I don't think that you would be able to use seek() unless the fields were of a fixed length in bytes. Each character is encoded as Unicode which is two bytes so going from character to character would be easy using readByte() but word to word would be impossible as all the fields would be different lengths.
For instance the following code would print out each character that is contained in the file to the console.
Code:
import java.io.*;
public class R{
public static void main(String[] args){
File f;
RandomAccessFile raf = null;
try{
f = new File("C:" + File.separator + "U.txt");
raf = new RandomAccessFile(f,"r");
}catch(IOException e){;}
char filechar = 0;
for(;;){
try{
System.out.print(filechar = (char) raf.readByte());
}catch(IOException e){
System.exit(0);
}
}
}
}