropriate folders.

While there you may also wish to download my Greek Verb Help hypertext app. This was not reviewed, though it is far more widely useful (and, from what I hear, popular), as it relies neither on Hypercard nor on any particular textbook.