• No results found

6  Extended activity analyses and design of Phrases 2

6.2  The content of the new version

6.2.1 The activity module:  Shopping ‐ overview 

With the new model, new activities could be added to the system, but for research purposes it was still the activity Shopping that was looked into. The new recordings provided 1 hour and 38 minutes of new data from carpet shopping (50.48 min), camcorder purchase (10 min), bikini shopping (7 min) and supermarket shopping from the GSLC (31.19 min).

The new data, together with the data from the games and food shops + role-play evaluation led to a new activity structure for the shop vocabulary. This structure was in some sort of sequential order, but it was still presumed that many of the sub-activities and actions between entering the store and exiting it did not always come in the same order, something that was also supported by Tykesson-Bergman’s research (Tykesson-Bergman, 2006).

Vocabulary structure for shopping: Deciding where to go, Shops, Shopping list, Meeting people, Finding your way in the shop, Asking for item, Asking for information/help, Size, Evaluation of size, Colour, Evaluation of colour, Price, Amount, Numbers, Properties of merchandise, General evaluation, Hesitation about purchase, Buying, Bag/trolley, Paying, Money.

123 

A decision was made that each group should include 18 expressions or less. It was still important to include a variety of expressions, in order to find out which ones were preferred by the participants who were going to try the vocabulary. At the same time it was considered important to limit the number of expressions in each group, to not put too high a cognitive load on the users.

Table 6.2. The origins of the expressions in the extended shopping vocabulary 

Where to go    Size Properties of merchandise

                   

                   

                   

   

Shops  Evaluation of size General evaluation 

                   

                   

                   

   

Shopping list (empty)  Colour Hesitation about purchase

                   

                   

                   

   

Meeting people  Evaluation of colour Buying

                   

                   

                   

   

Finding your way  Price Bag / trolley 

                   

                   

                   

   

Ask for item  Amount Paying

                   

                   

                   

   

Ask for information/help Numbers Money

                   

                   

                   

 

  Games and food     New recordings GSLC in general  Own idea 

The content of the vocabulary came from different sources: GSLC Food and Games shops, the new recordings, the researcher’s own ideas grounded in experience and from the GSLC in general. The latter expressions originated from the new structure and the researcher’s ideas about what would be useful, but the expressions were checked against the language corpus in order to find utterances that were as frequent as possible and that would sound natural. Table 6.2 gives an overview over the origins of the expressions in each group. No group is without elements from recordings of natural speech, except the shopping list that is supposed to be filled by the users. The groups that contain requests for items or information, buying and paying are the ones where most expressions have to do with direct interaction between customer and shop assistant. Many expressions for these groups also come from the new recordings, where this kind of interaction also took place. Other groups are more oriented towards interactions between customers shopping

124 

together, such as general evaluation, evaluation of size and colour and hesitation about purchase. Most of the additions that stem solely from the researcher have to do with words that may be less frequent, but crucial when needed, such as names for specific shops, colour, size, numbers and money.

6.2.2 The addition of a quickfire module 

The new recordings provided the activity related part of the vocabulary with additional actions and expressions, but they were still related mainly to shopping. Many of the expressions derived from the shopping corpus, including the new recordings, were relational or directed towards “small talk”, and although both relevant and useful during shopping, such expressions are not specific for service encounters but can be used in almost any situation. One important step towards a more universal vocabulary for Swedish was to provide a special quickfire module for expressions of this kind.

In order to not invent the wheel again, it was important to see what quickfire modules looked like in other systems, while at the same time grounding the module in spoken Swedish. The Contact prototype (File et al., 2003; Todman et al., 2008), as well as the TALK system (Todman, 2000) were used as models for the kinds of expressions to include, but the expressions were taken from the Gothenburg Spoken Language Corpus, GSLC. The quickfire module in the Contact system includes expressions under the following 21 headings: Agree/Disagree, Good/Bad, Maybe, Interrupt, End, Change, Yes, Dunno, No, OK, Thanks, Hi, Wait, Oops, Hedge, Question, Don’t understand,

Feedback, Sympathy, Curse and Saying. The groups in TALK were similar, and both systems were well grounded in research as well as AAC practice (Todman et al., 1994), so using them as some kind of reference seemed a wise thing to do.

