The effect of silicate melts on the phase formation and shape change of $Al_2O_3$ has been studied. Although, a number of studies have previously been made on the phase formation and grain morphology during the dissolution of oxides in silicate melts, our understanding is still quite limited. This thesis consists of three series of investigation:1) formation of new compounds and their shape during the dissolution of $Al_2O_3$ in a $CaMgSiO_4$ melt(partII),2) entrapment of silicate glass matrix in $MgAl_2O_4$ grains and its equilibrium shape (partIII), and 3) microstructrue of sintered $Al_2O_3$ in silicate glasses(partIV). For the experiments in part II and partIII, sintered $Al_2O_3$ was packed with balanced $CaMgSiO_4$ glass powder and then annealed at $1600\,^\circ\!C$ for various times in air. Above the the melting temperature ($1505\,^\circ\!C$) of $CaMgSiO_4$ glass, $CaMgSiO_4$ melt was infiltrated into the sintered $Al_2O_3$. The dissolution of $Al_2O_3$ into te melt occured at the surface of $Al_2O_3$ specimen and $MgAl_2O_4$ spinel formed ahead of the $Al_2O_3$ at the beginning. During further dissolution, however, a layer of $CaO\cdot6Al_2O_3$ grains formed between $Al_2O_3$ and $MgAl_2O_4$ spinel. The growth of the spinel layer was possible through the growth of $CaO\cdot6Al_2O_3$ grains towards the $Al_2O_3$ grains and their redissolution towardd spinel grains.Such a heterogenuous dissolution was explained based on a phase diagram previously proposed. The growing front of $CaO\cdot6Al_2O_3$ grains was faceted and its dissolving front was irregular, implying that the dissolution shape was less govened by the surface energy anisotropy than the growth shape. After 13h annealing at $1600\,^\circ\!C$, the $MgAl_2O_4$ spinel and a liquid phase, as final reaction phases, remained in the microstructure. In part III, the entrapment process of liquid drops within faceted grains was observed and their equilibrium shape was determined in $MgAl_2O_4$-CMSA(CaMgSi...