First you should not use the datatype text for title, you are better of using varchar if at all possible. The same goes for feature. Second it seems a bit strange to have all the languages in the same table like that. There are two ways around it that I can think of:
CREATE TABLE articles (
article_id int(11) NOT NULL auto_increment,
article_date date NOT NULL default '0000-00-00',
article_author varchar(60) NOT NULL default '',
article_publish int(11) NOT NULL default '0',
article_cat int(11) NOT NULL default '0',
article_comments int(11) NOT NULL default '1',
PRIMARY KEY (article_id)
) ENGINE=InnoDB AUTO_INCREMENT=905 DEFAULT CHARSET=latin1;
CREATE TABLE articles_language (
article_id int(11) NOT NULL auto_increment,
language varchar(3) NOT NULL,
article_title varchar(200) NOT NULL,
article_body longtext NOT NULL,
article_feature_far varchar(50) NOT NULL,
PRIMARY KEY (article_id, language)
) ENGINE=InnoDB AUTO_INCREMENT=905 DEFAULT CHARSET=latin1;
The solution above will have all information that is related to all languages in one table and all information that are related to a single language in another table. The other solution is to create a new table for each language, but with the same basic idea as above. It may even be possible to use another charset for the tables that needs it if you work this way.
CREATE TABLE articles (
article_id int(11) NOT NULL auto_increment,
article_date date NOT NULL default '0000-00-00',
article_author varchar(60) NOT NULL default '',
article_publish int(11) NOT NULL default '0',
article_cat int(11) NOT NULL default '0',
article_comments int(11) NOT NULL default '1',
PRIMARY KEY (article_id)
) ENGINE=InnoDB AUTO_INCREMENT=905 DEFAULT CHARSET=latin1;
CREATE TABLE articles_eng (
article_id int(11) NOT NULL auto_increment,
article_title varchar(200) NOT NULL,
article_body longtext NOT NULL,
article_feature_far varchar(50) NOT NULL,
PRIMARY KEY (article_id)
) ENGINE=InnoDB AUTO_INCREMENT=905 DEFAULT CHARSET=latin1;
Note that these suggestions is to handle the database better, it won't take less space. But let's face it, 20 MB in a database is not much.