Try ROT-7: t(20) → 13=m h(8) → 1=a m(13) → 6=f y(25) → 18=r l(12) → 5=e Word = m a f r e → "mafre"? Not English.
So maybe the whole phrase is Arabic names in English letters but encoded.
But the phrase length is: thmyl (5) brnamj (6) ywr (3) frydwm (6) mhkr (4) alakhdr (7) thmyl brnamj ywr frydwm mhkr alakhdr
t → g h → u m → z y → l l → y
Given “alakhdr” → if we apply ROT-3: a→x, l→i, a→x, k→h, h→e, d→a, r→o → “xixheao” no. Try ROT-7: t(20) → 13=m h(8) → 1=a
Given “alakhdr” clearly looks like “al-akhdar”, I’d say the phrase might be:
Try ROT-8: t(20)→12=l h(8)→0 (a)?? No, mod26: 8-8=0=a, m(13)→5=e, y(25)→17=q, l(12)→4=d → "l a e q d" no. But the phrase length is: thmyl (5) brnamj
But if each word is a separate Caesar shift: