We discuss the linearity and discreteness of amalgamated products of linear wordhyperbolic groups. In particular, we prove that the double of an Anosov group along a maximal cyclic subgroup is always linear, and we construct examples of such groups which do not admit any discrete and faithful representation in rank 1. We also build new examples of non-linear word hyberbolic groups, elaborating on a previous work of Canary-Stover-Tsouvalas.Theorem 1.3. Let Γ 1 and Γ 2 be lattices in Sp(k, 1), k 2, and w i be a cyclic subgroup of Γ i , i = 1, 2. Assume that w 1 ∈ Γ 1 and w 2 ∈ Γ 2 have different translation lengths in the symmetric space of Sp(k, 1). Then every linear representation of Γ 1 * w1=w2 Γ 2 maps Γ 1 to a finite group.We say that a group H is linear if it admits a faithful representation into GL(d, k) for some d ∈ N and some field k. As a corollary of Theorem 1.3 we have:Corollary 1.4. Let Γ 1 and Γ 2 be lattices in Sp(k, 1), k 2. Let W be an infinite finitely generated group and i 1 , i 2 be embeddings of W into Γ 1 and Γ 2 respectively. Assume that there exists w ∈ W such that i 1 (w) and i 2 (w) have different translation lengths. Then Γ 1 * W Γ 2 is not linear.The above theorem does not apply to doubles of a quaternionic lattice Γ over a subgroup W . However, even these tend to be non-linear for a larger group W .Theorem 1.5. Let Γ be a uniform lattice in Sp(k, 1), k 2 and W be a proper subgroup of Γ which is not the intersection of Γ with an algebraic subgroup of Sp(k, 1). Then Γ * W Γ is not linear.Finally, we point out the importance of the Anosov assumption in Theorem 1.2 by proving the following:Theorem 1.6. For n 3, there exist maximal cyclic subgroups w of SL(n, Z) for which the double of SL(n, Z) along w is not linear. 1.3. Discreteness and Anosov property. The subgroup Γ * w Γ of G in Theorem 1.2 has no reason to be discrete, and our work leaves open the following question: Question 1.7. Let Γ be an Anosov group and w a maximal cyclic subgroup of Γ. Is Γ * w Γ Anosov (possibly in some larger group)?A weaker statement that we strongly believe to be true is the following: Conjecture 1.8. Let Γ be an Anosov group and w a maximal cyclic subgroup of Γ. Then there exists a finite index subgroup Γ ′ of Γ containing w such that Γ ′ * w Γ ′ admits an Anosov representation.Remark 1.9. Contrary to what one might naively think, if Γ ′ is a proper finite index subgroup of Γ with w ∈ Γ ′ , the group Γ ′ * w Γ ′ is not a finite index subgroup of Γ * w Γ so that Conjecture 1.8 does not imply a positive answer to Question 1.7.