When you downloaded the data, you had the option of ArcGRID, GridFloat and IMG. Any of those is readable by GDAL. Of those formats, I prefer the IMG as it's a single file. ArcGRID is a really old format that requires a complete directory structure and is quite fragile (you can easily break them using standard file managers without realising).
That said, you have Arc GRID data. When you extracted, you should have something like:
USGS_NED_2_n70w147_ArcGrid
├── grdn70w147_2
│ ├── dblbnd.adf
│ ├── hdr.adf
│ ├── metadata.xml
│ ├── prj.adf
│ ├── sta.adf
│ ├── w001001.adf
│ └── w001001x.adf
├── grdn70w147_2.aux.xml
├── info
│ ├── arc0000.dat
│ ├── arc0000.nit
│ ├── arc0001.dat
│ ├── arc0001.nit
│ └── arc.dir
├── n70w147
│ ├── n70w147.cpg
│ ├── n70w147.dbf
│ ├── n70w147.prj
│ ├── n70w147.shp
│ └── n70w147.shx
├── NED_DataDictionary.url
├── readme.pdf
├── SpatialMetadata.url
├── USGS_NED_2_n70w147_ArcGrid_meta.html
├── USGS_NED_2_n70w147_ArcGrid_meta.txt
├── USGS_NED_2_n70w147_ArcGrid_meta.xml
└── USGS_NED_2_n70w147_ArcGrid_thumb.jpg
You would then run gdal_translate and pass either the directory name, the hdr.adf or the w001001.adf. I just use the directory name, it's shorter...
gdal_translate -of XYZ USGS_NED_2_n70w147_ArcGrid/grdn70w147_2 grdn70w147_2.xyz.txt
gdal_translate -of XYZ USGS_NED_2_n70w147_ArcGrid/grdn70w147_2/hdr.adf grdn70w147_2.xyz.txt
gdal_translate -of XYZ USGS_NED_2_n70w147_ArcGrid/grdn70w147_2/w001001.adf grdn70w147_2.xyz.txt