VVA CLASSIFICATION RESOURCES

Homepage
Philip Tagg, updated 2010-06-
14


General info

These resources are primarily intended for anyone wanting (or having) to carry out music reception tests based on unguided association procedures (Chapter 6 in Music’s Meanings). I hope that the VVA response grids included in these resources (both the overview and the detailed version) can give some idea of how VVAs (verbal-visual associations, i.e. single concepts derived from your test responses) can be organised so you can easily find out and present how much of what respondents imagined when hearing the reception test examples you gave them. It's obviously better if you group, say, romance together with love than with alphabetical neighbours like Rome, Rommel or Romania, more useful if love is nearer romance than lousey, louts, low-life or Lwenbru

Responses in the form of unguided associations obviously need to be discretised into individual concepts so that, for example, love in an original response like <The femme fatale whispers "I love you" while waving her cigarette holder over their drinks> and romance in <Typical Hollywood romance or thriller from 1950s in black and white> can both be categorised as indicative of the same (or similar) love and romance connotations in response to the same music. That sort of classification may be relatively unproblematic but some concepts, not least proper names, are not so simple. How, for example, would you categorise Lwenbru? Does it sort under Drinks and other comestibles because it's a beer (category 2642 in the detailed grid), or under Germany (for obvious reasons, 3716 in the grid) or under Advert (8231, because your respondent refers to a TV commercial), or do you count it in all three? There's no room here to account for "polysemic VVAs", "context-contingent VVAs" or for any of the other problems involved, nor to discuss possible solutions. Instead I respectfully refer readers to pages 107-152 (esp. 125, ff.) in Ten Little Title Tunes for a more substantial treatment of these issues. 

Still, one point may be worth making: you obviously have to think musogenically, not just verbally, when dealing with VVAs. The word abandon, for example, means both leave in the lurch (e.g. an abandoned child, 1236) and letting yourself go (with great abandon, no holds barred, yippee!, 105). Those two states require very different music, as do over (as in Over the Rainbow), over (riding over the prairie), over (as in the party's over) and over (as in a dark cloud over the city). Response words arise out of the music and should be treated accordingly.

If you are conducting a reception test as part of your graduate or postgraduate research I think you would find good use for all the resources listed as links at the top of this page. If you're including a reception test in an undergraduate assignment you'll probably get by with just the two online grids (see next) and the Excel files (see later on).


Top VVA response grids

Two response grids, both based on the VVA taxonomy in Ten Little Title Tunes, are included in these online resources: [1] a Basic VVA response grid; [2] a Detailed VVA response grid.

The original VVA taxonomy has four levels of categorisation and most VVAs were given a  suitable four-digit code. Romance, for example, is in category 1112 which it shares with lots of love but not with just love (1111, could be brotherly or parental), nor with tender or gentle (1117). However, despite those important musogenic nuances of love (romance is not always tender and it is definitely unhealthy to confuse parental with romantic love), all those related concepts do belong to the same three-digit category 111 (love and kindness) which is distinct from other positive three-digit categories like Joy and festivity (113) or Lightness and openness (115) and at the other end of the affective spectrum from 121 (Emnity and aggression) or 125 (Darkness, encumbrance, clandestinity and miasma). 

Top Basic grid

The Basic VVA response grid gives a good overview of the principal categories in the VVA taxonomy from Ten Little Title Tunes. It lists the numbers and labels of all the three-digit categories in the four-digit hierarchy. Hovering with your cursor for second over almost any of the highlighted numbers or letters reveals examples of concepts contained below that higher-level of abstraction (two- or three-digits). For example, hovering over 111 (Love & kindness) under 11 (Positive affect) reveals the text "incl. romance, sensuality, sensitivity, tenderness" while hovering over 121 (Emnity) under 12 (Negative affect) displays the message "hate, rage, aggression, implacability, etc." - a different kettle of musical fish if ever there was.

Clicking rather than just hovering over the sort of links just mentioned will take you to the relevant place in the Detailed response grid where you can check categories at the four-digit level, as explained in the paragraph before last (here) and next. 

Top Detailed grid

Both grids are based on 8,442 VVAs collected in the early 1980s from over 600 individuals (mainly Swedes and Latin Americans) responding to ten different film and TV title tunes. The totality of those respondents' imagination on hearing those pieces is too complex to classify in any semantically exhaustive way, so the Detailed VVA response grid offered here is presented solely as a source for ideas. It is in no way intended as some sort of watertight taxonomy. More importantly, the grid deals with just an infinitesimal part of all the VVAs imaginable in response to any other music heard by other populations at other times and in other places (see Cultural specificity caveats). Of course, the more the grid presented here can help sort any issues of response classification the better, but I have absolutely no illusion that it can do much more than just offer a few ideas: 1% of something is, I believe, better than 99% of nothing. Here's a concrete example of the problem.

