1 Introduction

The Infrared Space Observatory (ISO) mission has provided an unprecedented view of the interstellar medium (ISM) in galaxies from the near-infrared (NIR) continuum between 3 and 5 $\mu$m to the C⁺ fine-structure transition at 158 $\mu$m and beyond. Infrared spectra (3 to 12 $\mu$m) obtained by ISO-PHOT reveal the mid-infrared (MIR) emission from normal galaxies to be characterized by the Aromatic Features in Emission (AFEs) at 3.3, 6.2, 7.7, 8.6, and 11.3 $\mu$m, and an underlying continuum which contributes about half the luminosity in the 3 to 12 $\mu$m range (Helou et al. 2000).

However, the ISO data have also raised many questions about the ISM constituents. In particular, the nature of the dust responsible for the AFEs is not yet clear, and the debate over three-dimensional Very Small Grains (VSGs), two-dimensional Polycyclic Aromatic Hydrocarbon molecules (PAHs), or Amorphous Hydrogenated Carbon particles (HACs), and their relative contribution in different environments is still underway (Jenniskens & Désert 1993; Lu 1998; Cesarsky et al. 1998). Moreover, the relationship between the carriers of the AFEs and the underlying continuum remains obscure (Boulanger et al. 1996; Helou et al. 2000).

A related question is the relative contributions of dust in the ISM, stellar photospheres, and circumstellar dust to the MIR radiation in spirals and ellipticals. IRAS data already convincingly showed that the MIR emission in spirals is mainly due to a continuum from small grains transiently heated to high temperatures and AFEs. On the other hand, in elliptical galaxies the ISM-to-stellar ratio is generally low, so that photospheric and circumstellar emission from evolved red giants may contribute significantly to the MIR (Knapp et al. 1992; Mazzei & de Zotti 1994). It has been argued, though, that the same IRAS color-color relation (e.g., Helou 1986) holds for ellipticals and spirals, which implies a similar ISM origin (Sauvage & Thuan 1994).

The continuum component between 3 and 5 $\mu$m can be extrapolated fairly well to the continuum level at 9-10 $\mu$m (Helou et al. 2000), which suggests that some of the 3-5 $\mu$m continuum must be produced by the ISM. Nevertheless, a portion of the 3-5 $\mu$m continuum must arise from stellar photospheres since they dominate the emission at 1-2 $\mu$m. Although the signal-to-noise in the 3-5 $\mu$m ISO spectra is relatively low, that is the spectral region where the MIR spectra of the 43 galaxies observed by Helou et al. (2000) show the most significant galaxy-to-galaxy variation. To understand the physical origin of this, and assess the relationship of the MIR continuum to the ubiquitous AFEs, it is necessary to separate the ISM contribution from that of stellar photospheres and circumstellar dust. To this end, we have acquired JHK$L^\prime$ photometry of the galaxies studied in Helou et al. (2000), which constitutes one of the ISO Normal Galaxy Key Projects. A variety of ISO observations have been secured for this sample, including mid-infrared spectra (Helou et al. 2000), ionic and atomic fine-structure line fluxes (Malhotra et al. 1997; Malhotra et al. 2001), mid-infrared maps at 7 and 15 $\mu$m, and a small number of 4.5 $\mu$m images (Dale et al. 2000).

The rest of the paper is organized as follows: Sect. 2 describes the observations, reduction, and photometric calibration. We compare the JHKL colors of the sample galaxies with their ISO properties in Sect. 3, and with photodissociation-region models in Sect. 4. Section 5 discusses the nature of the 4  $\mu$m continuum, and our interpretation of these results in terms of "active'' and "passive'' star formation.