DE102012102157A1 - Surface texture recording with microphone - Google Patents



By attaching a Term status flag red, amber or green to each piece of information cf. Efficient collision detection for composite finite element simulation of cuts in deformable bodies. It is among other things in this respect that technical dictionaries generally fail to meet the expectations of professional translators. It may be desirable to have them displayed all at once, although they are not part of a virtual collocations field. Apart from linguistic iJ.

UK Kurven Features Comparison Tables


The ultra-soft fabrication feels simply delicate and heavenly on skin that you will hardly notice its existence! Being light and st. Heavy enough to avoid tipping or being pushed over Easy to clean; Machine washable after removing the rubber stoppers on the bottom Microwavable with medium power level after removing the rubber stoppers on the bottom Size: Red; White; Black Classy and functional; it fits right in with your home decor.

One for Pets The Kurve Raised Pet Bowl, Small, Black is a good product to be included in the shopping list, visitors can see product details by clicking the Read more button, Visitor can also read One for Pets The Kurve Raised Pet Bow kurven costumer reviews from customers that have already purchased this product, happy shopping. The soft jersey material of the fabric allows internal cooling within the top to keep you cool and fresh!

The racerback design adds additional support and durability. The stretchable, breathable material allows any body figures to easily slip on and off. With just the essential amount of compression, this tank top will limit any bounces or discomforts during your exercise sessions. As an alternative, this top also wo kurven. KURVE provides sexy seamless styles to a vast variety of niche markets; basics, dancewear, club wear, sportswear, plus size, shape wear, and accessories.

All of which is conveniently available under one virtual roof. Tubular knit outdoor mask to protect our adventurers from outside elements, such as wind, dust, dirt, and snow Perfect for hikers, bikers, runners, and yogis.

UK review Sonos post-boxing day deals: Post at Kurven Features Reviews. Establishing Great Look and Fee kurven. One for Pets The Kurve Raised Pet Bow kurven Features UK Heavy enough to avoid tipping or being pushed over Easy to clean; Machine washable after removing the rubber stoppers on the bottom Microwavable with medium power level after removing the rubber stoppers on the bottom Size: All of our fashions are designed in Lo kurven.

B Who will the users of the termbank be and what are their respective needs? C How can the termbank best reflect the setting in which it is to be used? Admittedly, a clear-cut separation of these aspects is impossible. For reasons of clarity, however, we will try to address these questions individually in the following sections, always taking into account the various needs of the four groups of translators identified and discussed in our last report cf.

In their daily work, translators inevitably come up with a lot of terminological information that cannot be found in any dictionary. Until recently, the standard procedure for "processing" that information was to stick it in a folder or at best put it on a file card.

What translators need is a tool that allows them to easily store valuable information for future reference and have it at their fingertips whenever the need arises.

Conceivably, all four groups of translators might also wish to make their terminology or parts of it available to colleagues or customers. Having all the information ready in a termbank makes it a lot easier to distribute print-outs or soft copies. Such a purpose would have to be supported by the implementation of comprehensive filter and printing facilities. As far as international organizations and particularly companies are concerned, the purpose of a termbank can be more complex than with freelancers or agencies.

Here, one of the main purposes of a termbank might be to centralize hitherto scattered terminological information and to make it available to translators, terminologists and anyone else who needs it. This would first help avoid double work, because one translator would not have to reinvent the wheel by tackling translation problems already solved by another, and second promote the consistent use of terminology not only among translators, but also by other parties involved in the communication process.

Who could these possible beneficiaries be? Here, the decision to invest in a multilingual termbank is rarely taken with only the translators in mind. Rather, the termbank is intended to support internal communication as a whole. In this context, it is important to note that the needs of the various users are not always compatible.

Terminologists, for instance, usually adopt a more systematic approach than do translators. While the translator's terminology work is mainly based on individual texts, the terminologist's task is to cover the whole body of specialized words relating to a particular subject.