Let's say I've included an extract of industrial music in a reception test and that one of the respondents writes <Dystopian robot in a disused factory> and that I discretise the response into five VVAs: [1] dystopian, [2] robot, [3] in (yes!), [4] disused and [5] factory. Opening the Detailed VVA grid and using my browser's Find function (Cntrl-F) I uncover only two of those concepts in the grid: in (in category 3020 - indoors rather than out, a small but significant musogenic difference) and factory (category 353 - Urban buildings and locations). What do I do with the other three VVAs in the response? 

Dystopian is the hardest nut to crack. It's clearly negative (category 12 in the Basic grid overview), with connotations of oppression and darkness (125), but dystopias are always set in the not-so-distant future (389) and dystopian is also a literary and cinematic genre (84). Frankly, I probably wouldn't worry too much about dystopia as a genre but I would consider entering some sort of cross-reference to the concept under Future time. Still, I would probably end up by putting dystopia primarily in category 1250 with darkness and  gloom.

Robot is slightly easier. Assuming it's like Robocop or the Terminator, this robot is a single 'male' being (211) rather than just a machine (2660) but it's a being that doesn't fit in any of the existing 211 (single male) subcategories. The only solution to that problem is to invent the new four-digit category 211S ('S' for Sci-Fi) to include male-gendered robots, humanoids, cyborgs, extraterrestrials and suchlike, or to open one under 201 (201S, for example) to include all such beings, culturally gendered or not. Finally, disused is not too difficult to classify since it clearly belongs to category 125 (includes decaying, dirty, rotting, ill); or, failing that, 128 (includes mess, shambles, chaos). 

To get a general idea of the VVA taxonomy, either use the Basic grid or click here to scroll through the detailed table.

For ideas about classifying particular response concepts (VVAs):
  [1] open the detailed grid and use your browser's FIND and FIND NEXT functions (Cntrl-F in Firefox) to locate occurrences of the VVA you're looking for;
  [2] check the position of any occurrence of interest in relation to the surrounding four-, three-, two- or one-digit categories. Then decide if its placement in the grid strikes you as useful or not.

Tip. Just enter the first part of a word if you don't find exactly what you're looking for.
      For example, enter |embrac| rather than |embrace| or |embracing|
      because the concept might be listed as either one or the other.

Warning. As the dystopian robot example shows, you should not be surprised if you don't find the VVA you're looking for in the Detailed response grid. It's based on responses from 0.0000001% of the world population to 0.00000001% of the music circulating in the early 1980s when the reception tests were conducted (see Cultural specificity caveats). In fact, you might well find you need to construct your own grid for a very different set of musical, cultural and social conditions. In which case I recommend the blank but headed Excel spreadsheets explained next.


Top Excel DIY response grids  
 

For reasons already explained, you may well find it more practical to construct your own grid for classifying VVAs. If you have only ten or fifteen respondents you can probably do everything 'by hand' but as soon as you have much more than that and an average of 3 or 4 VVAs per person it's definitely worth keeping track of responses in a spreadsheet or database. In such cases you'll need three tables. If you don't know how to set up a spreadsheet or database, or if you want to make life easier you can downlod three templates, either one at a time (1  2  3) or all three packed into one ZIP file.

Table 1 should include the unique but anonymous identity (a number will do) of each respondent and any demographic or personal information of relevance to your study, e.g. age, gender, nationality, education, whether they consider themselves to be fans of the music you're asking them to respond to (if that's important to your study), if they're musicians (if you're interested in that aspect) or politically active (if that's part of what you're investigating). A spreadsheet like Table 1 is easy to construct using Microsoft Office Excel. Perhaps you can adapt this one to your needs? You should  start with either Table 1 or Table 2 when you input your data. You don't need Table 1 if there's only one music example in your reception test (see explanations for Table 2).

Table 2 should include the unique but anonymous identity of each respondent and the identity of each example you're testing (if there's more than one). It should also contain the complete original answer given for each music example by each respondent. Maybe this table will be useful for that purpose. No interpretation is required on your part to input data into Tables 1 and 2. You'll need both Tables 1 and 2 if you have more than one or two examples to save yourself the trouble of re-entering demographic information for each respondent for each music example. If you only have one or two examples you can include all the demographic and personal data in Table 2 and skip Table 1.

