Testing/Results

So the testing went slightly different than I had expected it to from my methodology.

The outlook of my final questionnaire remained the exact same. The first problem was with quantity of participants, as it got closer and closer to the dates that I had wanted to test, volunteers dropped out due to other work/commitments. The result was a test upon 16 participants as opposed to 24, two groups of 8. Although I had initially stated 24 to leave room for any no show's or dropouts, I had not expected that many people to not make the experiment. As it became apparent that the groups were becoming less and more and more people could not make one specific time, it came to the point that I had to change the layout of the experiment as not everyone could make a single showing in the Hannah Maclure centre.

Instead of testing on everyone at the same time, I decided to create a 'personal' version of the test. I chose 2 days in which I could sit in Abertay Universities Whitespace all day and let people know where I'd be for them to come and test, on top of this for a couple people I personally went to for them to partake as they could not make it to the university.

The end result was for the video to be played through my macbook laptop, audio presented through studio headphones with the questionnaire presented to them on an apple ipad as an editable pdf where they could swipe on the likert scales for an accurate measurement of emotional state.

This actually turned out to be a more concise experiment in my opinion than the original method. Although the group of participants tested on was less than I would have liked, the test itself was no less conclusive with enough data feedback to accurately prove my hypothesis that a manipulated musical theme over the course of a film clip could change the audience's opinion of a character.

The test itself was able to be more in depth as the viewer was able to watch the clip at their own leisure, allowing them to pause the clip at the end of each musical cue if they wished to catch up and think a wee bit more about their answers.

Below is an example of a filled out questionnaire.

Once all data it was gathered into a spreadsheet for each excerpt as shown here:

The mean value of each scale within each cue was then calculated that could be formed into graphical evidence as show here:

What do these results mean? (The following is the initial result write up that I have written for my dissertation:

For the initial dissection of these results, the Emotional – Emotionless scale will not be taken into account as these results will be further examined on their own. 

The graphical results clearly conclude a confirmation of the hypothesis, that an audience’s perception of a character in a film may be changed through the manipulation of an established musical theme, in both excerpts. In excerpt one, other than the initial first cue result of the innocent adjective, all of the left hand adjectives in cue one exhibit a mean value of between one and two before gradually rising to between four and five in cue five. In excerpt one, the first of these scales, Unalarming – Threatening, produce the most conclusive evidence gathered with a variance of 2.64 between the mean of cue one (1.95) and five (4.59) producing a clear indication in shift of perception between the initial cue and the last. Excerpt two, exhibits similar results with a variable of 2.20 between the first and last musical cues, with means of 2.01 and 4.21. Both of the excerpts show a strong feeling of indifference either slightly above or slightly below the mean of three during cue three, where the music crosses over from a fragile, innocent tone to a more sinister, uncontrolled tone or vice versa.

In excerpt one, the scale of Innocent – Menacing demonstrate the least absolute conclusive result out of these five scales with a variance of 2.33 between cue ones mean of 2.18 and cue fives mean of 4.51. Although this is the lowest of variance, the information gathered still constructs conclusive evidence that there is a large shift in perception between cues one and five with cue three demonstrating almost complete indifference of feeling with a mean average of 3.10. This same scale in excerpt two provides the second most conclusive results under Unalarming – Threatening as the difference between the means of cues one and five is 2.17.

Excerpt two on each of the likert scales underperforms in examination compared to excerpt one. This does not conclude that excerpt two does not demonstrate a change in perception or a conclusive result as there is still a distinct decline in figures from the right hand adjective to the left. Contained – Uncontrolled exhibit the closest of these varying figures as the second cue rises from a mean of 3.38 to 3.56 before declining. Cues three and four have a very similar mean of 2.78 and 2.75 before dropping to 2.19. The participants feelings here were that the character was never in complete control of himself throughout excerpt two. Although the excerpt begins with the most chaotic rendition of the musical cue, the character is contained to his car making him only a threat to himself whereas upon entering the hotel, the musical cue begins to stray away from dissonance, there is still an element in tonal uncertainty that paired with the characters introduction to civilisation may perceive him as in less control of his emotional state. Once the character is once again contained in an isolated space, the musical theme is capable of perceiving the character in a controlled, fragile state.

As excerpt two begins with the instant musical theme implying a chaotic disturbance in the characters frame of mind with little visual clue as to why the character may feel that way, the experiments volunteers may have found it hard to place exactly how out of control or aggressive the character is in the initial cue before making up their mind over time as cue two is beginning to take effect. If this is the case, there is no reason as to why the evidence does not continue to prove the initial hypothesis as the viewers perception clearly changes once they are happy with their conceived idea of the characters state of mind.