Consequently, a termbank should be flexible enough to accommodate both approaches, i. To be precise, this means that apart from full-blown entries it should also be possible to save entries which contain only minimal but nevertheless important information without wasting disk space. Since their job is the production of documentation as opposed to translations , they are of course not so much interested in Transfer information as in the unilingual part of the termbank.

This again underlines the usefulness of making a distinction between these two types of information. In order not to overload the screen with information that a particular target group does not need anyway, it would be helpful to have a "hide" function for the transfer part.

This would help technical writers concentrate on the information they are looking for, namely meaning definitions, Encyclopaedic information and possibly even terms they are not supposed to use in their documentation. The different needs of varying users need to be seen in the context of the increasing interaction between translation, technical writing and LSP communication; this general development calls for a flexible termbank catering for a wide variety of processing needs in multilingual communication.

In fact, within companies, there are even more people who could be considered possible users of a termbank. Designers, for example, might want to check whether a particular term or acronym has already been used by someone else and, if so, what its meaning definition is. Again, this would be a major contribution to the consistent use of terminology throughout the company, insofar as a designer who wants to label a newly developed feature and finds out that the term or acronym he had in mind is already being used for something else can now think of alternatives before going public so to speak.

Unlike designers, who, like technical writers are mainly interested in the unilingual part, other employees might want to consult the transfer part of a termbank. Sometimes, draughtspeople, for instance, are encouraged to deal with seemingly simple translation tasks, such as labelling their drawings in different languages. Having a technical background, they of course hope to find straightforward information and do not wish to be bothered with descriptions such as transitive or intransitive.

At this point, it becomes particularly clear that the interests of the various users are not always compatible and that compromises will have to be made on the basis of the situational context. In some cases, external users could be given access to the termbank. This is very much in line with the increasingly frequent practice of companies, agencies and even organizations such as the European Union of keeping the number of their in-house translators to a minimum and having part of the translations done externally.

However, the group of possible external users comprises not only translators. As far as international organizations are concerned, the termbank might even be open to the public e. In any case, the logical prerequisite for this is that the termbank be technically accessible from outside. Alternatively, if external users are to be provided with print- outs, the termbank's software should feature powerful filter and printing facilities.

As soon as a termbank is accessible from outside, it may of course also be subject to unwanted manipulation. For this reason, an important requirement is reliable security functions and access restrictions. In companies it may be advisable to make a distinction between "public" information and confidential information that should only be available internally.

A major consequence is that a termbank must provide different views on the available terminological information. There are a few general requirements which are applicable to all user groups. Moreover, the user interface should be as easy to use as possible; it has been our experience that especially in international organizations and established companies, there are long-serving translators who are not necessarily familiar with computer- assisted translation tools. In order to help them overcome possible inhibitions, any termbank should be designed with maximum user-friendliness.

By this means, it will also be easier for newly employed translators to make themselves familiar with the termbank. An interesting observation we made in the course of our investigations has to do with the "terminological behaviour" of translators in relation to the language s they work with. In a group of translators all of them German native speakers , those responsible for French seemed to favour an exhaustive approach to terminology work, i.

The English translators, on the contrary, seemed to prefer a rather more minimalist approach, storing only information they had to dig out themselves because they were unable to find it anywhere else. The question as to whether this pattern can be generalized and whether it has perhaps to do with the way different languages are mentally represented could be an interesting subject of psycholinguistic studies.

Considered in the aggregate, it should have become clear that if a termbank is to meet the varying needs of its potential users, it must be highly flexible in terms of its structure and features. By this we mean factors that have a bearing on the termbank's design and structure, such as matters relating to the working languages and text types, clients and fields of activity, existing terminology and, last but not least, money.

From a general point of view, these languages determine what alphabets e. Greek a termbank tool needs to support. More specifically, the languages also influence the entry structure. In our investigations, we came across two typical patterns as far as the division of labour within a group of translators is concerned.

In the first pattern, each translator was responsible for only one language pair, e. German - English, while the translations into French were done by somebody else. In that case it would make sense to have the entries structured according to language pairs. In the second pattern, all translators worked with German plus at least two of the languages English, French and Spanish.

