Wow! good review of alternatives for the 12thmove
(vote for the French Tarrasch ?)
I can't resist to share some lines I recorded after one game in rapid play with french GM-editor-trainer-paper writer David Marciano. (I have a good score against him beside of complety lost games

)
12...Bxf4 13.Nxf4 Qd6!? (13...Ne4 main line; 13...Ng4 same ideas than previous post; 13...Qb6 maybe Watson will favour this move; 13...Qc7!?! where I was beatufully crushed in Khamrakulov-Laurain,Condom 2002)
Now White can play 14.g3 ; 14. Qc1 or 14.Ne2 (and not 14.Qd2?? Ne4)
A) 14.g3 Qb4 (14..g6!? 15.Re1 Qb4 worth a try ; 14...e5 15.dxe5 Nxe5 16.Nxe5 Qxe5 Qb3 and small advantage to White ; 14...Bd7 Psakhis doesn't trust it but I am not convinced by the analysis ; 14...Ng4?! Matamoros's idea given karlzen on NJ Chess forum and by Emms in FrenchTarrasch book) 15.Rb1 Bd7 16.Re1 Qb6 (16...Rae8 then 17.Bc2 or 17.a3 Qb6, no future for Black) 17.Bc2 Rac8=
One idea to remind is the fork with e5, but must be calculated precisely. 17...Rae8 18.b4 Nxd4 19.Qxd4 Qxd4 20.Nxd4 e5 21.Nfe6 Bxe6 22.Rxe5 +=
or 18...a6 19.h4 g6 20.a3 Kg7 21.Qd2 Nxd4 22.Qxd4 Qxd4 23.Nxd4 e5 24.Nfe6! Bxe6 25.Rxe5 += Vokarev-Volkov,St Petersburg 2000 (Psakhis Nd2)
B) 14.Qc1 Ng4!? (Dom) (not 14...Ne4 transposing to main lines or good lines for White after g3 or Ne2) 15.g3 e5
C) 14.Ne2 and now following David's advice, 14...Ng4 is the good move and not 14...Bd7 (Marciano-Laurain,rapid SteFoy d'Aigrefeuille 2004 ; Petrov-Radjabov,Moscou 1997 (Psakhis Nd2);Hakki-Knaak,Novi Sad 1990 (Psakhis Nd2)) . Then 14...Ng4 15.h3 (15.Ng3?! e5 16.dxe5 Ngxe5 17.Nxe5 Nxe5 and Black has a good game according to Psakhis because of misplaced knight on g3) Rxf3 16.hxg4 Rf7 17.f4 Qb4! forgotten move (17..Bd7 18.a3 Guseinov-Yakymov,St Petersburg 2001 (Psakhis)) 18.Rf3 g6
Note: it's better to play Qd6 before Ng4: 13...Ng4 14.Qd2! Qd6 15.g3 Bd7 (15...e5?! Zapata-Matamoros,Seville 1992 : basis idea for Qd6: play Qd6-e5-Qh6-and an attack) 16.h3!? or 16.Rfe1