Re‐enacting of shop conversations through role‐play

107 

on which one to re-enact, read the instructions for that specific scene and then started the role-playing.

5.1.5 Criteria for replication of interactions from GSLC in role‐play 

For a role-play interaction to be regarded as a replication of an interaction in GSLC, it had to meet certain criteria.

A full replication of an interaction from GSLC was said to occur when:

• All communicative acts that were used by the customer in the original interaction were also used by the VOCA-user in role-play.

• The role-play participant using the VOCA asked for the same item and/or information from the shop assistant.

• The result of the interaction was the same, i.e. with regard to what got purchased.

A partial replication of an interaction from GSLC was said to occur when:

• At least 50% of the communicative acts that were used by the customer in the original interaction were also used by the VOCA-user in role-play.

• The role-play participant using the VOCA asked for the same item and/or information from the shop assistant.

• The result of the interaction was the same, i.e. with regard to what got purchased.

5.1.6 Improvisation in role‐play 

The role-play sessions were to a great extent improvised, an important feature for making them as close to real shopping activities as possible. The instructions to the participants made them ask for specific things or give pre-determined answers (such as that a game or book was not in stock), but the conversations were not scripted. It was not decided beforehand what VOCAs to use in what scenes or who should play the role as customer or shop assistant in a specific scene. How it turned out can be seen in table 5.4, and the implications are that certain questions can be answered through the resulting role-play conversations and others can not. Due to the mixture of participants, VOCAs and shops, it is not possible to make a direct comparison between the efficiency of specific VOCAs, there are too many confounding factors. On the other hand, the improvisations may have participated in creating role-play sessions that were closer to real life, which is also not scripted.

108 

Table 5.4. Shopping for Star Wars Technical facts in GSLC and role‐play 

Customer expr.   GSLC, Shop: A7904051  Speakout 3  Imagetalk3 

Greeting  Hallå (hi)  Hej (2) (hello) 

Hej då (bye) 

Hej (2)  Hej då  Request for 

item 

Jag letar efter en bok som heter  starwars technical facts (I’m  looking for a book called STF) 

Jag skulle vilja ha  starwars technical facts. 

(I would like to have  STF) 

Har ni  (do you  have)  +  ((leaves a  note))  Request for 

information 

Nä man kan beställa den eller  nåt kanske (no you could order it  or something maybe) 

Vad tar du för (..) annars cirkus  (what do you take for it (...)  otherwise approx.) 

Du har inte den (you don’t have  it) 

Finns inte att få tag på alltså  eller (impossible to get is it) 

Har ni inte den tjocka  (don’t you have the fat  one) 

Vad kostar dom ungefär  (how much are they  approximately) 

Har ni det (do  you have it) 

Var har ni det  (Where do you  have it) 

Ni får inte in  dom snart  heller (you  don’t get it  soon either)  Acknowledge‐

ment 

Okej tack (2) (okay thanks)  Tack för hjälpen (thanks  for your help) 

Tack (thank  you) 

Affirm/Confirm  Ja  (yes)  Ja (2) (yes)   

Rejection/ 

denial 

  Nej boken (no the book) 

Nej (no) 

Nej (2) (no) 

Feedback  m (2);    Jaha okej (oh okay)    Okej (okay) 

Informing  (provision of   information) 

Står inga priser på men  (Can’t  see any prices but) 

Ja jag bara undrar (yes I’m just  wondering) 

Såg den bara i England så tänkte  jag måste finnas nånstans (saw it  in England thought it has to be  somewhere) 

Jag var inne på Wettergrens och  dom kollade konstigt på mig  bara (I was at W and they just  looked at me funny) 

Jag har sett boken i  England (I have seen the  book in England)  Jag har köpt den där (I  have bought it there)  Jag var på wettergrens  men de tittade konstigt  på mig (3) (I was at W  and they looked at me  funny) 

Specification  Hittar inte den där lite fetare  varianten (can’t find the  somewhat fatter variety)  Technical facts 

Starwars  Technical facts 

Correction    Sett (seen)   

The conversation includes complications in that the customer asks for information that the shop assistant does not already have, but has to consult the computer to get. The customer offers additional, redundant information about where he has seen the book and where else he has looked for it. There is also a long pause in the original recording, when the shop assistant had to take a telephone call. Table 5.4 shows the expressions used in

109 

the original conversation and in two of the role-play sessions where the scene was replicated. The expressions are displayed according to the communicative acts that were performed.

Table 5.4 shows that all the communicative acts that the customer used in the actual conversation also were used in at least one of the VOCA conversations. For greeting, acknowledgement, affirmation, rejection/denial and feedback, standard short phrases were used in all cases. Requests for the book and request for information were performed in all three instances, but the way they were expressed varied.

The participant using the SpeakOut wrote his expressions from the keyboard and was able to express whatever he wanted. He used complete, regular sentences, whereas the original speaker expressed himself more vaguely, with more hedging and, in three of four requests for information, with the word order for statements instead of that for

questions. Seven of the eight communicative acts that were used in the original

conversation were also expressed by the VOCA-using customer in this interaction. Since he requested the same item, with the same result as in the original conversation, it can be seen as a partially replicated interaction, with respect to the criteria in 5.1.5.

The participant using the Nokia Smartphone with Imagetalk only had a compressed version of Phrases 1, with a small number of pre-prepared phrases at her disposal. The smartphone also had an on-screen keyboard that she found hard to use. Since the instructions for the VOCA users were to prepare the names of the things they wanted to buy before starting the role-play, the Imagetalk user had prepared a written note with the name “Star Wars Technical facts” that she handed to the shop assistant, as well as saying

“Har ni (Do you have)” with the device. The Imagetalk had a number of general requests for information that were used in the conversation. Just like the user of the keyboard device, she requested the same items and got the same result as the original customer, but she only used 5 of the 8 communicative acts from the original conversation. The

interaction met the criteria for a partially replicated interaction, but at the lower end of the scale.

The most important difference between the SpeakOut and the Imagetalk sessions was with respect to informing and specification. The person using the Imagetalk did not provide any such information, while the person using the SpeakOut provided much of the same information that the original speaker had done. This information was not prepared in the Imagetalk and it proved cumbersome to write with the small on-screen keyboard and also to make the synthetic speech say what had been written. In such circumstances it was perhaps natural to leave out this, in many ways redundant, information altogether.

110  Replication of the GSLC interactions 

These were the results of the 17 role-play sessions where the participants tried to replicate one of the conversations in GSLC, through following the instructions of the instruction cards:

In seven of the interactions a VOCA with Phrases 1 was used. Three of these interactions met the criteria for full replication (see 5.1.5), and the remaining four met the criteria for partial replication. The mean percentage of replicated communicative acts was 87.

Of four interactions where keyboard VOCAs were used, all four met the criteria for partial replication. The percentage of replicated communicative acts was 74.

Finally, in seven role-play conversations where other VOCAs with dynamic displays were used, one interaction met the criteria for full replication of the original conversation, the other six met the criteria for partial replication. The mean percentage of replicated communicative acts was 71. Thus, of 18 studied role-play interactions, all met the criteria for partial replication of the GSLC-conversations, and three met the criteria for full replication.