For them it seemed reasonable to create multilingual entries in their termbank, which, however, makes them difficult to read. It should be noted that these problems only occur when entry-oriented software is being used. As we shall see later, a relational termbank can combine these patterns. As far as text types are concerned, the question is to what extent the texts contain standard paragraphs or repetitive phrases.

If there is a high proportion of standardized text as in official resolutions, for example the termbank should feature a translation memory that saves the translators the task of typing in the same phrases again and again.

The work of agency translators is focused on the various clients or projects they are dealing with. Company translators need to structure their work according to the various lines their company deals in.

These criteria are imperative for translators, as they wish to organize terminological information with a view to retrieving it in accordance with the particular translation job which might be arriving on their desk at any given point in time. In general, people will have built up a certain stock of terminology in whatever form which will become the starting point for the new tool. The termbank should therefore provide facilities for the smooth import of such assets, especially if they are available in machine-readable form.

The users should also ask themselves if there is any internationally standardized terminology or other information that might make the first steps easier. While the need for a termbank is generally acknowledged within international organizations, industrial translators find it difficult to convince management of the usefulness of such a tool.

It is true that the setting-up of a termbank requires substantial resources software, manpower, maintenance ; however, the advantages clearly outweigh the disadvantages: However, many of the facilities described in the foregoing are not necessarily relevant to these target groups.

For this reason, it might be worth considering the possibility of offering a "TWB Deluxe" version for international organizations and companies, and a "TWB light" version for small offices. This way, all potential users could benefit from the advantages of a computer-assisted terminology tool.

As we have seen in the previous chapter, the termbank could leaving aside translators conceivably be of practical use to a large number of different people.

While we are trying to take their needs into consideration, too, one must be aware that within the framework of TWB, the main emphasis is clearly on translationally relevant information. Since then, however, a lot of water has flowed under the bridge and it will be necessary to reconsider them on the basis of our latest findings. Each category will be looked at in terms of contents, structure where applicable and possible difficulties the user might encounter.

Particularly thorny problems that might even relate to information clusters will be dealt with separately in chapter 4. As we shall see, a relational termbank as envisaged in TWB can help solve many a problem posed by entry-oriented databases, while at the same time reducing the amount of work to be put in by the user.

This section is of course not so much directed at the user as at those who have the know-how to programme and implement a state-of-the-art termbank. For obvious reasons, Administrative information should also be seen as a distinct part. It should be noted that the following overview is in no way intended to stress the importance of certain categories over others.

It is true that Transfer equivalents can be expected to be the most frequently consulted category. Given the appropriate circumstances, however, other types of information might be equally important. What has not been taken up elsewhere, and what is new about our approach, is the inclusion of Transfer Comments and Encyclopaedic information. In our empirical research, we have gained strong evidence that all of the information categories described below are of great translational relevance.

They should therefore be taken as a clear recommendation for any termbank. Those whose terminology work is still in its infancy should not be deterred by the wealth of this information. It is important to note that not all of these categories have to be filled for each and every entry. However, they should be available when a need for them is perceived. This can be particularly important when a term has different meanings in two or more domains e.

Also, being an attribute, this field contains machine-readable values, which means that the computer can collect and print out all the lemmata relating to a particular domain cf.

The problematic thing is the question as to what values i. In the course of our studies, we have come to realize that there is and can be no generally valid classification for each and every domain of human knowledge. Interviews with the head terminologist of the Bundesministerium fiir Arbeit und Sozialordnung Federal Ministry for Employment and Social Affairs have shown that their domain classification is quite incompatible with the classification adopted by any other Federal Ministry.

Any domain is multidimensional by nature and cannot be forced into a fixed, generally applicable classification system1. This means that before embarking upon terminology work, each user group should thoroughly analyze their field of activity and classify their domain s accordingly.

