To some extent it depends upon what you want to do with the results.
If you read the file as a database and end up with fields for Phone1 and Phone2 and then want to output each with common additional fields, one option would be to create a summary that produced a record using the Phone1 field plus other fields required. Change the default name of the summary field to, say, Phone within the summary definition.
Then, since you have V9, duplicate that summary to create a second summary with the same format but replace Phone1 with Phone2 and rename that also to Phone. (save the second summary definition at least once before renaming the field in the second summary to avoid confusing thing in the cache memory.)
You can then use a Project Export to export both summaries to the same output file (or wherever). It would likely need to be sorted of course.
There are likely to be other approaches as well but this one is the first that came to mind.
Does it help?
If it is true that every phone number is the same number of digits and the number of phone numbers per line is constant (or at max a certain number), you could use a multiple column region. For example if there are 10 phone number max per line and each one is 12 digits with a coma you could say there are 10 columns per line with 13 characters per column. Cases where there were only 8 phone numbers per line wouldn't cause a problem.
If there is a variant number of phone numbers with no limit to the number per row, then it gets somewhat messy. It's possible, but messy.