With the old style cursor that is used in this study (the new style cursors are the arcpy.da. cursors), you can parse a table row by row (= line by line) and, for each row, get the value of an attribute field (=column) based on its name.
In the case of your script, 'cur', 'row', 'text1' and 'text2' are the names of variables (you could change these names without affecting your script).
cur is an object used to reads table "tabelle" entity by entity (line by line)
row is an object used to access the content of each line
text1 and text2 are variable (probably string) used to store the content of one cell (a specific column of a given row)
"Tabelle" is also a variable, so it must have been defined above this code snippet. However, Objekt and Text must be field names that already exist in the table "tabelle". Writing 'row.objekt' is equivalent to 'row.getValue("objekt")' If you use a table with other field names, those names must be adapted. Note that, if you don't do anything with text1 and text2 inside the "while" loop, then text1 and text2 will only store the values of "text" and "objekt" for the last row of your table, because each assignment to an existing variable overwrites the previous value.