In this context, it seems to be advisable not to have too many domains on the same level, but to limit the number of domains and divide them into subdomains which can be freely combined with any first-level domain. The translators of an international plant engineering company, for instance, resolved to structure their terminology according to the general fields of technology, EDP, electronics, and commercial affairs on the one hand, and the fields of activity of the company's main trading partners, e.

They emphasized that it was the free combination of the general domains with the more specialized subdomains which best helped to support the overall structuring of their termbank information. This example also goes to show that any classification must be adapted to the specific needs of the user group concerned; inter- national classifications are hardly ever of use here. The original idea of making this field an attribute and giving definitions of the type a dog ; a member of the canine family had to be discarded for the sake of user-friendliness.

We found that for translators in everyday practice, it is hardly possible to come across such formal definitions. In addition, they are of little help, which is also true of the definitions of international standards.

A translationally relevant "definition" is a description or explication of meaning, which may take more than a single sentence. Often such descriptions need to be compiled from a number of technical texts. It scarcely needs pointing out that writing useful and understandable definitions is anything but easy. We shall therefore limit ourselves to saying that the definition of a technical term should at least give some indication as to the form and function of the 1 Incidentally, this confirms our findings with respect to a classification of the encyclopaedia cf.

It should be straightforward and to the point, because additional information can be found in the Encyclopaedic Unit to which the head term is linked. Unfortunately, this category also poses enormous problems with respect to representation within the termbank.

Not all collocations are as simple as the ones quoted above. There are others whose representation or lemmatization is anything but obvious: As this involves tricky problems of representation, the issue will be dealt with separately in chapter 4.

Wherever several transfer equivalents, synonyms or variants are listed for a term, they need to be differentiated with respect to their register e. US; in-house only topicality e.

It some cases, warnings of the misleading use of a term may need to be made explicit; in others, it may be important to indicate the termbank administrator's in-house ban on a given term and recommendation of another. For this reason, the termbank group has come out in favour of making a distinction between Short Usage and a Usage Comment, the former being an attribute with several values to be defined by the user cf.

Just like the domain classification, the Short Usage values should be defined on the basis of the user's needs.

These relationships can be of the following kind: There are of course much more complex relationships comprising more than two levels, such as the following: However, the representation of synonymy in the termbank is not as easy as this sentence suggests. Our first report cf. To a certain extent, usage information see above can help the translator to choose the appropriate term from the selection offered him.

However, in the course of our studies we have found that it does not suffice simply to include synonyms in the termbank by indexing them as lemmata and adding information on their usage. This is due to the fact that multiple compounding leads to comprehensive synonym clusters, such as the following which may be extended by a wide range of near-synonyms: Schadstoffreinigung Schadstoffemissionsreinigung Schadstoffemissionsminderung Schadstoftbehandlung Schadstoffemissionsbehandlung Schadstoffemissionsverringerung Schadstoffentgiftung Schadstoffemissionsentgiftung Schadstoffemissionsreduzierung The problem here is that all of these terms should be directly accessible, 1.

At the same time, it is impossible to elaborate all information categories for each and every synonym. What is more, it would not be possible to link the information categories elaborated for one of the terms to all other terms of the cluster either, as this would put excessive demands on the link network. Against this background, it was decided to opt for the so-called kernel relation, whereby lemmatized, but less elaborated terms are linked to a fully elaborated kernel term.

The user would be automatically referred to the kernel term for further information. However, it must be made quite clear that, in such a case, the information retrieved is on the kernel term and not on the search term itself. This could be supported by explicitly mentioning the kernel term in the definition as a reminder , e. Kernel relations should be handled very flexibly.

It should be possible to provide the less elaborated term with its own definition, if this differs from that of the kernel. The same is true of all other categories. Partially elaborated terms could share information where appropriate and have their own where they differ. The representational problems in connection with synonymy, near-synonymy and referential identity will be dealt with in chapter 4. TWC for three-way catalyst; NTB for non-tariff barrier Since variants are not always interchangeable, they must be accompanied by usage information.