The scale of Emotional – Emotionless was left out of the resulting discussion above, as it provides the least conclusive of results to the hypothesis, however, this does not make the results invalid to the experiment. Although excerpt one ends with the male character standing in his room expressionless before holding a shard of broken mirror in a trancelike state, the character at no point loses his emotional depth in terms of inner misery and frustration. Although the character may have begun the excerpt in a controlled state of mind, this does not make him any more or any less emotional, neither does this change through his distress as at all times the character feels pain and misery. In both excerpts, the character may end in contemplation, whether that is in reminiscing what has previously happened to him and how he will proceed in life or the contemplation of himself being a threat to others or himself, these state of minds are still backed by an emotive force which provides a character with a human element at all times whether he is an uncontrollable threat or is innocent and sane. Excerpt ones results all fall within a mean average between 1.85 and 2.11, concluding that at no time the character presents himself as emotionless. Excerpt two exhibits a range between 1.66 and 2.79, again at no point bridging into an emotionless state but interestingly does show a slow decline into an absolute emotional state between cues two and five as the music becomes more delicate and harmonically close, this explaining that although the character is never presented as emotionless, that he does become more emotionally involved as the excerpt progresses.

Twelve of the participating volunteers were aged twenty-five or under, the remaining four ranging form thirty-one to sixty-one. Eight volunteers either study or studied audio at a formal educational level, four regard themselves as musicians or can read sheet music and four have no previous musical knowledge. Through analysis of results, age and musical background did not play a factor in the change of perception and was left out of the graphical analysis as each scaled posed similar results regardless of this information.

Eleven out of the sixteen participants regarded the character to be of threatening behaviour when asked how they felt for him either at the end or beginning of the clip depending on which excerpt was viewed. Interestingly, out of the five that did not believe the character to be of threat at all, similar comments were made whereby they believed the character to be in a state of emotional significance or felt that the character although did not become threatening, did pull through a strong emotional struggle within himself. This is largely supported through the participants interpretation of possible plot to each of the excerpts where the most common themes were; the character going through a mental breakdown, the loss of family to an accident, the loss of the characters wife, the descent into insanity, running away from life, a wanted man, personal distress, the contemplation of life and the contemplation of suicide. These themes give a range of emotional states that may be felt for a character at either end of the scale before they were turned around through the use of manipulated musical themes.

To conclude:

Through practical development in compositional technique and study of its application to film music, the experiment displayed two film excerpts that prove the hypothesis that an audience’s perception of a single character may change throughout a film clip through the use of an established musical theme that is manipulated throughout the excerpt. Excerpt one displayed a character theme that began melodic and harmonically close to the tonal centre before spiralling into a dissonant breakdown of controlled chaos whereas excerpt two swapped this around, beginning with little harmonic tonality. Five out of the six likert rating scales given to an audience concluded results that there was a gradual change in perception through the five musical cues heard. Excerpt one established a variance between cues one and five of over 2.33 in a five-point scale for each of the scales but one, demonstrating a clear shift from one perception to another. Excerpt two’s results were less favourable with a variable result of over 1.19 for each of the rating scales but one. Although this spread of data provides a tighter curve of result, graphically the evidence continues to example a clear change in perception from the beginning of the excerpt until the end, the audience possibly feeling reluctant to commit to a true immersive emotion initially as the musical cue begins so harmonically detached.

If this experiment were to be further tested or reworked for a larger audience the first change to be made would be in the compositional styles of excerpt one and two. Although two varying pieces of music could in theory manipulate the concluding results as certain styles or genres of music may create a preconceived mood or perception for the subject on screen, they may provide more coherent results, as the composed theme is not written entirely for one excerpt before being edited for the second. This initial method of testing proved to perform strongly in the first excerpt but did not have the same strength of result upon the second due to the fact that the audience was not presented with an initially recognisable, conventional or tonal theme either melodically, harmonically or rhythmically to grasp on to and associate with the character instantly, instead the viewer had to become accustomed to the theme in which by cue two, they could compose an accurate judgment on their perception.

To examine a full change in perception of a character, using a full feature film to subtly change the viewers understanding of the characters inner emotions would allow for a substantial analysis of musical representation of those emotions in a considerably greater narrative setting. Within a feature length film, music could not only perceive the character in more than one light but could steadily change that character throughout the whole feature creating a distinct manipulation of personality from the first scene to the last. Within this experiment, five musical cues were presented within a three-minute space that does not portray how an audience would commonly hear music in film. Within a feature film music would be interspersed between dialogue and other narrative devices such as transitions, camera shots, action sequences and so on, allowing a much greater time frame for a character to become who the film dictates he should be and a much longer time for that to be further shaped or altered.