Let's say a user can have multiple phone numbers.
I can understand that the first table below is a violation of 1NF, as the userID=2 is repeated.
+--------+-------+
| userID | phone |
+--------+-------+
| 1 | 1010 |
| 2 | 1020 |
| 2 | 1021 |
| 3 | 1030 |
+--------+-------+
But is this second table also a violation of 1NF ? Yes, it appears to be a bad, inflexible design - but is it violating 1NF ?
+--------+--------+--------+
| userID | phone1 | phone2 |
+--------+--------+--------+
| 1 | 1010 | |
| 2 | 1020 | 1021 |
| 3 | 1030 | |
+--------+--------+--------+

What happens to your second design when you have a third telephone number? It may not be a violation of 1NF but it's poor design. I, for example, have a home telephone number, a work telephone number, a mobile telephone number and a fax number. The ERP program which I use at work has a 'user contacts' table which has fields for each of those numbers. This allows for easier output but can sometimes lead to problems if a person has more than one number of a given type.
The standard way of handling multiple telephone numbers is to have a separate table with fields owner, telephone number and description.