They should all have lemma status, because if they are merely included as information for one particular head term, they might not be found.

The representational problems arising from this will be discussed in chapter 4. For this reason, it was decided to include context examples in the termbank, in order to save translators the time-consuming work of finding such passages in the original texts. With respect to this category, the danger is that it be misused as a holdall for all kinds of information which actually belongs to another category, especially longer collocations or phrases and suggested translations.

The user can only benefit from this category if it demonstrates how the expert would actually put it. To that end, it should contain typical examples to illustrate the use of a given term in a self-contained passage of at least one or two entire sentences. Hannonization The single market programme also eased the process of legislation by reducing the harmonization of laws and regulations to the minimum needed for health, safety and consumer protection and relying beyond that on mutual recognition by member states of each other's standards and regulations.

Ideally, the context examples should be accompanied by source information cf. Catalyst to catalyze vb , catalyser n , catalysis n , catalytic adj , catalytically adv. It might even contain "adopted" members of the family: Smell smell n , to smell vb , smelly adv , olfactory adj , olfaction n. This sort of information has turned out to be extremely helpful, especially with respect to the production phase.

Note that in order to avoid misinterpretations, the family members should be accompanied by an indication of the appropriate word class. Here again, the question is whether the information should be provided in a free-text field which means that those to be directly accessible would have to be entered twice or whether each family member should have lemma status which could lead to an inflation of links. Again, as in the case of synonymy, the kernel relation seems to be the appropriate solution.

It offers the advantage of lemmatization while avoiding a possible breakdown of the system due to excessive interlinkage.

It should be an attribute containing values defined on the basis of the user's needs cf. In combination with the domain attribute, text type will provide the user with helpful information and selection options. Short Grammar contains information given in any useful unilingual dictionary, i. In English, macrokinetics although a plural form goes with a verb in the singular: This need not only apply to irregular cases. Users who are not familiar with foreign languages but wish to use the multilingual termbank cf.

When several equivalents are given, they need to be accompanied by Short Usage values. The transfer relation is often far from straightforward, so that information in the form of a Transfer Comment is required. While in German the fuel consumption of a vehicle is expressed in liters per km, English uses the proportion of miles per gallon. In the tests we ran, users unanimously confirmed the usefulness of Transfer Comments. TCs cover a variety of translationally relevant problems.

In order to avoid confusion, they need to be categorized and well structured. Possible criteria for categorization could be the following; cross-cultural differences numbers and figures LSP conventions The categorization must clearly be guided by the user's needs.

For greater clarity, the TC could be divided into three components: Die englische Verwendungsweise ist oft weniger prazise. Apart from classification and structure, the following points should be considered in the elaboration of TCs: However, while corning up with the transfer equivalents, they need to be linked to the respective source terms.

In order to help the translator to quickly solve problems of comprehension and production, this information is divided into small, manageable units so-called Encyclopaedic Units, EUs focusing on particular aspects of the domain in question and covering a number of terms so-called Encyclopaedic Terms, ETs that are interrelated. The ETs are lemmatized in the termbank and it is via these terms that the EUs can be accessed.

Contrary to our original expectation, the Encyclopaedia should not be envisaged as an introductory tool for translators who have no experience whatsoever in the corresponding domain. Texts for the uninitiated reader must be written in a redundant way, very much in the style of a textbook. However, this would run counter to the purpose the Encyclopaedia is meant to fulfil, i. One may consider providing some units which explain in detail the context in which an aspect of a domain e.

More important, however, terminologists and translators should concentrate on providing concise EUs, perhaps accompanied by bibliographical references on introductory literature for beginners. Our studies and interviews have shown that what users want is short, problem- oriented information on domain-specific technicalities; what they do not want is to read long passages of text before detecting the piece of information they are looking for.

Consequently, the EUs should be concerned with individual problems and be clearly structured. In fact, we have found that in each domain there are certain aspects which generally confront the user with difficulties. Even experienced translators have told us that they keep looking up certain pieces of information time and again.

