You can use a rel Canonical to avoid Duplicate content errors on Google.

For each article on your client's web site add a meta in head with the url of the origin article.

    <link rel="canonical" href="https://original-website.com/article-url" />

For the official source, you can check this page :

https://moz.com/learn/seo/duplicate-content

About your update, this definition given by Google makes clear if you copy 1000 words of 3000 without rewriting it, this will be Duplicate content.

> Duplicate content generally refers to substantive blocks of content within or
> across domains that either completely match other content or are
> appreciably similar.

https://support.google.com/webmasters/answer/66359?hl=en

The only way you have not to make DC is to rewrite those articles or add a Canonical. So, Yes it's still ok to add Canonical tag ;)

**An easier way** could be to add a `no-index` meta tag on your pages this way you are sure your copyed articles don't harm original ones SEO.