1: \begin{abstract}

2: We consider a joint asymptotic framework for studying

3: semi-nonpara\-metric regression models where (finite-dimensional)

4: Euclidean parameters and (infinite-dimensional) functional parameters

5: are both of interest. The class of models in consideration share a

6: partially linear structure and are estimated in two general contexts:

7: (i) quasi-likelihood and (ii) true likelihood. We first show that the

8: Euclidean estimator and (pointwise) functional estimator, which are

9: re-scaled at different rates, jointly converge to a zero-mean Gaussian

10: vector. This weak convergence result reveals a surprising \textit{joint

11: asymptotics phenomenon}: these two estimators are asymptotically

12: independent. A~major goal of this paper is to gain first-hand insights

13: into the above phenomenon. Moreover, a likelihood ratio testing is

14: proposed for a set of joint local hypotheses, where a new version of

15: the Wilks phenomenon [\textit{Ann. Math. Stat.} \textbf{9} (1938)

16: 60--62; \textit{Ann. Statist.} \textbf{1} (2001) 153--193] is

17: unveiled. A~novel

18: technical tool, called a \textit{joint Bahadur representation}, is

19: developed for studying these joint asymptotics results.

20: \end{abstract}