Such aspects must be identified and stored as EUs. In the field of catalytic converter technology the three major chemical reactions in a catalytic converter are a case in point; they convert carbon monoxide CO and hydrocarbons HC , by means of oxygen, into the harmless chemical substances carbon dioxide C02 and water H20 , and nitrogen oxides NOx into nitrogen N2 and carbon dioxide C The conversion of harmful substances into harmless ones is a major aspect of the catalytic converter and mentioned in many a technical text on the subject.

The user should therefore be able to view the chemical formulae at a single glance: Moreover, we suggest that the type of information be explicitly stated to increase user-friendliness. An EU would then take the following form: This can serve to elaborate a classification system for the Encyclopaedia, e. However, a Chinese proverb has it that a picture sometimes says more than a thousand words. With a view to providing more user-friendliness, one should therefore seek possibilities whereby graphics and diagrams can be made part of the Encyclopaedia.

A term might be entered preliminarily, because the user is not yet sure about its reliability but nevertheless wants to retain it.

It should be left to the users' imagination and needs as to what labels are preferred e. With the help of this category, the terminologist can easily retrieve all lemmata which have yet to be approved. This may be instructive as to how reliable the information is. What is more, however, it indicates the person to address in the event that the user wishes to make any comments.

This category is useful only if the user knows the persons concerned. The Terminologist field may be relevant for all information categories. Depending on how authoritative the source is, it also indicates the degree of reliability. Should the translator receive a negative feedback on a particular translation, the critic can be referred to the source.

Source information can be attached to any unilingual, Transfer or Encyclopaedic information category such as equivalents, synonyms, collocations etc. The field should be unlimited in length; for most sources, however, it will suffice to give a short form whose full form can be found in a list of sources. For external users, this information is of course not very useful and should therefore be hidden. Ideally, source information should be combined with bibliographical references.

It is probably best to have a maximum of two dates: Just like Terminologist information, the Date category contains useful information, which is not necessary at every given moment; it therefore does not need to be displayed automatically. It must be possible, however, to look it up via the retrieval window. There might even be a single Administrative information button, which can be clicked and then delivers both the Terminologist and the Date field.

They may even serve as a basis for work with currently available commercial terminology software. However, these programmes have an entry-based structure and confront the user with a number of disadvantages: Some of them only provide a pre-configured structure that cannot be adapted to the user's needs. In spite of the greater flexibility this allows, it has a major drawback. Depending on the individual layout arrangement, information categories may relate to either one lemma, or several lemmata, or even to the whole entry.

Confusion on the part of the user as to which information refers to what lemma is thus preprogrammed. What is more, users are rarely capable of making full use of the potential of an open system.

We witnessed that they tend to get involved in a long process of termbank configuration and endless negotiations, trying to gain the general agreement of all eventual users. They sometimes end up with a cycle of different prototype versions, time-consuming, belated modifications and no clear concept for the users to comply with. The potential for irritation is exacerbated in both systems by the fact that they offer multilingual entries.

On account of their complex structure, such entries are difficult to read. Another liability common to such systems is that information relevant for several entries e. Furthermore, there is the difficulty of accounting for complex asymetrical interrelationships, as, other than the relational database, entry-based termbanks cannot establish strictly bilateral relationships for every pair of a cluster of synonyms or equivalents.

Thus it is not possible to make it clear that in a cluster of synonyms, terml may be synonymous with term2 and term2 with term3, but that term3 is not synonymous with terml cf. In the same way, it is far from easy to deal with the phenomenon of bilingual divergence cf.

Finally, a negative feature is that, in some cases, fields are restricted to a limited number of characters. The termbank designed in the framework of TWB, by contrast, compares favourably with its modular and relational structure.

By modular we mean that a distinction is made between Unilingual information and Transfer information and Encyclopaedic and Administrative information , the advantages of which have already been described in chapter 1 cf. The main difference to available termbanks, however, is the relational model underlying the database as seen by the user. The relational approach means that the terminological data as a whole is treated as a set of entities with relationships that hold between them.

