3. Constitute parts by identifying meaning units.

Once you have a sense of the whole, decide on its most appropriate parts by identifying the units into which the meanings in the natural language sample can be readily divided. The size and scope of your meaning unit depends on the phenomenon at-hand and the type of research being conducted. There are often multiple and sometimes overlapping meanings being relayed in natural language, so it is of use to consider multiple ways of identifying the most meaningful units in a natural language response. Raven’s Eye is programmed to accommodate the free exploration and selection of multiple kinds and sizes of meaning units1. As with getting a sense of the whole, practice leads to rapid completion of this procedure.

Overt meaning units are generally easy to identify, because they are often based on the preset conditions of the research as designed by the researcher, or conventionalities arising from writing or speaking. For instance, if the research is in the form of a survey, then overt meaning units are likely formed by each question asked. Or, if the research concerns document analysis, the overt meaning units might be the chapters, sections, or paragraphs of the document. If overt meaning units are the focus of your study, your data will likely already manifest their division as separate columns in your spreadsheet2.

Sometimes, however, one’s sense of the whole indicates that other, less overt meaning units are present in the natural language response. Those interested in construct validation from a positivistic quantitative attitude, or those conducting investigation with a phenomenological attitude, will want to take precautions to ensure that the meaning units they employ do indeed coincide with the general units of meaning being relayed in the experience. If this is the case with your project, we provide the following considerations as standardized and fruitful means of identifying non-overt meaning units in your data.

3.1 Consider subdividing or combining portions of the natural language sample.

Sometimes, interesting meaning units can be identified by altering the original format of your natural language sample. If, as you consider them, your meaning units appear to constitute all of the natural language contained in more than one column, you can either merge the columns in your spreadsheet, or ensure to that you select the columns together when you analyze them in Raven’s Eye. If, however, multiple meaning units occur within one column’s cell of natural language data, or a given meaning unit is distributed in portions of cells in more than one column, then some data manipulation is in order. In these cases, bracket according to the actions noted in 3.4.

3.2 Consider response-based characteristics.

If your sense of the whole leads you to believe that the meanings units in your natural language sample can best be elucidated by the specific terms used in it, you might divide your meaning unit based on the presence or proportionality of specific words in specific responses or sections of the natural language sample, and then paste these into a new column for comparison. For instance, if you are trying to understand consumer sentiment in response to an advertisement, you might divide your natural language sample based on whether or not a particular response contains specific overrepresented and emotionally positive words, or specific overrepresented and emotionally negative words. To do so, follow the actions noted in 3.4

3.3 Consider states of geists and other variables.

Using the variables dropdown menu, select one specific state or class of one variable for analysis. In the subsample comparisons table that appears, visually examine the Variable Overrepresentation and Overrepresentation columns (the 7th and 8th columns, respectively). Look for apparent and substantial changes in the form or proportionality of gists in the subsample created by the state or class of the variable that you have selected. Also, briefly inspect the metadata for apparent and substantial differences arising from the specific state or class of variable selected. Continue selecting differing states or classes of each of your variables and visually inspecting the resultant analyses until you are satisfied that your meaning units either remain fairly stable across states or classes of each variable, or change based on the particular state or class of a given variable3. If you identify substantial differences in word patterns or metadata based on the state or class of your variable, consider creating meaning units based on such variation, as described in 3.4.

3.4 Bracket accordingly.

When you have arrived at a set of meaning units for your natural language sample, bracket to account for those meaningful units of the natural language response that are not otherwise already reflected by separate columns in your spreadsheet data.

Raven’s Eye operationalizes the concept of bracketing for meaning units by creating additional responses as new columns in your original spreadsheet, which can then be subjected to separate analyses. We recommend that such newly created columns or meaning units be derived from the actual linguistic features of the natural language sample, or observable environmental or demographic conditions involved in the acquisition of the sample4. To aid in later recall and replication, we further recommend that each column’s label be brief, and descriptive of the meaning unit it reflects. You are otherwise free to bracket in any way that best reflects the apparent form of your data, or the function of your project, or both.

To bracket, create a sufficient number of columns in your original spreadsheet, so that each column represents an independent meaning unit, and distribute the original natural language sample across them accordingly by copying and pasting separate meaning units into separate columns. Once you have created and named a new meaning unit (and thus created and labeled a new column in your original spreadsheet), you will need to populate each case (and corresponding row in your newly created column) of that meaning unit. To populate them, you may either proceed row-by-row and copy and paste them sequentially into the appropriate cell of the newly created column one case at a time, or utilize the filter feature of your spreadsheet program to copy and paste multiple cases at once.

For instance, primacy factors in language expression can often be identified by separating the first phrase, sentence, or paragraph in each natural language response or document section from the remainder of that response or section. To analyze primacy, then, you would isolate the incipit as a primal meaning unit by highlighting with your cursor the appropriate text in the cell of in the original column of your spreadsheet, then copy it to a corresponding cell in a new column. You would then repeat this process for the incipit of each applicable case or row. Then, when you upload your newly altered spreadsheet to Raven’s Eye, the incipit can now be analyzed independently from the rest of your natural language sample.

3.5 Reiterate, as needed.

If the process of bracketing appears to reveal additional functional meaning units, repeat substeps 3.1-3.4 until they are generally accounted for in your spreadsheet data.

If, after you have considered your meaning units, you have not changed the original structure of the data in your spreadsheet, you may proceed. If, however, you repartitioned it based on the considerations in this step, you will need to upload it to Raven’s Eye as a separate project, categorize your columns, and select it for analysis5. By default, you will arrive at the main table of your newly uploaded project. If you have substantially revised your meaning units, it is advisable to revisit step 2 and proceed from that point in the procedures.

Many researchers may find the needs of their projects met upon completion of this procedure. If such is the case in your project, you can download the results you have been otherwise visually inspecting, and produce messages, tables, graphs, or other visualizations in your spreadsheet, word processing, or other programs.

Notes.

1 In this procedure, we both functionalize and expand Giorgi’s (2012) corresponding step. With Quantitative Phenomenology, the meaning unit no longer needs to be arbitrary, or governed wholly by practical necessities arising from human limitations related to reading comprehension and information processing.

2 Due to definitional features of validity from a positivistic quantitative attitude, such is often the case for non-phenomenological research.

3 This does not have to be an exhaustive examination at this stage. However, you may also consider selecting combinations of variables, or several of their states or classes, or both, at the same time.

4 This is a general recommendation.

5 In this way, your major data transformations are demarcated by file, and thus facilitate easier replication and extension in subsequent projects.