The process of selecting the Swedish expressions to include involved examining the 235 most common words, word pairs, word triplets and word quadruplets from the GSLC, compiled in the book “Talspråksfrekvenser” (Spoken language frequencies) (Allwood, 2000).

The total amount of tokens/words in the corpus at that point was 1,263,408. To begin with, 49 words, 63 word pairs, 70 word triplets and 46 word quadruplets were found to be meaningful, able to stand by themselves and falling within the range of expressions needed for the quickfire section. In a number of cases the meanings of several expressions were the same or similar. During the selection process an expression that could have multiple meanings was then favoured over expressions that were more limited. Attempts were made to sort the expressions into the Contact quickfire structure, resulting in some groups getting 40 expressions, some 2, 7 or 17 and some none. Based on the expressions found in the corpus, the content of the Contact quickfire, the phatic expressions used in the role-play sessions with version 1 of Phrases, and an attempt to include discourse markers for taking and handing over the turn, a quickfire structure with

125 

11 groups, each containing 12 expressions, was created. A twelfth group was used for the most common words/expressions that were to be available at all times.

In the quickfire module the expressions had the functions that can be seen in table 6.3

Table 6.3. The functions of the expressions in the quickfire module of Phrases 2  Menu button  Functions of the expressions 

Får jag (May I)  Turn taking, Ask somebody to wait, Request 

Men (But)  Continued turn taking 

Hej (Hello)   Greeting, Express joy over meeting, Prepare closing, Wish the other well, Wish to see the other again  

Tack (Thank you)  Thanking, Thanking for something, Congratulating, Express that something is good/enough, Reply to thanks 

Bra/Dåligt (Good/Bad)  Express satisfaction, Express dissatisfaction, Express regret 

Jaha (I see)  Give feedback

Oj (Oh)  Express alarm, Express surprise, Express dislike, Apologize, Admit mistake, Express pain

Ja (Yes)  Confirmation, Agreement 

Kanske (Maybe)   

Express hesitation, Hesitate about confirmation, Hesitate about denial, Reject suggestion 

Nej (No)  Denial, Rejection, Regret, Dissociation 

Väl (I think)  Positive turn closing, Neutral turn closing, Turn closing question 

Snabba ord (Quick words)  Positive, Neutral, Negative 

Since the quickfire module was constructed through a mixture of quantitative data, experience, examples from other sources and linguistic intuition, the process is hardly replicable and the section has to derive its validity through testing and use rather than from how it was constructed.

A list of all the expressions in Phrases 2 can be found in appendix E.

6.2.3 Prototyping with Toolbook Instructor 

At the stage were the new version of the vocabulary was constructed, planning was also made for the evaluation of the vocabulary. That individuals who were experienced users of AAC systems were to be involved in the evaluation was one important feature.

Another one was to find a way to compare the rate of selecting the pre-stored phrases to the rate of writing the same phrases with a keyboard. A logging system was envisioned to help accomplish this, but software that could log keystrokes was hard to build on, since the computers considered them to be spyware, an inconvenience that was hard to get around. There was also a need for software that could contain the new vocabulary and be used in new role-play sessions and testing that involved participants without speech impairments, who were to complement the evaluations involving experienced VOCA

126 

users. A decision was then made to use Toolbook Instructor 6.5 (later 9), an as a prototyping tool, since this authoring tool was well known by the author and had the possibility to include logging to the applications created with its help. For the participants who already were using VOCAs, the new vocabulary would be added to their existing systems.

Figure 6.2 shows the prototype software, containing a text-based version of the Phrases 2 vocabulary. Used together with a speech synthesis that can speak the contents of the clipboard, the prototype software functioned as a VOCA in a Windows computer.

Figure 6.2, The prototype software with the vocabulary Phrases 2. 