Lemmata Attributes Free-text categories Any lemma, i. The set of values is agreed by the users and allows the computer to filter out specific information. For instance, the machine may be asked to collect and perhaps print out all lemmata in which the attribute Tenn Status has the value Red. Free-text categories, by contrast, may contain any number of any characters. However, it should still be possible to select, say, all Definitions which contain the character string cylinder in their free text.

The following list shows what type of entity the information categories described in section 3. The following relations are possible: Thus a lemma can be linked to several free-text fields, each containing different types of information: A Definition may, for instance, be related to various synonymous lernrnata cf. A Collocation can be linked to more than one of its constituent terms and a head term to various collocations. In this way, information does not need to be entered twice or even several times, which considerably reduces redundancy in termbank storage and the amount of work to be done.

It is not stored as a full entry, but nevertheless appears to the user as though it were. All the other categories giving information on that particular term and linked up to it will be displayed on the screen in the form of an entry. Similarly, while the contents of the Definition field, i.

Which entities exactly are displayed as a virtual entry should be definable by the user cf. We quite realize that this is only a general description of the underlying database model. However, this section was meant to give the reader a general idea of what goes on behind the scenes. The important point to realize it that the TWB concept may spare the user a cumbersome trial and error-like configuration process by delivering software with an appropriate database structure and a well defined and ready-to-use interface - both effectively geared towards translation purposes on the basis of a careful analysis of current translation practice research in LSP theory the comparison of existing termbanks extensive user requirement studies accompanying the design process and a long evaluation phase testing several prototype versions.

The user would thus be able to concentrate on terminology work proper. The flexibility would not confuse him in the process of setting up the termbank, but might come into play at a later point - i. It may be regarded as a kind of "shopping list" compiled on the basis of our own experience and the wishes expressed by the translators we spoke to during our investigations cf.

Since we are concerned with a termbank for translation purposes, we will not discuss the general principles of user-friendliness window technique, mouse operation, use and abuse of colours etc. The actual implementation of some of these features may perhaps give today's computer programmers a headache; but there are always those Nintendo whizz-kids The users should be given the option of adapting the entry format to their needs. It would be helpful if each information category were displayed in its own window, which could be positioned by the user anywhere on the workspace.

Conceivably, there could be three entry formats: All translators have pointed out the importance of being able to store an unlimited number of characters in each information field, and also of having an unlimited number of free-text fields.

Certain commercial programmes impose restrictions in this respect, and these hamper efficient work. As regards characters, the termbank should equally support "exotic" alphabets, such as Greek, Arabic, Cyrillic etc.

Programmers might be relieved to know that there is no need to develop sophisticated routines for the reversal of transfer directions. Translators must be aware as most of them are that the information they feed into the computer has to be unidirectional and language-pair specific, so that simply to reverse the direction does not make much sense.

Indeed, there is no way around the problem of elaborating information for each transfer direction individually. At the same time, it is of prime importance that navigating between languages be as easy as browsing Unilingual information cf.

Finally, as has been pointed out before, any termbank intended for a group of users must feature extensive networking facilities to support in-group communication. This clearly shows that there can only be an individual "user standard", and not a standard as such.

It is therefore of the utmost importance that the input of data be as easy as possible. A lot of terminological information is and has to be elaborated in the course of translation. The translator must be able to quickly access the termbank from the word processor, enter the appropriate information, and go back instantaneously. This should be the procedure for straightforward entries. However, sometimes the elaboration of an entry takes more time than the translator can afford to spend at that particular moment.

For such cases, it would be helpful to save all new entries as having the status Red by default, unless specifically stated otherwise. Alternatively, there could be some sort of electronic notepad where provisional information can be jotted down for later elaboration a similar function is envisaged in TWB under the name of "private termbank".

