Pseudogenes can be expressed if they have a functional promoter. The thing that makes them pseudogenes is the lack of a known functional protein. See Wikipedia for an overview. There is some evidence that pseudogenes may have a conserved functional role as targets of miRNA; see the Pandolfi lab's recent Nature publication on that topic. Pseudogenes are definitely found on Affymetrix arrays, particularly the older versions (e.g. M430) that were designed against earlier EST collections.

I wouldn't call it "normal" that pseudogenes are at the top of a list, but to get better help you should specify some important details like:

1. what are the general properties of your experiment
2. do you mean top by statistical significance, by fold-change, or both
3. Are the coding versions of those psuedogenes also differentially expressed (DE)?
4. are your pseudogenes the only things DE?

AltAnalyze will return expression values for all Ensembl genes linked to exon array probesets. Affy is quite liberal in assigning probesets to any putative expressed regions in the genome. I would make sure that the expression of house keeping genes look to be highly expressed or even include other CEL files from GEO for similar cell types to make sure that expression is somewhat similar. This will help ensure that there is not some issues with sample prep or hybridization. Some quick RT-PCR assays can help to verify the top few genes.

Having said that, the biology you are observing may be significant. New evidence suggests that many pseudogenes can produce long non-coding RNAs that act as anti-sense RNAs that bind to complementary spliced mRNAs produced in the genome as double stranded RNAs (dsRNA). These dsRNAs will be targets of Dicer and the RISC complex. See some recent papers on long ncRNAs, for example:

Tam OH, Aravin AA, Stein P, Girard A, Murchison EP, Cheloufi S, Hodges E, Anger M, Sachidanandam R, Schultz RM, et al. 2008. Pseudogene-derived small interfering RNAs regulate gene expression in mouse oocytes. Nature 453: 534–538.

Some pseduogenes are so labeled because they are disrupted by an inserted transposon. When this happens, the transposon's promoter, often very strong (think of LMTV and MMTV promoters), can support transcription of the partial or pseudogene. So, if your experimental condition is one where transposon activity is increased, perhaps this is a real result. This would not be unheard of in Drosophila or plants. Which organism are you studying?

David's answer is very good. There are different types of pseudogenes or different reasons that label is attached to that genetic entity. I'd take a look at the genomic context of these genes. Are they all near Alu elements (in human) or B elements (mouse)? Are they all near telomeres? In other words, back away from the detail of the genes and try to take in a bigger picture - maybe there is something there.