6.2.4 Properties of the prototype software  

The prototype software was built to contain the second version of the vocabulary Phrases.

6.2.4.1 Structure 

The quickfire module in the prototype software (at the upper right hand of the screen) was always visible. So were the 18 menu buttons of the activity module (at the upper left).

The 18 activity related expressions could be replaced by an on-screen keyboard, but could be brought back with one click on one of the menu buttons.

In the quickfire module, 12 expressions were always available with one click. When one of the menu buttons was selected, twelve utterances that were related to that menu button were displayed. When an utterance was selected, it got spoken, and then the message buttons changed back to display the twelve default messages again.

The prototype software had an on-screen keyboard that could be brought out with one click, and also be hidden with one click when the user presses a menu button for an activity. The quickfire expressions were possible to use together with the keyboard, and as a default they only got spoken, not sent to the message field.

127 

The prototype software contained six main groups of activities, that each could hold 12 sub-activities. For this reason, the shopping vocabulary in Phrases 2 had to be divided into two main sections (when used in the prototype software).

Figure 6.3. The prototype software with its on‐screen keyboard 

The activity called Affären (the Shop) contained the prototypical shop activities: Finding your way in the shops, Properties of merchandise, Asking for item, Asking for

information/help, Shopping list, Price, Amount, Numbers, Buying, Bag/trolley, Paying and Money.

The other part was called Handla och Tycka (Shopping & Evaluating) and contained the rest of the sub-activities. One part had to do with actions preceding or surrounding the actual shopping: Deciding where to go, How to get there, Shops, Meeting people.

Another part had to do with decisions and evaluations during the shopping, subjects that could be addressed by a shopper and his/her companion: Size, Evaluation of size, Colour, Evaluation of colour, General evaluation and Hesitation about purchase. Ideally, these sub-activities should not have been separated from the other sub-activities in the shop, and when the structure was included in a pre-existing vocabulary without the same restraints, they were not. 

The use of colour 

In the prototype software colours were used to discriminate between the menu buttons and the buttons containing the expressions. The menu buttons were all purple, except the six main menu buttons that specified the activity from which the expressions were selected. In the menu buttons, the text was written with capital letters, to serve as further discrimination from the buttons containing the expressions.

128 

Colours were also used in Phrases 2, to specify some of the functions of the expressions.

Positive expressions were light green and negative expressions were light red. Questions were light blue and phrase starters that had to be continued were orange. Statements and neutral expressions were light greyish purple. The reason to use colours was to make it easier for the users to find the desired expressions. Colours in AAC displays can have an impact on visual processing and memory (Wilkinson & Jagaroo, 2004; Light & Drager, 2007). It can also enhance speed and accuracy, at least in children (Wilkinson, Carlin &

Jagaroo, 2006).

Editing  

In the prototype software, all expressions were possible to edit through pressing the edit-button and navigating to the desired page, without exiting the program. It was an

important part of the evaluation of Phrases 2 to encourage users to change the expressions they did not like to others that they preferred.

Modes  

It was always possible to change between a mode called Prata (Talk) and a mode called SMS-style in the prototype software The Talk mode meant that all expressions were spoken when clicked on or pressed (with a touch screen). The SMS-mode meant that all expressions were sent to the message field. There was a difference between the two different types of expressions, because the Quickfire expressions were always spoken and the Activity-related phrases were always sent to the message field, giving the following result:

In the Talk mode, the Quickfire expressions were spoken when pressed or clicked on, but not sent to the message field. The Activity-related phrases were also spoken when pressed, and at the same time sent to the message field.

In the SMS-mode all expressions were sent to the message field, but only the quickfire expressions were spoken directly. It was always possible to let the speech synthesis speak the content of the message field through pressing the Speak-button.

6.2.5 Development of the prototype softwarethroughout the evaluations  During the evaluations that are described in chapters 7-11, there was a constant development of the prototype software. Different versions and variants were developed in order to make it usable by participants who needed or asked for different kinds of accommodations and so that it could be used in the experiment described in chapter 9.

 

129