capture a file and separate columns and add sth using arrays

This approach avoids awk but using bash's IFS to separate the fields in FILE:

FILE="/etc/passwd"
while IFS=: read Username pass uid gid Realname Homedir Loginshell; do
    result=$(printf 'Username: %s Realname: %s Homedir: %s Loginshell: %s' "$Username" "$Realname" "$Homedir" "$Loginshell")
    echo "$result"
done <"$FILE"

To supply the file name as an argument (as per question 2):

if [ "$#" -ne "1" ]; then
    echo "Usage: `basename $0` <input-file>"
    exit 1
fi
FILE=$1

while IFS=: read Username pass uid gid Realname Homedir Loginshell; do
    result=$(printf 'Username: %s Realname: %s Homedir: %s Loginshell: %s' "$Username" "$Realname" "$Homedir" "$Loginshell")
    echo "$result"
done <"$FILE"

If you want to use arrays, then:

#!/bin/bash
if [ "$#" -ne "1" ]; then
    echo "Usage: `basename $0` <input-file>"
    exit 1
fi
FILE=$1

#declare -a fields
IFS=:
while read line; do
    fields=($line)
    result=$(printf 'Username: %s Realname: %s Homedir: %s Loginshell: %s' "${fields[0]}" "${fields[4]}" "${fields[5]}" "${fields[6]}")
    echo "$result"
done <"$FILE"

The above reads /etc/passwd one line at a time. With IFS set to a colon, each line is then converted to an array using fields=($line). To avoid surprises, you may want to reset IFS to its default value before executing any bash code that depends on bash's normal word separation rules.

Extra:

Here is an attempt at pretty-printing the output:

#!/bin/bash
if [ "$#" -ne "1" ]; then
    echo "Usage: `basename $0` <input-file>"
    exit 1
fi
FILE=$1

{
echo "========:========:=======:=========="
echo "Username:Realname:Homedir:Loginshell"
echo "========:========:=======:=========="

IFS=:
while read line; do
    fields=($line)
    echo "${fields[0]}:${fields[4]}:${fields[5]}:${fields[6]}"
done < "$FILE"
} | column -nts:

By default, column separates fields by whitespace. However, the realname output field can contain spaces. So, the -s: option is used so that column uses colon-separated fields on input. To assure that headings lined up with the data fields, both headings and data are submitted to the same instance of column.