This would be particularly helpful for users who have only read access to the termbank and come across information they would like to comment upon. Their contribution to the termbank's quality should not be underestimated see also the section on updating in ch. Each time the user hits the Enter key to save a new lemma, an automatic routine should check whether this lemma or a similar one already exists in the termbank and inform the user if such is the case. This means that if the user types in, say, air- brake, the routine should find airbrake or air brake providing they are lemmatized, of course.

At the same time, however, the user in quest of a particular spelling, for example, should be able to "fine-tune" the search pattern. This jump facility would allow the user to derive maximum benefit from the cross- references between individual entries see also the section on recorders and the interplay of multilingual information. I They also pointed out that the on-line help should be problem-oriented rather than structured in a top-down manner.

In fact, certain software manufacturers have since I I included special sections in their manuals containing questions frequently posed in their hotline services. This could be the starting-point for problem-oriented on-line help. It was even suggested that one could have a quick-reference help function for experienced users and a comprehensive one for beginners. For companies dealing with confidential information, the termbank should provide a facility whereby access to such information can be restricted to certain users.

Quite obviously, the definition of different access levels is even more important when it comes to controlling the process of terminology manipulation.

The provision of access restrictions together with a variety of freely combinable filter options helps cover the need to extract selected parts of the available termbank information. Finally, the termbank should feature comprehensive printing facilities allowing the user to print out any part of the termbank without having to rely on additional printing software.

Sometimes it is simply much more convenient to work with hard copies, especially when excerpts are to be distributed to other people or small parts of the termbank are to be updated. Conversely, the administrator s might wish to inform the users about any changes made or about newly implemented features.

In view of the astonishing speed of technological progress, the termbank should provide facilities for a smooth software upgrade. This is of particular importance for international organizations or big companies whose termbank serves a large number of users and therefore cannot be switched off at a moment's notice. One of them is the "double checker", which searches for double entries as already mentioned in section 3.

However, the user should be able to apply this checker at any time, not only when a new entry is being saved. That way, the termbank's contents could be streamlined from time to time.

In this context, which affects the whole complex of homonymy and polysemy, it is important to note that the double checker should realize that suspension in the technical sense and suspension in the legal sense are in no way doubles, but belong to different domains.

From a technical point of view, this should not be too great a problem, because the lemma suspension would be linked to two different domain attributes and two different definitions. In those very rare cases when polysemy is found within a single domain, the different meanings will be dealt with in a single definition, so that there would be no major problem in that regard.

There might be cases, however, where a polysemous term has two meanings within the same domain and where it would make much more sense to have two separate entries.

In such a case, a mini-hitlist of definitions should be provided for, which comes up with the term, and on the basis of the definitions, the user may then select one of the entries relating to the different meanings.

This is a useful way of handling the retrieval of intra-domain polysemy; naturally, the "double checker" should always be designed as an interactive routine, as double entries within the same domain are here deliberately placed and should not be automatically removed in the process of updating. Another useful tool would be a checker comparing a given text with the termbank's contents. Such a comparison checker could be helpful from various points of view, i.

As far as orthography is concerned, the termbank should provide a tool that makes it possible to combine its contents i. The target text could then be checked for misspellings and, where appropriate, alternatives could be offered. This is particularly important with respect to navigation. Sometimes the user might explore the termbank's contents by making use of the cross-references. A history recorder could retain, say, the last ten steps taken and display them to the user perhaps by double-clicking on a tape symbol when the need arises.

By selecting any lemma in the ensuing hitlist, the user should be able to go back to previous entries. Recording could also be useful in another sense, namely for finding out if there are any terms that users keep looking up without finding them, either because access to the particular piece of information is not obvious enough, or else because the term is not part of the termbank in the first place.

It might be fruitful to consider the possibility of integrating such programmes into the TWB termbank; after all, these tools should go hand in hand. One of them is the translation memory as has been developed within the framework of TWB, i.

The source text must be divided into segments, usually sentences or parts of sentences. When the translator activates a text segment, it is looked up in the translation memory.

If it already exists, its translation is retrieved and displayed as a suggested translation which can either be taken over in the target text straight away, or be adapted beforehand.