Table 3 should include the unique number identity of each respondent, the identity of each example you're testing (if you're testing more than one) and each single VVA that you extract from the complete original answers collected in Table 2. This Excel table can house that data. Since, as already shown, you would probably want romance to sort under the same category as love rather than with its alphabetical neighbours Roman and Romania, you'll need to number your categories so you can, using Excel's column sorting facility (top right in Excel), sort the data by category code so as to see at a glance how much of what (love and romance, for example) was imagined in conjunction with which tune by which respondents.  

Top Tips

  1. If you're unsure about how to discretise complete responses into single VVAs, read pp. 122-147 in Ten Little Title Tunes and/or check the VVA grids as suggested above.
  2. When entering respondent IDs, tune IDs and category codes into Table 3, and if you're using numbers, start each code with a letter (e.g. |R01| for respondent #1, |C1234| for category 1234, |T1| for tune #1)  so that Excel doesn't treat them as numbers to be added up or averaged.
  3. If you have more than 9 respondents or songs, put a zero in front of numbers under 10 so that Excel can sort your data in the right order, or else 2 will be listed after 19 whereas 02 will be in the right place. If you have more than 99 respondents, you'll need to number the ones under 100 with one or two leading zeros (e.g. 001, 002..., 010, 011..., 099, 100, etc.).
  4. You can adapt the columns in the Excel files according to your own needs. Of course, delete any template data I've included by way of example.
  5. It's quite a good idea, if you want to keep column labels at the top of your Excel spreadsheets, to preceed each label with an exclamation mark because |!| sorts alphabetically before anything you can produce on your keyboard except for Space.
  6. Make good use of the column indexing facility in Excel. Remember that you can sort on one column, say on VVAs, to get data into that order then sort on another, say Category number, to produce a listing in order of  VVA category with each individual VVA in alphabetical order inside each category.  

TopCultural specificity caveats

The resources presented here are largely based on a very limited and specific set of cultural circumstances: the imagination of 562 Scandinavians aged 15-60 and 45 Latin Americans aged mainly 20-30, male and female, mostly interested in music, and responding, in the early to mid 1980s, to ten extracts of stylistically mainstream title themes from principally UK or US film or TV circulating in those media at various times between 1951 and 1985. Such a high degree of cultural specificity implies the following.

  1. Historical location categories 387 and 388 (recent history and today/modern), altered here to fit the year 2010, will be in constant need of adjustment. Obviously, in 1983 or 1984 the 1970s were recent and the 1980s up-to-date, today and modern.
  2. Geographical categories 37 (location) and 87 (production location) are subdivided quite ethnocentrically due to the specificity of the musical and/or audiovisual material on which the grid is based and of the respondents' own cultural environment.
  3. The under-respresentation of women in the material is indicative of the responses collected and of the sort of audiovisual production imagined in conjunction with the ten test pieces. This imbalance is discussed under the heading Gender and ideology on pages 666-679 in Ten Little Title Tunes.
  4. Since the music examples used in the reception test were related to moving images and since subjects were explicitly asked to include visual elements in their responses, it is possible that the VVAs at the base of the materials presented here have more of a visual than, say, auditory, tactile, spatial or kinetic bias. The epistemological and methodological reasons for such visual bias are set out on pp. 108-110 in Ten Little Title Tunes and on pp. 83-100 in Kojak - 50 Seconds of TV Music.

Moreover, remembering that it took a large team of well-financed experts a good decade to come up with the UNICODE system of international character encoding for computers, it would be absurd to think that a single individual with a full-time job could come up with an interculturally viable taxonomy for every imaginable visual-verbal response to music. That's why I urge those interested in classifying reception test responses from unguided association procedures to either adapt (add, delete, alter) the grids provided here and here or to construct their own (see here).

Philip Tagg
Huddersfield (UK),
8-14 June 2010


Top TAXONOMY: DEFINITION

Taxonomy is used here in the sense of  a scheme of classification (Oxford Concise English Dictionary, 1995) arranged hierarchically. Taxonomies are constructed on  supertype-subtype relationships& [T]he subtype& has the same properties& as the supertype plus one or more additional properties. For example, car (category 2652 in our VVA taxonomy)  is a subtype of vehicle (category 265).  So any car is also a vehicle, but not every vehicle is a car (Wikipedia: Taxonomy [100614]).

 

 

 

 

 

 

 

 

 


Top